blob: 57a62da4623aaac724b4119a2611b69896c36f4a [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 Tarreaua5614042016-03-14 00:16:53 +01005 version 1.5.16
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreaua5614042016-03-14 00:16:53 +01007 2016/03/13
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
11specified above. It does not provide any hint, example or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013The summary below is meant to help you search sections by name and navigate
14through 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
22 sometimes useful to prefix all output lines (logs, console outs) with 3
23 closing angle brackets ('>>>') in order to help get the difference between
24 inputs and outputs when it can become ambiguous. If you add sections,
25 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
341.2.1. The Request line
351.2.2. The request headers
361.3. HTTP response
371.3.1. The Response line
381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
422.2. Time format
Patrick Mezard35da19c2010-06-12 17:02:47 +0200432.3. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020044
453. Global parameters
463.1. Process management and security
473.2. Performance tuning
483.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100493.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200503.5. Peers
Willy Tarreauc57f0e22009-05-10 13:12:33 +020051
524. Proxies
534.1. Proxy keywords matrix
544.2. Alphabetically sorted keywords reference
55
Willy Tarreau086fbf52012-09-24 20:34:51 +0200565. Bind and Server options
575.1. Bind options
585.2. Server and default-server options
Willy Tarreauc57f0e22009-05-10 13:12:33 +020059
606. HTTP header manipulation
61
Willy Tarreau74ca5042013-06-11 23:12:07 +0200627. Using ACLs and fetching samples
637.1. ACL basics
647.1.1. Matching booleans
657.1.2. Matching integers
667.1.3. Matching strings
677.1.4. Matching regular expressions (regexes)
687.1.5. Matching arbitrary data blocks
697.1.6. Matching IPv4 and IPv6 addresses
707.2. Using ACLs to form conditions
717.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200727.3.1. Converters
737.3.2. Fetching samples from internal states
747.3.3. Fetching samples at Layer 4
757.3.4. Fetching samples at Layer 5
767.3.5. Fetching samples from buffer contents (Layer 6)
777.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200787.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020079
808. Logging
818.1. Log levels
828.2. Log formats
838.2.1. Default log format
848.2.2. TCP log format
858.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100868.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100878.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200888.3. Advanced logging options
898.3.1. Disabling logging of external tests
908.3.2. Logging before waiting for the session to terminate
918.3.3. Raising log level upon errors
928.3.4. Disabling logging of successful connections
938.4. Timing events
948.5. Session state at disconnection
958.6. Non-printable characters
968.7. Capturing HTTP cookies
978.8. Capturing HTTP headers
988.9. Examples of logs
99
1009. Statistics and monitoring
1019.1. CSV format
1029.2. Unix Socket commands
103
104
1051. Quick reminder about HTTP
106----------------------------
107
108When haproxy is running in HTTP mode, both the request and the response are
109fully analyzed and indexed, thus it becomes possible to build matching criteria
110on almost anything found in the contents.
111
112However, it is important to understand how HTTP requests and responses are
113formed, and how HAProxy decomposes them. It will then become easier to write
114correct rules and to debug existing configurations.
115
116
1171.1. The HTTP transaction model
118-------------------------------
119
120The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100121to one and only one response. Traditionally, a TCP connection is established
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122from the client to the server, a request is sent by the client on the
123connection, the server responds and the connection is closed. A new request
124will involve a new connection :
125
126 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
127
128In this mode, called the "HTTP close" mode, there are as many connection
129establishments as there are HTTP transactions. Since the connection is closed
130by the server after the response, the client does not need to know the content
131length.
132
133Due to the transactional nature of the protocol, it was possible to improve it
134to avoid closing a connection between two subsequent transactions. In this mode
135however, it is mandatory that the server indicates the content length for each
136response so that the client does not wait indefinitely. For this, a special
137header is used: "Content-length". This mode is called the "keep-alive" mode :
138
139 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
140
141Its advantages are a reduced latency between transactions, and less processing
142power required on the server side. It is generally better than the close mode,
143but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200144a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200145
146A last improvement in the communications is the pipelining mode. It still uses
147keep-alive, but the client does not wait for the first response to send the
148second request. This is useful for fetching large number of images composing a
149page :
150
151 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
152
153This can obviously have a tremendous benefit on performance because the network
154latency is eliminated between subsequent requests. Many HTTP agents do not
155correctly support pipelining since there is no way to associate a response with
156the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100157server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158
Willy Tarreau70dffda2014-01-30 03:07:23 +0100159By default HAProxy operates in keep-alive mode with regards to persistent
160connections: for each connection it processes each request and response, and
161leaves the connection idle on both sides between the end of a response and the
162start of a new request.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200163
Willy Tarreau70dffda2014-01-30 03:07:23 +0100164HAProxy supports 5 connection modes :
165 - keep alive : all requests and responses are processed (default)
166 - tunnel : only the first request and response are processed,
167 everything else is forwarded with no analysis.
168 - passive close : tunnel with "Connection: close" added in both directions.
169 - server close : the server-facing connection is closed after the response.
170 - forced close : the connection is actively closed after end of response.
171
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172
1731.2. HTTP request
174-----------------
175
176First, let's consider this HTTP request :
177
178 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100179 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200180 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
181 2 Host: www.mydomain.com
182 3 User-agent: my small browser
183 4 Accept: image/jpeg, image/gif
184 5 Accept: image/png
185
186
1871.2.1. The Request line
188-----------------------
189
190Line 1 is the "request line". It is always composed of 3 fields :
191
192 - a METHOD : GET
193 - a URI : /serv/login.php?lang=en&profile=2
194 - a version tag : HTTP/1.1
195
196All of them are delimited by what the standard calls LWS (linear white spaces),
197which are commonly spaces, but can also be tabs or line feeds/carriage returns
198followed by spaces/tabs. The method itself cannot contain any colon (':') and
199is limited to alphabetic letters. All those various combinations make it
200desirable that HAProxy performs the splitting itself rather than leaving it to
201the user to write a complex or inaccurate regular expression.
202
203The URI itself can have several forms :
204
205 - A "relative URI" :
206
207 /serv/login.php?lang=en&profile=2
208
209 It is a complete URL without the host part. This is generally what is
210 received by servers, reverse proxies and transparent proxies.
211
212 - An "absolute URI", also called a "URL" :
213
214 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
215
216 It is composed of a "scheme" (the protocol name followed by '://'), a host
217 name or address, optionally a colon (':') followed by a port number, then
218 a relative URI beginning at the first slash ('/') after the address part.
219 This is generally what proxies receive, but a server supporting HTTP/1.1
220 must accept this form too.
221
222 - a star ('*') : this form is only accepted in association with the OPTIONS
223 method and is not relayable. It is used to inquiry a next hop's
224 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100225
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200226 - an address:port combination : 192.168.0.12:80
227 This is used with the CONNECT method, which is used to establish TCP
228 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
229 other protocols too.
230
231In a relative URI, two sub-parts are identified. The part before the question
232mark is called the "path". It is typically the relative path to static objects
233on the server. The part after the question mark is called the "query string".
234It is mostly used with GET requests sent to dynamic scripts and is very
235specific to the language, framework or application in use.
236
237
2381.2.2. The request headers
239--------------------------
240
241The headers start at the second line. They are composed of a name at the
242beginning of the line, immediately followed by a colon (':'). Traditionally,
243an LWS is added after the colon but that's not required. Then come the values.
244Multiple identical headers may be folded into one single line, delimiting the
245values with commas, provided that their order is respected. This is commonly
246encountered in the "Cookie:" field. A header may span over multiple lines if
247the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
248define a total of 3 values for the "Accept:" header.
249
250Contrary to a common mis-conception, header names are not case-sensitive, and
251their values are not either if they refer to other header names (such as the
252"Connection:" header).
253
254The end of the headers is indicated by the first empty line. People often say
255that it's a double line feed, which is not exact, even if a double line feed
256is one valid form of empty line.
257
258Fortunately, HAProxy takes care of all these complex combinations when indexing
259headers, checking values and counting them, so there is no reason to worry
260about the way they could be written, but it is important not to accuse an
261application of being buggy if it does unusual, valid things.
262
263Important note:
264 As suggested by RFC2616, HAProxy normalizes headers by replacing line breaks
265 in the middle of headers by LWS in order to join multi-line headers. This
266 is necessary for proper analysis and helps less capable HTTP parsers to work
267 correctly and not to be fooled by such complex constructs.
268
269
2701.3. HTTP response
271------------------
272
273An HTTP response looks very much like an HTTP request. Both are called HTTP
274messages. Let's consider this HTTP response :
275
276 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100277 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200278 1 HTTP/1.1 200 OK
279 2 Content-length: 350
280 3 Content-Type: text/html
281
Willy Tarreau816b9792009-09-15 21:25:21 +0200282As a special case, HTTP supports so called "Informational responses" as status
283codes 1xx. These messages are special in that they don't convey any part of the
284response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100285continue to post its request for instance. In the case of a status 100 response
286the requested information will be carried by the next non-100 response message
287following the informational one. This implies that multiple responses may be
288sent to a single request, and that this only works when keep-alive is enabled
289(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
290correctly forward and skip them, and only process the next non-100 response. As
291such, these messages are neither logged nor transformed, unless explicitly
292state otherwise. Status 101 messages indicate that the protocol is changing
293over the same connection and that haproxy must switch to tunnel mode, just as
294if a CONNECT had occurred. Then the Upgrade header would contain additional
295information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200296
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200297
2981.3.1. The Response line
299------------------------
300
301Line 1 is the "response line". It is always composed of 3 fields :
302
303 - a version tag : HTTP/1.1
304 - a status code : 200
305 - a reason : OK
306
307The status code is always 3-digit. The first digit indicates a general status :
Willy Tarreau816b9792009-09-15 21:25:21 +0200308 - 1xx = informational message to be skipped (eg: 100, 101)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200309 - 2xx = OK, content is following (eg: 200, 206)
310 - 3xx = OK, no content following (eg: 302, 304)
311 - 4xx = error caused by the client (eg: 401, 403, 404)
312 - 5xx = error caused by the server (eg: 500, 502, 503)
313
314Please refer to RFC2616 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100315"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200316found there, but it's a common practice to respect the well-established
317messages. It can be composed of one or multiple words, such as "OK", "Found",
318or "Authentication Required".
319
320Haproxy may emit the following status codes by itself :
321
322 Code When / reason
323 200 access to stats page, and when replying to monitoring requests
324 301 when performing a redirection, depending on the configured code
325 302 when performing a redirection, depending on the configured code
326 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100327 307 when performing a redirection, depending on the configured code
328 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200329 400 for an invalid or too large request
330 401 when an authentication is required to perform the action (when
331 accessing the stats page)
332 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
333 408 when the request timeout strikes before the request is complete
334 500 when haproxy encounters an unrecoverable internal error, such as a
335 memory allocation failure, which should never happen
336 502 when the server returns an empty, invalid or incomplete response, or
337 when an "rspdeny" filter blocks the response.
338 503 when no server was available to handle the request, or in response to
339 monitoring requests which match the "monitor fail" condition
340 504 when the response timeout strikes before the server responds
341
342The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3434.2).
344
345
3461.3.2. The response headers
347---------------------------
348
349Response headers work exactly like request headers, and as such, HAProxy uses
350the same parsing function for both. Please refer to paragraph 1.2.2 for more
351details.
352
353
3542. Configuring HAProxy
355----------------------
356
3572.1. Configuration file format
358------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200359
360HAProxy's configuration process involves 3 major sources of parameters :
361
362 - the arguments from the command-line, which always take precedence
363 - the "global" section, which sets process-wide parameters
364 - the proxies sections which can take form of "defaults", "listen",
365 "frontend" and "backend".
366
Willy Tarreau0ba27502007-12-24 16:55:16 +0100367The configuration file syntax consists in lines beginning with a keyword
368referenced in this manual, optionally followed by one or several parameters
369delimited by spaces. If spaces have to be entered in strings, then they must be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100370preceded by a backslash ('\') to be escaped. Backslashes also have to be
Willy Tarreau0ba27502007-12-24 16:55:16 +0100371escaped by doubling them.
372
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200373
3742.2. Time format
375----------------
376
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100377Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100378values are generally expressed in milliseconds (unless explicitly stated
379otherwise) but may be expressed in any other unit by suffixing the unit to the
380numeric value. It is important to consider this because it will not be repeated
381for every keyword. Supported units are :
382
383 - us : microseconds. 1 microsecond = 1/1000000 second
384 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
385 - s : seconds. 1s = 1000ms
386 - m : minutes. 1m = 60s = 60000ms
387 - h : hours. 1h = 60m = 3600s = 3600000ms
388 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
389
390
Patrick Mezard35da19c2010-06-12 17:02:47 +02003912.3. Examples
392-------------
393
394 # Simple configuration for an HTTP proxy listening on port 80 on all
395 # interfaces and forwarding requests to a single backend "servers" with a
396 # single server "server1" listening on 127.0.0.1:8000
397 global
398 daemon
399 maxconn 256
400
401 defaults
402 mode http
403 timeout connect 5000ms
404 timeout client 50000ms
405 timeout server 50000ms
406
407 frontend http-in
408 bind *:80
409 default_backend servers
410
411 backend servers
412 server server1 127.0.0.1:8000 maxconn 32
413
414
415 # The same configuration defined with a single listen block. Shorter but
416 # less expressive, especially in HTTP mode.
417 global
418 daemon
419 maxconn 256
420
421 defaults
422 mode http
423 timeout connect 5000ms
424 timeout client 50000ms
425 timeout server 50000ms
426
427 listen http-in
428 bind *:80
429 server server1 127.0.0.1:8000 maxconn 32
430
431
432Assuming haproxy is in $PATH, test these configurations in a shell with:
433
Willy Tarreauccb289d2010-12-11 20:19:38 +0100434 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200435
436
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004373. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200438--------------------
439
440Parameters in the "global" section are process-wide and often OS-specific. They
441are generally set once for all and do not need being changed once correct. Some
442of them have command-line equivalents.
443
444The following keywords are supported in the "global" section :
445
446 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200447 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200448 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200449 - crt-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200450 - daemon
451 - gid
452 - group
453 - log
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100454 - log-send-hostname
Willy Tarreau6a06a402007-07-15 20:15:28 +0200455 - nbproc
456 - pidfile
457 - uid
458 - ulimit-n
459 - user
Willy Tarreaufbee7132007-10-18 13:53:22 +0200460 - stats
Emeric Brun850efd52014-01-29 12:24:34 +0100461 - ssl-server-verify
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +0200462 - node
463 - description
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100464 - unix-bind
Willy Tarreaud72758d2010-01-12 10:42:19 +0100465
Willy Tarreau6a06a402007-07-15 20:15:28 +0200466 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200467 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200468 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200469 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100470 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100471 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100472 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200473 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200474 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200475 - maxsslrate
Willy Tarreau6a06a402007-07-15 20:15:28 +0200476 - noepoll
477 - nokqueue
478 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100479 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300480 - nogetaddrinfo
Willy Tarreaufe255b72007-10-14 23:09:26 +0200481 - spread-checks
Willy Tarreau27a674e2009-08-17 07:23:33 +0200482 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200483 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100484 - tune.comp.maxlevel
Willy Tarreau193b8c62012-11-22 00:17:38 +0100485 - tune.http.cookielen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200486 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100487 - tune.idletimer
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100488 - tune.maxaccept
489 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200490 - tune.maxrewrite
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200491 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100492 - tune.rcvbuf.client
493 - tune.rcvbuf.server
494 - tune.sndbuf.client
495 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100496 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100497 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200498 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100499 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200500 - tune.ssl.default-dh-param
William Lallemanda509e4c2012-11-07 16:54:34 +0100501 - tune.zlib.memlevel
502 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100503
Willy Tarreau6a06a402007-07-15 20:15:28 +0200504 * Debugging
505 - debug
506 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200507
508
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005093.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200510------------------------------------
511
Emeric Brunc8e8d122012-10-02 18:42:10 +0200512ca-base <dir>
513 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200514 relative path is used with "ca-file" or "crl-file" directives. Absolute
515 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200516
Willy Tarreau6a06a402007-07-15 20:15:28 +0200517chroot <jail dir>
518 Changes current directory to <jail dir> and performs a chroot() there before
519 dropping privileges. This increases the security level in case an unknown
520 vulnerability would be exploited, since it would make it very hard for the
521 attacker to exploit the system. This only works when the process is started
522 with superuser privileges. It is important to ensure that <jail_dir> is both
523 empty and unwritable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100524
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100525cpu-map <"all"|"odd"|"even"|process_num> <cpu-set>...
526 On Linux 2.6 and above, it is possible to bind a process to a specific CPU
527 set. This means that the process will never run on other CPUs. The "cpu-map"
528 directive specifies CPU sets for process sets. The first argument is the
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100529 process number to bind. This process must have a number between 1 and 32 or
530 64, depending on the machine's word size, and any process IDs above nbproc
531 are ignored. It is possible to specify all processes at once using "all",
532 only odd numbers using "odd" or even numbers using "even", just like with the
533 "bind-process" directive. The second and forthcoming arguments are CPU sets.
534 Each CPU set is either a unique number between 0 and 31 or 63 or a range with
535 two such numbers delimited by a dash ('-'). Multiple CPU numbers or ranges
536 may be specified, and the processes will be allowed to bind to all of them.
537 Obviously, multiple "cpu-map" directives may be specified. Each "cpu-map"
538 directive will replace the previous ones when they overlap.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100539
Emeric Brunc8e8d122012-10-02 18:42:10 +0200540crt-base <dir>
541 Assigns a default directory to fetch SSL certificates from when a relative
542 path is used with "crtfile" directives. Absolute locations specified after
543 "crtfile" prevail and ignore "crt-base".
544
Willy Tarreau6a06a402007-07-15 20:15:28 +0200545daemon
546 Makes the process fork into background. This is the recommended mode of
547 operation. It is equivalent to the command line "-D" argument. It can be
548 disabled by the command line "-db" argument.
549
550gid <number>
551 Changes the process' group ID to <number>. It is recommended that the group
552 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
553 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100554 Note that if haproxy is started from a user having supplementary groups, it
555 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200556 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100557
Willy Tarreau6a06a402007-07-15 20:15:28 +0200558group <group name>
559 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
560 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100561
Willy Tarreaudc2695c2014-06-27 18:10:07 +0200562log <address> [len <length>] <facility> [max level [min level]]
Willy Tarreau6a06a402007-07-15 20:15:28 +0200563 Adds a global syslog server. Up to two global servers can be defined. They
564 will receive logs for startups and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100565 configured with "log global".
566
567 <address> can be one of:
568
Willy Tarreau2769aa02007-12-27 18:26:09 +0100569 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100570 no port is specified, 514 is used by default (the standard syslog
571 port).
572
David du Colombier24bb5f52011-03-17 10:40:23 +0100573 - An IPv6 address followed by a colon and optionally a UDP port. If
574 no port is specified, 514 is used by default (the standard syslog
575 port).
576
Robert Tsai81ae1952007-12-05 10:47:29 +0100577 - A filesystem path to a UNIX domain socket, keeping in mind
578 considerations for chroot (be sure the path is accessible inside
579 the chroot) and uid/gid (be sure the path is appropriately
580 writeable).
581
Willy Tarreaudad36a32013-03-11 01:20:04 +0100582 Any part of the address string may reference any number of environment
583 variables by preceding their name with a dollar sign ('$') and
584 optionally enclosing them with braces ('{}'), similarly to what is done
585 in Bourne shell.
586
Willy Tarreaudc2695c2014-06-27 18:10:07 +0200587 <length> is an optional maximum line length. Log lines larger than this value
588 will be truncated before being sent. The reason is that syslog
589 servers act differently on log line length. All servers support the
590 default value of 1024, but some servers simply drop larger lines
591 while others do log them. If a server supports long lines, it may
592 make sense to set this value here in order to avoid truncating long
593 lines. Similarly, if a server drops long lines, it is preferable to
594 truncate them before sending them. Accepted values are 80 to 65535
595 inclusive. The default value of 1024 is generally fine for all
596 standard usages. Some specific cases of long captures or
597 JSON-formated logs may require larger values.
598
Robert Tsai81ae1952007-12-05 10:47:29 +0100599 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200600
601 kern user mail daemon auth syslog lpr news
602 uucp cron auth2 ftp ntp audit alert cron2
603 local0 local1 local2 local3 local4 local5 local6 local7
604
605 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200606 all messages are sent. If a maximum level is specified, only messages with a
607 severity at least as important as this level will be sent. An optional minimum
608 level can be specified. If it is set, logs emitted with a more severe level
609 than this one will be capped to this level. This is used to avoid sending
610 "emerg" messages on all terminals on some default syslog configurations.
611 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200612
Cyril Bontédc4d9032012-04-08 21:57:39 +0200613 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200614
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100615log-send-hostname [<string>]
616 Sets the hostname field in the syslog header. If optional "string" parameter
617 is set the header is set to the string contents, otherwise uses the hostname
618 of the system. Generally used if one is not relaying logs through an
619 intermediate syslog server or for simply customizing the hostname printed in
620 the logs.
621
Kevinm48936af2010-12-22 16:08:21 +0000622log-tag <string>
623 Sets the tag field in the syslog header to this string. It defaults to the
624 program name as launched from the command line, which usually is "haproxy".
625 Sometimes it can be useful to differentiate between multiple processes
626 running on the same host.
627
Willy Tarreau6a06a402007-07-15 20:15:28 +0200628nbproc <number>
629 Creates <number> processes when going daemon. This requires the "daemon"
630 mode. By default, only one process is created, which is the recommended mode
631 of operation. For systems limited to small sets of file descriptors per
632 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
633 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
634
635pidfile <pidfile>
636 Writes pids of all daemons into file <pidfile>. This option is equivalent to
637 the "-p" command line argument. The file must be accessible to the user
638 starting the process. See also "daemon".
639
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100640stats bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200641 Limits the stats socket to a certain set of processes numbers. By default the
642 stats socket is bound to all processes, causing a warning to be emitted when
643 nbproc is greater than 1 because there is no way to select the target process
644 when connecting. However, by using this setting, it becomes possible to pin
645 the stats socket to a specific set of processes, typically the first one. The
646 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100647 the number of processes used. The maximum process ID depends on the machine's
Willy Tarreauae302532014-05-07 19:22:24 +0200648 word size (32 or 64). A better option consists in using the "process" setting
649 of the "stats socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +0200650
Willy Tarreau610f04b2014-02-13 11:36:41 +0100651ssl-default-bind-ciphers <ciphers>
652 This setting is only available when support for OpenSSL was built in. It sets
653 the default string describing the list of cipher algorithms ("cipher suite")
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300654 that are negotiated during the SSL/TLS handshake for all "bind" lines which
Willy Tarreau610f04b2014-02-13 11:36:41 +0100655 do not explicitly define theirs. The format of the string is defined in
656 "man 1 ciphers" from OpenSSL man pages, and can be for instance a string such
657 as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). Please check the
658 "bind" keyword for more information.
659
Emeric Brun42a3e202014-10-30 15:56:50 +0100660ssl-default-bind-options [<option>]...
661 This setting is only available when support for OpenSSL was built in. It sets
662 default ssl-options to force on all "bind" lines. Please check the "bind"
663 keyword to see available options.
664
665 Example:
666 global
667 ssl-default-bind-options no-sslv3 no-tls-tickets
668
Willy Tarreau610f04b2014-02-13 11:36:41 +0100669ssl-default-server-ciphers <ciphers>
670 This setting is only available when support for OpenSSL was built in. It
671 sets the default string describing the list of cipher algorithms that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300672 negotiated during the SSL/TLS handshake with the server, for all "server"
Willy Tarreau610f04b2014-02-13 11:36:41 +0100673 lines which do not explicitly define theirs. The format of the string is
674 defined in "man 1 ciphers". Please check the "server" keyword for more
675 information.
676
Emeric Brun42a3e202014-10-30 15:56:50 +0100677ssl-default-server-options [<option>]...
678 This setting is only available when support for OpenSSL was built in. It sets
679 default ssl-options to force on all "server" lines. Please check the "server"
680 keyword to see available options.
681
Emeric Brun850efd52014-01-29 12:24:34 +0100682ssl-server-verify [none|required]
683 The default behavior for SSL verify on servers side. If specified to 'none',
684 servers certificates are not verified. The default is 'required' except if
685 forced using cmdline option '-dV'.
686
Willy Tarreauabb175f2012-09-24 12:43:26 +0200687stats socket [<address:port>|<path>] [param*]
688 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
689 Connections to this socket will return various statistics outputs and even
690 allow some commands to be issued to change some runtime settings. Please
691 consult section 9.2 "Unix Socket commands" for more details.
Willy Tarreau6162db22009-10-10 17:13:00 +0200692
Willy Tarreauabb175f2012-09-24 12:43:26 +0200693 All parameters supported by "bind" lines are supported, for instance to
694 restrict access to some users or their access rights. Please consult
695 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200696
697stats timeout <timeout, in milliseconds>
698 The default timeout on the stats socket is set to 10 seconds. It is possible
699 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +0100700 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200701
702stats maxconn <connections>
703 By default, the stats socket is limited to 10 concurrent connections. It is
704 possible to change this value with "stats maxconn".
705
Willy Tarreau6a06a402007-07-15 20:15:28 +0200706uid <number>
707 Changes the process' user ID to <number>. It is recommended that the user ID
708 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
709 be started with superuser privileges in order to be able to switch to another
710 one. See also "gid" and "user".
711
712ulimit-n <number>
713 Sets the maximum number of per-process file-descriptors to <number>. By
714 default, it is automatically computed, so it is recommended not to use this
715 option.
716
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100717unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
718 [ group <group> ] [ gid <gid> ]
719
720 Fixes common settings to UNIX listening sockets declared in "bind" statements.
721 This is mainly used to simplify declaration of those UNIX sockets and reduce
722 the risk of errors, since those settings are most commonly required but are
723 also process-specific. The <prefix> setting can be used to force all socket
724 path to be relative to that directory. This might be needed to access another
725 component's chroot. Note that those paths are resolved before haproxy chroots
726 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
727 all have the same meaning as their homonyms used by the "bind" statement. If
728 both are specified, the "bind" statement has priority, meaning that the
729 "unix-bind" settings may be seen as process-wide default settings.
730
Willy Tarreau6a06a402007-07-15 20:15:28 +0200731user <user name>
732 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
733 See also "uid" and "group".
734
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +0200735node <name>
736 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
737
738 This statement is useful in HA configurations where two or more processes or
739 servers share the same IP address. By setting a different node-name on all
740 nodes, it becomes easy to immediately spot what server is handling the
741 traffic.
742
743description <text>
744 Add a text that describes the instance.
745
746 Please note that it is required to escape certain characters (# for example)
747 and this text is inserted into a html page so you should avoid using
748 "<" and ">" characters.
749
Willy Tarreau6a06a402007-07-15 20:15:28 +0200750
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007513.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200752-----------------------
753
Willy Tarreau1746eec2014-04-25 10:46:47 +0200754max-spread-checks <delay in milliseconds>
755 By default, haproxy tries to spread the start of health checks across the
756 smallest health check interval of all the servers in a farm. The principle is
757 to avoid hammering services running on the same server. But when using large
758 check intervals (10 seconds or more), the last servers in the farm take some
759 time before starting to be tested, which can be a problem. This parameter is
760 used to enforce an upper bound on delay between the first and the last check,
761 even if the servers' check intervals are larger. When servers run with
762 shorter intervals, their intervals will be respected though.
763
Willy Tarreau6a06a402007-07-15 20:15:28 +0200764maxconn <number>
765 Sets the maximum per-process number of concurrent connections to <number>. It
766 is equivalent to the command-line argument "-n". Proxies will stop accepting
767 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +0200768 automatically adjusted according to this value. See also "ulimit-n". Note:
769 the "select" poller cannot reliably use more than 1024 file descriptors on
770 some platforms. If your platform only supports select and reports "select
771 FAILED" on startup, you need to reduce maxconn until it works (slightly
772 below 500 in general).
Willy Tarreau6a06a402007-07-15 20:15:28 +0200773
Willy Tarreau81c25d02011-09-07 15:17:21 +0200774maxconnrate <number>
775 Sets the maximum per-process number of connections per second to <number>.
776 Proxies will stop accepting connections when this limit is reached. It can be
777 used to limit the global capacity regardless of each frontend capacity. It is
778 important to note that this can only be used as a service protection measure,
779 as there will not necessarily be a fair share between frontends when the
780 limit is reached, so it's a good idea to also limit each frontend to some
781 value close to its expected share. Also, lowering tune.maxaccept can improve
782 fairness.
783
William Lallemandd85f9172012-11-09 17:05:39 +0100784maxcomprate <number>
785 Sets the maximum per-process input compression rate to <number> kilobytes
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300786 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +0100787 level will be decreased during the session. If the maximum is reached at the
788 beginning of a session, the session will not compress at all. If the maximum
789 is not reached, the compression level will be increased up to
790 tune.comp.maxlevel. A value of zero means there is no limit, this is the
791 default value.
792
William Lallemand072a2bf2012-11-20 17:01:01 +0100793maxcompcpuusage <number>
794 Sets the maximum CPU usage HAProxy can reach before stopping the compression
795 for new requests or decreasing the compression level of current requests.
796 It works like 'maxcomprate' but measures CPU usage instead of incoming data
797 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
798 case of multiple processes (nbproc > 1), each process manages its individual
799 usage. A value of 100 disable the limit. The default value is 100. Setting
800 a lower value will prevent the compression work from slowing the whole
801 process down and from introducing high latencies.
802
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100803maxpipes <number>
804 Sets the maximum per-process number of pipes to <number>. Currently, pipes
805 are only used by kernel-based tcp splicing. Since a pipe contains two file
806 descriptors, the "ulimit-n" value will be increased accordingly. The default
807 value is maxconn/4, which seems to be more than enough for most heavy usages.
808 The splice code dynamically allocates and releases pipes, and can fall back
809 to standard copy, so setting this value too low may only impact performance.
810
Willy Tarreau93e7c002013-10-07 18:51:07 +0200811maxsessrate <number>
812 Sets the maximum per-process number of sessions per second to <number>.
813 Proxies will stop accepting connections when this limit is reached. It can be
814 used to limit the global capacity regardless of each frontend capacity. It is
815 important to note that this can only be used as a service protection measure,
816 as there will not necessarily be a fair share between frontends when the
817 limit is reached, so it's a good idea to also limit each frontend to some
818 value close to its expected share. Also, lowering tune.maxaccept can improve
819 fairness.
820
Willy Tarreau403edff2012-09-06 11:58:37 +0200821maxsslconn <number>
822 Sets the maximum per-process number of concurrent SSL connections to
823 <number>. By default there is no SSL-specific limit, which means that the
824 global maxconn setting will apply to all connections. Setting this limit
825 avoids having openssl use too much memory and crash when malloc returns NULL
826 (since it unfortunately does not reliably check for such conditions). Note
827 that the limit applies both to incoming and outgoing connections, so one
828 connection which is deciphered then ciphered accounts for 2 SSL connections.
829
Willy Tarreaue43d5322013-10-07 20:01:52 +0200830maxsslrate <number>
831 Sets the maximum per-process number of SSL sessions per second to <number>.
832 SSL listeners will stop accepting connections when this limit is reached. It
833 can be used to limit the global SSL CPU usage regardless of each frontend
834 capacity. It is important to note that this can only be used as a service
835 protection measure, as there will not necessarily be a fair share between
836 frontends when the limit is reached, so it's a good idea to also limit each
837 frontend to some value close to its expected share. It is also important to
838 note that the sessions are accounted before they enter the SSL stack and not
839 after, which also protects the stack against bad handshakes. Also, lowering
840 tune.maxaccept can improve fairness.
841
William Lallemand9d5f5482012-11-07 16:12:57 +0100842maxzlibmem <number>
843 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
844 When the maximum amount is reached, future sessions will not compress as long
845 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +0100846 The default value is 0. The value is available in bytes on the UNIX socket
847 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
848 "ZlibMemUsage" in bytes.
849
Willy Tarreau6a06a402007-07-15 20:15:28 +0200850noepoll
851 Disables the use of the "epoll" event polling system on Linux. It is
852 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +0100853 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +0200854
855nokqueue
856 Disables the use of the "kqueue" event polling system on BSD. It is
857 equivalent to the command-line argument "-dk". The next polling system
858 used will generally be "poll". See also "nopoll".
859
860nopoll
861 Disables the use of the "poll" event polling system. It is equivalent to the
862 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +0100863 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +0100864 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +0200865
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100866nosplice
867 Disables the use of kernel tcp splicing between sockets on Linux. It is
868 equivalent to the command line argument "-dS". Data will then be copied
869 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100870 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100871 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
872 be used. This option makes it easier to globally disable kernel splicing in
873 case of doubt. See also "option splice-auto", "option splice-request" and
874 "option splice-response".
875
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300876nogetaddrinfo
877 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
878 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
879
Willy Tarreaufe255b72007-10-14 23:09:26 +0200880spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +0900881 Sometimes it is desirable to avoid sending agent and health checks to
882 servers at exact intervals, for instance when many logical servers are
883 located on the same physical server. With the help of this parameter, it
884 becomes possible to add some randomness in the check interval between 0
885 and +/- 50%. A value between 2 and 5 seems to show good results. The
886 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +0200887
Willy Tarreau27a674e2009-08-17 07:23:33 +0200888tune.bufsize <number>
889 Sets the buffer size to this size (in bytes). Lower values allow more
890 sessions to coexist in the same amount of RAM, and higher values allow some
891 applications with very large cookies to work. The default value is 16384 and
892 can be changed at build time. It is strongly recommended not to change this
893 from the default value, as very low values will break some services such as
894 statistics, and values larger than default size will increase memory usage,
895 possibly causing the system to run out of memory. At least the global maxconn
896 parameter should be decreased by the same factor as this one is increased.
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +0400897 If HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
898 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
899 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +0200900
Willy Tarreau43961d52010-10-04 20:39:20 +0200901tune.chksize <number>
902 Sets the check buffer size to this size (in bytes). Higher values may help
903 find string or regex patterns in very large pages, though doing so may imply
904 more memory and CPU usage. The default value is 16384 and can be changed at
905 build time. It is not recommended to change this value, but to use better
906 checks whenever possible.
907
William Lallemandf3747832012-11-09 12:33:10 +0100908tune.comp.maxlevel <number>
909 Sets the maximum compression level. The compression level affects CPU
910 usage during compression. This value affects CPU usage during compression.
911 Each session using compression initializes the compression algorithm with
912 this value. The default value is 1.
913
Willy Tarreau193b8c62012-11-22 00:17:38 +0100914tune.http.cookielen <number>
915 Sets the maximum length of captured cookies. This is the maximum value that
916 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
917 will automatically be truncated to this one. It is important not to set too
918 high a value because all cookie captures still allocate this size whatever
919 their configured value (they share a same pool). This value is per request
920 per response, so the memory allocated is twice this value per connection.
921 When not specified, the limit is set to 63 characters. It is recommended not
922 to change this value.
923
Willy Tarreauac1932d2011-10-24 19:14:41 +0200924tune.http.maxhdr <number>
925 Sets the maximum number of headers in a request. When a request comes with a
926 number of headers greater than this value (including the first line), it is
927 rejected with a "400 Bad Request" status code. Similarly, too large responses
928 are blocked with "502 Bad Gateway". The default value is 101, which is enough
929 for all usages, considering that the widely deployed Apache server uses the
930 same limit. It can be useful to push this limit further to temporarily allow
931 a buggy application to work by the time it gets fixed. Keep in mind that each
932 new header consumes 32bits of memory for each session, so don't push this
933 limit too high.
934
Willy Tarreau7e312732014-02-12 16:35:14 +0100935tune.idletimer <timeout>
936 Sets the duration after which haproxy will consider that an empty buffer is
937 probably associated with an idle stream. This is used to optimally adjust
938 some packet sizes while forwarding large and small data alternatively. The
939 decision to use splice() or to send large buffers in SSL is modulated by this
940 parameter. The value is in milliseconds between 0 and 65535. A value of zero
941 means that haproxy will not try to detect idle streams. The default is 1000,
942 which seems to correctly detect end user pauses (eg: read a page before
943 clicking). There should be not reason for changing this value. Please check
944 tune.ssl.maxrecord below.
945
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100946tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +0100947 Sets the maximum number of consecutive connections a process may accept in a
948 row before switching to other work. In single process mode, higher numbers
949 give better performance at high connection rates. However in multi-process
950 modes, keeping a bit of fairness between processes generally is better to
951 increase performance. This value applies individually to each listener, so
952 that the number of processes a listener is bound to is taken into account.
953 This value defaults to 64. In multi-process mode, it is divided by twice
954 the number of processes the listener is bound to. Setting this value to -1
955 completely disables the limitation. It should normally not be needed to tweak
956 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100957
958tune.maxpollevents <number>
959 Sets the maximum amount of events that can be processed at once in a call to
960 the polling system. The default value is adapted to the operating system. It
961 has been noticed that reducing it below 200 tends to slightly decrease
962 latency at the expense of network bandwidth, and increasing it above 200
963 tends to trade latency for slightly increased bandwidth.
964
Willy Tarreau27a674e2009-08-17 07:23:33 +0200965tune.maxrewrite <number>
966 Sets the reserved buffer space to this size in bytes. The reserved space is
967 used for header rewriting or appending. The first reads on sockets will never
968 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
969 bufsize, though that does not make much sense since there are rarely large
970 numbers of headers to add. Setting it too high prevents processing of large
971 requests or responses. Setting it too low prevents addition of new headers
972 to already large requests or to POST requests. It is generally wise to set it
973 to about 1024. It is automatically readjusted to half of bufsize if it is
974 larger than that. This means you don't have to worry about it when changing
975 bufsize.
976
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200977tune.pipesize <number>
978 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
979 are the default size for the system. But sometimes when using TCP splicing,
980 it can improve performance to increase pipe sizes, especially if it is
981 suspected that pipes are not filled and that many calls to splice() are
982 performed. This has an impact on the kernel's memory footprint, so this must
983 not be changed if impacts are not understood.
984
Willy Tarreaue803de22010-01-21 17:43:04 +0100985tune.rcvbuf.client <number>
986tune.rcvbuf.server <number>
987 Forces the kernel socket receive buffer size on the client or the server side
988 to the specified value in bytes. This value applies to all TCP/HTTP frontends
989 and backends. It should normally never be set, and the default size (0) lets
990 the kernel autotune this value depending on the amount of available memory.
991 However it can sometimes help to set it to very low values (eg: 4096) in
992 order to save kernel memory by preventing it from buffering too large amounts
993 of received data. Lower values will significantly increase CPU usage though.
994
995tune.sndbuf.client <number>
996tune.sndbuf.server <number>
997 Forces the kernel socket send buffer size on the client or the server side to
998 the specified value in bytes. This value applies to all TCP/HTTP frontends
999 and backends. It should normally never be set, and the default size (0) lets
1000 the kernel autotune this value depending on the amount of available memory.
1001 However it can sometimes help to set it to very low values (eg: 4096) in
1002 order to save kernel memory by preventing it from buffering too large amounts
1003 of received data. Lower values will significantly increase CPU usage though.
1004 Another use case is to prevent write timeouts with extremely slow clients due
1005 to the kernel waiting for a large part of the buffer to be read before
1006 notifying haproxy again.
1007
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001008tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001009 Sets the size of the global SSL session cache, in a number of blocks. A block
1010 is large enough to contain an encoded session without peer certificate.
1011 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001012 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001013 200 bytes of memory. The default value may be forced at build time, otherwise
1014 defaults to 20000. When the cache is full, the most idle entries are purged
1015 and reassigned. Higher values reduce the occurrence of such a purge, hence
1016 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1017 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001018 and are shared between all processes if "nbproc" is greater than 1. Setting
1019 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001020
Emeric Brun8dc60392014-05-09 13:52:00 +02001021tune.ssl.force-private-cache
1022 This boolean disables SSL session cache sharing between all processes. It
1023 should normally not be used since it will force many renegotiations due to
1024 clients hitting a random process. But it may be required on some operating
1025 systems where none of the SSL cache synchronization method may be used. In
1026 this case, adding a first layer of hash-based load balancing before the SSL
1027 layer might limit the impact of the lack of session sharing.
1028
Emeric Brun4f65bff2012-11-16 15:11:00 +01001029tune.ssl.lifetime <timeout>
1030 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001031 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001032 does not guarantee that sessions will last that long, because if the cache is
1033 full, the longest idle sessions will be purged despite their configured
1034 lifetime. The real usefulness of this setting is to prevent sessions from
1035 being used for too long.
1036
Willy Tarreaubfd59462013-02-21 07:46:09 +01001037tune.ssl.maxrecord <number>
1038 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1039 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1040 data only once it has received a full record. With large records, it means
1041 that clients might have to download up to 16kB of data before starting to
1042 process them. Limiting the value can improve page load times on browsers
1043 located over high latency or low bandwidth networks. It is suggested to find
1044 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1045 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1046 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1047 2859 gave good results during tests. Use "strace -e trace=write" to find the
Willy Tarreau7e312732014-02-12 16:35:14 +01001048 best value. Haproxy will automatically switch to this setting after an idle
1049 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001050
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001051tune.ssl.default-dh-param <number>
1052 Sets the maximum size of the Diffie-Hellman parameters used for generating
1053 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1054 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1055 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1056 this maximum value. Default value if 1024. Only 1024 or higher values are
1057 allowed. Higher values will increase the CPU load, and values greater than
1058 1024 bits are not supported by Java 7 and earlier clients. This value is not
1059 used if static Diffie-Hellman parameters are supplied via the certificate file.
1060
William Lallemanda509e4c2012-11-07 16:54:34 +01001061tune.zlib.memlevel <number>
1062 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001063 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001064 state. A value of 1 uses minimum memory but is slow and reduces compression
1065 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
1066 between 1 and 9. The default value is 8.
1067
1068tune.zlib.windowsize <number>
1069 Sets the window size (the size of the history buffer) as a parameter of the
1070 zlib initialization for each session. Larger values of this parameter result
1071 in better compression at the expense of memory usage. Can be a value between
1072 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001073
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010743.3. Debugging
1075--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001076
1077debug
1078 Enables debug mode which dumps to stdout all exchanges, and disables forking
1079 into background. It is the equivalent of the command-line argument "-d". It
1080 should never be used in a production configuration since it may prevent full
1081 system startup.
1082
1083quiet
1084 Do not display any message during startup. It is equivalent to the command-
1085 line argument "-q".
1086
Emeric Brunf099e792010-09-27 12:05:28 +02001087
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010010883.4. Userlists
1089--------------
1090It is possible to control access to frontend/backend/listen sections or to
1091http stats by allowing only authenticated and authorized users. To do this,
1092it is required to create at least one userlist and to define users.
1093
1094userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001095 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001096 used to store authentication & authorization data for independent customers.
1097
1098group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001099 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001100 attach users to this group by using a comma separated list of names
1101 proceeded by "users" keyword.
1102
Cyril Bontéf0c60612010-02-06 14:44:47 +01001103user <username> [password|insecure-password <password>]
1104 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001105 Adds user <username> to the current userlist. Both secure (encrypted) and
1106 insecure (unencrypted) passwords can be used. Encrypted passwords are
Cyril Bonté78caf842010-03-10 22:41:43 +01001107 evaluated using the crypt(3) function so depending of the system's
1108 capabilities, different algorithms are supported. For example modern Glibc
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001109 based Linux system supports MD5, SHA-256, SHA-512 and of course classic,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001110 DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001111
1112
1113 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001114 userlist L1
1115 group G1 users tiger,scott
1116 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001117
Cyril Bontéf0c60612010-02-06 14:44:47 +01001118 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1119 user scott insecure-password elgato
1120 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001121
Cyril Bontéf0c60612010-02-06 14:44:47 +01001122 userlist L2
1123 group G1
1124 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001125
Cyril Bontéf0c60612010-02-06 14:44:47 +01001126 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1127 user scott insecure-password elgato groups G1,G2
1128 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001129
1130 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001131
Emeric Brunf099e792010-09-27 12:05:28 +02001132
11333.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001134----------
Emeric Brunf099e792010-09-27 12:05:28 +02001135It is possible to synchronize server entries in stick tables between several
1136haproxy instances over TCP connections in a multi-master fashion. Each instance
1137pushes its local updates and insertions to remote peers. Server IDs are used to
1138identify servers remotely, so it is important that configurations look similar
1139or at least that the same IDs are forced on each server on all participants.
1140Interrupted exchanges are automatically detected and recovered from the last
1141known point. In addition, during a soft restart, the old process connects to
1142the new one using such a TCP connection to push all its entries before the new
1143process tries to connect to other peers. That ensures very fast replication
1144during a reload, it typically takes a fraction of a second even for large
1145tables.
1146
1147peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001148 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001149 which is referenced by one or more stick-tables.
1150
Willy Tarreau34d05b02015-05-01 20:02:17 +02001151disabled
1152 Disables a peers section. It disables both listening and any synchronization
1153 related to this section. This is provided to disable synchronization of stick
1154 tables without having to comment out all "peers" references.
1155
1156enable
1157 This re-enables a disabled peers section which was previously disabled.
1158
Emeric Brunf099e792010-09-27 12:05:28 +02001159peer <peername> <ip>:<port>
1160 Defines a peer inside a peers section.
1161 If <peername> is set to the local peer name (by default hostname, or forced
1162 using "-L" command line option), haproxy will listen for incoming remote peer
1163 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1164 to join the remote peer, and <peername> is used at the protocol level to
1165 identify and validate the remote peer on the server side.
1166
1167 During a soft restart, local peer <ip>:<port> is used by the old instance to
1168 connect the new one and initiate a complete replication (teaching process).
1169
1170 It is strongly recommended to have the exact same peers declaration on all
1171 peers and to only rely on the "-L" command line argument to change the local
1172 peer name. This makes it easier to maintain coherent configuration files
1173 across all peers.
1174
Willy Tarreaudad36a32013-03-11 01:20:04 +01001175 Any part of the address string may reference any number of environment
1176 variables by preceding their name with a dollar sign ('$') and optionally
1177 enclosing them with braces ('{}'), similarly to what is done in Bourne shell.
1178
Cyril Bontédc4d9032012-04-08 21:57:39 +02001179 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001180 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001181 peer haproxy1 192.168.0.1:1024
1182 peer haproxy2 192.168.0.2:1024
1183 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001184
1185 backend mybackend
1186 mode tcp
1187 balance roundrobin
1188 stick-table type ip size 20k peers mypeers
1189 stick on src
1190
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001191 server srv1 192.168.0.30:80
1192 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001193
1194
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011954. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001196----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001197
Willy Tarreau6a06a402007-07-15 20:15:28 +02001198Proxy configuration can be located in a set of sections :
1199 - defaults <name>
1200 - frontend <name>
1201 - backend <name>
1202 - listen <name>
1203
1204A "defaults" section sets default parameters for all other sections following
1205its declaration. Those default parameters are reset by the next "defaults"
1206section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001207section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001208
1209A "frontend" section describes a set of listening sockets accepting client
1210connections.
1211
1212A "backend" section describes a set of servers to which the proxy will connect
1213to forward incoming connections.
1214
1215A "listen" section defines a complete proxy with its frontend and backend
1216parts combined in one section. It is generally useful for TCP-only traffic.
1217
Willy Tarreau0ba27502007-12-24 16:55:16 +01001218All proxy names must be formed from upper and lower case letters, digits,
1219'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1220case-sensitive, which means that "www" and "WWW" are two different proxies.
1221
1222Historically, all proxy names could overlap, it just caused troubles in the
1223logs. Since the introduction of content switching, it is mandatory that two
1224proxies with overlapping capabilities (frontend/backend) have different names.
1225However, it is still permitted that a frontend and a backend share the same
1226name, as this configuration seems to be commonly encountered.
1227
1228Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1229and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001230bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001231protocol, and can interact with it by allowing, blocking, switching, adding,
1232modifying, or removing arbitrary contents in requests or responses, based on
1233arbitrary criteria.
1234
Willy Tarreau70dffda2014-01-30 03:07:23 +01001235In HTTP mode, the processing applied to requests and responses flowing over
1236a connection depends in the combination of the frontend's HTTP options and
1237the backend's. HAProxy supports 5 connection modes :
1238
1239 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1240 requests and responses are processed, and connections remain open but idle
1241 between responses and new requests.
1242
1243 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1244 1.0 to 1.5-dev21 : only the first request and response are processed, and
1245 everything else is forwarded with no analysis at all. This mode should not
1246 be used as it creates lots of trouble with logging and HTTP processing.
1247
1248 - PCL: passive close ("option httpclose") : exactly the same as tunnel mode,
1249 but with "Connection: close" appended in both directions to try to make
1250 both ends close after the first request/response exchange.
1251
1252 - SCL: server close ("option http-server-close") : the server-facing
1253 connection is closed after the end of the response is received, but the
1254 client-facing connection remains open.
1255
1256 - FCL: forced close ("option forceclose") : the connection is actively closed
1257 after the end of the response.
1258
1259The effective mode that will be applied to a connection passing through a
1260frontend and a backend can be determined by both proxy modes according to the
1261following matrix, but in short, the modes are symmetric, keep-alive is the
1262weakest option and force close is the strongest.
1263
1264 Backend mode
1265
1266 | KAL | TUN | PCL | SCL | FCL
1267 ----+-----+-----+-----+-----+----
1268 KAL | KAL | TUN | PCL | SCL | FCL
1269 ----+-----+-----+-----+-----+----
1270 TUN | TUN | TUN | PCL | SCL | FCL
1271 Frontend ----+-----+-----+-----+-----+----
1272 mode PCL | PCL | PCL | PCL | FCL | FCL
1273 ----+-----+-----+-----+-----+----
1274 SCL | SCL | SCL | FCL | SCL | FCL
1275 ----+-----+-----+-----+-----+----
1276 FCL | FCL | FCL | FCL | FCL | FCL
1277
Willy Tarreau0ba27502007-12-24 16:55:16 +01001278
Willy Tarreau70dffda2014-01-30 03:07:23 +01001279
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012804.1. Proxy keywords matrix
1281--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001282
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001283The following list of keywords is supported. Most of them may only be used in a
1284limited set of section types. Some of them are marked as "deprecated" because
1285they are inherited from an old syntax which may be confusing or functionally
1286limited, and there are new recommended keywords to replace them. Keywords
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001287marked with "(*)" can be optionally inverted using the "no" prefix, eg. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001288option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02001289and must be disabled for a specific instance. Such options may also be prefixed
1290with "default" in order to restore default settings regardless of what has been
1291specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001292
Willy Tarreau6a06a402007-07-15 20:15:28 +02001293
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001294 keyword defaults frontend listen backend
1295------------------------------------+----------+----------+---------+---------
1296acl - X X X
1297appsession - - X X
1298backlog X X X -
1299balance X - X X
1300bind - X X -
1301bind-process X X X X
1302block - X X X
1303capture cookie - X X -
1304capture request header - X X -
1305capture response header - X X -
1306clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02001307compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001308contimeout (deprecated) X - X X
1309cookie X - X X
1310default-server X - X X
1311default_backend X X X -
1312description - X X X
1313disabled X X X X
1314dispatch - - X X
1315enabled X X X X
1316errorfile X X X X
1317errorloc X X X X
1318errorloc302 X X X X
1319-- keyword -------------------------- defaults - frontend - listen -- backend -
1320errorloc303 X X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001321force-persist - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001322fullconn X - X X
1323grace X X X X
1324hash-type X - X X
1325http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01001326http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02001327http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001328http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02001329http-response - X X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02001330http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001331id - X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001332ignore-persist - X X X
William Lallemand0f99e342011-10-12 17:50:54 +02001333log (*) X X X X
Willy Tarreau14104732015-01-07 14:55:17 +01001334log-format X X X -
Willy Tarreauc35362a2014-04-25 13:58:37 +02001335max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001336maxconn X X X -
1337mode X X X X
1338monitor fail - X X -
1339monitor-net X X X -
1340monitor-uri X X X -
1341option abortonclose (*) X - X X
1342option accept-invalid-http-request (*) X X X -
1343option accept-invalid-http-response (*) X - X X
1344option allbackups (*) X - X X
1345option checkcache (*) X - X X
1346option clitcpka (*) X X X -
1347option contstats (*) X X X -
1348option dontlog-normal (*) X X X -
1349option dontlognull (*) X X X -
1350option forceclose (*) X X X X
1351-- keyword -------------------------- defaults - frontend - listen -- backend -
1352option forwardfor X X X X
Willy Tarreaub17e8b92015-05-01 22:40:51 +02001353option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01001354option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02001355option http-no-delay (*) X X X X
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02001356option http-pretend-keepalive (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001357option http-server-close (*) X X X X
Willy Tarreau02bce8b2014-01-30 00:15:28 +01001358option http-tunnel (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001359option http-use-proxy-header (*) X X X -
1360option httpchk X - X X
1361option httpclose (*) X X X X
1362option httplog X X X X
1363option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001364option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02001365option ldap-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001366option log-health-checks (*) X - X X
1367option log-separate-errors (*) X X X -
1368option logasap (*) X X X -
1369option mysql-check X - X X
1370option nolinger (*) X X X X
1371option originalto X X X X
1372option persist (*) X - X X
Baptiste Assmanne1444222015-10-12 20:22:55 +02001373option pgsql-check X - X X
1374option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001375option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02001376option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001377option smtpchk X - X X
1378option socket-stats (*) X X X -
1379option splice-auto (*) X X X X
1380option splice-request (*) X X X X
1381option splice-response (*) X X X X
1382option srvtcpka (*) X - X X
1383option ssl-hello-chk X - X X
1384-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01001385option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001386option tcp-smart-accept (*) X X X -
1387option tcp-smart-connect (*) X - X X
1388option tcpka X X X X
1389option tcplog X X X X
1390option transparent (*) X - X X
1391persist rdp-cookie X - X X
1392rate-limit sessions X X X -
1393redirect - X X X
1394redisp (deprecated) X - X X
1395redispatch (deprecated) X - X X
1396reqadd - X X X
1397reqallow - X X X
1398reqdel - X X X
1399reqdeny - X X X
1400reqiallow - X X X
1401reqidel - X X X
1402reqideny - X X X
1403reqipass - X X X
1404reqirep - X X X
1405reqisetbe - X X X
1406reqitarpit - X X X
1407reqpass - X X X
1408reqrep - X X X
1409-- keyword -------------------------- defaults - frontend - listen -- backend -
1410reqsetbe - X X X
1411reqtarpit - X X X
1412retries X - X X
1413rspadd - X X X
1414rspdel - X X X
1415rspdeny - X X X
1416rspidel - X X X
1417rspideny - X X X
1418rspirep - X X X
1419rsprep - X X X
1420server - - X X
1421source X - X X
1422srvtimeout (deprecated) X - X X
Baptiste Assmann5c88d6e2015-10-12 20:30:24 +02001423stats admin - X X X
1424stats auth X X X X
1425stats enable X X X X
1426stats hide-version X X X X
1427stats http-request - X X X
1428stats realm X X X X
1429stats refresh X X X X
1430stats scope X X X X
1431stats show-desc X X X X
1432stats show-legends X X X X
1433stats show-node X X X X
1434stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001435-- keyword -------------------------- defaults - frontend - listen -- backend -
1436stick match - - X X
1437stick on - - X X
1438stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02001439stick store-response - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001440stick-table - - X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02001441tcp-check connect - - X X
1442tcp-check expect - - X X
1443tcp-check send - - X X
1444tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02001445tcp-request connection - X X -
1446tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02001447tcp-request inspect-delay - X X X
Emeric Brun0a3b67f2010-09-24 15:34:53 +02001448tcp-response content - - X X
1449tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001450timeout check X - X X
1451timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02001452timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001453timeout clitimeout (deprecated) X X X -
1454timeout connect X - X X
1455timeout contimeout (deprecated) X - X X
1456timeout http-keep-alive X X X X
1457timeout http-request X X X X
1458timeout queue X - X X
1459timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02001460timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001461timeout srvtimeout (deprecated) X - X X
1462timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02001463timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001464transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01001465unique-id-format X X X -
1466unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001467use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02001468use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001469------------------------------------+----------+----------+---------+---------
1470 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02001471
Willy Tarreau0ba27502007-12-24 16:55:16 +01001472
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014734.2. Alphabetically sorted keywords reference
1474---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001475
1476This section provides a description of each keyword and its usage.
1477
1478
1479acl <aclname> <criterion> [flags] [operator] <value> ...
1480 Declare or complete an access list.
1481 May be used in sections : defaults | frontend | listen | backend
1482 no | yes | yes | yes
1483 Example:
1484 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1485 acl invalid_src src_port 0:1023
1486 acl local_dst hdr(host) -i localhost
1487
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001488 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001489
1490
Cyril Bontéb21570a2009-11-29 20:04:48 +01001491appsession <cookie> len <length> timeout <holdtime>
1492 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001493 Define session stickiness on an existing application cookie.
1494 May be used in sections : defaults | frontend | listen | backend
1495 no | no | yes | yes
1496 Arguments :
1497 <cookie> this is the name of the cookie used by the application and which
1498 HAProxy will have to learn for each new session.
1499
Cyril Bontéb21570a2009-11-29 20:04:48 +01001500 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01001501 checked in each cookie value.
1502
1503 <holdtime> this is the time after which the cookie will be removed from
1504 memory if unused. If no unit is specified, this time is in
1505 milliseconds.
1506
Cyril Bontébf47aeb2009-10-15 00:15:40 +02001507 request-learn
1508 If this option is specified, then haproxy will be able to learn
1509 the cookie found in the request in case the server does not
1510 specify any in response. This is typically what happens with
1511 PHPSESSID cookies, or when haproxy's session expires before
1512 the application's session and the correct server is selected.
1513 It is recommended to specify this option to improve reliability.
1514
Cyril Bontéb21570a2009-11-29 20:04:48 +01001515 prefix When this option is specified, haproxy will match on the cookie
1516 prefix (or URL parameter prefix). The appsession value is the
1517 data following this prefix.
1518
1519 Example :
1520 appsession ASPSESSIONID len 64 timeout 3h prefix
1521
1522 This will match the cookie ASPSESSIONIDXXXX=XXXXX,
1523 the appsession value will be XXXX=XXXXX.
1524
1525 mode This option allows to change the URL parser mode.
1526 2 modes are currently supported :
1527 - path-parameters :
1528 The parser looks for the appsession in the path parameters
1529 part (each parameter is separated by a semi-colon), which is
1530 convenient for JSESSIONID for example.
1531 This is the default mode if the option is not set.
1532 - query-string :
1533 In this mode, the parser will look for the appsession in the
1534 query string.
1535
Willy Tarreau0ba27502007-12-24 16:55:16 +01001536 When an application cookie is defined in a backend, HAProxy will check when
1537 the server sets such a cookie, and will store its value in a table, and
1538 associate it with the server's identifier. Up to <length> characters from
1539 the value will be retained. On each connection, haproxy will look for this
Cyril Bontéb21570a2009-11-29 20:04:48 +01001540 cookie both in the "Cookie:" headers, and as a URL parameter (depending on
1541 the mode used). If a known value is found, the client will be directed to the
1542 server associated with this value. Otherwise, the load balancing algorithm is
Willy Tarreau0ba27502007-12-24 16:55:16 +01001543 applied. Cookies are automatically removed from memory when they have been
1544 unused for a duration longer than <holdtime>.
1545
1546 The definition of an application cookie is limited to one per backend.
1547
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001548 Note : Consider not using this feature in multi-process mode (nbproc > 1)
1549 unless you know what you do : memory is not shared between the
1550 processes, which can result in random behaviours.
1551
Willy Tarreau0ba27502007-12-24 16:55:16 +01001552 Example :
1553 appsession JSESSIONID len 52 timeout 3h
1554
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001555 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
1556 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001557
1558
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001559backlog <conns>
1560 Give hints to the system about the approximate listen backlog desired size
1561 May be used in sections : defaults | frontend | listen | backend
1562 yes | yes | yes | no
1563 Arguments :
1564 <conns> is the number of pending connections. Depending on the operating
1565 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02001566 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001567
1568 In order to protect against SYN flood attacks, one solution is to increase
1569 the system's SYN backlog size. Depending on the system, sometimes it is just
1570 tunable via a system parameter, sometimes it is not adjustable at all, and
1571 sometimes the system relies on hints given by the application at the time of
1572 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
1573 to the listen() syscall. On systems which can make use of this value, it can
1574 sometimes be useful to be able to specify a different value, hence this
1575 backlog parameter.
1576
1577 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
1578 used as a hint and the system accepts up to the smallest greater power of
1579 two, and never more than some limits (usually 32768).
1580
1581 See also : "maxconn" and the target operating system's tuning guide.
1582
1583
Willy Tarreau0ba27502007-12-24 16:55:16 +01001584balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02001585balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001586 Define the load balancing algorithm to be used in a backend.
1587 May be used in sections : defaults | frontend | listen | backend
1588 yes | no | yes | yes
1589 Arguments :
1590 <algorithm> is the algorithm used to select a server when doing load
1591 balancing. This only applies when no persistence information
1592 is available, or when a connection is redispatched to another
1593 server. <algorithm> may be one of the following :
1594
1595 roundrobin Each server is used in turns, according to their weights.
1596 This is the smoothest and fairest algorithm when the server's
1597 processing time remains equally distributed. This algorithm
1598 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02001599 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08001600 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02001601 large farms, when a server becomes up after having been down
1602 for a very short time, it may sometimes take a few hundreds
1603 requests for it to be re-integrated into the farm and start
1604 receiving traffic. This is normal, though very rare. It is
1605 indicated here in case you would have the chance to observe
1606 it, so that you don't worry.
1607
1608 static-rr Each server is used in turns, according to their weights.
1609 This algorithm is as similar to roundrobin except that it is
1610 static, which means that changing a server's weight on the
1611 fly will have no effect. On the other hand, it has no design
1612 limitation on the number of servers, and when a server goes
1613 up, it is always immediately reintroduced into the farm, once
1614 the full map is recomputed. It also uses slightly less CPU to
1615 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01001616
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01001617 leastconn The server with the lowest number of connections receives the
1618 connection. Round-robin is performed within groups of servers
1619 of the same load to ensure that all servers will be used. Use
1620 of this algorithm is recommended where very long sessions are
1621 expected, such as LDAP, SQL, TSE, etc... but is not very well
1622 suited for protocols using short sessions such as HTTP. This
1623 algorithm is dynamic, which means that server weights may be
1624 adjusted on the fly for slow starts for instance.
1625
Willy Tarreauf09c6602012-02-13 17:12:08 +01001626 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001627 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01001628 identifier to the highest (see server parameter "id"), which
1629 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02001630 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01001631 not make sense to use this algorithm without setting maxconn.
1632 The purpose of this algorithm is to always use the smallest
1633 number of servers so that extra servers can be powered off
1634 during non-intensive hours. This algorithm ignores the server
1635 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02001636 or IMAP than HTTP, though it can be useful there too. In
1637 order to use this algorithm efficiently, it is recommended
1638 that a cloud controller regularly checks server usage to turn
1639 them off when unused, and regularly checks backend queue to
1640 turn new servers on when the queue inflates. Alternatively,
1641 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01001642
Willy Tarreau0ba27502007-12-24 16:55:16 +01001643 source The source IP address is hashed and divided by the total
1644 weight of the running servers to designate which server will
1645 receive the request. This ensures that the same client IP
1646 address will always reach the same server as long as no
1647 server goes down or up. If the hash result changes due to the
1648 number of running servers changing, many clients will be
1649 directed to a different server. This algorithm is generally
1650 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001651 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01001652 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001653 static by default, which means that changing a server's
1654 weight on the fly will have no effect, but this can be
1655 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001656
Oskar Stolc8dc41842012-05-19 10:19:54 +01001657 uri This algorithm hashes either the left part of the URI (before
1658 the question mark) or the whole URI (if the "whole" parameter
1659 is present) and divides the hash value by the total weight of
1660 the running servers. The result designates which server will
1661 receive the request. This ensures that the same URI will
1662 always be directed to the same server as long as no server
1663 goes up or down. This is used with proxy caches and
1664 anti-virus proxies in order to maximize the cache hit rate.
1665 Note that this algorithm may only be used in an HTTP backend.
1666 This algorithm is static by default, which means that
1667 changing a server's weight on the fly will have no effect,
1668 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001669
Oskar Stolc8dc41842012-05-19 10:19:54 +01001670 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001671 "depth", both followed by a positive integer number. These
1672 options may be helpful when it is needed to balance servers
1673 based on the beginning of the URI only. The "len" parameter
1674 indicates that the algorithm should only consider that many
1675 characters at the beginning of the URI to compute the hash.
1676 Note that having "len" set to 1 rarely makes sense since most
1677 URIs start with a leading "/".
1678
1679 The "depth" parameter indicates the maximum directory depth
1680 to be used to compute the hash. One level is counted for each
1681 slash in the request. If both parameters are specified, the
1682 evaluation stops when either is reached.
1683
Willy Tarreau0ba27502007-12-24 16:55:16 +01001684 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001685 the query string of each HTTP GET request.
1686
1687 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02001688 request entity will be searched for the parameter argument,
1689 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02001690 ('?') in the URL. The message body will only start to be
1691 analyzed once either the advertised amount of data has been
1692 received or the request buffer is full. In the unlikely event
1693 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02001694 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02001695 be randomly balanced if at all. This keyword used to support
1696 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001697
1698 If the parameter is found followed by an equal sign ('=') and
1699 a value, then the value is hashed and divided by the total
1700 weight of the running servers. The result designates which
1701 server will receive the request.
1702
1703 This is used to track user identifiers in requests and ensure
1704 that a same user ID will always be sent to the same server as
1705 long as no server goes up or down. If no value is found or if
1706 the parameter is not found, then a round robin algorithm is
1707 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001708 backend. This algorithm is static by default, which means
1709 that changing a server's weight on the fly will have no
1710 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001711
Cyril Bontédc4d9032012-04-08 21:57:39 +02001712 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
1713 request. Just as with the equivalent ACL 'hdr()' function,
1714 the header name in parenthesis is not case sensitive. If the
1715 header is absent or if it does not contain any value, the
1716 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01001717
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001718 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01001719 reducing the hash algorithm to the main domain part with some
1720 specific headers such as 'Host'. For instance, in the Host
1721 value "haproxy.1wt.eu", only "1wt" will be considered.
1722
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001723 This algorithm is static by default, which means that
1724 changing a server's weight on the fly will have no effect,
1725 but this can be changed using "hash-type".
1726
Emeric Brun736aa232009-06-30 17:56:00 +02001727 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02001728 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02001729 The RDP cookie <name> (or "mstshash" if omitted) will be
1730 looked up and hashed for each incoming TCP request. Just as
1731 with the equivalent ACL 'req_rdp_cookie()' function, the name
1732 is not case-sensitive. This mechanism is useful as a degraded
1733 persistence mode, as it makes it possible to always send the
1734 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001735 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02001736 used instead.
1737
1738 Note that for this to work, the frontend must ensure that an
1739 RDP cookie is already present in the request buffer. For this
1740 you must use 'tcp-request content accept' rule combined with
1741 a 'req_rdp_cookie_cnt' ACL.
1742
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001743 This algorithm is static by default, which means that
1744 changing a server's weight on the fly will have no effect,
1745 but this can be changed using "hash-type".
1746
Cyril Bontédc4d9032012-04-08 21:57:39 +02001747 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09001748
Willy Tarreau0ba27502007-12-24 16:55:16 +01001749 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001750 algorithms. Right now, only "url_param" and "uri" support an
1751 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001752
Willy Tarreau3cd9af22009-03-15 14:06:41 +01001753 The load balancing algorithm of a backend is set to roundrobin when no other
1754 algorithm, mode nor option have been set. The algorithm may only be set once
1755 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001756
1757 Examples :
1758 balance roundrobin
1759 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001760 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01001761 balance hdr(User-Agent)
1762 balance hdr(host)
1763 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001764
1765 Note: the following caveats and limitations on using the "check_post"
1766 extension with "url_param" must be considered :
1767
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001768 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001769 to determine if the parameters will be found in the body or entity which
1770 may contain binary data. Therefore another method may be required to
1771 restrict consideration of POST requests that have no URL parameters in
1772 the body. (see acl reqideny http_end)
1773
1774 - using a <max_wait> value larger than the request buffer size does not
1775 make sense and is useless. The buffer size is set at build time, and
1776 defaults to 16 kB.
1777
1778 - Content-Encoding is not supported, the parameter search will probably
1779 fail; and load balancing will fall back to Round Robin.
1780
1781 - Expect: 100-continue is not supported, load balancing will fall back to
1782 Round Robin.
1783
1784 - Transfer-Encoding (RFC2616 3.6.1) is only supported in the first chunk.
1785 If the entire parameter value is not present in the first chunk, the
1786 selection of server is undefined (actually, defined by how little
1787 actually appeared in the first chunk).
1788
1789 - This feature does not support generation of a 100, 411 or 501 response.
1790
1791 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001792 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001793 white space or control characters are found, indicating the end of what
1794 might be a URL parameter list. This is probably not a concern with SGML
1795 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001796
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001797 See also : "dispatch", "cookie", "appsession", "transparent", "hash-type" and
1798 "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001799
1800
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001801bind [<address>]:<port_range> [, ...] [param*]
1802bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001803 Define one or several listening addresses and/or ports in a frontend.
1804 May be used in sections : defaults | frontend | listen | backend
1805 no | yes | yes | no
1806 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001807 <address> is optional and can be a host name, an IPv4 address, an IPv6
1808 address, or '*'. It designates the address the frontend will
1809 listen on. If unset, all IPv4 addresses of the system will be
1810 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01001811 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01001812 Optionally, an address family prefix may be used before the
1813 address to force the family regardless of the address format,
1814 which can be useful to specify a path to a unix socket with
1815 no slash ('/'). Currently supported prefixes are :
1816 - 'ipv4@' -> address is always IPv4
1817 - 'ipv6@' -> address is always IPv6
1818 - 'unix@' -> address is a path to a local unix socket
Willy Tarreaub4fca5d2014-07-08 00:37:50 +02001819 - 'abns@' -> address is in abstract namespace (Linux only).
1820 Note: since abstract sockets are not "rebindable", they
1821 do not cope well with multi-process mode during
1822 soft-restart, so it is better to avoid them if
1823 nbproc is greater than 1. The effect is that if the
1824 new process fails to start, only one of the old ones
1825 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01001826 - 'fd@<n>' -> use file descriptor <n> inherited from the
1827 parent. The fd must be bound and may or may not already
1828 be listening.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001829 Any part of the address string may reference any number of
1830 environment variables by preceding their name with a dollar
1831 sign ('$') and optionally enclosing them with braces ('{}'),
1832 similarly to what is done in Bourne shell.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001833
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001834 <port_range> is either a unique TCP port, or a port range for which the
1835 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001836 above. The port is mandatory for TCP listeners. Note that in
1837 the case of an IPv6 address, the port is always the number
1838 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001839 - a numerical port (ex: '80')
1840 - a dash-delimited ports range explicitly stating the lower
1841 and upper bounds (ex: '2000-2100') which are included in
1842 the range.
1843
1844 Particular care must be taken against port ranges, because
1845 every <address:port> couple consumes one socket (= a file
1846 descriptor), so it's easy to consume lots of descriptors
1847 with a simple range, and to run out of sockets. Also, each
1848 <address:port> couple must be used only once among all
1849 instances running on a same system. Please note that binding
1850 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001851 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001852 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001853
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001854 <path> is a UNIX socket path beginning with a slash ('/'). This is
1855 alternative to the TCP listening port. Haproxy will then
1856 receive UNIX connections on the socket located at this place.
1857 The path must begin with a slash and by default is absolute.
1858 It can be relative to the prefix defined by "unix-bind" in
1859 the global section. Note that the total length of the prefix
1860 followed by the socket path cannot exceed some system limits
1861 for UNIX sockets, which commonly are set to 107 characters.
1862
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001863 <param*> is a list of parameters common to all sockets declared on the
1864 same line. These numerous parameters depend on OS and build
1865 options and have a complete section dedicated to them. Please
1866 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001867
Willy Tarreau0ba27502007-12-24 16:55:16 +01001868 It is possible to specify a list of address:port combinations delimited by
1869 commas. The frontend will then listen on all of these addresses. There is no
1870 fixed limit to the number of addresses and ports which can be listened on in
1871 a frontend, as well as there is no limit to the number of "bind" statements
1872 in a frontend.
1873
1874 Example :
1875 listen http_proxy
1876 bind :80,:443
1877 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001878 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01001879
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001880 listen http_https_proxy
1881 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02001882 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001883
Willy Tarreau24709282013-03-10 21:32:12 +01001884 listen http_https_proxy_explicit
1885 bind ipv6@:80
1886 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
1887 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
1888
Willy Tarreaudad36a32013-03-11 01:20:04 +01001889 listen external_bind_app1
1890 bind fd@${FD_APP1}
1891
Willy Tarreauf0c95fc2015-09-27 15:03:15 +02001892 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
1893 sun_path length is used for the address length. Some other programs
1894 such as socat use the string length only by default. Pass the option
1895 ",unix-tightsocklen=0" to any abstract socket definition in socat to
1896 make it compatible with HAProxy's.
1897
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001898 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001899 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001900
1901
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001902bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001903 Limit visibility of an instance to a certain set of processes numbers.
1904 May be used in sections : defaults | frontend | listen | backend
1905 yes | yes | yes | yes
1906 Arguments :
1907 all All process will see this instance. This is the default. It
1908 may be used to override a default value.
1909
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001910 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001911 option may be combined with other numbers.
1912
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001913 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001914 option may be combined with other numbers. Do not use it
1915 with less than 2 processes otherwise some instances might be
1916 missing from all processes.
1917
Willy Tarreau110ecc12012-11-15 17:50:01 +01001918 number The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001919 whose values must all be between 1 and 32 or 64 depending on
Willy Tarreau102df612014-05-07 23:56:38 +02001920 the machine's word size. If a proxy is bound to process
1921 numbers greater than the configured global.nbproc, it will
1922 either be forced to process #1 if a single process was
1923 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001924
1925 This keyword limits binding of certain instances to certain processes. This
1926 is useful in order not to have too many processes listening to the same
1927 ports. For instance, on a dual-core machine, it might make sense to set
1928 'nbproc 2' in the global section, then distributes the listeners among 'odd'
1929 and 'even' instances.
1930
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001931 At the moment, it is not possible to reference more than 32 or 64 processes
1932 using this keyword, but this should be more than enough for most setups.
1933 Please note that 'all' really means all processes regardless of the machine's
1934 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001935
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001936 Each "bind" line may further be limited to a subset of the proxy's processes,
1937 please consult the "process" bind keyword in section 5.1.
1938
Willy Tarreaue56c4f12014-09-16 13:21:03 +02001939 When a frontend has no explicit "bind-process" line, it tries to bind to all
1940 the processes referenced by its "bind" lines. That means that frontends can
1941 easily adapt to their listeners' processes.
1942
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001943 If some backends are referenced by frontends bound to other processes, the
1944 backend automatically inherits the frontend's processes.
1945
1946 Example :
1947 listen app_ip1
1948 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001949 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001950
1951 listen app_ip2
1952 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001953 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001954
1955 listen management
1956 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001957 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001958
Willy Tarreau110ecc12012-11-15 17:50:01 +01001959 listen management
1960 bind 10.0.0.4:80
1961 bind-process 1-4
1962
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001963 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001964
1965
Willy Tarreau0ba27502007-12-24 16:55:16 +01001966block { if | unless } <condition>
1967 Block a layer 7 request if/unless a condition is matched
1968 May be used in sections : defaults | frontend | listen | backend
1969 no | yes | yes | yes
1970
1971 The HTTP request will be blocked very early in the layer 7 processing
1972 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001973 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02001974 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01001975 conditions are met or not met. There is no fixed limit to the number of
1976 "block" statements per instance.
1977
1978 Example:
1979 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1980 acl invalid_src src_port 0:1023
1981 acl local_dst hdr(host) -i localhost
1982 block if invalid_src || local_dst
1983
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001984 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001985
1986
1987capture cookie <name> len <length>
1988 Capture and log a cookie in the request and in the response.
1989 May be used in sections : defaults | frontend | listen | backend
1990 no | yes | yes | no
1991 Arguments :
1992 <name> is the beginning of the name of the cookie to capture. In order
1993 to match the exact name, simply suffix the name with an equal
1994 sign ('='). The full name will appear in the logs, which is
1995 useful with application servers which adjust both the cookie name
1996 and value (eg: ASPSESSIONXXXXX).
1997
1998 <length> is the maximum number of characters to report in the logs, which
1999 include the cookie name, the equal sign and the value, all in the
2000 standard "name=value" form. The string will be truncated on the
2001 right if it exceeds <length>.
2002
2003 Only the first cookie is captured. Both the "cookie" request headers and the
2004 "set-cookie" response headers are monitored. This is particularly useful to
2005 check for application bugs causing session crossing or stealing between
2006 users, because generally the user's cookies can only change on a login page.
2007
2008 When the cookie was not presented by the client, the associated log column
2009 will report "-". When a request does not cause a cookie to be assigned by the
2010 server, a "-" is reported in the response column.
2011
2012 The capture is performed in the frontend only because it is necessary that
2013 the log format does not change for a given frontend depending on the
2014 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002015 "capture cookie" statement in a frontend. The maximum capture length is set
2016 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2017 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002018
2019 Example:
2020 capture cookie ASPSESSION len 32
2021
2022 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002023 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002024
2025
2026capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002027 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002028 May be used in sections : defaults | frontend | listen | backend
2029 no | yes | yes | no
2030 Arguments :
2031 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002032 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002033 appear in the requests, with the first letter of each word in
2034 upper case. The header name will not appear in the logs, only the
2035 value is reported, but the position in the logs is respected.
2036
2037 <length> is the maximum number of characters to extract from the value and
2038 report in the logs. The string will be truncated on the right if
2039 it exceeds <length>.
2040
Willy Tarreau4460d032012-11-21 23:37:37 +01002041 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002042 value will be added to the logs between braces ('{}'). If multiple headers
2043 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002044 in the same order they were declared in the configuration. Non-existent
2045 headers will be logged just as an empty string. Common uses for request
2046 header captures include the "Host" field in virtual hosting environments, the
2047 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002048 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002049 environments to find where the request came from.
2050
2051 Note that when capturing headers such as "User-agent", some spaces may be
2052 logged, making the log analysis more difficult. Thus be careful about what
2053 you log if you know your log parser is not smart enough to rely on the
2054 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002055
Willy Tarreau0900abb2012-11-22 00:21:46 +01002056 There is no limit to the number of captured request headers nor to their
2057 length, though it is wise to keep them low to limit memory usage per session.
2058 In order to keep log format consistent for a same frontend, header captures
2059 can only be declared in a frontend. It is not possible to specify a capture
2060 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002061
2062 Example:
2063 capture request header Host len 15
2064 capture request header X-Forwarded-For len 15
Cyril Bonté694574f2015-10-26 22:37:39 +01002065 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002066
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002067 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002068 about logging.
2069
2070
2071capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002072 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002073 May be used in sections : defaults | frontend | listen | backend
2074 no | yes | yes | no
2075 Arguments :
2076 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002077 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002078 appear in the response, with the first letter of each word in
2079 upper case. The header name will not appear in the logs, only the
2080 value is reported, but the position in the logs is respected.
2081
2082 <length> is the maximum number of characters to extract from the value and
2083 report in the logs. The string will be truncated on the right if
2084 it exceeds <length>.
2085
Willy Tarreau4460d032012-11-21 23:37:37 +01002086 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002087 result will be added to the logs between braces ('{}') after the captured
2088 request headers. If multiple headers are captured, they will be delimited by
2089 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002090 the configuration. Non-existent headers will be logged just as an empty
2091 string. Common uses for response header captures include the "Content-length"
2092 header which indicates how many bytes are expected to be returned, the
2093 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002094
Willy Tarreau0900abb2012-11-22 00:21:46 +01002095 There is no limit to the number of captured response headers nor to their
2096 length, though it is wise to keep them low to limit memory usage per session.
2097 In order to keep log format consistent for a same frontend, header captures
2098 can only be declared in a frontend. It is not possible to specify a capture
2099 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002100
2101 Example:
2102 capture response header Content-length len 9
2103 capture response header Location len 15
2104
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002105 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002106 about logging.
2107
2108
Cyril Bontéf0c60612010-02-06 14:44:47 +01002109clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002110 Set the maximum inactivity time on the client side.
2111 May be used in sections : defaults | frontend | listen | backend
2112 yes | yes | yes | no
2113 Arguments :
2114 <timeout> is the timeout value is specified in milliseconds by default, but
2115 can be in any other unit if the number is suffixed by the unit,
2116 as explained at the top of this document.
2117
2118 The inactivity timeout applies when the client is expected to acknowledge or
2119 send data. In HTTP mode, this timeout is particularly important to consider
2120 during the first phase, when the client sends the request, and during the
2121 response while it is reading data sent by the server. The value is specified
2122 in milliseconds by default, but can be in any other unit if the number is
2123 suffixed by the unit, as specified at the top of this document. In TCP mode
2124 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2125 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002126 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002127 losses by specifying timeouts that are slightly above multiples of 3 seconds
2128 (eg: 4 or 5 seconds).
2129
2130 This parameter is specific to frontends, but can be specified once for all in
2131 "defaults" sections. This is in fact one of the easiest solutions not to
2132 forget about it. An unspecified timeout results in an infinite timeout, which
2133 is not recommended. Such a usage is accepted and works but reports a warning
2134 during startup because it may results in accumulation of expired sessions in
2135 the system if the system's timeouts are not configured either.
2136
2137 This parameter is provided for compatibility but is currently deprecated.
2138 Please use "timeout client" instead.
2139
Willy Tarreau036fae02008-01-06 13:24:40 +01002140 See also : "timeout client", "timeout http-request", "timeout server", and
2141 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002142
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002143compression algo <algorithm> ...
2144compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002145compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002146 Enable HTTP compression.
2147 May be used in sections : defaults | frontend | listen | backend
2148 yes | yes | yes | yes
2149 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002150 algo is followed by the list of supported compression algorithms.
2151 type is followed by the list of MIME types that will be compressed.
2152 offload makes haproxy work as a compression offloader only (see notes).
2153
2154 The currently supported algorithms are :
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002155 identity this is mostly for debugging, and it was useful for developing
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002156 the compression feature. Identity does not apply any change on
2157 data.
2158
2159 gzip applies gzip compression. This setting is only available when
2160 support for zlib was built in.
2161
2162 deflate same as gzip, but with deflate algorithm and zlib format.
2163 Note that this algorithm has ambiguous support on many browsers
2164 and no support at all from recent ones. It is strongly
2165 recommended not to use it for anything else than experimentation.
2166 This setting is only available when support for zlib was built
2167 in.
2168
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002169 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002170 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002171 If backend servers support HTTP compression, these directives
2172 will be no-op: haproxy will see the compressed response and will not
2173 compress again. If backend servers do not support HTTP compression and
2174 there is Accept-Encoding header in request, haproxy will compress the
2175 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002176
2177 The "offload" setting makes haproxy remove the Accept-Encoding header to
2178 prevent backend servers from compressing responses. It is strongly
2179 recommended not to do this because this means that all the compression work
2180 will be done on the single point where haproxy is located. However in some
2181 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002182 with broken HTTP compression implementation which can't be turned off.
2183 In that case haproxy can be used to prevent that gateway from emitting
2184 invalid payloads. In this case, simply removing the header in the
2185 configuration does not work because it applies before the header is parsed,
2186 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreau4cf57af2014-07-12 16:37:02 +02002187 then be used for such scenarios. Note: for now, the "offload" setting is
2188 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002189
William Lallemand05097442012-11-20 12:14:28 +01002190 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002191 * the request does not advertise a supported compression algorithm in the
2192 "Accept-Encoding" header
2193 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002194 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002195 * response header "Transfer-Encoding" contains "chunked" (Temporary
2196 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002197 * response contain neither a "Content-Length" header nor a
2198 "Transfer-Encoding" whose last value is "chunked"
2199 * response contains a "Content-Type" header whose first value starts with
2200 "multipart"
2201 * the response contains the "no-transform" value in the "Cache-control"
2202 header
2203 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2204 and later
2205 * The response contains a "Content-Encoding" header, indicating that the
2206 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002207
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002208 Note: The compression does not rewrite Etag headers, and does not emit the
2209 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002210
William Lallemand82fe75c2012-10-23 10:25:10 +02002211 Examples :
2212 compression algo gzip
2213 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002214
Cyril Bontéf0c60612010-02-06 14:44:47 +01002215contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002216 Set the maximum time to wait for a connection attempt to a server to succeed.
2217 May be used in sections : defaults | frontend | listen | backend
2218 yes | no | yes | yes
2219 Arguments :
2220 <timeout> is the timeout value is specified in milliseconds by default, but
2221 can be in any other unit if the number is suffixed by the unit,
2222 as explained at the top of this document.
2223
2224 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002225 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002226 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01002227 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
2228 connect timeout also presets the queue timeout to the same value if this one
2229 has not been specified. Historically, the contimeout was also used to set the
2230 tarpit timeout in a listen section, which is not possible in a pure frontend.
2231
2232 This parameter is specific to backends, but can be specified once for all in
2233 "defaults" sections. This is in fact one of the easiest solutions not to
2234 forget about it. An unspecified timeout results in an infinite timeout, which
2235 is not recommended. Such a usage is accepted and works but reports a warning
2236 during startup because it may results in accumulation of failed sessions in
2237 the system if the system's timeouts are not configured either.
2238
2239 This parameter is provided for backwards compatibility but is currently
2240 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2241 instead.
2242
2243 See also : "timeout connect", "timeout queue", "timeout tarpit",
2244 "timeout server", "contimeout".
2245
2246
Willy Tarreau55165fe2009-05-10 12:02:55 +02002247cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002248 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2249 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002250 Enable cookie-based persistence in a backend.
2251 May be used in sections : defaults | frontend | listen | backend
2252 yes | no | yes | yes
2253 Arguments :
2254 <name> is the name of the cookie which will be monitored, modified or
2255 inserted in order to bring persistence. This cookie is sent to
2256 the client via a "Set-Cookie" header in the response, and is
2257 brought back by the client in a "Cookie" header in all requests.
2258 Special care should be taken to choose a name which does not
2259 conflict with any likely application cookie. Also, if the same
2260 backends are subject to be used by the same clients (eg:
2261 HTTP/HTTPS), care should be taken to use different cookie names
2262 between all backends if persistence between them is not desired.
2263
2264 rewrite This keyword indicates that the cookie will be provided by the
2265 server and that haproxy will have to modify its value to set the
2266 server's identifier in it. This mode is handy when the management
2267 of complex combinations of "Set-cookie" and "Cache-control"
2268 headers is left to the application. The application can then
2269 decide whether or not it is appropriate to emit a persistence
Lukas Tribus3082bde2016-02-03 18:09:37 +01002270 cookie. Since all responses should be monitored, this mode
2271 doesn't work in HTTP tunnel mode. Unless the application
2272 behaviour is very complex and/or broken, it is advised not to
2273 start with this mode for new deployments. This keyword is
2274 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002275
2276 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02002277 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002278
Willy Tarreaua79094d2010-08-31 22:54:15 +02002279 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002280 server. When used without the "preserve" option, if the server
2281 emits a cookie with the same name, it will be remove before
2282 processing. For this reason, this mode can be used to upgrade
2283 existing configurations running in the "rewrite" mode. The cookie
2284 will only be a session cookie and will not be stored on the
2285 client's disk. By default, unless the "indirect" option is added,
2286 the server will see the cookies emitted by the client. Due to
2287 caching effects, it is generally wise to add the "nocache" or
2288 "postonly" keywords (see below). The "insert" keyword is not
2289 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002290
2291 prefix This keyword indicates that instead of relying on a dedicated
2292 cookie for the persistence, an existing one will be completed.
2293 This may be needed in some specific environments where the client
2294 does not support more than one single cookie and the application
2295 already needs it. In this case, whenever the server sets a cookie
2296 named <name>, it will be prefixed with the server's identifier
2297 and a delimiter. The prefix will be removed from all client
2298 requests so that the server still finds the cookie it emitted.
2299 Since all requests and responses are subject to being modified,
Lukas Tribus3082bde2016-02-03 18:09:37 +01002300 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02002301 not compatible with "rewrite" and "insert". Note: it is highly
2302 recommended not to use "indirect" with "prefix", otherwise server
2303 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002304
Willy Tarreaua79094d2010-08-31 22:54:15 +02002305 indirect When this option is specified, no cookie will be emitted to a
2306 client which already has a valid one for the server which has
2307 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002308 it will be removed, unless the "preserve" option is also set. In
2309 "insert" mode, this will additionally remove cookies from the
2310 requests transmitted to the server, making the persistence
2311 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02002312 Note: it is highly recommended not to use "indirect" with
2313 "prefix", otherwise server cookie updates would not be sent to
2314 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002315
2316 nocache This option is recommended in conjunction with the insert mode
2317 when there is a cache between the client and HAProxy, as it
2318 ensures that a cacheable response will be tagged non-cacheable if
2319 a cookie needs to be inserted. This is important because if all
2320 persistence cookies are added on a cacheable home page for
2321 instance, then all customers will then fetch the page from an
2322 outer cache and will all share the same persistence cookie,
2323 leading to one server receiving much more traffic than others.
2324 See also the "insert" and "postonly" options.
2325
2326 postonly This option ensures that cookie insertion will only be performed
2327 on responses to POST requests. It is an alternative to the
2328 "nocache" option, because POST responses are not cacheable, so
2329 this ensures that the persistence cookie will never get cached.
2330 Since most sites do not need any sort of persistence before the
2331 first POST which generally is a login request, this is a very
2332 efficient method to optimize caching without risking to find a
2333 persistence cookie in the cache.
2334 See also the "insert" and "nocache" options.
2335
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002336 preserve This option may only be used with "insert" and/or "indirect". It
2337 allows the server to emit the persistence cookie itself. In this
2338 case, if a cookie is found in the response, haproxy will leave it
2339 untouched. This is useful in order to end persistence after a
2340 logout request for instance. For this, the server just has to
2341 emit a cookie with an invalid value (eg: empty) or with a date in
2342 the past. By combining this mechanism with the "disable-on-404"
2343 check option, it is possible to perform a completely graceful
2344 shutdown because users will definitely leave the server after
2345 they logout.
2346
Willy Tarreau4992dd22012-05-31 21:02:17 +02002347 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
2348 when a cookie is inserted. This attribute is used so that a
2349 user agent doesn't share the cookie with non-HTTP components.
2350 Please check RFC6265 for more information on this attribute.
2351
2352 secure This option tells haproxy to add a "Secure" cookie attribute when
2353 a cookie is inserted. This attribute is used so that a user agent
2354 never emits this cookie over non-secure channels, which means
2355 that a cookie learned with this flag will be presented only over
2356 SSL/TLS connections. Please check RFC6265 for more information on
2357 this attribute.
2358
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002359 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002360 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01002361 name. If the domain begins with a dot, the browser is allowed to
2362 use it for any host ending with that name. It is also possible to
2363 specify several domain names by invoking this option multiple
2364 times. Some browsers might have small limits on the number of
2365 domains, so be careful when doing that. For the record, sending
2366 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002367
Willy Tarreau996a92c2010-10-13 19:30:47 +02002368 maxidle This option allows inserted cookies to be ignored after some idle
2369 time. It only works with insert-mode cookies. When a cookie is
2370 sent to the client, the date this cookie was emitted is sent too.
2371 Upon further presentations of this cookie, if the date is older
2372 than the delay indicated by the parameter (in seconds), it will
2373 be ignored. Otherwise, it will be refreshed if needed when the
2374 response is sent to the client. This is particularly useful to
2375 prevent users who never close their browsers from remaining for
2376 too long on the same server (eg: after a farm size change). When
2377 this option is set and a cookie has no date, it is always
2378 accepted, but gets refreshed in the response. This maintains the
2379 ability for admins to access their sites. Cookies that have a
2380 date in the future further than 24 hours are ignored. Doing so
2381 lets admins fix timezone issues without risking kicking users off
2382 the site.
2383
2384 maxlife This option allows inserted cookies to be ignored after some life
2385 time, whether they're in use or not. It only works with insert
2386 mode cookies. When a cookie is first sent to the client, the date
2387 this cookie was emitted is sent too. Upon further presentations
2388 of this cookie, if the date is older than the delay indicated by
2389 the parameter (in seconds), it will be ignored. If the cookie in
2390 the request has no date, it is accepted and a date will be set.
2391 Cookies that have a date in the future further than 24 hours are
2392 ignored. Doing so lets admins fix timezone issues without risking
2393 kicking users off the site. Contrary to maxidle, this value is
2394 not refreshed, only the first visit date counts. Both maxidle and
2395 maxlife may be used at the time. This is particularly useful to
2396 prevent users who never close their browsers from remaining for
2397 too long on the same server (eg: after a farm size change). This
2398 is stronger than the maxidle method in that it forces a
2399 redispatch after some absolute delay.
2400
Willy Tarreau0ba27502007-12-24 16:55:16 +01002401 There can be only one persistence cookie per HTTP backend, and it can be
2402 declared in a defaults section. The value of the cookie will be the value
2403 indicated after the "cookie" keyword in a "server" statement. If no cookie
2404 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002405
Willy Tarreau0ba27502007-12-24 16:55:16 +01002406 Examples :
2407 cookie JSESSIONID prefix
2408 cookie SRV insert indirect nocache
2409 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02002410 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01002411
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002412 See also : "appsession", "balance source", "capture cookie", "server"
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002413 and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002414
Willy Tarreau983e01e2010-01-11 18:42:06 +01002415
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002416default-server [param*]
2417 Change default options for a server in a backend
2418 May be used in sections : defaults | frontend | listen | backend
2419 yes | no | yes | yes
2420 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01002421 <param*> is a list of parameters for this server. The "default-server"
2422 keyword accepts an important number of options and has a complete
2423 section dedicated to it. Please refer to section 5 for more
2424 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002425
Willy Tarreau983e01e2010-01-11 18:42:06 +01002426 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002427 default-server inter 1000 weight 13
2428
2429 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01002430
Willy Tarreau983e01e2010-01-11 18:42:06 +01002431
Willy Tarreau0ba27502007-12-24 16:55:16 +01002432default_backend <backend>
2433 Specify the backend to use when no "use_backend" rule has been matched.
2434 May be used in sections : defaults | frontend | listen | backend
2435 yes | yes | yes | no
2436 Arguments :
2437 <backend> is the name of the backend to use.
2438
2439 When doing content-switching between frontend and backends using the
2440 "use_backend" keyword, it is often useful to indicate which backend will be
2441 used when no rule has matched. It generally is the dynamic backend which
2442 will catch all undetermined requests.
2443
Willy Tarreau0ba27502007-12-24 16:55:16 +01002444 Example :
2445
2446 use_backend dynamic if url_dyn
2447 use_backend static if url_css url_img extension_img
2448 default_backend dynamic
2449
Willy Tarreau2769aa02007-12-27 18:26:09 +01002450 See also : "use_backend", "reqsetbe", "reqisetbe"
2451
Willy Tarreau0ba27502007-12-24 16:55:16 +01002452
Baptiste Assmann27f51342013-10-09 06:51:49 +02002453description <string>
2454 Describe a listen, frontend or backend.
2455 May be used in sections : defaults | frontend | listen | backend
2456 no | yes | yes | yes
2457 Arguments : string
2458
2459 Allows to add a sentence to describe the related object in the HAProxy HTML
2460 stats page. The description will be printed on the right of the object name
2461 it describes.
2462 No need to backslash spaces in the <string> arguments.
2463
2464
Willy Tarreau0ba27502007-12-24 16:55:16 +01002465disabled
2466 Disable a proxy, frontend or backend.
2467 May be used in sections : defaults | frontend | listen | backend
2468 yes | yes | yes | yes
2469 Arguments : none
2470
2471 The "disabled" keyword is used to disable an instance, mainly in order to
2472 liberate a listening port or to temporarily disable a service. The instance
2473 will still be created and its configuration will be checked, but it will be
2474 created in the "stopped" state and will appear as such in the statistics. It
2475 will not receive any traffic nor will it send any health-checks or logs. It
2476 is possible to disable many instances at once by adding the "disabled"
2477 keyword in a "defaults" section.
2478
2479 See also : "enabled"
2480
2481
Willy Tarreau5ce94572010-06-07 14:35:41 +02002482dispatch <address>:<port>
2483 Set a default server address
2484 May be used in sections : defaults | frontend | listen | backend
2485 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02002486 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02002487
2488 <address> is the IPv4 address of the default server. Alternatively, a
2489 resolvable hostname is supported, but this name will be resolved
2490 during start-up.
2491
2492 <ports> is a mandatory port specification. All connections will be sent
2493 to this port, and it is not permitted to use port offsets as is
2494 possible with normal servers.
2495
Willy Tarreau787aed52011-04-15 06:45:37 +02002496 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02002497 server can take the connection. In the past it was used to forward non
2498 persistent connections to an auxiliary load balancer. Due to its simple
2499 syntax, it has also been used for simple TCP relays. It is recommended not to
2500 use it for more clarity, and to use the "server" directive instead.
2501
2502 See also : "server"
2503
2504
Willy Tarreau0ba27502007-12-24 16:55:16 +01002505enabled
2506 Enable a proxy, frontend or backend.
2507 May be used in sections : defaults | frontend | listen | backend
2508 yes | yes | yes | yes
2509 Arguments : none
2510
2511 The "enabled" keyword is used to explicitly enable an instance, when the
2512 defaults has been set to "disabled". This is very rarely used.
2513
2514 See also : "disabled"
2515
2516
2517errorfile <code> <file>
2518 Return a file contents instead of errors generated by HAProxy
2519 May be used in sections : defaults | frontend | listen | backend
2520 yes | yes | yes | yes
2521 Arguments :
2522 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002523 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002524
2525 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002526 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01002527 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01002528 error pages, and to use absolute paths, since files are read
2529 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002530
2531 It is important to understand that this keyword is not meant to rewrite
2532 errors returned by the server, but errors detected and returned by HAProxy.
2533 This is why the list of supported errors is limited to a small set.
2534
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002535 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2536
Willy Tarreau0ba27502007-12-24 16:55:16 +01002537 The files are returned verbatim on the TCP socket. This allows any trick such
2538 as redirections to another URL or site, as well as tricks to clean cookies,
2539 force enable or disable caching, etc... The package provides default error
2540 files returning the same contents as default errors.
2541
Willy Tarreau59140a22009-02-22 12:02:30 +01002542 The files should not exceed the configured buffer size (BUFSIZE), which
2543 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
2544 not to put any reference to local contents (eg: images) in order to avoid
2545 loops between the client and HAProxy when all servers are down, causing an
2546 error to be returned instead of an image. For better HTTP compliance, it is
2547 recommended that all header lines end with CR-LF and not LF alone.
2548
Willy Tarreau0ba27502007-12-24 16:55:16 +01002549 The files are read at the same time as the configuration and kept in memory.
2550 For this reason, the errors continue to be returned even when the process is
2551 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01002552 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002553 403 status code and interrogating a blocked URL.
2554
2555 See also : "errorloc", "errorloc302", "errorloc303"
2556
Willy Tarreau59140a22009-02-22 12:02:30 +01002557 Example :
2558 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau2705a612014-05-23 17:38:34 +02002559 errorfile 408 /dev/null # workaround Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01002560 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
2561 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
2562
Willy Tarreau2769aa02007-12-27 18:26:09 +01002563
2564errorloc <code> <url>
2565errorloc302 <code> <url>
2566 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2567 May be used in sections : defaults | frontend | listen | backend
2568 yes | yes | yes | yes
2569 Arguments :
2570 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002571 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002572
2573 <url> it is the exact contents of the "Location" header. It may contain
2574 either a relative URI to an error page hosted on the same site,
2575 or an absolute URI designating an error page on another site.
2576 Special care should be given to relative URIs to avoid redirect
2577 loops if the URI itself may generate the same error (eg: 500).
2578
2579 It is important to understand that this keyword is not meant to rewrite
2580 errors returned by the server, but errors detected and returned by HAProxy.
2581 This is why the list of supported errors is limited to a small set.
2582
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002583 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2584
Willy Tarreau2769aa02007-12-27 18:26:09 +01002585 Note that both keyword return the HTTP 302 status code, which tells the
2586 client to fetch the designated URL using the same HTTP method. This can be
2587 quite problematic in case of non-GET methods such as POST, because the URL
2588 sent to the client might not be allowed for something other than GET. To
2589 workaround this problem, please use "errorloc303" which send the HTTP 303
2590 status code, indicating to the client that the URL must be fetched with a GET
2591 request.
2592
2593 See also : "errorfile", "errorloc303"
2594
2595
2596errorloc303 <code> <url>
2597 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2598 May be used in sections : defaults | frontend | listen | backend
2599 yes | yes | yes | yes
2600 Arguments :
2601 <code> is the HTTP status code. Currently, HAProxy is capable of
2602 generating codes 400, 403, 408, 500, 502, 503, and 504.
2603
2604 <url> it is the exact contents of the "Location" header. It may contain
2605 either a relative URI to an error page hosted on the same site,
2606 or an absolute URI designating an error page on another site.
2607 Special care should be given to relative URIs to avoid redirect
2608 loops if the URI itself may generate the same error (eg: 500).
2609
2610 It is important to understand that this keyword is not meant to rewrite
2611 errors returned by the server, but errors detected and returned by HAProxy.
2612 This is why the list of supported errors is limited to a small set.
2613
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002614 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2615
Willy Tarreau2769aa02007-12-27 18:26:09 +01002616 Note that both keyword return the HTTP 303 status code, which tells the
2617 client to fetch the designated URL using the same HTTP GET method. This
2618 solves the usual problems associated with "errorloc" and the 302 code. It is
2619 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002620 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002621
2622 See also : "errorfile", "errorloc", "errorloc302"
2623
2624
Willy Tarreau4de91492010-01-22 19:10:05 +01002625force-persist { if | unless } <condition>
2626 Declare a condition to force persistence on down servers
2627 May be used in sections: defaults | frontend | listen | backend
2628 no | yes | yes | yes
2629
2630 By default, requests are not dispatched to down servers. It is possible to
2631 force this using "option persist", but it is unconditional and redispatches
2632 to a valid server if "option redispatch" is set. That leaves with very little
2633 possibilities to force some requests to reach a server which is artificially
2634 marked down for maintenance operations.
2635
2636 The "force-persist" statement allows one to declare various ACL-based
2637 conditions which, when met, will cause a request to ignore the down status of
2638 a server and still try to connect to it. That makes it possible to start a
2639 server, still replying an error to the health checks, and run a specially
2640 configured browser to test the service. Among the handy methods, one could
2641 use a specific source IP address, or a specific cookie. The cookie also has
2642 the advantage that it can easily be added/removed on the browser from a test
2643 page. Once the service is validated, it is then possible to open the service
2644 to the world by returning a valid response to health checks.
2645
2646 The forced persistence is enabled when an "if" condition is met, or unless an
2647 "unless" condition is met. The final redispatch is always disabled when this
2648 is used.
2649
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002650 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002651 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01002652
2653
Willy Tarreau2769aa02007-12-27 18:26:09 +01002654fullconn <conns>
2655 Specify at what backend load the servers will reach their maxconn
2656 May be used in sections : defaults | frontend | listen | backend
2657 yes | no | yes | yes
2658 Arguments :
2659 <conns> is the number of connections on the backend which will make the
2660 servers use the maximal number of connections.
2661
Willy Tarreau198a7442008-01-17 12:05:32 +01002662 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01002663 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01002664 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01002665 load. The server will then always accept at least <minconn> connections,
2666 never more than <maxconn>, and the limit will be on the ramp between both
2667 values when the backend has less than <conns> concurrent connections. This
2668 makes it possible to limit the load on the servers during normal loads, but
2669 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002670 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002671
Willy Tarreaufbb78422011-06-05 15:38:35 +02002672 Since it's hard to get this value right, haproxy automatically sets it to
2673 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01002674 backend (based on "use_backend" and "default_backend" rules). That way it's
2675 safe to leave it unset. However, "use_backend" involving dynamic names are
2676 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02002677
Willy Tarreau2769aa02007-12-27 18:26:09 +01002678 Example :
2679 # The servers will accept between 100 and 1000 concurrent connections each
2680 # and the maximum of 1000 will be reached when the backend reaches 10000
2681 # connections.
2682 backend dynamic
2683 fullconn 10000
2684 server srv1 dyn1:80 minconn 100 maxconn 1000
2685 server srv2 dyn2:80 minconn 100 maxconn 1000
2686
2687 See also : "maxconn", "server"
2688
2689
2690grace <time>
2691 Maintain a proxy operational for some time after a soft stop
2692 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01002693 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01002694 Arguments :
2695 <time> is the time (by default in milliseconds) for which the instance
2696 will remain operational with the frontend sockets still listening
2697 when a soft-stop is received via the SIGUSR1 signal.
2698
2699 This may be used to ensure that the services disappear in a certain order.
2700 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002701 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01002702 needed by the equipment to detect the failure.
2703
2704 Note that currently, there is very little benefit in using this parameter,
2705 and it may in fact complicate the soft-reconfiguration process more than
2706 simplify it.
2707
Willy Tarreau0ba27502007-12-24 16:55:16 +01002708
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002709hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002710 Specify a method to use for mapping hashes to servers
2711 May be used in sections : defaults | frontend | listen | backend
2712 yes | no | yes | yes
2713 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04002714 <method> is the method used to select a server from the hash computed by
2715 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002716
Bhaskar98634f02013-10-29 23:30:51 -04002717 map-based the hash table is a static array containing all alive servers.
2718 The hashes will be very smooth, will consider weights, but
2719 will be static in that weight changes while a server is up
2720 will be ignored. This means that there will be no slow start.
2721 Also, since a server is selected by its position in the array,
2722 most mappings are changed when the server count changes. This
2723 means that when a server goes up or down, or when a server is
2724 added to a farm, most connections will be redistributed to
2725 different servers. This can be inconvenient with caches for
2726 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01002727
Bhaskar98634f02013-10-29 23:30:51 -04002728 consistent the hash table is a tree filled with many occurrences of each
2729 server. The hash key is looked up in the tree and the closest
2730 server is chosen. This hash is dynamic, it supports changing
2731 weights while the servers are up, so it is compatible with the
2732 slow start feature. It has the advantage that when a server
2733 goes up or down, only its associations are moved. When a
2734 server is added to the farm, only a few part of the mappings
2735 are redistributed, making it an ideal method for caches.
2736 However, due to its principle, the distribution will never be
2737 very smooth and it may sometimes be necessary to adjust a
2738 server's weight or its ID to get a more balanced distribution.
2739 In order to get the same distribution on multiple load
2740 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002741 same IDs. Note: consistent hash uses sdbm and avalanche if no
2742 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04002743
2744 <function> is the hash function to be used :
2745
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002746 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04002747 reimplementation of ndbm) database library. It was found to do
2748 well in scrambling bits, causing better distribution of the keys
2749 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002750 function with good distribution, unless the total server weight
2751 is a multiple of 64, in which case applying the avalanche
2752 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04002753
2754 djb2 this function was first proposed by Dan Bernstein many years ago
2755 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002756 function provides a better distribution than sdbm. It generally
2757 works well with text-based inputs though it can perform extremely
2758 poorly with numeric-only input or when the total server weight is
2759 a multiple of 33, unless the avalanche modifier is also used.
2760
Willy Tarreaua0f42712013-11-14 14:30:35 +01002761 wt6 this function was designed for haproxy while testing other
2762 functions in the past. It is not as smooth as the other ones, but
2763 is much less sensible to the input data set or to the number of
2764 servers. It can make sense as an alternative to sdbm+avalanche or
2765 djb2+avalanche for consistent hashing or when hashing on numeric
2766 data such as a source IP address or a visitor identifier in a URL
2767 parameter.
2768
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002769 <modifier> indicates an optional method applied after hashing the key :
2770
2771 avalanche This directive indicates that the result from the hash
2772 function above should not be used in its raw form but that
2773 a 4-byte full avalanche hash must be applied first. The
2774 purpose of this step is to mix the resulting bits from the
2775 previous hash in order to avoid any undesired effect when
2776 the input contains some limited values or when the number of
2777 servers is a multiple of one of the hash's components (64
2778 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
2779 result less predictable, but it's also not as smooth as when
2780 using the original function. Some testing might be needed
2781 with some workloads. This hash is one of the many proposed
2782 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002783
Bhaskar98634f02013-10-29 23:30:51 -04002784 The default hash type is "map-based" and is recommended for most usages. The
2785 default function is "sdbm", the selection of a function should be based on
2786 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002787
2788 See also : "balance", "server"
2789
2790
Willy Tarreau0ba27502007-12-24 16:55:16 +01002791http-check disable-on-404
2792 Enable a maintenance mode upon HTTP/404 response to health-checks
2793 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01002794 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01002795 Arguments : none
2796
2797 When this option is set, a server which returns an HTTP code 404 will be
2798 excluded from further load-balancing, but will still receive persistent
2799 connections. This provides a very convenient method for Web administrators
2800 to perform a graceful shutdown of their servers. It is also important to note
2801 that a server which is detected as failed while it was in this mode will not
2802 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
2803 will immediately be reinserted into the farm. The status on the stats page
2804 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01002805 option only works in conjunction with the "httpchk" option. If this option
2806 is used with "http-check expect", then it has precedence over it so that 404
2807 responses will still be considered as soft-stop.
2808
2809 See also : "option httpchk", "http-check expect"
2810
2811
2812http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002813 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01002814 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02002815 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01002816 Arguments :
2817 <match> is a keyword indicating how to look for a specific pattern in the
2818 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002819 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01002820 exclamation mark ("!") to negate the match. Spaces are allowed
2821 between the exclamation mark and the keyword. See below for more
2822 details on the supported keywords.
2823
2824 <pattern> is the pattern to look for. It may be a string or a regular
2825 expression. If the pattern contains spaces, they must be escaped
2826 with the usual backslash ('\').
2827
2828 By default, "option httpchk" considers that response statuses 2xx and 3xx
2829 are valid, and that others are invalid. When "http-check expect" is used,
2830 it defines what is considered valid or invalid. Only one "http-check"
2831 statement is supported in a backend. If a server fails to respond or times
2832 out, the check obviously fails. The available matches are :
2833
2834 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002835 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002836 response's status code is exactly this string. If the
2837 "status" keyword is prefixed with "!", then the response
2838 will be considered invalid if the status code matches.
2839
2840 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002841 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002842 response's status code matches the expression. If the
2843 "rstatus" keyword is prefixed with "!", then the response
2844 will be considered invalid if the status code matches.
2845 This is mostly used to check for multiple codes.
2846
2847 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002848 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002849 response's body contains this exact string. If the
2850 "string" keyword is prefixed with "!", then the response
2851 will be considered invalid if the body contains this
2852 string. This can be used to look for a mandatory word at
2853 the end of a dynamic page, or to detect a failure when a
2854 specific error appears on the check page (eg: a stack
2855 trace).
2856
2857 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002858 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002859 response's body matches this expression. If the "rstring"
2860 keyword is prefixed with "!", then the response will be
2861 considered invalid if the body matches the expression.
2862 This can be used to look for a mandatory word at the end
2863 of a dynamic page, or to detect a failure when a specific
2864 error appears on the check page (eg: a stack trace).
2865
2866 It is important to note that the responses will be limited to a certain size
2867 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
2868 Thus, too large responses may not contain the mandatory pattern when using
2869 "string" or "rstring". If a large response is absolutely required, it is
2870 possible to change the default max size by setting the global variable.
2871 However, it is worth keeping in mind that parsing very large responses can
2872 waste some CPU cycles, especially when regular expressions are used, and that
2873 it is always better to focus the checks on smaller resources.
2874
Cyril Bontéa448e162015-01-30 00:07:07 +01002875 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
2876 will automatically append a "Connection: close" header, meaning that this
2877 header should not be present in the request provided by "option httpchk".
2878
Willy Tarreaubd741542010-03-16 18:46:54 +01002879 Last, if "http-check expect" is combined with "http-check disable-on-404",
2880 then this last one has precedence when the server responds with 404.
2881
2882 Examples :
2883 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002884 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01002885
2886 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002887 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01002888
2889 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002890 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01002891
2892 # check that we have a correct hexadecimal tag before /html
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002893 http-check expect rstring <!--tag:[0-9a-f]*</html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01002894
Willy Tarreaubd741542010-03-16 18:46:54 +01002895 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01002896
2897
Willy Tarreauef781042010-01-27 11:53:01 +01002898http-check send-state
2899 Enable emission of a state header with HTTP health checks
2900 May be used in sections : defaults | frontend | listen | backend
2901 yes | no | yes | yes
2902 Arguments : none
2903
2904 When this option is set, haproxy will systematically send a special header
2905 "X-Haproxy-Server-State" with a list of parameters indicating to each server
2906 how they are seen by haproxy. This can be used for instance when a server is
2907 manipulated without access to haproxy and the operator needs to know whether
2908 haproxy still sees it up or not, or if the server is the last one in a farm.
2909
2910 The header is composed of fields delimited by semi-colons, the first of which
2911 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
2912 checks on the total number before transition, just as appears in the stats
2913 interface. Next headers are in the form "<variable>=<value>", indicating in
2914 no specific order some values available in the stats interface :
2915 - a variable "name", containing the name of the backend followed by a slash
2916 ("/") then the name of the server. This can be used when a server is
2917 checked in multiple backends.
2918
2919 - a variable "node" containing the name of the haproxy node, as set in the
2920 global "node" variable, otherwise the system's hostname if unspecified.
2921
2922 - a variable "weight" indicating the weight of the server, a slash ("/")
2923 and the total weight of the farm (just counting usable servers). This
2924 helps to know if other servers are available to handle the load when this
2925 one fails.
2926
2927 - a variable "scur" indicating the current number of concurrent connections
2928 on the server, followed by a slash ("/") then the total number of
2929 connections on all servers of the same backend.
2930
2931 - a variable "qcur" indicating the current number of requests in the
2932 server's queue.
2933
2934 Example of a header received by the application server :
2935 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
2936 scur=13/22; qcur=0
2937
2938 See also : "option httpchk", "http-check disable-on-404"
2939
Willy Tarreauccbcc372012-12-27 12:37:57 +01002940http-request { allow | deny | tarpit | auth [realm <realm>] | redirect <rule> |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02002941 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002942 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06002943 replace-header <name> <match-regex> <replace-fmt> |
2944 replace-value <name> <match-regex> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02002945 set-tos <tos> | set-mark <mark> |
2946 add-acl(<file name>) <key fmt> |
2947 del-acl(<file name>) <key fmt> |
2948 del-map(<file name>) <key fmt> |
2949 set-map(<file name>) <key fmt> <value fmt>
2950 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01002951 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002952 Access control for Layer 7 requests
2953
2954 May be used in sections: defaults | frontend | listen | backend
2955 no | yes | yes | yes
2956
Willy Tarreau20b0de52012-12-24 15:45:22 +01002957 The http-request statement defines a set of rules which apply to layer 7
2958 processing. The rules are evaluated in their declaration order when they are
2959 met in a frontend, listen or backend section. Any rule may optionally be
2960 followed by an ACL-based condition, in which case it will only be evaluated
2961 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002962
Willy Tarreau20b0de52012-12-24 15:45:22 +01002963 The first keyword is the rule's action. Currently supported actions include :
2964 - "allow" : this stops the evaluation of the rules and lets the request
2965 pass the check. No further "http-request" rules are evaluated.
2966
2967 - "deny" : this stops the evaluation of the rules and immediately rejects
2968 the request and emits an HTTP 403 error. No further "http-request" rules
2969 are evaluated.
2970
Willy Tarreauccbcc372012-12-27 12:37:57 +01002971 - "tarpit" : this stops the evaluation of the rules and immediately blocks
2972 the request without responding for a delay specified by "timeout tarpit"
2973 or "timeout connect" if the former is not set. After that delay, if the
2974 client is still connected, an HTTP error 500 is returned so that the
2975 client does not suspect it has been tarpitted. Logs will report the flags
2976 "PT". The goal of the tarpit rule is to slow down robots during an attack
2977 when they're limited on the number of concurrent requests. It can be very
2978 efficient against very dumb robots, and will significantly reduce the
2979 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002980 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreauccbcc372012-12-27 12:37:57 +01002981 front firewall to support insane number of concurrent connections.
2982
Willy Tarreau20b0de52012-12-24 15:45:22 +01002983 - "auth" : this stops the evaluation of the rules and immediately responds
2984 with an HTTP 401 or 407 error code to invite the user to present a valid
2985 user name and password. No further "http-request" rules are evaluated. An
2986 optional "realm" parameter is supported, it sets the authentication realm
2987 that is returned with the response (typically the application's name).
2988
Willy Tarreau81499eb2012-12-27 12:19:02 +01002989 - "redirect" : this performs an HTTP redirection based on a redirect rule.
2990 This is exactly the same as the "redirect" statement except that it
2991 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01002992 "http-request" rules and that these rules use the "log-format" strings.
2993 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01002994
Willy Tarreau20b0de52012-12-24 15:45:22 +01002995 - "add-header" appends an HTTP header field whose name is specified in
2996 <name> and whose value is defined by <fmt> which follows the log-format
2997 rules (see Custom Log Format in section 8.2.4). This is particularly
2998 useful to pass connection-specific information to the server (eg: the
2999 client's SSL certificate), or to combine several headers into one. This
3000 rule is not final, so it is possible to add other similar rules. Note
3001 that header addition is performed immediately, so one rule might reuse
3002 the resulting header from a previous rule.
3003
3004 - "set-header" does the same as "add-header" except that the header name
3005 is first removed if it existed. This is useful when passing security
3006 information to the server, where the header must not be manipulated by
Willy Tarreaufcf75d22015-01-21 20:39:27 +01003007 external users. Note that the new value is computed before the removal so
3008 it is possible to concatenate a value to an existing header.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003009
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003010 - "del-header" removes all HTTP header fields whose name is specified in
3011 <name>.
3012
Sasha Pachev218f0642014-06-16 12:05:59 -06003013 - "replace-header" matches the regular expression in all occurrences of
3014 header field <name> according to <match-regex>, and replaces them with
3015 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3016 and work like in <fmt> arguments in "add-header". The match is only
3017 case-sensitive. It is important to understand that this action only
3018 considers whole header lines, regardless of the number of values they
3019 may contain. This usage is suited to headers naturally containing commas
3020 in their value, such as If-Modified-Since and so on.
3021
3022 Example:
3023
3024 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
3025
3026 applied to:
3027
3028 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3029
3030 outputs:
3031
3032 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3033
3034 assuming the backend IP is 192.168.1.20
3035
3036 - "replace-value" works like "replace-header" except that it matches the
3037 regex against every comma-delimited value of the header field <name>
3038 instead of the entire header. This is suited for all headers which are
3039 allowed to carry more than one value. An example could be the Accept
3040 header.
3041
3042 Example:
3043
3044 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
3045
3046 applied to:
3047
3048 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
3049
3050 outputs:
3051
3052 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
3053
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003054 - "set-nice" sets the "nice" factor of the current request being processed.
3055 It only has effect against the other requests being processed at the same
3056 time. The default value is 0, unless altered by the "nice" setting on the
3057 "bind" line. The accepted range is -1024..1024. The higher the value, the
3058 nicest the request will be. Lower values will make the request more
3059 important than other ones. This can be useful to improve the speed of
3060 some requests, or lower the priority of non-important requests. Using
3061 this setting without prior experimentation can cause some major slowdown.
3062
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003063 - "set-log-level" is used to change the log level of the current request
3064 when a certain condition is met. Valid levels are the 8 syslog levels
3065 (see the "log" keyword) plus the special level "silent" which disables
3066 logging for this request. This rule is not final so the last matching
3067 rule wins. This rule can be useful to disable health checks coming from
3068 another equipment.
3069
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003070 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3071 the client to the value passed in <tos> on platforms which support this.
3072 This value represents the whole 8 bits of the IP TOS field, and can be
3073 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3074 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3075 bits are always 0. This can be used to adjust some routing behaviour on
3076 border routers based on some information from the request. See RFC 2474,
3077 2597, 3260 and 4594 for more information.
3078
Willy Tarreau51347ed2013-06-11 19:34:13 +02003079 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3080 client to the value passed in <mark> on platforms which support it. This
3081 value is an unsigned 32 bit value which can be matched by netfilter and
3082 by the routing table. It can be expressed both in decimal or hexadecimal
3083 format (prefixed by "0x"). This can be useful to force certain packets to
3084 take a different route (for example a cheaper network path for bulk
3085 downloads). This works on Linux kernels 2.6.32 and above and requires
3086 admin privileges.
3087
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003088 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3089 from a file (even a dummy empty file). The file name of the ACL to be
3090 updated is passed between parentheses. It takes one argument: <key fmt>,
3091 which follows log-format rules, to collect content of the new entry. It
3092 performs a lookup in the ACL before insertion, to avoid duplicated (or
3093 more) values. This lookup is done by a linear search and can be expensive
3094 with large lists! It is the equivalent of the "add acl" command from the
3095 stats socket, but can be triggered by an HTTP request.
3096
3097 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3098 from a file (even a dummy empty file). The file name of the ACL to be
3099 updated is passed between parentheses. It takes one argument: <key fmt>,
3100 which follows log-format rules, to collect content of the entry to delete.
3101 It is the equivalent of the "del acl" command from the stats socket, but
3102 can be triggered by an HTTP request.
3103
3104 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3105 from a file (even a dummy empty file). The file name of the MAP to be
3106 updated is passed between parentheses. It takes one argument: <key fmt>,
3107 which follows log-format rules, to collect content of the entry to delete.
3108 It takes one argument: "file name" It is the equivalent of the "del map"
3109 command from the stats socket, but can be triggered by an HTTP request.
3110
3111 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3112 from a file (even a dummy empty file). The file name of the MAP to be
3113 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3114 which follows log-format rules, used to collect MAP key, and <value fmt>,
3115 which follows log-format rules, used to collect content for the new entry.
3116 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3117 more) values. This lookup is done by a linear search and can be expensive
3118 with large lists! It is the equivalent of the "set map" command from the
3119 stats socket, but can be triggered by an HTTP request.
3120
Willy Tarreau20b0de52012-12-24 15:45:22 +01003121 There is no limit to the number of http-request statements per instance.
3122
3123 It is important to know that http-request rules are processed very early in
3124 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
3125 rules. That way, headers added by "add-header"/"set-header" are visible by
3126 almost all further ACL rules.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003127
3128 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01003129 acl nagios src 192.168.129.3
3130 acl local_net src 192.168.0.0/16
3131 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003132
Cyril Bonté78caf842010-03-10 22:41:43 +01003133 http-request allow if nagios
3134 http-request allow if local_net auth_ok
3135 http-request auth realm Gimme if local_net auth_ok
3136 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003137
Cyril Bonté78caf842010-03-10 22:41:43 +01003138 Example:
3139 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01003140 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003141
Willy Tarreau20b0de52012-12-24 15:45:22 +01003142 Example:
3143 http-request set-header X-Haproxy-Current-Date %T
3144 http-request set-header X-SSL %[ssl_fc]
Willy Tarreaud3a93a92015-08-27 17:15:05 +02003145 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
Willy Tarreau20b0de52012-12-24 15:45:22 +01003146 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
3147 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
3148 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
3149 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
3150 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
3151 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
3152
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003153 Example:
3154 acl key req.hdr(X-Add-Acl-Key) -m found
3155 acl add path /addacl
3156 acl del path /delacl
3157
3158 acl myhost hdr(Host) -f myhost.lst
3159
3160 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
3161 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
3162
3163 Example:
3164 acl value req.hdr(X-Value) -m found
3165 acl setmap path /setmap
3166 acl delmap path /delmap
3167
3168 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3169
3170 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
3171 http-request del-map(map.lst) %[src] if delmap
3172
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02003173 See also : "stats http-request", section 3.4 about userlists and section 7
3174 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01003175
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003176http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003177 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003178 replace-header <name> <regex-match> <replace-fmt> |
3179 replace-value <name> <regex-match> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003180 set-log-level <level> | set-mark <mark> | set-tos <tos> |
3181 add-acl(<file name>) <key fmt> |
3182 del-acl(<file name>) <key fmt> |
3183 del-map(<file name>) <key fmt> |
3184 set-map(<file name>) <key fmt> <value fmt>
3185 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02003186 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003187 Access control for Layer 7 responses
3188
3189 May be used in sections: defaults | frontend | listen | backend
3190 no | yes | yes | yes
3191
3192 The http-response statement defines a set of rules which apply to layer 7
3193 processing. The rules are evaluated in their declaration order when they are
3194 met in a frontend, listen or backend section. Any rule may optionally be
3195 followed by an ACL-based condition, in which case it will only be evaluated
3196 if the condition is true. Since these rules apply on responses, the backend
3197 rules are applied first, followed by the frontend's rules.
3198
3199 The first keyword is the rule's action. Currently supported actions include :
3200 - "allow" : this stops the evaluation of the rules and lets the response
3201 pass the check. No further "http-response" rules are evaluated for the
3202 current section.
3203
3204 - "deny" : this stops the evaluation of the rules and immediately rejects
3205 the response and emits an HTTP 502 error. No further "http-response"
3206 rules are evaluated.
3207
3208 - "add-header" appends an HTTP header field whose name is specified in
3209 <name> and whose value is defined by <fmt> which follows the log-format
3210 rules (see Custom Log Format in section 8.2.4). This may be used to send
3211 a cookie to a client for example, or to pass some internal information.
3212 This rule is not final, so it is possible to add other similar rules.
3213 Note that header addition is performed immediately, so one rule might
3214 reuse the resulting header from a previous rule.
3215
3216 - "set-header" does the same as "add-header" except that the header name
3217 is first removed if it existed. This is useful when passing security
3218 information to the server, where the header must not be manipulated by
3219 external users.
3220
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003221 - "del-header" removes all HTTP header fields whose name is specified in
3222 <name>.
3223
Sasha Pachev218f0642014-06-16 12:05:59 -06003224 - "replace-header" matches the regular expression in all occurrences of
3225 header field <name> according to <match-regex>, and replaces them with
3226 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3227 and work like in <fmt> arguments in "add-header". The match is only
3228 case-sensitive. It is important to understand that this action only
3229 considers whole header lines, regardless of the number of values they
3230 may contain. This usage is suited to headers naturally containing commas
3231 in their value, such as Set-Cookie, Expires and so on.
3232
3233 Example:
3234
3235 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
3236
3237 applied to:
3238
3239 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
3240
3241 outputs:
3242
3243 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
3244
3245 assuming the backend IP is 192.168.1.20.
3246
3247 - "replace-value" works like "replace-header" except that it matches the
3248 regex against every comma-delimited value of the header field <name>
3249 instead of the entire header. This is suited for all headers which are
3250 allowed to carry more than one value. An example could be the Accept
3251 header.
3252
3253 Example:
3254
3255 http-response replace-value Cache-control ^public$ private
3256
3257 applied to:
3258
3259 Cache-Control: max-age=3600, public
3260
3261 outputs:
3262
3263 Cache-Control: max-age=3600, private
3264
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003265 - "set-nice" sets the "nice" factor of the current request being processed.
3266 It only has effect against the other requests being processed at the same
3267 time. The default value is 0, unless altered by the "nice" setting on the
3268 "bind" line. The accepted range is -1024..1024. The higher the value, the
3269 nicest the request will be. Lower values will make the request more
3270 important than other ones. This can be useful to improve the speed of
3271 some requests, or lower the priority of non-important requests. Using
3272 this setting without prior experimentation can cause some major slowdown.
3273
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003274 - "set-log-level" is used to change the log level of the current request
3275 when a certain condition is met. Valid levels are the 8 syslog levels
3276 (see the "log" keyword) plus the special level "silent" which disables
3277 logging for this request. This rule is not final so the last matching
3278 rule wins. This rule can be useful to disable health checks coming from
3279 another equipment.
3280
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003281 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3282 the client to the value passed in <tos> on platforms which support this.
3283 This value represents the whole 8 bits of the IP TOS field, and can be
3284 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3285 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3286 bits are always 0. This can be used to adjust some routing behaviour on
3287 border routers based on some information from the request. See RFC 2474,
3288 2597, 3260 and 4594 for more information.
3289
Willy Tarreau51347ed2013-06-11 19:34:13 +02003290 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3291 client to the value passed in <mark> on platforms which support it. This
3292 value is an unsigned 32 bit value which can be matched by netfilter and
3293 by the routing table. It can be expressed both in decimal or hexadecimal
3294 format (prefixed by "0x"). This can be useful to force certain packets to
3295 take a different route (for example a cheaper network path for bulk
3296 downloads). This works on Linux kernels 2.6.32 and above and requires
3297 admin privileges.
3298
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003299 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3300 from a file (even a dummy empty file). The file name of the ACL to be
3301 updated is passed between parentheses. It takes one argument: <key fmt>,
3302 which follows log-format rules, to collect content of the new entry. It
3303 performs a lookup in the ACL before insertion, to avoid duplicated (or
3304 more) values. This lookup is done by a linear search and can be expensive
3305 with large lists! It is the equivalent of the "add acl" command from the
3306 stats socket, but can be triggered by an HTTP response.
3307
3308 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3309 from a file (even a dummy empty file). The file name of the ACL to be
3310 updated is passed between parentheses. It takes one argument: <key fmt>,
3311 which follows log-format rules, to collect content of the entry to delete.
3312 It is the equivalent of the "del acl" command from the stats socket, but
3313 can be triggered by an HTTP response.
3314
3315 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3316 from a file (even a dummy empty file). The file name of the MAP to be
3317 updated is passed between parentheses. It takes one argument: <key fmt>,
3318 which follows log-format rules, to collect content of the entry to delete.
3319 It takes one argument: "file name" It is the equivalent of the "del map"
3320 command from the stats socket, but can be triggered by an HTTP response.
3321
3322 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3323 from a file (even a dummy empty file). The file name of the MAP to be
3324 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3325 which follows log-format rules, used to collect MAP key, and <value fmt>,
3326 which follows log-format rules, used to collect content for the new entry.
3327 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3328 more) values. This lookup is done by a linear search and can be expensive
3329 with large lists! It is the equivalent of the "set map" command from the
3330 stats socket, but can be triggered by an HTTP response.
3331
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003332 There is no limit to the number of http-response statements per instance.
3333
Godbach09250262013-07-02 01:19:15 +08003334 It is important to know that http-response rules are processed very early in
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003335 the HTTP processing, before "reqdel" or "reqrep" rules. That way, headers
3336 added by "add-header"/"set-header" are visible by almost all further ACL
3337 rules.
3338
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003339 Example:
3340 acl key_acl res.hdr(X-Acl-Key) -m found
3341
3342 acl myhost hdr(Host) -f myhost.lst
3343
3344 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3345 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3346
3347 Example:
3348 acl value res.hdr(X-Value) -m found
3349
3350 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3351
3352 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
3353 http-response del-map(map.lst) %[src] if ! value
3354
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003355 See also : "http-request", section 3.4 about userlists and section 7 about
3356 ACL usage.
3357
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02003358
Mark Lamourinec2247f02012-01-04 13:02:01 -05003359http-send-name-header [<header>]
3360 Add the server name to a request. Use the header string given by <header>
3361
3362 May be used in sections: defaults | frontend | listen | backend
3363 yes | no | yes | yes
3364
3365 Arguments :
3366
3367 <header> The header string to use to send the server name
3368
3369 The "http-send-name-header" statement causes the name of the target
3370 server to be added to the headers of an HTTP request. The name
3371 is added with the header string proved.
3372
3373 See also : "server"
3374
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003375id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02003376 Set a persistent ID to a proxy.
3377 May be used in sections : defaults | frontend | listen | backend
3378 no | yes | yes | yes
3379 Arguments : none
3380
3381 Set a persistent ID for the proxy. This ID must be unique and positive.
3382 An unused ID will automatically be assigned if unset. The first assigned
3383 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003384
3385
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003386ignore-persist { if | unless } <condition>
3387 Declare a condition to ignore persistence
3388 May be used in sections: defaults | frontend | listen | backend
3389 no | yes | yes | yes
3390
3391 By default, when cookie persistence is enabled, every requests containing
3392 the cookie are unconditionally persistent (assuming the target server is up
3393 and running).
3394
3395 The "ignore-persist" statement allows one to declare various ACL-based
3396 conditions which, when met, will cause a request to ignore persistence.
3397 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003398 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003399 persistence for a specific User-Agent (for example, some web crawler bots).
3400
3401 Combined with "appsession", it can also help reduce HAProxy memory usage, as
3402 the appsession table won't grow if persistence is ignored.
3403
3404 The persistence is ignored when an "if" condition is met, or unless an
3405 "unless" condition is met.
3406
3407 See also : "force-persist", "cookie", and section 7 about ACL usage.
3408
3409
Willy Tarreau2769aa02007-12-27 18:26:09 +01003410log global
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003411log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02003412no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01003413 Enable per-instance logging of events and traffic.
3414 May be used in sections : defaults | frontend | listen | backend
3415 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02003416
3417 Prefix :
3418 no should be used when the logger list must be flushed. For example,
3419 if you don't want to inherit from the default logger list. This
3420 prefix does not allow arguments.
3421
Willy Tarreau2769aa02007-12-27 18:26:09 +01003422 Arguments :
3423 global should be used when the instance's logging parameters are the
3424 same as the global ones. This is the most common usage. "global"
3425 replaces <address>, <facility> and <level> with those of the log
3426 entries found in the "global" section. Only one "log global"
3427 statement may be used per instance, and this form takes no other
3428 parameter.
3429
3430 <address> indicates where to send the logs. It takes the same format as
3431 for the "global" section's logs, and can be one of :
3432
3433 - An IPv4 address optionally followed by a colon (':') and a UDP
3434 port. If no port is specified, 514 is used by default (the
3435 standard syslog port).
3436
David du Colombier24bb5f52011-03-17 10:40:23 +01003437 - An IPv6 address followed by a colon (':') and optionally a UDP
3438 port. If no port is specified, 514 is used by default (the
3439 standard syslog port).
3440
Willy Tarreau2769aa02007-12-27 18:26:09 +01003441 - A filesystem path to a UNIX domain socket, keeping in mind
3442 considerations for chroot (be sure the path is accessible
3443 inside the chroot) and uid/gid (be sure the path is
3444 appropriately writeable).
3445
Willy Tarreaudad36a32013-03-11 01:20:04 +01003446 Any part of the address string may reference any number of
3447 environment variables by preceding their name with a dollar
3448 sign ('$') and optionally enclosing them with braces ('{}'),
3449 similarly to what is done in Bourne shell.
3450
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003451 <length> is an optional maximum line length. Log lines larger than this
3452 value will be truncated before being sent. The reason is that
3453 syslog servers act differently on log line length. All servers
3454 support the default value of 1024, but some servers simply drop
3455 larger lines while others do log them. If a server supports long
3456 lines, it may make sense to set this value here in order to avoid
3457 truncating long lines. Similarly, if a server drops long lines,
3458 it is preferable to truncate them before sending them. Accepted
3459 values are 80 to 65535 inclusive. The default value of 1024 is
3460 generally fine for all standard usages. Some specific cases of
3461 long captures or JSON-formated logs may require larger values.
3462
Willy Tarreau2769aa02007-12-27 18:26:09 +01003463 <facility> must be one of the 24 standard syslog facilities :
3464
3465 kern user mail daemon auth syslog lpr news
3466 uucp cron auth2 ftp ntp audit alert cron2
3467 local0 local1 local2 local3 local4 local5 local6 local7
3468
3469 <level> is optional and can be specified to filter outgoing messages. By
3470 default, all messages are sent. If a level is specified, only
3471 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003472 will be sent. An optional minimum level can be specified. If it
3473 is set, logs emitted with a more severe level than this one will
3474 be capped to this level. This is used to avoid sending "emerg"
3475 messages on all terminals on some default syslog configurations.
3476 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003477
3478 emerg alert crit err warning notice info debug
3479
William Lallemand0f99e342011-10-12 17:50:54 +02003480 It is important to keep in mind that it is the frontend which decides what to
3481 log from a connection, and that in case of content switching, the log entries
3482 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003483
3484 However, backend log declaration define how and where servers status changes
3485 will be logged. Level "notice" will be used to indicate a server going up,
3486 "warning" will be used for termination signals and definitive service
3487 termination, and "alert" will be used for when a server goes down.
3488
3489 Note : According to RFC3164, messages are truncated to 1024 bytes before
3490 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003491
3492 Example :
3493 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003494 log 127.0.0.1:514 local0 notice # only send important events
3495 log 127.0.0.1:514 local0 notice notice # same but limit output level
Willy Tarreaudad36a32013-03-11 01:20:04 +01003496 log ${LOCAL_SYSLOG}:514 local0 notice # send to local server
3497
Willy Tarreau2769aa02007-12-27 18:26:09 +01003498
William Lallemand48940402012-01-30 16:47:22 +01003499log-format <string>
Willy Tarreau14104732015-01-07 14:55:17 +01003500 Specifies the log format string to use for traffic logs
3501 May be used in sections: defaults | frontend | listen | backend
3502 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01003503
Willy Tarreau14104732015-01-07 14:55:17 +01003504 This directive specifies the log format string that will be used for all logs
3505 resulting from traffic passing through the frontend using this line. If the
3506 directive is used in a defaults section, all subsequent frontends will use
3507 the same log format. Please see section 8.2.4 which covers the log format
3508 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01003509
Willy Tarreau2769aa02007-12-27 18:26:09 +01003510
Willy Tarreauc35362a2014-04-25 13:58:37 +02003511max-keep-alive-queue <value>
3512 Set the maximum server queue size for maintaining keep-alive connections
3513 May be used in sections: defaults | frontend | listen | backend
3514 yes | no | yes | yes
3515
3516 HTTP keep-alive tries to reuse the same server connection whenever possible,
3517 but sometimes it can be counter-productive, for example if a server has a lot
3518 of connections while other ones are idle. This is especially true for static
3519 servers.
3520
3521 The purpose of this setting is to set a threshold on the number of queued
3522 connections at which haproxy stops trying to reuse the same server and prefers
3523 to find another one. The default value, -1, means there is no limit. A value
3524 of zero means that keep-alive requests will never be queued. For very close
3525 servers which can be reached with a low latency and which are not sensible to
3526 breaking keep-alive, a low value is recommended (eg: local static server can
3527 use a value of 10 or less). For remote servers suffering from a high latency,
3528 higher values might be needed to cover for the latency and/or the cost of
3529 picking a different server.
3530
3531 Note that this has no impact on responses which are maintained to the same
3532 server consecutively to a 401 response. They will still go to the same server
3533 even if they have to be queued.
3534
3535 See also : "option http-server-close", "option prefer-last-server", server
3536 "maxconn" and cookie persistence.
3537
3538
Willy Tarreau2769aa02007-12-27 18:26:09 +01003539maxconn <conns>
3540 Fix the maximum number of concurrent connections on a frontend
3541 May be used in sections : defaults | frontend | listen | backend
3542 yes | yes | yes | no
3543 Arguments :
3544 <conns> is the maximum number of concurrent connections the frontend will
3545 accept to serve. Excess connections will be queued by the system
3546 in the socket's listen queue and will be served once a connection
3547 closes.
3548
3549 If the system supports it, it can be useful on big sites to raise this limit
3550 very high so that haproxy manages connection queues, instead of leaving the
3551 clients with unanswered connection attempts. This value should not exceed the
3552 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann75f8c902016-03-06 23:34:31 +01003553 of tune.bufsize (16kB by default) each, as well as some other data resulting
3554 in about 33 kB of RAM being consumed per established connection. That means
3555 that a medium system equipped with 1GB of RAM can withstand around
3556 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003557
3558 Also, when <conns> is set to large values, it is possible that the servers
3559 are not sized to accept such loads, and for this reason it is generally wise
3560 to assign them some reasonable connection limits.
3561
Vincent Bernat6341be52012-06-27 17:18:30 +02003562 By default, this value is set to 2000.
3563
Willy Tarreau2769aa02007-12-27 18:26:09 +01003564 See also : "server", global section's "maxconn", "fullconn"
3565
3566
3567mode { tcp|http|health }
3568 Set the running mode or protocol of the instance
3569 May be used in sections : defaults | frontend | listen | backend
3570 yes | yes | yes | yes
3571 Arguments :
3572 tcp The instance will work in pure TCP mode. A full-duplex connection
3573 will be established between clients and servers, and no layer 7
3574 examination will be performed. This is the default mode. It
3575 should be used for SSL, SSH, SMTP, ...
3576
3577 http The instance will work in HTTP mode. The client request will be
3578 analyzed in depth before connecting to any server. Any request
3579 which is not RFC-compliant will be rejected. Layer 7 filtering,
3580 processing and switching will be possible. This is the mode which
3581 brings HAProxy most of its value.
3582
3583 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02003584 to incoming connections and close the connection. Alternatively,
3585 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
3586 instead. Nothing will be logged in either case. This mode is used
3587 to reply to external components health checks. This mode is
3588 deprecated and should not be used anymore as it is possible to do
3589 the same and even better by combining TCP or HTTP modes with the
3590 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003591
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003592 When doing content switching, it is mandatory that the frontend and the
3593 backend are in the same mode (generally HTTP), otherwise the configuration
3594 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003595
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003596 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003597 defaults http_instances
3598 mode http
3599
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003600 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003601
Willy Tarreau0ba27502007-12-24 16:55:16 +01003602
Cyril Bontéf0c60612010-02-06 14:44:47 +01003603monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01003604 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003605 May be used in sections : defaults | frontend | listen | backend
3606 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01003607 Arguments :
3608 if <cond> the monitor request will fail if the condition is satisfied,
3609 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003610 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01003611 are met, for instance a low number of servers both in a
3612 backend and its backup.
3613
3614 unless <cond> the monitor request will succeed only if the condition is
3615 satisfied, and will fail otherwise. Such a condition may be
3616 based on a test on the presence of a minimum number of active
3617 servers in a list of backends.
3618
3619 This statement adds a condition which can force the response to a monitor
3620 request to report a failure. By default, when an external component queries
3621 the URI dedicated to monitoring, a 200 response is returned. When one of the
3622 conditions above is met, haproxy will return 503 instead of 200. This is
3623 very useful to report a site failure to an external component which may base
3624 routing advertisements between multiple sites on the availability reported by
3625 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003626 criterion. Note that "monitor fail" only works in HTTP mode. Both status
3627 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003628
3629 Example:
3630 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01003631 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01003632 acl site_dead nbsrv(dynamic) lt 2
3633 acl site_dead nbsrv(static) lt 2
3634 monitor-uri /site_alive
3635 monitor fail if site_dead
3636
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003637 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003638
3639
3640monitor-net <source>
3641 Declare a source network which is limited to monitor requests
3642 May be used in sections : defaults | frontend | listen | backend
3643 yes | yes | yes | no
3644 Arguments :
3645 <source> is the source IPv4 address or network which will only be able to
3646 get monitor responses to any request. It can be either an IPv4
3647 address, a host name, or an address followed by a slash ('/')
3648 followed by a mask.
3649
3650 In TCP mode, any connection coming from a source matching <source> will cause
3651 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003652 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01003653 forwarding the connection to a remote server.
3654
3655 In HTTP mode, a connection coming from a source matching <source> will be
3656 accepted, the following response will be sent without waiting for a request,
3657 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
3658 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02003659 running without forwarding the request to a backend server. Note that this
3660 response is sent in raw format, without any transformation. This is important
3661 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003662
Willy Tarreau82569f92012-09-27 23:48:56 +02003663 Monitor requests are processed very early, just after tcp-request connection
3664 ACLs which are the only ones able to block them. These connections are short
3665 lived and never wait for any data from the client. They cannot be logged, and
3666 it is the intended purpose. They are only used to report HAProxy's health to
3667 an upper component, nothing more. Please note that "monitor fail" rules do
3668 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01003669
Willy Tarreau95cd2832010-03-04 23:36:33 +01003670 Last, please note that only one "monitor-net" statement can be specified in
3671 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003672
Willy Tarreau2769aa02007-12-27 18:26:09 +01003673 Example :
3674 # addresses .252 and .253 are just probing us.
3675 frontend www
3676 monitor-net 192.168.0.252/31
3677
3678 See also : "monitor fail", "monitor-uri"
3679
3680
3681monitor-uri <uri>
3682 Intercept a URI used by external components' monitor requests
3683 May be used in sections : defaults | frontend | listen | backend
3684 yes | yes | yes | no
3685 Arguments :
3686 <uri> is the exact URI which we want to intercept to return HAProxy's
3687 health status instead of forwarding the request.
3688
3689 When an HTTP request referencing <uri> will be received on a frontend,
3690 HAProxy will not forward it nor log it, but instead will return either
3691 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
3692 conditions defined with "monitor fail". This is normally enough for any
3693 front-end HTTP probe to detect that the service is UP and running without
3694 forwarding the request to a backend server. Note that the HTTP method, the
3695 version and all headers are ignored, but the request must at least be valid
3696 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
3697
3698 Monitor requests are processed very early. It is not possible to block nor
3699 divert them using ACLs. They cannot be logged either, and it is the intended
3700 purpose. They are only used to report HAProxy's health to an upper component,
3701 nothing more. However, it is possible to add any number of conditions using
3702 "monitor fail" and ACLs so that the result can be adjusted to whatever check
3703 can be imagined (most often the number of available servers in a backend).
3704
3705 Example :
3706 # Use /haproxy_test to report haproxy's status
3707 frontend www
3708 mode http
3709 monitor-uri /haproxy_test
3710
3711 See also : "monitor fail", "monitor-net"
3712
Willy Tarreau0ba27502007-12-24 16:55:16 +01003713
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003714option abortonclose
3715no option abortonclose
3716 Enable or disable early dropping of aborted requests pending in queues.
3717 May be used in sections : defaults | frontend | listen | backend
3718 yes | no | yes | yes
3719 Arguments : none
3720
3721 In presence of very high loads, the servers will take some time to respond.
3722 The per-instance connection queue will inflate, and the response time will
3723 increase respective to the size of the queue times the average per-session
3724 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01003725 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003726 the queue, and slowing down other users, and the servers as well, because the
3727 request will eventually be served, then aborted at the first error
3728 encountered while delivering the response.
3729
3730 As there is no way to distinguish between a full STOP and a simple output
3731 close on the client side, HTTP agents should be conservative and consider
3732 that the client might only have closed its output channel while waiting for
3733 the response. However, this introduces risks of congestion when lots of users
3734 do the same, and is completely useless nowadays because probably no client at
3735 all will close the session while waiting for the response. Some HTTP agents
Willy Tarreaud72758d2010-01-12 10:42:19 +01003736 support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003737 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01003738 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003739 of being the single component to break rare but valid traffic is extremely
3740 low, which adds to the temptation to be able to abort a session early while
3741 still not served and not pollute the servers.
3742
3743 In HAProxy, the user can choose the desired behaviour using the option
3744 "abortonclose". By default (without the option) the behaviour is HTTP
3745 compliant and aborted requests will be served. But when the option is
3746 specified, a session with an incoming channel closed will be aborted while
3747 it is still possible, either pending in the queue for a connection slot, or
3748 during the connection establishment if the server has not yet acknowledged
3749 the connection request. This considerably reduces the queue size and the load
3750 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01003751 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003752
3753 If this option has been enabled in a "defaults" section, it can be disabled
3754 in a specific instance by prepending the "no" keyword before it.
3755
3756 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
3757
3758
Willy Tarreau4076a152009-04-02 15:18:36 +02003759option accept-invalid-http-request
3760no option accept-invalid-http-request
3761 Enable or disable relaxing of HTTP request parsing
3762 May be used in sections : defaults | frontend | listen | backend
3763 yes | yes | yes | no
3764 Arguments : none
3765
Willy Tarreau55645552015-05-01 13:26:00 +02003766 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02003767 means that invalid characters in header names are not permitted and cause an
3768 error to be returned to the client. This is the desired behaviour as such
3769 forbidden characters are essentially used to build attacks exploiting server
3770 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3771 server will emit invalid header names for whatever reason (configuration,
3772 implementation) and the issue will not be immediately fixed. In such a case,
3773 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01003774 even if that does not make sense, by specifying this option. Similarly, the
3775 list of characters allowed to appear in a URI is well defined by RFC3986, and
3776 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
3777 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
3778 not allowed at all. Haproxy always blocks a number of them (0..32, 127). The
Willy Tarreau55645552015-05-01 13:26:00 +02003779 remaining ones are blocked by default unless this option is enabled. This
3780 option also relaxes the test on the HTTP version format, it allows multiple
3781 digits for both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02003782
3783 This option should never be enabled by default as it hides application bugs
3784 and open security breaches. It should only be deployed after a problem has
3785 been confirmed.
3786
3787 When this option is enabled, erroneous header names will still be accepted in
3788 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01003789 analysis using the "show errors" request on the UNIX stats socket. Similarly,
3790 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02003791 also helps confirming that the issue has been solved.
3792
3793 If this option has been enabled in a "defaults" section, it can be disabled
3794 in a specific instance by prepending the "no" keyword before it.
3795
3796 See also : "option accept-invalid-http-response" and "show errors" on the
3797 stats socket.
3798
3799
3800option accept-invalid-http-response
3801no option accept-invalid-http-response
3802 Enable or disable relaxing of HTTP response parsing
3803 May be used in sections : defaults | frontend | listen | backend
3804 yes | no | yes | yes
3805 Arguments : none
3806
Willy Tarreau55645552015-05-01 13:26:00 +02003807 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02003808 means that invalid characters in header names are not permitted and cause an
3809 error to be returned to the client. This is the desired behaviour as such
3810 forbidden characters are essentially used to build attacks exploiting server
3811 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3812 server will emit invalid header names for whatever reason (configuration,
3813 implementation) and the issue will not be immediately fixed. In such a case,
3814 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau55645552015-05-01 13:26:00 +02003815 even if that does not make sense, by specifying this option. This option also
3816 relaxes the test on the HTTP version format, it allows multiple digits for
3817 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02003818
3819 This option should never be enabled by default as it hides application bugs
3820 and open security breaches. It should only be deployed after a problem has
3821 been confirmed.
3822
3823 When this option is enabled, erroneous header names will still be accepted in
3824 responses, but the complete response will be captured in order to permit
3825 later analysis using the "show errors" request on the UNIX stats socket.
3826 Doing this also helps confirming that the issue has been solved.
3827
3828 If this option has been enabled in a "defaults" section, it can be disabled
3829 in a specific instance by prepending the "no" keyword before it.
3830
3831 See also : "option accept-invalid-http-request" and "show errors" on the
3832 stats socket.
3833
3834
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003835option allbackups
3836no option allbackups
3837 Use either all backup servers at a time or only the first one
3838 May be used in sections : defaults | frontend | listen | backend
3839 yes | no | yes | yes
3840 Arguments : none
3841
3842 By default, the first operational backup server gets all traffic when normal
3843 servers are all down. Sometimes, it may be preferred to use multiple backups
3844 at once, because one will not be enough. When "option allbackups" is enabled,
3845 the load balancing will be performed among all backup servers when all normal
3846 ones are unavailable. The same load balancing algorithm will be used and the
3847 servers' weights will be respected. Thus, there will not be any priority
3848 order between the backup servers anymore.
3849
3850 This option is mostly used with static server farms dedicated to return a
3851 "sorry" page when an application is completely offline.
3852
3853 If this option has been enabled in a "defaults" section, it can be disabled
3854 in a specific instance by prepending the "no" keyword before it.
3855
3856
3857option checkcache
3858no option checkcache
Godbach7056a352013-12-11 20:01:07 +08003859 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003860 May be used in sections : defaults | frontend | listen | backend
3861 yes | no | yes | yes
3862 Arguments : none
3863
3864 Some high-level frameworks set application cookies everywhere and do not
3865 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003866 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003867 high risk of session crossing or stealing between users traversing the same
3868 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003869 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003870
3871 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003872 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01003873 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003874 response to check if there's a risk of caching a cookie on a client-side
3875 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01003876 to the client are :
3877 - all those without "Set-Cookie" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003878 - all those with a return code other than 200, 203, 206, 300, 301, 410,
Willy Tarreau198a7442008-01-17 12:05:32 +01003879 provided that the server has not set a "Cache-control: public" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003880 - all those that come from a POST request, provided that the server has not
3881 set a 'Cache-Control: public' header ;
3882 - those with a 'Pragma: no-cache' header
3883 - those with a 'Cache-control: private' header
3884 - those with a 'Cache-control: no-store' header
3885 - those with a 'Cache-control: max-age=0' header
3886 - those with a 'Cache-control: s-maxage=0' header
3887 - those with a 'Cache-control: no-cache' header
3888 - those with a 'Cache-control: no-cache="set-cookie"' header
3889 - those with a 'Cache-control: no-cache="set-cookie,' header
3890 (allowing other fields after set-cookie)
3891
3892 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01003893 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003894 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003895 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003896 that admins are informed that there's something to be fixed.
3897
3898 Due to the high impact on the application, the application should be tested
3899 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003900 good practice to always activate it during tests, even if it is not used in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003901 production, as it will report potentially dangerous application behaviours.
3902
3903 If this option has been enabled in a "defaults" section, it can be disabled
3904 in a specific instance by prepending the "no" keyword before it.
3905
3906
3907option clitcpka
3908no option clitcpka
3909 Enable or disable the sending of TCP keepalive packets on the client side
3910 May be used in sections : defaults | frontend | listen | backend
3911 yes | yes | yes | no
3912 Arguments : none
3913
3914 When there is a firewall or any session-aware component between a client and
3915 a server, and when the protocol involves very long sessions with long idle
3916 periods (eg: remote desktops), there is a risk that one of the intermediate
3917 components decides to expire a session which has remained idle for too long.
3918
3919 Enabling socket-level TCP keep-alives makes the system regularly send packets
3920 to the other end of the connection, leaving it active. The delay between
3921 keep-alive probes is controlled by the system only and depends both on the
3922 operating system and its tuning parameters.
3923
3924 It is important to understand that keep-alive packets are neither emitted nor
3925 received at the application level. It is only the network stacks which sees
3926 them. For this reason, even if one side of the proxy already uses keep-alives
3927 to maintain its connection alive, those keep-alive packets will not be
3928 forwarded to the other side of the proxy.
3929
3930 Please note that this has nothing to do with HTTP keep-alive.
3931
3932 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
3933 client side of a connection, which should help when session expirations are
3934 noticed between HAProxy and a client.
3935
3936 If this option has been enabled in a "defaults" section, it can be disabled
3937 in a specific instance by prepending the "no" keyword before it.
3938
3939 See also : "option srvtcpka", "option tcpka"
3940
3941
Willy Tarreau0ba27502007-12-24 16:55:16 +01003942option contstats
3943 Enable continuous traffic statistics updates
3944 May be used in sections : defaults | frontend | listen | backend
3945 yes | yes | yes | no
3946 Arguments : none
3947
3948 By default, counters used for statistics calculation are incremented
3949 only when a session finishes. It works quite well when serving small
3950 objects, but with big ones (for example large images or archives) or
3951 with A/V streaming, a graph generated from haproxy counters looks like
3952 a hedgehog. With this option enabled counters get incremented continuously,
3953 during a whole session. Recounting touches a hotpath directly so
3954 it is not enabled by default, as it has small performance impact (~0.5%).
3955
3956
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003957option dontlog-normal
3958no option dontlog-normal
3959 Enable or disable logging of normal, successful connections
3960 May be used in sections : defaults | frontend | listen | backend
3961 yes | yes | yes | no
3962 Arguments : none
3963
3964 There are large sites dealing with several thousand connections per second
3965 and for which logging is a major pain. Some of them are even forced to turn
3966 logs off and cannot debug production issues. Setting this option ensures that
3967 normal connections, those which experience no error, no timeout, no retry nor
3968 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
3969 mode, the response status code is checked and return codes 5xx will still be
3970 logged.
3971
3972 It is strongly discouraged to use this option as most of the time, the key to
3973 complex issues is in the normal logs which will not be logged here. If you
3974 need to separate logs, see the "log-separate-errors" option instead.
3975
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003976 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003977 logging.
3978
3979
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003980option dontlognull
3981no option dontlognull
3982 Enable or disable logging of null connections
3983 May be used in sections : defaults | frontend | listen | backend
3984 yes | yes | yes | no
3985 Arguments : none
3986
3987 In certain environments, there are components which will regularly connect to
3988 various systems to ensure that they are still alive. It can be the case from
3989 another load balancer as well as from monitoring systems. By default, even a
3990 simple port probe or scan will produce a log. If those connections pollute
3991 the logs too much, it is possible to enable option "dontlognull" to indicate
3992 that a connection on which no data has been transferred will not be logged,
Willy Tarreau1c3a6122015-05-01 15:37:53 +02003993 which typically corresponds to those probes. Note that errors will still be
3994 returned to the client and accounted for in the stats. If this is not what is
3995 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003996
3997 It is generally recommended not to use this option in uncontrolled
3998 environments (eg: internet), otherwise scans and other malicious activities
3999 would not be logged.
4000
4001 If this option has been enabled in a "defaults" section, it can be disabled
4002 in a specific instance by prepending the "no" keyword before it.
4003
Willy Tarreau1c3a6122015-05-01 15:37:53 +02004004 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
4005 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004006
4007
4008option forceclose
4009no option forceclose
4010 Enable or disable active connection closing after response is transferred.
4011 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01004012 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004013 Arguments : none
4014
4015 Some HTTP servers do not necessarily close the connections when they receive
4016 the "Connection: close" set by "option httpclose", and if the client does not
4017 close either, then the connection remains open till the timeout expires. This
4018 causes high number of simultaneous connections on the servers and shows high
4019 global session times in the logs.
4020
4021 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01004022 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01004023 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004024
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004025 This option may also be combined with "option http-pretend-keepalive", which
4026 will disable sending of the "Connection: close" header, but will still cause
4027 the connection to be closed once the whole response is received.
4028
Cyril Bonté653dcd62014-02-20 00:13:15 +01004029 This option disables and replaces any previous "option httpclose", "option
4030 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004031
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004032 If this option has been enabled in a "defaults" section, it can be disabled
4033 in a specific instance by prepending the "no" keyword before it.
4034
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004035 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004036
4037
Willy Tarreau87cf5142011-08-19 22:57:24 +02004038option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004039 Enable insertion of the X-Forwarded-For header to requests sent to servers
4040 May be used in sections : defaults | frontend | listen | backend
4041 yes | yes | yes | yes
4042 Arguments :
4043 <network> is an optional argument used to disable this option for sources
4044 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02004045 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01004046 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004047
4048 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
4049 their client address. This is sometimes annoying when the client's IP address
4050 is expected in server logs. To solve this problem, the well-known HTTP header
4051 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
4052 This header contains a value representing the client's IP address. Since this
4053 header is always appended at the end of the existing header list, the server
4054 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02004055 the server's manual to find how to enable use of this standard header. Note
4056 that only the last occurrence of the header must be used, since it is really
4057 possible that the client has already brought one.
4058
Willy Tarreaud72758d2010-01-12 10:42:19 +01004059 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02004060 the default "X-Forwarded-For". This can be useful where you might already
Willy Tarreaud72758d2010-01-12 10:42:19 +01004061 have a "X-Forwarded-For" header from a different application (eg: stunnel),
4062 and you need preserve it. Also if your backend server doesn't use the
Ross Westaf72a1d2008-08-03 10:51:45 +02004063 "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers
4064 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01004065
4066 Sometimes, a same HAProxy instance may be shared between a direct client
4067 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4068 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4069 header for a known source address or network by adding the "except" keyword
4070 followed by the network address. In this case, any source IP matching the
4071 network will not cause an addition of this header. Most common uses are with
4072 private networks or 127.0.0.1.
4073
Willy Tarreau87cf5142011-08-19 22:57:24 +02004074 Alternatively, the keyword "if-none" states that the header will only be
4075 added if it is not present. This should only be used in perfectly trusted
4076 environment, as this might cause a security issue if headers reaching haproxy
4077 are under the control of the end-user.
4078
Willy Tarreauc27debf2008-01-06 08:57:02 +01004079 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02004080 least one of them uses it, the header will be added. Note that the backend's
4081 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02004082 both are defined. In the case of the "if-none" argument, if at least one of
4083 the frontend or the backend does not specify it, it wants the addition to be
4084 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004085
Ross Westaf72a1d2008-08-03 10:51:45 +02004086 Examples :
Willy Tarreauc27debf2008-01-06 08:57:02 +01004087 # Public HTTP address also used by stunnel on the same machine
4088 frontend www
4089 mode http
4090 option forwardfor except 127.0.0.1 # stunnel already adds the header
4091
Ross Westaf72a1d2008-08-03 10:51:45 +02004092 # Those servers want the IP Address in X-Client
4093 backend www
4094 mode http
4095 option forwardfor header X-Client
4096
Willy Tarreau87cf5142011-08-19 22:57:24 +02004097 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01004098 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01004099
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004100
Willy Tarreau1c3a6122015-05-01 15:37:53 +02004101option http-ignore-probes
4102no option http-ignore-probes
4103 Enable or disable logging of null connections and request timeouts
4104 May be used in sections : defaults | frontend | listen | backend
4105 yes | yes | yes | no
4106 Arguments : none
4107
4108 Recently some browsers started to implement a "pre-connect" feature
4109 consisting in speculatively connecting to some recently visited web sites
4110 just in case the user would like to visit them. This results in many
4111 connections being established to web sites, which end up in 408 Request
4112 Timeout if the timeout strikes first, or 400 Bad Request when the browser
4113 decides to close them first. These ones pollute the log and feed the error
4114 counters. There was already "option dontlognull" but it's insufficient in
4115 this case. Instead, this option does the following things :
4116 - prevent any 400/408 message from being sent to the client if nothing
4117 was received over a connection before it was closed ;
4118 - prevent any log from being emitted in this situation ;
4119 - prevent any error counter from being incremented
4120
4121 That way the empty connection is silently ignored. Note that it is better
4122 not to use this unless it is clear that it is needed, because it will hide
4123 real problems. The most common reason for not receiving a request and seeing
4124 a 408 is due to an MTU inconsistency between the client and an intermediary
4125 element such as a VPN, which blocks too large packets. These issues are
4126 generally seen with POST requests as well as GET with large cookies. The logs
4127 are often the only way to detect them.
4128
4129 If this option has been enabled in a "defaults" section, it can be disabled
4130 in a specific instance by prepending the "no" keyword before it.
4131
4132 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
4133
4134
Willy Tarreau16bfb022010-01-16 19:48:41 +01004135option http-keep-alive
4136no option http-keep-alive
4137 Enable or disable HTTP keep-alive from client to server
4138 May be used in sections : defaults | frontend | listen | backend
4139 yes | yes | yes | yes
4140 Arguments : none
4141
Willy Tarreau70dffda2014-01-30 03:07:23 +01004142 By default HAProxy operates in keep-alive mode with regards to persistent
4143 connections: for each connection it processes each request and response, and
4144 leaves the connection idle on both sides between the end of a response and the
4145 start of a new request. This mode may be changed by several options such as
4146 "option http-server-close", "option forceclose", "option httpclose" or
4147 "option http-tunnel". This option allows to set back the keep-alive mode,
4148 which can be useful when another mode was used in a defaults section.
4149
4150 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
4151 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01004152 network) and the fastest session reuse on the server side at the expense
4153 of maintaining idle connections to the servers. In general, it is possible
4154 with this option to achieve approximately twice the request rate that the
4155 "http-server-close" option achieves on small objects. There are mainly two
4156 situations where this option may be useful :
4157
4158 - when the server is non-HTTP compliant and authenticates the connection
4159 instead of requests (eg: NTLM authentication)
4160
4161 - when the cost of establishing the connection to the server is significant
4162 compared to the cost of retrieving the associated object from the server.
4163
4164 This last case can happen when the server is a fast static server of cache.
4165 In this case, the server will need to be properly tuned to support high enough
4166 connection counts because connections will last until the client sends another
4167 request.
4168
4169 If the client request has to go to another backend or another server due to
4170 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01004171 immediately be closed and a new one re-opened. Option "prefer-last-server" is
4172 available to try optimize server selection so that if the server currently
4173 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01004174
4175 In general it is preferred to use "option http-server-close" with application
4176 servers, and some static servers might benefit from "option http-keep-alive".
4177
4178 At the moment, logs will not indicate whether requests came from the same
4179 session or not. The accept date reported in the logs corresponds to the end
4180 of the previous request, and the request time corresponds to the time spent
4181 waiting for a new request. The keep-alive request time is still bound to the
4182 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4183 not set.
4184
Cyril Bonté653dcd62014-02-20 00:13:15 +01004185 This option disables and replaces any previous "option httpclose", "option
4186 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01004187 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01004188 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004189
4190 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01004191 "option prefer-last-server", "option http-pretend-keepalive",
4192 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004193
4194
Willy Tarreau96e31212011-05-30 18:10:30 +02004195option http-no-delay
4196no option http-no-delay
4197 Instruct the system to favor low interactive delays over performance in HTTP
4198 May be used in sections : defaults | frontend | listen | backend
4199 yes | yes | yes | yes
4200 Arguments : none
4201
4202 In HTTP, each payload is unidirectional and has no notion of interactivity.
4203 Any agent is expected to queue data somewhat for a reasonably low delay.
4204 There are some very rare server-to-server applications that abuse the HTTP
4205 protocol and expect the payload phase to be highly interactive, with many
4206 interleaved data chunks in both directions within a single request. This is
4207 absolutely not supported by the HTTP specification and will not work across
4208 most proxies or servers. When such applications attempt to do this through
4209 haproxy, it works but they will experience high delays due to the network
4210 optimizations which favor performance by instructing the system to wait for
4211 enough data to be available in order to only send full packets. Typical
4212 delays are around 200 ms per round trip. Note that this only happens with
4213 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
4214 affected.
4215
4216 When "option http-no-delay" is present in either the frontend or the backend
4217 used by a connection, all such optimizations will be disabled in order to
4218 make the exchanges as fast as possible. Of course this offers no guarantee on
4219 the functionality, as it may break at any other place. But if it works via
4220 HAProxy, it will work as fast as possible. This option should never be used
4221 by default, and should never be used at all unless such a buggy application
4222 is discovered. The impact of using this option is an increase of bandwidth
4223 usage and CPU usage, which may significantly lower performance in high
4224 latency environments.
4225
4226
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004227option http-pretend-keepalive
4228no option http-pretend-keepalive
4229 Define whether haproxy will announce keepalive to the server or not
4230 May be used in sections : defaults | frontend | listen | backend
4231 yes | yes | yes | yes
4232 Arguments : none
4233
4234 When running with "option http-server-close" or "option forceclose", haproxy
4235 adds a "Connection: close" header to the request forwarded to the server.
4236 Unfortunately, when some servers see this header, they automatically refrain
4237 from using the chunked encoding for responses of unknown length, while this
4238 is totally unrelated. The immediate effect is that this prevents haproxy from
4239 maintaining the client connection alive. A second effect is that a client or
4240 a cache could receive an incomplete response without being aware of it, and
4241 consider the response complete.
4242
4243 By setting "option http-pretend-keepalive", haproxy will make the server
4244 believe it will keep the connection alive. The server will then not fall back
4245 to the abnormal undesired above. When haproxy gets the whole response, it
4246 will close the connection with the server just as it would do with the
4247 "forceclose" option. That way the client gets a normal response and the
4248 connection is correctly closed on the server side.
4249
4250 It is recommended not to enable this option by default, because most servers
4251 will more efficiently close the connection themselves after the last packet,
4252 and release its buffers slightly earlier. Also, the added packet on the
4253 network could slightly reduce the overall peak performance. However it is
4254 worth noting that when this option is enabled, haproxy will have slightly
4255 less work to do. So if haproxy is the bottleneck on the whole architecture,
4256 enabling this option might save a few CPU cycles.
4257
4258 This option may be set both in a frontend and in a backend. It is enabled if
4259 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004260 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02004261 keepalive to be announced to the server and close to be announced to the
4262 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004263
4264 If this option has been enabled in a "defaults" section, it can be disabled
4265 in a specific instance by prepending the "no" keyword before it.
4266
Willy Tarreau16bfb022010-01-16 19:48:41 +01004267 See also : "option forceclose", "option http-server-close", and
4268 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004269
Willy Tarreauc27debf2008-01-06 08:57:02 +01004270
Willy Tarreaub608feb2010-01-02 22:47:18 +01004271option http-server-close
4272no option http-server-close
4273 Enable or disable HTTP connection closing on the server side
4274 May be used in sections : defaults | frontend | listen | backend
4275 yes | yes | yes | yes
4276 Arguments : none
4277
Willy Tarreau70dffda2014-01-30 03:07:23 +01004278 By default HAProxy operates in keep-alive mode with regards to persistent
4279 connections: for each connection it processes each request and response, and
4280 leaves the connection idle on both sides between the end of a response and
4281 the start of a new request. This mode may be changed by several options such
4282 as "option http-server-close", "option forceclose", "option httpclose" or
4283 "option http-tunnel". Setting "option http-server-close" enables HTTP
4284 connection-close mode on the server side while keeping the ability to support
4285 HTTP keep-alive and pipelining on the client side. This provides the lowest
4286 latency on the client side (slow network) and the fastest session reuse on
4287 the server side to save server resources, similarly to "option forceclose".
4288 It also permits non-keepalive capable servers to be served in keep-alive mode
4289 to the clients if they conform to the requirements of RFC2616. Please note
4290 that some servers do not always conform to those requirements when they see
4291 "Connection: close" in the request. The effect will be that keep-alive will
4292 never be used. A workaround consists in enabling "option
4293 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004294
4295 At the moment, logs will not indicate whether requests came from the same
4296 session or not. The accept date reported in the logs corresponds to the end
4297 of the previous request, and the request time corresponds to the time spent
4298 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01004299 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4300 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004301
4302 This option may be set both in a frontend and in a backend. It is enabled if
4303 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004304 It disables and replaces any previous "option httpclose", "option forceclose",
4305 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01004306 ("Proxies") to see how this option combines with others when frontend and
4307 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004308
4309 If this option has been enabled in a "defaults" section, it can be disabled
4310 in a specific instance by prepending the "no" keyword before it.
4311
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004312 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01004313 "option httpclose", "option http-keep-alive", and
4314 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004315
4316
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004317option http-tunnel
4318no option http-tunnel
4319 Disable or enable HTTP connection processing after first transaction
4320 May be used in sections : defaults | frontend | listen | backend
4321 yes | yes | yes | yes
4322 Arguments : none
4323
Willy Tarreau70dffda2014-01-30 03:07:23 +01004324 By default HAProxy operates in keep-alive mode with regards to persistent
4325 connections: for each connection it processes each request and response, and
4326 leaves the connection idle on both sides between the end of a response and
4327 the start of a new request. This mode may be changed by several options such
4328 as "option http-server-close", "option forceclose", "option httpclose" or
4329 "option http-tunnel".
4330
4331 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004332 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01004333 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
4334 is normally not needed anymore unless in very specific cases such as when
4335 using an in-house protocol that looks like HTTP but is not compatible, or
4336 just to log one request per client in order to reduce log size. Note that
4337 everything which works at the HTTP level, including header parsing/addition,
4338 cookie processing or content switching will only work for the first request
4339 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004340
4341 If this option has been enabled in a "defaults" section, it can be disabled
4342 in a specific instance by prepending the "no" keyword before it.
4343
4344 See also : "option forceclose", "option http-server-close",
4345 "option httpclose", "option http-keep-alive", and
4346 "1.1. The HTTP transaction model".
4347
4348
Willy Tarreau88d349d2010-01-25 12:15:43 +01004349option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01004350no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01004351 Make use of non-standard Proxy-Connection header instead of Connection
4352 May be used in sections : defaults | frontend | listen | backend
4353 yes | yes | yes | no
4354 Arguments : none
4355
4356 While RFC2616 explicitly states that HTTP/1.1 agents must use the
4357 Connection header to indicate their wish of persistent or non-persistent
4358 connections, both browsers and proxies ignore this header for proxied
4359 connections and make use of the undocumented, non-standard Proxy-Connection
4360 header instead. The issue begins when trying to put a load balancer between
4361 browsers and such proxies, because there will be a difference between what
4362 haproxy understands and what the client and the proxy agree on.
4363
4364 By setting this option in a frontend, haproxy can automatically switch to use
4365 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribus3082bde2016-02-03 18:09:37 +01004366 defined here as one where the URI begins with neither a '/' nor a '*'. This
4367 is incompatible with the HTTP tunnel mode. Note that this option can only be
4368 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01004369
Willy Tarreau844a7e72010-01-31 21:46:18 +01004370 Also, when this option is set, a request which requires authentication will
4371 automatically switch to use proxy authentication headers if it is itself a
4372 proxied request. That makes it possible to check or enforce authentication in
4373 front of an existing proxy.
4374
Willy Tarreau88d349d2010-01-25 12:15:43 +01004375 This option should normally never be used, except in front of a proxy.
4376
4377 See also : "option httpclose", "option forceclose" and "option
4378 http-server-close".
4379
4380
Willy Tarreaud63335a2010-02-26 12:56:52 +01004381option httpchk
4382option httpchk <uri>
4383option httpchk <method> <uri>
4384option httpchk <method> <uri> <version>
4385 Enable HTTP protocol to check on the servers health
4386 May be used in sections : defaults | frontend | listen | backend
4387 yes | no | yes | yes
4388 Arguments :
4389 <method> is the optional HTTP method used with the requests. When not set,
4390 the "OPTIONS" method is used, as it generally requires low server
4391 processing and is easy to filter out from the logs. Any method
4392 may be used, though it is not recommended to invent non-standard
4393 ones.
4394
4395 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
4396 which is accessible by default on almost any server, but may be
4397 changed to any other URI. Query strings are permitted.
4398
4399 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
4400 but some servers might behave incorrectly in HTTP 1.0, so turning
4401 it to HTTP/1.1 may sometimes help. Note that the Host field is
4402 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
4403 after "\r\n" following the version string.
4404
4405 By default, server health checks only consist in trying to establish a TCP
4406 connection. When "option httpchk" is specified, a complete HTTP request is
4407 sent once the TCP connection is established, and responses 2xx and 3xx are
4408 considered valid, while all other ones indicate a server failure, including
4409 the lack of any response.
4410
4411 The port and interval are specified in the server configuration.
4412
4413 This option does not necessarily require an HTTP backend, it also works with
4414 plain TCP backends. This is particularly useful to check simple scripts bound
4415 to some dedicated ports using the inetd daemon.
4416
4417 Examples :
4418 # Relay HTTPS traffic to Apache instance and check service availability
4419 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
4420 backend https_relay
4421 mode tcp
4422 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
4423 server apache1 192.168.1.1:443 check port 80
4424
Simon Hormanafc47ee2013-11-25 10:46:35 +09004425 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
4426 "option pgsql-check", "http-check" and the "check", "port" and
4427 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01004428
4429
Willy Tarreauc27debf2008-01-06 08:57:02 +01004430option httpclose
4431no option httpclose
4432 Enable or disable passive HTTP connection closing
4433 May be used in sections : defaults | frontend | listen | backend
4434 yes | yes | yes | yes
4435 Arguments : none
4436
Willy Tarreau70dffda2014-01-30 03:07:23 +01004437 By default HAProxy operates in keep-alive mode with regards to persistent
4438 connections: for each connection it processes each request and response, and
4439 leaves the connection idle on both sides between the end of a response and
4440 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01004441 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01004442 "option http-tunnel".
4443
4444 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
4445 if a "Connection: close" header is already set in each direction, and will
4446 add one if missing. Each end should react to this by actively closing the TCP
4447 connection after each transfer, thus resulting in a switch to the HTTP close
4448 mode. Any "Connection" header different from "close" will also be removed.
4449 Note that this option is deprecated since what it does is very cheap but not
4450 reliable. Using "option http-server-close" or "option forceclose" is strongly
4451 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004452
4453 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004454 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01004455 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
4456 it is possible to use the "option forceclose" which actively closes the
4457 request connection once the server responds. Option "forceclose" also
4458 releases the server connection earlier because it does not have to wait for
4459 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004460
4461 This option may be set both in a frontend and in a backend. It is enabled if
4462 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004463 It disables and replaces any previous "option http-server-close",
4464 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01004465 check section 4 ("Proxies") to see how this option combines with others when
4466 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004467
4468 If this option has been enabled in a "defaults" section, it can be disabled
4469 in a specific instance by prepending the "no" keyword before it.
4470
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004471 See also : "option forceclose", "option http-server-close" and
4472 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01004473
4474
Emeric Brun3a058f32009-06-30 18:26:00 +02004475option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004476 Enable logging of HTTP request, session state and timers
4477 May be used in sections : defaults | frontend | listen | backend
4478 yes | yes | yes | yes
Emeric Brun3a058f32009-06-30 18:26:00 +02004479 Arguments :
4480 clf if the "clf" argument is added, then the output format will be
4481 the CLF format instead of HAProxy's default HTTP format. You can
4482 use this when you need to feed HAProxy's logs through a specific
4483 log analyser which only support the CLF format and which is not
4484 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004485
4486 By default, the log output format is very poor, as it only contains the
4487 source and destination addresses, and the instance name. By specifying
4488 "option httplog", each log line turns into a much richer format including,
4489 but not limited to, the HTTP request, the connection timers, the session
4490 status, the connections numbers, the captured headers and cookies, the
4491 frontend, backend and server name, and of course the source address and
4492 ports.
4493
4494 This option may be set either in the frontend or the backend.
4495
PiBa-NL211c2e92014-12-11 21:31:54 +01004496 Specifying only "option httplog" will automatically clear the 'clf' mode
4497 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02004498
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004499 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004500
Willy Tarreau55165fe2009-05-10 12:02:55 +02004501
4502option http_proxy
4503no option http_proxy
4504 Enable or disable plain HTTP proxy mode
4505 May be used in sections : defaults | frontend | listen | backend
4506 yes | yes | yes | yes
4507 Arguments : none
4508
4509 It sometimes happens that people need a pure HTTP proxy which understands
4510 basic proxy requests without caching nor any fancy feature. In this case,
4511 it may be worth setting up an HAProxy instance with the "option http_proxy"
4512 set. In this mode, no server is declared, and the connection is forwarded to
4513 the IP address and port found in the URL after the "http://" scheme.
4514
4515 No host address resolution is performed, so this only works when pure IP
4516 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribus3082bde2016-02-03 18:09:37 +01004517 it will probably be used only by experts who know they need exactly it. This
4518 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02004519
4520 If this option has been enabled in a "defaults" section, it can be disabled
4521 in a specific instance by prepending the "no" keyword before it.
4522
4523 Example :
4524 # this backend understands HTTP proxy requests and forwards them directly.
4525 backend direct_forward
4526 option httpclose
4527 option http_proxy
4528
4529 See also : "option httpclose"
4530
Willy Tarreau211ad242009-10-03 21:45:07 +02004531
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004532option independent-streams
4533no option independent-streams
4534 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004535 May be used in sections : defaults | frontend | listen | backend
4536 yes | yes | yes | yes
4537 Arguments : none
4538
4539 By default, when data is sent over a socket, both the write timeout and the
4540 read timeout for that socket are refreshed, because we consider that there is
4541 activity on that socket, and we have no other means of guessing if we should
4542 receive data or not.
4543
4544 While this default behaviour is desirable for almost all applications, there
4545 exists a situation where it is desirable to disable it, and only refresh the
4546 read timeout if there are incoming data. This happens on sessions with large
4547 timeouts and low amounts of exchanged data such as telnet session. If the
4548 server suddenly disappears, the output data accumulates in the system's
4549 socket buffers, both timeouts are correctly refreshed, and there is no way
4550 to know the server does not receive them, so we don't timeout. However, when
4551 the underlying protocol always echoes sent data, it would be enough by itself
4552 to detect the issue using the read timeout. Note that this problem does not
4553 happen with more verbose protocols because data won't accumulate long in the
4554 socket buffers.
4555
4556 When this option is set on the frontend, it will disable read timeout updates
4557 on data sent to the client. There probably is little use of this case. When
4558 the option is set on the backend, it will disable read timeout updates on
4559 data sent to the server. Doing so will typically break large HTTP posts from
4560 slow lines, so use it with caution.
4561
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004562 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004563 with a spelling mistake. This spelling is still supported but
4564 deprecated.
4565
Willy Tarreauce887fd2012-05-12 12:50:00 +02004566 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004567
4568
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02004569option ldap-check
4570 Use LDAPv3 health checks for server testing
4571 May be used in sections : defaults | frontend | listen | backend
4572 yes | no | yes | yes
4573 Arguments : none
4574
4575 It is possible to test that the server correctly talks LDAPv3 instead of just
4576 testing that it accepts the TCP connection. When this option is set, an
4577 LDAPv3 anonymous simple bind message is sent to the server, and the response
4578 is analyzed to find an LDAPv3 bind response message.
4579
4580 The server is considered valid only when the LDAP response contains success
4581 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
4582
4583 Logging of bind requests is server dependent see your documentation how to
4584 configure it.
4585
4586 Example :
4587 option ldap-check
4588
4589 See also : "option httpchk"
4590
4591
Willy Tarreau211ad242009-10-03 21:45:07 +02004592option log-health-checks
4593no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02004594 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02004595 May be used in sections : defaults | frontend | listen | backend
4596 yes | no | yes | yes
4597 Arguments : none
4598
Willy Tarreaubef1b322014-05-13 21:01:39 +02004599 By default, failed health check are logged if server is UP and successful
4600 health checks are logged if server is DOWN, so the amount of additional
4601 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02004602
Willy Tarreaubef1b322014-05-13 21:01:39 +02004603 When this option is enabled, any change of the health check status or to
4604 the server's health will be logged, so that it becomes possible to know
4605 that a server was failing occasional checks before crashing, or exactly when
4606 it failed to respond a valid HTTP status, then when the port started to
4607 reject connections, then when the server stopped responding at all.
4608
4609 Note that status changes not caused by health checks (eg: enable/disable on
4610 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02004611
Willy Tarreaubef1b322014-05-13 21:01:39 +02004612 See also: "option httpchk", "option ldap-check", "option mysql-check",
4613 "option pgsql-check", "option redis-check", "option smtpchk",
4614 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02004615
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004616
4617option log-separate-errors
4618no option log-separate-errors
4619 Change log level for non-completely successful connections
4620 May be used in sections : defaults | frontend | listen | backend
4621 yes | yes | yes | no
4622 Arguments : none
4623
4624 Sometimes looking for errors in logs is not easy. This option makes haproxy
4625 raise the level of logs containing potentially interesting information such
4626 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
4627 level changes from "info" to "err". This makes it possible to log them
4628 separately to a different file with most syslog daemons. Be careful not to
4629 remove them from the original file, otherwise you would lose ordering which
4630 provides very important information.
4631
4632 Using this option, large sites dealing with several thousand connections per
4633 second may log normal traffic to a rotating buffer and only archive smaller
4634 error logs.
4635
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004636 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004637 logging.
4638
Willy Tarreauc27debf2008-01-06 08:57:02 +01004639
4640option logasap
4641no option logasap
4642 Enable or disable early logging of HTTP requests
4643 May be used in sections : defaults | frontend | listen | backend
4644 yes | yes | yes | no
4645 Arguments : none
4646
4647 By default, HTTP requests are logged upon termination so that the total
4648 transfer time and the number of bytes appear in the logs. When large objects
4649 are being transferred, it may take a while before the request appears in the
4650 logs. Using "option logasap", the request gets logged as soon as the server
4651 sends the complete headers. The only missing information in the logs will be
4652 the total number of bytes which will indicate everything except the amount
4653 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004654 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01004655 "Content-Length" response header so that the logs at least indicate how many
4656 bytes are expected to be transferred.
4657
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004658 Examples :
4659 listen http_proxy 0.0.0.0:80
4660 mode http
4661 option httplog
4662 option logasap
4663 log 192.168.2.200 local3
4664
4665 >>> Feb 6 12:14:14 localhost \
4666 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
4667 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
4668 "GET /image.iso HTTP/1.0"
4669
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004670 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01004671 logging.
4672
4673
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004674option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004675 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004676 May be used in sections : defaults | frontend | listen | backend
4677 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004678 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004679 <username> This is the username which will be used when connecting to MySQL
4680 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004681 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004682
4683 If you specify a username, the check consists of sending two MySQL packet,
4684 one Client Authentication packet, and one QUIT packet, to correctly close
4685 MySQL session. We then parse the MySQL Handshake Initialisation packet and/or
4686 Error packet. It is a basic but useful test which does not produce error nor
4687 aborted connect on the server. However, it requires adding an authorization
4688 in the MySQL table, like this :
4689
4690 USE mysql;
4691 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
4692 FLUSH PRIVILEGES;
4693
4694 If you don't specify a username (it is deprecated and not recommended), the
4695 check only consists in parsing the Mysql Handshake Initialisation packet or
4696 Error packet, we don't send anything in this mode. It was reported that it
4697 can generate lockout if check is too frequent and/or if there is not enough
4698 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
4699 value as if a connection is established successfully within fewer than MySQL
4700 "max_connect_errors" attempts after a previous connection was interrupted,
4701 the error count for the host is cleared to zero. If HAProxy's server get
4702 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
4703
4704 Remember that this does not check database presence nor database consistency.
4705 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004706
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02004707 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004708
4709 Most often, an incoming MySQL server needs to see the client's IP address for
4710 various purposes, including IP privilege matching and connection logging.
4711 When possible, it is often wise to masquerade the client's IP address when
4712 connecting to the server using the "usesrc" argument of the "source" keyword,
4713 which requires the cttproxy feature to be compiled in, and the MySQL server
4714 to route the client via the machine hosting haproxy.
4715
4716 See also: "option httpchk"
4717
4718
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004719option nolinger
4720no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004721 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004722 May be used in sections: defaults | frontend | listen | backend
4723 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004724 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004725
4726 When clients or servers abort connections in a dirty way (eg: they are
4727 physically disconnected), the session timeouts triggers and the session is
4728 closed. But it will remain in FIN_WAIT1 state for some time in the system,
4729 using some resources and possibly limiting the ability to establish newer
4730 connections.
4731
4732 When this happens, it is possible to activate "option nolinger" which forces
4733 the system to immediately remove any socket's pending data on close. Thus,
4734 the session is instantly purged from the system's tables. This usually has
4735 side effects such as increased number of TCP resets due to old retransmits
4736 getting immediately rejected. Some firewalls may sometimes complain about
4737 this too.
4738
4739 For this reason, it is not recommended to use this option when not absolutely
4740 needed. You know that you need it when you have thousands of FIN_WAIT1
4741 sessions on your system (TIME_WAIT ones do not count).
4742
4743 This option may be used both on frontends and backends, depending on the side
4744 where it is required. Use it on the frontend for clients, and on the backend
4745 for servers.
4746
4747 If this option has been enabled in a "defaults" section, it can be disabled
4748 in a specific instance by prepending the "no" keyword before it.
4749
4750
Willy Tarreau55165fe2009-05-10 12:02:55 +02004751option originalto [ except <network> ] [ header <name> ]
4752 Enable insertion of the X-Original-To header to requests sent to servers
4753 May be used in sections : defaults | frontend | listen | backend
4754 yes | yes | yes | yes
4755 Arguments :
4756 <network> is an optional argument used to disable this option for sources
4757 matching <network>
4758 <name> an optional argument to specify a different "X-Original-To"
4759 header name.
4760
4761 Since HAProxy can work in transparent mode, every request from a client can
4762 be redirected to the proxy and HAProxy itself can proxy every request to a
4763 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
4764 be lost. This is annoying when you want access rules based on destination ip
4765 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
4766 added by HAProxy to all requests sent to the server. This header contains a
4767 value representing the original destination IP address. Since this must be
4768 configured to always use the last occurrence of this header only. Note that
4769 only the last occurrence of the header must be used, since it is really
4770 possible that the client has already brought one.
4771
4772 The keyword "header" may be used to supply a different header name to replace
4773 the default "X-Original-To". This can be useful where you might already
4774 have a "X-Original-To" header from a different application, and you need
4775 preserve it. Also if your backend server doesn't use the "X-Original-To"
4776 header and requires different one.
4777
4778 Sometimes, a same HAProxy instance may be shared between a direct client
4779 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4780 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4781 header for a known source address or network by adding the "except" keyword
4782 followed by the network address. In this case, any source IP matching the
4783 network will not cause an addition of this header. Most common uses are with
4784 private networks or 127.0.0.1.
4785
4786 This option may be specified either in the frontend or in the backend. If at
4787 least one of them uses it, the header will be added. Note that the backend's
4788 setting of the header subargument takes precedence over the frontend's if
4789 both are defined.
4790
Willy Tarreau55165fe2009-05-10 12:02:55 +02004791 Examples :
4792 # Original Destination address
4793 frontend www
4794 mode http
4795 option originalto except 127.0.0.1
4796
4797 # Those servers want the IP Address in X-Client-Dst
4798 backend www
4799 mode http
4800 option originalto header X-Client-Dst
4801
Willy Tarreau87cf5142011-08-19 22:57:24 +02004802 See also : "option httpclose", "option http-server-close",
4803 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02004804
4805
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004806option persist
4807no option persist
4808 Enable or disable forced persistence on down servers
4809 May be used in sections: defaults | frontend | listen | backend
4810 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004811 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004812
4813 When an HTTP request reaches a backend with a cookie which references a dead
4814 server, by default it is redispatched to another server. It is possible to
4815 force the request to be sent to the dead server first using "option persist"
4816 if absolutely needed. A common use case is when servers are under extreme
4817 load and spend their time flapping. In this case, the users would still be
4818 directed to the server they opened the session on, in the hope they would be
4819 correctly served. It is recommended to use "option redispatch" in conjunction
4820 with this option so that in the event it would not be possible to connect to
4821 the server at all (server definitely dead), the client would finally be
4822 redirected to another valid server.
4823
4824 If this option has been enabled in a "defaults" section, it can be disabled
4825 in a specific instance by prepending the "no" keyword before it.
4826
Willy Tarreau4de91492010-01-22 19:10:05 +01004827 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004828
4829
Willy Tarreau0c122822013-12-15 18:49:01 +01004830option pgsql-check [ user <username> ]
4831 Use PostgreSQL health checks for server testing
4832 May be used in sections : defaults | frontend | listen | backend
4833 yes | no | yes | yes
4834 Arguments :
4835 <username> This is the username which will be used when connecting to
4836 PostgreSQL server.
4837
4838 The check sends a PostgreSQL StartupMessage and waits for either
4839 Authentication request or ErrorResponse message. It is a basic but useful
4840 test which does not produce error nor aborted connect on the server.
4841 This check is identical with the "mysql-check".
4842
4843 See also: "option httpchk"
4844
4845
Willy Tarreau9420b122013-12-15 18:58:25 +01004846option prefer-last-server
4847no option prefer-last-server
4848 Allow multiple load balanced requests to remain on the same server
4849 May be used in sections: defaults | frontend | listen | backend
4850 yes | no | yes | yes
4851 Arguments : none
4852
4853 When the load balancing algorithm in use is not deterministic, and a previous
4854 request was sent to a server to which haproxy still holds a connection, it is
4855 sometimes desirable that subsequent requests on a same session go to the same
4856 server as much as possible. Note that this is different from persistence, as
4857 we only indicate a preference which haproxy tries to apply without any form
4858 of warranty. The real use is for keep-alive connections sent to servers. When
4859 this option is used, haproxy will try to reuse the same connection that is
4860 attached to the server instead of rebalancing to another server, causing a
4861 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01004862 not make much sense to use this in combination with hashing algorithms. Note,
4863 haproxy already automatically tries to stick to a server which sends a 401 or
4864 to a proxy which sends a 407 (authentication required). This is mandatory for
4865 use with the broken NTLM authentication challenge, and significantly helps in
4866 troubleshooting some faulty applications. Option prefer-last-server might be
4867 desirable in these environments as well, to avoid redistributing the traffic
4868 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01004869
4870 If this option has been enabled in a "defaults" section, it can be disabled
4871 in a specific instance by prepending the "no" keyword before it.
4872
4873 See also: "option http-keep-alive"
4874
4875
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004876option redispatch
4877no option redispatch
4878 Enable or disable session redistribution in case of connection failure
4879 May be used in sections: defaults | frontend | listen | backend
4880 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004881 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004882
4883 In HTTP mode, if a server designated by a cookie is down, clients may
4884 definitely stick to it because they cannot flush the cookie, so they will not
4885 be able to access the service anymore.
4886
4887 Specifying "option redispatch" will allow the proxy to break their
4888 persistence and redistribute them to a working server.
4889
4890 It also allows to retry last connection to another server in case of multiple
4891 connection failures. Of course, it requires having "retries" set to a nonzero
4892 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01004893
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004894 This form is the preferred form, which replaces both the "redispatch" and
4895 "redisp" keywords.
4896
4897 If this option has been enabled in a "defaults" section, it can be disabled
4898 in a specific instance by prepending the "no" keyword before it.
4899
Willy Tarreau4de91492010-01-22 19:10:05 +01004900 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004901
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004902
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02004903option redis-check
4904 Use redis health checks for server testing
4905 May be used in sections : defaults | frontend | listen | backend
4906 yes | no | yes | yes
4907 Arguments : none
4908
4909 It is possible to test that the server correctly talks REDIS protocol instead
4910 of just testing that it accepts the TCP connection. When this option is set,
4911 a PING redis command is sent to the server, and the response is analyzed to
4912 find the "+PONG" response message.
4913
4914 Example :
4915 option redis-check
4916
4917 See also : "option httpchk"
4918
4919
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004920option smtpchk
4921option smtpchk <hello> <domain>
4922 Use SMTP health checks for server testing
4923 May be used in sections : defaults | frontend | listen | backend
4924 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01004925 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004926 <hello> is an optional argument. It is the "hello" command to use. It can
4927 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
4928 values will be turned into the default command ("HELO").
4929
4930 <domain> is the domain name to present to the server. It may only be
4931 specified (and is mandatory) if the hello command has been
4932 specified. By default, "localhost" is used.
4933
4934 When "option smtpchk" is set, the health checks will consist in TCP
4935 connections followed by an SMTP command. By default, this command is
4936 "HELO localhost". The server's return code is analyzed and only return codes
4937 starting with a "2" will be considered as valid. All other responses,
4938 including a lack of response will constitute an error and will indicate a
4939 dead server.
4940
4941 This test is meant to be used with SMTP servers or relays. Depending on the
4942 request, it is possible that some servers do not log each connection attempt,
4943 so you may want to experiment to improve the behaviour. Using telnet on port
4944 25 is often easier than adjusting the configuration.
4945
4946 Most often, an incoming SMTP server needs to see the client's IP address for
4947 various purposes, including spam filtering, anti-spoofing and logging. When
4948 possible, it is often wise to masquerade the client's IP address when
4949 connecting to the server using the "usesrc" argument of the "source" keyword,
4950 which requires the cttproxy feature to be compiled in.
4951
4952 Example :
4953 option smtpchk HELO mydomain.org
4954
4955 See also : "option httpchk", "source"
4956
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004957
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02004958option socket-stats
4959no option socket-stats
4960
4961 Enable or disable collecting & providing separate statistics for each socket.
4962 May be used in sections : defaults | frontend | listen | backend
4963 yes | yes | yes | no
4964
4965 Arguments : none
4966
4967
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004968option splice-auto
4969no option splice-auto
4970 Enable or disable automatic kernel acceleration on sockets in both directions
4971 May be used in sections : defaults | frontend | listen | backend
4972 yes | yes | yes | yes
4973 Arguments : none
4974
4975 When this option is enabled either on a frontend or on a backend, haproxy
4976 will automatically evaluate the opportunity to use kernel tcp splicing to
4977 forward data between the client and the server, in either direction. Haproxy
4978 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004979 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004980 are not much aggressive in order to limit excessive use of splicing. This
4981 option requires splicing to be enabled at compile time, and may be globally
4982 disabled with the global option "nosplice". Since splice uses pipes, using it
4983 requires that there are enough spare pipes.
4984
4985 Important note: kernel-based TCP splicing is a Linux-specific feature which
4986 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
4987 transfer data between sockets without copying these data to user-space, thus
4988 providing noticeable performance gains and CPU cycles savings. Since many
4989 early implementations are buggy, corrupt data and/or are inefficient, this
4990 feature is not enabled by default, and it should be used with extreme care.
4991 While it is not possible to detect the correctness of an implementation,
4992 2.6.29 is the first version offering a properly working implementation. In
4993 case of doubt, splicing may be globally disabled using the global "nosplice"
4994 keyword.
4995
4996 Example :
4997 option splice-auto
4998
4999 If this option has been enabled in a "defaults" section, it can be disabled
5000 in a specific instance by prepending the "no" keyword before it.
5001
5002 See also : "option splice-request", "option splice-response", and global
5003 options "nosplice" and "maxpipes"
5004
5005
5006option splice-request
5007no option splice-request
5008 Enable or disable automatic kernel acceleration on sockets for requests
5009 May be used in sections : defaults | frontend | listen | backend
5010 yes | yes | yes | yes
5011 Arguments : none
5012
5013 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005014 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01005015 the client to the server. It might still use the recv/send scheme if there
5016 are no spare pipes left. This option requires splicing to be enabled at
5017 compile time, and may be globally disabled with the global option "nosplice".
5018 Since splice uses pipes, using it requires that there are enough spare pipes.
5019
5020 Important note: see "option splice-auto" for usage limitations.
5021
5022 Example :
5023 option splice-request
5024
5025 If this option has been enabled in a "defaults" section, it can be disabled
5026 in a specific instance by prepending the "no" keyword before it.
5027
5028 See also : "option splice-auto", "option splice-response", and global options
5029 "nosplice" and "maxpipes"
5030
5031
5032option splice-response
5033no option splice-response
5034 Enable or disable automatic kernel acceleration on sockets for responses
5035 May be used in sections : defaults | frontend | listen | backend
5036 yes | yes | yes | yes
5037 Arguments : none
5038
5039 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005040 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01005041 the server to the client. It might still use the recv/send scheme if there
5042 are no spare pipes left. This option requires splicing to be enabled at
5043 compile time, and may be globally disabled with the global option "nosplice".
5044 Since splice uses pipes, using it requires that there are enough spare pipes.
5045
5046 Important note: see "option splice-auto" for usage limitations.
5047
5048 Example :
5049 option splice-response
5050
5051 If this option has been enabled in a "defaults" section, it can be disabled
5052 in a specific instance by prepending the "no" keyword before it.
5053
5054 See also : "option splice-auto", "option splice-request", and global options
5055 "nosplice" and "maxpipes"
5056
5057
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005058option srvtcpka
5059no option srvtcpka
5060 Enable or disable the sending of TCP keepalive packets on the server side
5061 May be used in sections : defaults | frontend | listen | backend
5062 yes | no | yes | yes
5063 Arguments : none
5064
5065 When there is a firewall or any session-aware component between a client and
5066 a server, and when the protocol involves very long sessions with long idle
5067 periods (eg: remote desktops), there is a risk that one of the intermediate
5068 components decides to expire a session which has remained idle for too long.
5069
5070 Enabling socket-level TCP keep-alives makes the system regularly send packets
5071 to the other end of the connection, leaving it active. The delay between
5072 keep-alive probes is controlled by the system only and depends both on the
5073 operating system and its tuning parameters.
5074
5075 It is important to understand that keep-alive packets are neither emitted nor
5076 received at the application level. It is only the network stacks which sees
5077 them. For this reason, even if one side of the proxy already uses keep-alives
5078 to maintain its connection alive, those keep-alive packets will not be
5079 forwarded to the other side of the proxy.
5080
5081 Please note that this has nothing to do with HTTP keep-alive.
5082
5083 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
5084 server side of a connection, which should help when session expirations are
5085 noticed between HAProxy and a server.
5086
5087 If this option has been enabled in a "defaults" section, it can be disabled
5088 in a specific instance by prepending the "no" keyword before it.
5089
5090 See also : "option clitcpka", "option tcpka"
5091
5092
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005093option ssl-hello-chk
5094 Use SSLv3 client hello health checks for server testing
5095 May be used in sections : defaults | frontend | listen | backend
5096 yes | no | yes | yes
5097 Arguments : none
5098
5099 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
5100 possible to test that the server correctly talks SSL instead of just testing
5101 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
5102 SSLv3 client hello messages are sent once the connection is established to
5103 the server, and the response is analyzed to find an SSL server hello message.
5104 The server is considered valid only when the response contains this server
5105 hello message.
5106
5107 All servers tested till there correctly reply to SSLv3 client hello messages,
5108 and most servers tested do not even log the requests containing only hello
5109 messages, which is appreciable.
5110
Willy Tarreau763a95b2012-10-04 23:15:39 +02005111 Note that this check works even when SSL support was not built into haproxy
5112 because it forges the SSL message. When SSL support is available, it is best
5113 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005114
Willy Tarreau763a95b2012-10-04 23:15:39 +02005115 See also: "option httpchk", "check-ssl"
5116
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005117
Willy Tarreaued179852013-12-16 01:07:00 +01005118option tcp-check
5119 Perform health checks using tcp-check send/expect sequences
5120 May be used in sections: defaults | frontend | listen | backend
5121 yes | no | yes | yes
5122
5123 This health check method is intended to be combined with "tcp-check" command
5124 lists in order to support send/expect types of health check sequences.
5125
5126 TCP checks currently support 4 modes of operations :
5127 - no "tcp-check" directive : the health check only consists in a connection
5128 attempt, which remains the default mode.
5129
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005130 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005131 used to send a string along with a connection opening. With some
5132 protocols, it helps sending a "QUIT" message for example that prevents
5133 the server from logging a connection error for each health check. The
5134 check result will still be based on the ability to open the connection
5135 only.
5136
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005137 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01005138 The connection is opened and haproxy waits for the server to present some
5139 contents which must validate some rules. The check result will be based
5140 on the matching between the contents and the rules. This is suited for
5141 POP, IMAP, SMTP, FTP, SSH, TELNET.
5142
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005143 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005144 used to test a hello-type protocol. Haproxy sends a message, the server
5145 responds and its response is analysed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005146 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01005147 suited for protocols which require a binding or a request/response model.
5148 LDAP, MySQL, Redis and SSL are example of such protocols, though they
5149 already all have their dedicated checks with a deeper understanding of
5150 the respective protocols.
5151 In this mode, many questions may be sent and many answers may be
5152 analysed.
5153
5154 Examples :
5155 # perform a POP check (analyse only server's banner)
5156 option tcp-check
5157 tcp-check expect string +OK\ POP3\ ready
5158
5159 # perform an IMAP check (analyse only server's banner)
5160 option tcp-check
5161 tcp-check expect string *\ OK\ IMAP4\ ready
5162
5163 # look for the redis master server after ensuring it speaks well
5164 # redis protocol, then it exits properly.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005165 # (send a command then analyse the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01005166 option tcp-check
5167 tcp-check send PING\r\n
5168 tcp-check expect +PONG
5169 tcp-check send info\ replication\r\n
5170 tcp-check expect string role:master
5171 tcp-check send QUIT\r\n
5172 tcp-check expect string +OK
5173
5174 forge a HTTP request, then analyse the response
5175 (send many headers before analyzing)
5176 option tcp-check
5177 tcp-check send HEAD\ /\ HTTP/1.1\r\n
5178 tcp-check send Host:\ www.mydomain.com\r\n
5179 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
5180 tcp-check send \r\n
5181 tcp-check expect rstring HTTP/1\..\ (2..|3..)
5182
5183
5184 See also : "tcp-check expect", "tcp-check send"
5185
5186
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005187option tcp-smart-accept
5188no option tcp-smart-accept
5189 Enable or disable the saving of one ACK packet during the accept sequence
5190 May be used in sections : defaults | frontend | listen | backend
5191 yes | yes | yes | no
5192 Arguments : none
5193
5194 When an HTTP connection request comes in, the system acknowledges it on
5195 behalf of HAProxy, then the client immediately sends its request, and the
5196 system acknowledges it too while it is notifying HAProxy about the new
5197 connection. HAProxy then reads the request and responds. This means that we
5198 have one TCP ACK sent by the system for nothing, because the request could
5199 very well be acknowledged by HAProxy when it sends its response.
5200
5201 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
5202 sending this useless ACK on platforms which support it (currently at least
5203 Linux). It must not cause any problem, because the system will send it anyway
5204 after 40 ms if the response takes more time than expected to come.
5205
5206 During complex network debugging sessions, it may be desirable to disable
5207 this optimization because delayed ACKs can make troubleshooting more complex
5208 when trying to identify where packets are delayed. It is then possible to
5209 fall back to normal behaviour by specifying "no option tcp-smart-accept".
5210
5211 It is also possible to force it for non-HTTP proxies by simply specifying
5212 "option tcp-smart-accept". For instance, it can make sense with some services
5213 such as SMTP where the server speaks first.
5214
5215 It is recommended to avoid forcing this option in a defaults section. In case
5216 of doubt, consider setting it back to automatic values by prepending the
5217 "default" keyword before it, or disabling it using the "no" keyword.
5218
Willy Tarreaud88edf22009-06-14 15:48:17 +02005219 See also : "option tcp-smart-connect"
5220
5221
5222option tcp-smart-connect
5223no option tcp-smart-connect
5224 Enable or disable the saving of one ACK packet during the connect sequence
5225 May be used in sections : defaults | frontend | listen | backend
5226 yes | no | yes | yes
5227 Arguments : none
5228
5229 On certain systems (at least Linux), HAProxy can ask the kernel not to
5230 immediately send an empty ACK upon a connection request, but to directly
5231 send the buffer request instead. This saves one packet on the network and
5232 thus boosts performance. It can also be useful for some servers, because they
5233 immediately get the request along with the incoming connection.
5234
5235 This feature is enabled when "option tcp-smart-connect" is set in a backend.
5236 It is not enabled by default because it makes network troubleshooting more
5237 complex.
5238
5239 It only makes sense to enable it with protocols where the client speaks first
5240 such as HTTP. In other situations, if there is no data to send in place of
5241 the ACK, a normal ACK is sent.
5242
5243 If this option has been enabled in a "defaults" section, it can be disabled
5244 in a specific instance by prepending the "no" keyword before it.
5245
5246 See also : "option tcp-smart-accept"
5247
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005248
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005249option tcpka
5250 Enable or disable the sending of TCP keepalive packets on both sides
5251 May be used in sections : defaults | frontend | listen | backend
5252 yes | yes | yes | yes
5253 Arguments : none
5254
5255 When there is a firewall or any session-aware component between a client and
5256 a server, and when the protocol involves very long sessions with long idle
5257 periods (eg: remote desktops), there is a risk that one of the intermediate
5258 components decides to expire a session which has remained idle for too long.
5259
5260 Enabling socket-level TCP keep-alives makes the system regularly send packets
5261 to the other end of the connection, leaving it active. The delay between
5262 keep-alive probes is controlled by the system only and depends both on the
5263 operating system and its tuning parameters.
5264
5265 It is important to understand that keep-alive packets are neither emitted nor
5266 received at the application level. It is only the network stacks which sees
5267 them. For this reason, even if one side of the proxy already uses keep-alives
5268 to maintain its connection alive, those keep-alive packets will not be
5269 forwarded to the other side of the proxy.
5270
5271 Please note that this has nothing to do with HTTP keep-alive.
5272
5273 Using option "tcpka" enables the emission of TCP keep-alive probes on both
5274 the client and server sides of a connection. Note that this is meaningful
5275 only in "defaults" or "listen" sections. If this option is used in a
5276 frontend, only the client side will get keep-alives, and if this option is
5277 used in a backend, only the server side will get keep-alives. For this
5278 reason, it is strongly recommended to explicitly use "option clitcpka" and
5279 "option srvtcpka" when the configuration is split between frontends and
5280 backends.
5281
5282 See also : "option clitcpka", "option srvtcpka"
5283
Willy Tarreau844e3c52008-01-11 16:28:18 +01005284
5285option tcplog
5286 Enable advanced logging of TCP connections with session state and timers
5287 May be used in sections : defaults | frontend | listen | backend
5288 yes | yes | yes | yes
5289 Arguments : none
5290
5291 By default, the log output format is very poor, as it only contains the
5292 source and destination addresses, and the instance name. By specifying
5293 "option tcplog", each log line turns into a much richer format including, but
5294 not limited to, the connection timers, the session status, the connections
5295 numbers, the frontend, backend and server name, and of course the source
5296 address and ports. This option is useful for pure TCP proxies in order to
5297 find which of the client or server disconnects or times out. For normal HTTP
5298 proxies, it's better to use "option httplog" which is even more complete.
5299
5300 This option may be set either in the frontend or the backend.
5301
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005302 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005303
5304
Willy Tarreau844e3c52008-01-11 16:28:18 +01005305option transparent
5306no option transparent
5307 Enable client-side transparent proxying
5308 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01005309 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01005310 Arguments : none
5311
5312 This option was introduced in order to provide layer 7 persistence to layer 3
5313 load balancers. The idea is to use the OS's ability to redirect an incoming
5314 connection for a remote address to a local process (here HAProxy), and let
5315 this process know what address was initially requested. When this option is
5316 used, sessions without cookies will be forwarded to the original destination
5317 IP address of the incoming request (which should match that of another
5318 equipment), while requests with cookies will still be forwarded to the
5319 appropriate server.
5320
5321 Note that contrary to a common belief, this option does NOT make HAProxy
5322 present the client's IP to the server when establishing the connection.
5323
Willy Tarreaua1146052011-03-01 09:51:54 +01005324 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005325 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005326
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005327
Emeric Brun647caf12009-06-30 17:57:00 +02005328persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02005329persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02005330 Enable RDP cookie-based persistence
5331 May be used in sections : defaults | frontend | listen | backend
5332 yes | no | yes | yes
5333 Arguments :
5334 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02005335 default cookie name "msts" will be used. There currently is no
5336 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02005337
5338 This statement enables persistence based on an RDP cookie. The RDP cookie
5339 contains all information required to find the server in the list of known
5340 servers. So when this option is set in the backend, the request is analysed
5341 and if an RDP cookie is found, it is decoded. If it matches a known server
5342 which is still UP (or if "option persist" is set), then the connection is
5343 forwarded to this server.
5344
5345 Note that this only makes sense in a TCP backend, but for this to work, the
5346 frontend must have waited long enough to ensure that an RDP cookie is present
5347 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005348 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02005349 a single "listen" section.
5350
Willy Tarreau61e28f22010-05-16 22:31:05 +02005351 Also, it is important to understand that the terminal server will emit this
5352 RDP cookie only if it is configured for "token redirection mode", which means
5353 that the "IP address redirection" option is disabled.
5354
Emeric Brun647caf12009-06-30 17:57:00 +02005355 Example :
5356 listen tse-farm
5357 bind :3389
5358 # wait up to 5s for an RDP cookie in the request
5359 tcp-request inspect-delay 5s
5360 tcp-request content accept if RDP_COOKIE
5361 # apply RDP cookie persistence
5362 persist rdp-cookie
5363 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02005364 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02005365 balance rdp-cookie
5366 server srv1 1.1.1.1:3389
5367 server srv2 1.1.1.2:3389
5368
Simon Hormanab814e02011-06-24 14:50:20 +09005369 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
5370 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02005371
5372
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005373rate-limit sessions <rate>
5374 Set a limit on the number of new sessions accepted per second on a frontend
5375 May be used in sections : defaults | frontend | listen | backend
5376 yes | yes | yes | no
5377 Arguments :
5378 <rate> The <rate> parameter is an integer designating the maximum number
5379 of new sessions per second to accept on the frontend.
5380
5381 When the frontend reaches the specified number of new sessions per second, it
5382 stops accepting new connections until the rate drops below the limit again.
5383 During this time, the pending sessions will be kept in the socket's backlog
5384 (in system buffers) and haproxy will not even be aware that sessions are
5385 pending. When applying very low limit on a highly loaded service, it may make
5386 sense to increase the socket's backlog using the "backlog" keyword.
5387
5388 This feature is particularly efficient at blocking connection-based attacks
5389 or service abuse on fragile servers. Since the session rate is measured every
5390 millisecond, it is extremely accurate. Also, the limit applies immediately,
5391 no delay is needed at all to detect the threshold.
5392
5393 Example : limit the connection rate on SMTP to 10 per second max
5394 listen smtp
5395 mode tcp
5396 bind :25
5397 rate-limit sessions 10
Panagiotis Panagiotopoulos6f934442016-02-11 16:37:15 +02005398 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005399
Willy Tarreaua17c2d92011-07-25 08:16:20 +02005400 Note : when the maximum rate is reached, the frontend's status is not changed
5401 but its sockets appear as "WAITING" in the statistics if the
5402 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005403
5404 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
5405
5406
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005407redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
5408redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
5409redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005410 Return an HTTP redirection if/unless a condition is matched
5411 May be used in sections : defaults | frontend | listen | backend
5412 no | yes | yes | yes
5413
5414 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01005415 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005416
Willy Tarreau0140f252008-11-19 21:07:09 +01005417 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005418 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005419 the HTTP "Location" header. When used in an "http-request" rule,
5420 <loc> value follows the log-format rules and can include some
5421 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005422
5423 <pfx> With "redirect prefix", the "Location" header is built from the
5424 concatenation of <pfx> and the complete URI path, including the
5425 query string, unless the "drop-query" option is specified (see
5426 below). As a special case, if <pfx> equals exactly "/", then
5427 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005428 redirect to the same URL (for instance, to insert a cookie). When
5429 used in an "http-request" rule, <pfx> value follows the log-format
5430 rules and can include some dynamic values (see Custom Log Format
5431 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005432
5433 <sch> With "redirect scheme", then the "Location" header is built by
5434 concatenating <sch> with "://" then the first occurrence of the
5435 "Host" header, and then the URI path, including the query string
5436 unless the "drop-query" option is specified (see below). If no
5437 path is found or if the path is "*", then "/" is used instead. If
5438 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005439 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005440 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005441 HTTPS. When used in an "http-request" rule, <sch> value follows
5442 the log-format rules and can include some dynamic values (see
5443 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01005444
5445 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005446 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
5447 with 302 used by default if no code is specified. 301 means
5448 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmann6c7351b2015-08-03 11:42:50 +02005449 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005450 cache the redirection. 303 is equivalent to 302 except that the
5451 browser will fetch the location with a GET method. 307 is just
5452 like 302 but makes it clear that the same method must be reused.
5453 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01005454
5455 <option> There are several options which can be specified to adjust the
5456 expected behaviour of a redirection :
5457
5458 - "drop-query"
5459 When this keyword is used in a prefix-based redirection, then the
5460 location will be set without any possible query-string, which is useful
5461 for directing users to a non-secure page for instance. It has no effect
5462 with a location-type redirect.
5463
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005464 - "append-slash"
5465 This keyword may be used in conjunction with "drop-query" to redirect
5466 users who use a URL not ending with a '/' to the same one with the '/'.
5467 It can be useful to ensure that search engines will only see one URL.
5468 For this, a return code 301 is preferred.
5469
Willy Tarreau0140f252008-11-19 21:07:09 +01005470 - "set-cookie NAME[=value]"
5471 A "Set-Cookie" header will be added with NAME (and optionally "=value")
5472 to the response. This is sometimes used to indicate that a user has
5473 been seen, for instance to protect against some types of DoS. No other
5474 cookie option is added, so the cookie will be a session cookie. Note
5475 that for a browser, a sole cookie name without an equal sign is
5476 different from a cookie with an equal sign.
5477
5478 - "clear-cookie NAME[=]"
5479 A "Set-Cookie" header will be added with NAME (and optionally "="), but
5480 with the "Max-Age" attribute set to zero. This will tell the browser to
5481 delete this cookie. It is useful for instance on logout pages. It is
5482 important to note that clearing the cookie "NAME" will not remove a
5483 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
5484 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005485
5486 Example: move the login URL only to HTTPS.
5487 acl clear dst_port 80
5488 acl secure dst_port 8080
5489 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01005490 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01005491 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01005492 acl cookie_set hdr_sub(cookie) SEEN=1
5493
5494 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01005495 redirect prefix https://mysite.com if login_page !secure
5496 redirect prefix http://mysite.com drop-query if login_page !uid_given
5497 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01005498 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005499
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005500 Example: send redirects for request for articles without a '/'.
5501 acl missing_slash path_reg ^/article/[^/]*$
5502 redirect code 301 prefix / drop-query append-slash if missing_slash
5503
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005504 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01005505 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005506
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005507 Example: append 'www.' prefix in front of all hosts not having it
Baptiste Assmann4f661102016-03-06 23:42:52 +01005508 http-request redirect code 301 location \
5509 www.%[hdr(host)]%[capture.req.uri] unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005510
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005511 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005512
5513
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005514redisp (deprecated)
5515redispatch (deprecated)
5516 Enable or disable session redistribution in case of connection failure
5517 May be used in sections: defaults | frontend | listen | backend
5518 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005519 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005520
5521 In HTTP mode, if a server designated by a cookie is down, clients may
5522 definitely stick to it because they cannot flush the cookie, so they will not
5523 be able to access the service anymore.
5524
5525 Specifying "redispatch" will allow the proxy to break their persistence and
5526 redistribute them to a working server.
5527
5528 It also allows to retry last connection to another server in case of multiple
5529 connection failures. Of course, it requires having "retries" set to a nonzero
5530 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01005531
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005532 This form is deprecated, do not use it in any new configuration, use the new
5533 "option redispatch" instead.
5534
5535 See also : "option redispatch"
5536
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005537
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005538reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005539 Add a header at the end of the HTTP request
5540 May be used in sections : defaults | frontend | listen | backend
5541 no | yes | yes | yes
5542 Arguments :
5543 <string> is the complete line to be added. Any space or known delimiter
5544 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005545 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005546
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005547 <cond> is an optional matching condition built from ACLs. It makes it
5548 possible to ignore this rule when other conditions are not met.
5549
Willy Tarreau303c0352008-01-17 19:01:39 +01005550 A new line consisting in <string> followed by a line feed will be added after
5551 the last header of an HTTP request.
5552
5553 Header transformations only apply to traffic which passes through HAProxy,
5554 and not to traffic generated by HAProxy, such as health-checks or error
5555 responses.
5556
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005557 Example : add "X-Proto: SSL" to requests coming via port 81
5558 acl is-ssl dst_port 81
5559 reqadd X-Proto:\ SSL if is-ssl
5560
5561 See also: "rspadd", section 6 about HTTP header manipulation, and section 7
5562 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005563
5564
Willy Tarreau5321c422010-01-28 20:35:13 +01005565reqallow <search> [{if | unless} <cond>]
5566reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005567 Definitely allow an HTTP request if a line matches a regular expression
5568 May be used in sections : defaults | frontend | listen | backend
5569 no | yes | yes | yes
5570 Arguments :
5571 <search> is the regular expression applied to HTTP headers and to the
5572 request line. This is an extended regular expression. Parenthesis
5573 grouping is supported and no preliminary backslash is required.
5574 Any space or known delimiter must be escaped using a backslash
5575 ('\'). The pattern applies to a full line at a time. The
5576 "reqallow" keyword strictly matches case while "reqiallow"
5577 ignores case.
5578
Willy Tarreau5321c422010-01-28 20:35:13 +01005579 <cond> is an optional matching condition built from ACLs. It makes it
5580 possible to ignore this rule when other conditions are not met.
5581
Willy Tarreau303c0352008-01-17 19:01:39 +01005582 A request containing any line which matches extended regular expression
5583 <search> will mark the request as allowed, even if any later test would
5584 result in a deny. The test applies both to the request line and to request
5585 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005586 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005587
5588 It is easier, faster and more powerful to use ACLs to write access policies.
5589 Reqdeny, reqallow and reqpass should be avoided in new designs.
5590
5591 Example :
5592 # allow www.* but refuse *.local
5593 reqiallow ^Host:\ www\.
5594 reqideny ^Host:\ .*\.local
5595
Willy Tarreau5321c422010-01-28 20:35:13 +01005596 See also: "reqdeny", "block", section 6 about HTTP header manipulation, and
5597 section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005598
5599
Willy Tarreau5321c422010-01-28 20:35:13 +01005600reqdel <search> [{if | unless} <cond>]
5601reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005602 Delete all headers matching a regular expression in an HTTP request
5603 May be used in sections : defaults | frontend | listen | backend
5604 no | yes | yes | yes
5605 Arguments :
5606 <search> is the regular expression applied to HTTP headers and to the
5607 request line. This is an extended regular expression. Parenthesis
5608 grouping is supported and no preliminary backslash is required.
5609 Any space or known delimiter must be escaped using a backslash
5610 ('\'). The pattern applies to a full line at a time. The "reqdel"
5611 keyword strictly matches case while "reqidel" ignores case.
5612
Willy Tarreau5321c422010-01-28 20:35:13 +01005613 <cond> is an optional matching condition built from ACLs. It makes it
5614 possible to ignore this rule when other conditions are not met.
5615
Willy Tarreau303c0352008-01-17 19:01:39 +01005616 Any header line matching extended regular expression <search> in the request
5617 will be completely deleted. Most common use of this is to remove unwanted
5618 and/or dangerous headers or cookies from a request before passing it to the
5619 next servers.
5620
5621 Header transformations only apply to traffic which passes through HAProxy,
5622 and not to traffic generated by HAProxy, such as health-checks or error
5623 responses. Keep in mind that header names are not case-sensitive.
5624
5625 Example :
5626 # remove X-Forwarded-For header and SERVER cookie
5627 reqidel ^X-Forwarded-For:.*
5628 reqidel ^Cookie:.*SERVER=
5629
Willy Tarreau5321c422010-01-28 20:35:13 +01005630 See also: "reqadd", "reqrep", "rspdel", section 6 about HTTP header
5631 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005632
5633
Willy Tarreau5321c422010-01-28 20:35:13 +01005634reqdeny <search> [{if | unless} <cond>]
5635reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005636 Deny an HTTP request if a line matches a regular expression
5637 May be used in sections : defaults | frontend | listen | backend
5638 no | yes | yes | yes
5639 Arguments :
5640 <search> is the regular expression applied to HTTP headers and to the
5641 request line. This is an extended regular expression. Parenthesis
5642 grouping is supported and no preliminary backslash is required.
5643 Any space or known delimiter must be escaped using a backslash
5644 ('\'). The pattern applies to a full line at a time. The
5645 "reqdeny" keyword strictly matches case while "reqideny" ignores
5646 case.
5647
Willy Tarreau5321c422010-01-28 20:35:13 +01005648 <cond> is an optional matching condition built from ACLs. It makes it
5649 possible to ignore this rule when other conditions are not met.
5650
Willy Tarreau303c0352008-01-17 19:01:39 +01005651 A request containing any line which matches extended regular expression
5652 <search> will mark the request as denied, even if any later test would
5653 result in an allow. The test applies both to the request line and to request
5654 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005655 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005656
Willy Tarreauced27012008-01-17 20:35:34 +01005657 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005658 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01005659 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01005660
Willy Tarreau303c0352008-01-17 19:01:39 +01005661 It is easier, faster and more powerful to use ACLs to write access policies.
5662 Reqdeny, reqallow and reqpass should be avoided in new designs.
5663
5664 Example :
5665 # refuse *.local, then allow www.*
5666 reqideny ^Host:\ .*\.local
5667 reqiallow ^Host:\ www\.
5668
Willy Tarreau5321c422010-01-28 20:35:13 +01005669 See also: "reqallow", "rspdeny", "block", section 6 about HTTP header
5670 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005671
5672
Willy Tarreau5321c422010-01-28 20:35:13 +01005673reqpass <search> [{if | unless} <cond>]
5674reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005675 Ignore any HTTP request line matching a regular expression in next rules
5676 May be used in sections : defaults | frontend | listen | backend
5677 no | yes | yes | yes
5678 Arguments :
5679 <search> is the regular expression applied to HTTP headers and to the
5680 request line. This is an extended regular expression. Parenthesis
5681 grouping is supported and no preliminary backslash is required.
5682 Any space or known delimiter must be escaped using a backslash
5683 ('\'). The pattern applies to a full line at a time. The
5684 "reqpass" keyword strictly matches case while "reqipass" ignores
5685 case.
5686
Willy Tarreau5321c422010-01-28 20:35:13 +01005687 <cond> is an optional matching condition built from ACLs. It makes it
5688 possible to ignore this rule when other conditions are not met.
5689
Willy Tarreau303c0352008-01-17 19:01:39 +01005690 A request containing any line which matches extended regular expression
5691 <search> will skip next rules, without assigning any deny or allow verdict.
5692 The test applies both to the request line and to request headers. Keep in
5693 mind that URLs in request line are case-sensitive while header names are not.
5694
5695 It is easier, faster and more powerful to use ACLs to write access policies.
5696 Reqdeny, reqallow and reqpass should be avoided in new designs.
5697
5698 Example :
5699 # refuse *.local, then allow www.*, but ignore "www.private.local"
5700 reqipass ^Host:\ www.private\.local
5701 reqideny ^Host:\ .*\.local
5702 reqiallow ^Host:\ www\.
5703
Willy Tarreau5321c422010-01-28 20:35:13 +01005704 See also: "reqallow", "reqdeny", "block", section 6 about HTTP header
5705 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005706
5707
Willy Tarreau5321c422010-01-28 20:35:13 +01005708reqrep <search> <string> [{if | unless} <cond>]
5709reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005710 Replace a regular expression with a string in an HTTP request line
5711 May be used in sections : defaults | frontend | listen | backend
5712 no | yes | yes | yes
5713 Arguments :
5714 <search> is the regular expression applied to HTTP headers and to the
5715 request line. This is an extended regular expression. Parenthesis
5716 grouping is supported and no preliminary backslash is required.
5717 Any space or known delimiter must be escaped using a backslash
5718 ('\'). The pattern applies to a full line at a time. The "reqrep"
5719 keyword strictly matches case while "reqirep" ignores case.
5720
5721 <string> is the complete line to be added. Any space or known delimiter
5722 must be escaped using a backslash ('\'). References to matched
5723 pattern groups are possible using the common \N form, with N
5724 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005725 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005726
Willy Tarreau5321c422010-01-28 20:35:13 +01005727 <cond> is an optional matching condition built from ACLs. It makes it
5728 possible to ignore this rule when other conditions are not met.
5729
Willy Tarreau303c0352008-01-17 19:01:39 +01005730 Any line matching extended regular expression <search> in the request (both
5731 the request line and header lines) will be completely replaced with <string>.
5732 Most common use of this is to rewrite URLs or domain names in "Host" headers.
5733
5734 Header transformations only apply to traffic which passes through HAProxy,
5735 and not to traffic generated by HAProxy, such as health-checks or error
5736 responses. Note that for increased readability, it is suggested to add enough
5737 spaces between the request and the response. Keep in mind that URLs in
5738 request line are case-sensitive while header names are not.
5739
5740 Example :
5741 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04005742 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01005743 # replace "www.mydomain.com" with "www" in the host name.
5744 reqirep ^Host:\ www.mydomain.com Host:\ www
5745
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04005746 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", section 6 about
5747 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005748
5749
Willy Tarreau5321c422010-01-28 20:35:13 +01005750reqtarpit <search> [{if | unless} <cond>]
5751reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005752 Tarpit an HTTP request containing a line matching a regular expression
5753 May be used in sections : defaults | frontend | listen | backend
5754 no | yes | yes | yes
5755 Arguments :
5756 <search> is the regular expression applied to HTTP headers and to the
5757 request line. This is an extended regular expression. Parenthesis
5758 grouping is supported and no preliminary backslash is required.
5759 Any space or known delimiter must be escaped using a backslash
5760 ('\'). The pattern applies to a full line at a time. The
5761 "reqtarpit" keyword strictly matches case while "reqitarpit"
5762 ignores case.
5763
Willy Tarreau5321c422010-01-28 20:35:13 +01005764 <cond> is an optional matching condition built from ACLs. It makes it
5765 possible to ignore this rule when other conditions are not met.
5766
Willy Tarreau303c0352008-01-17 19:01:39 +01005767 A request containing any line which matches extended regular expression
5768 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01005769 be kept open for a pre-defined time, then will return an HTTP error 500 so
5770 that the attacker does not suspect it has been tarpitted. The status 500 will
5771 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01005772 delay is defined by "timeout tarpit", or "timeout connect" if the former is
5773 not set.
5774
5775 The goal of the tarpit is to slow down robots attacking servers with
5776 identifiable requests. Many robots limit their outgoing number of connections
5777 and stay connected waiting for a reply which can take several minutes to
5778 come. Depending on the environment and attack, it may be particularly
5779 efficient at reducing the load on the network and firewalls.
5780
Willy Tarreau5321c422010-01-28 20:35:13 +01005781 Examples :
Willy Tarreau303c0352008-01-17 19:01:39 +01005782 # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but
5783 # block all others.
5784 reqipass ^User-Agent:\.*(Mozilla|MSIE)
5785 reqitarpit ^User-Agent:
5786
Willy Tarreau5321c422010-01-28 20:35:13 +01005787 # block bad guys
5788 acl badguys src 10.1.0.3 172.16.13.20/28
5789 reqitarpit . if badguys
5790
5791 See also: "reqallow", "reqdeny", "reqpass", section 6 about HTTP header
5792 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005793
5794
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02005795retries <value>
5796 Set the number of retries to perform on a server after a connection failure
5797 May be used in sections: defaults | frontend | listen | backend
5798 yes | no | yes | yes
5799 Arguments :
5800 <value> is the number of times a connection attempt should be retried on
5801 a server when a connection either is refused or times out. The
5802 default value is 3.
5803
5804 It is important to understand that this value applies to the number of
5805 connection attempts, not full requests. When a connection has effectively
5806 been established to a server, there will be no more retry.
5807
5808 In order to avoid immediate reconnections to a server which is restarting,
5809 a turn-around timer of 1 second is applied before a retry occurs.
5810
5811 When "option redispatch" is set, the last retry may be performed on another
5812 server even if a cookie references a different server.
5813
5814 See also : "option redispatch"
5815
5816
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005817rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005818 Add a header at the end of the HTTP response
5819 May be used in sections : defaults | frontend | listen | backend
5820 no | yes | yes | yes
5821 Arguments :
5822 <string> is the complete line to be added. Any space or known delimiter
5823 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005824 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005825
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005826 <cond> is an optional matching condition built from ACLs. It makes it
5827 possible to ignore this rule when other conditions are not met.
5828
Willy Tarreau303c0352008-01-17 19:01:39 +01005829 A new line consisting in <string> followed by a line feed will be added after
5830 the last header of an HTTP response.
5831
5832 Header transformations only apply to traffic which passes through HAProxy,
5833 and not to traffic generated by HAProxy, such as health-checks or error
5834 responses.
5835
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005836 See also: "reqadd", section 6 about HTTP header manipulation, and section 7
5837 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005838
5839
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005840rspdel <search> [{if | unless} <cond>]
5841rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005842 Delete all headers matching a regular expression in an HTTP response
5843 May be used in sections : defaults | frontend | listen | backend
5844 no | yes | yes | yes
5845 Arguments :
5846 <search> is the regular expression applied to HTTP headers and to the
5847 response line. This is an extended regular expression, so
5848 parenthesis grouping is supported and no preliminary backslash
5849 is required. Any space or known delimiter must be escaped using
5850 a backslash ('\'). The pattern applies to a full line at a time.
5851 The "rspdel" keyword strictly matches case while "rspidel"
5852 ignores case.
5853
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005854 <cond> is an optional matching condition built from ACLs. It makes it
5855 possible to ignore this rule when other conditions are not met.
5856
Willy Tarreau303c0352008-01-17 19:01:39 +01005857 Any header line matching extended regular expression <search> in the response
5858 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005859 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01005860 client.
5861
5862 Header transformations only apply to traffic which passes through HAProxy,
5863 and not to traffic generated by HAProxy, such as health-checks or error
5864 responses. Keep in mind that header names are not case-sensitive.
5865
5866 Example :
5867 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02005868 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01005869
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005870 See also: "rspadd", "rsprep", "reqdel", section 6 about HTTP header
5871 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005872
5873
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005874rspdeny <search> [{if | unless} <cond>]
5875rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005876 Block an HTTP response if a line matches a regular expression
5877 May be used in sections : defaults | frontend | listen | backend
5878 no | yes | yes | yes
5879 Arguments :
5880 <search> is the regular expression applied to HTTP headers and to the
5881 response line. This is an extended regular expression, so
5882 parenthesis grouping is supported and no preliminary backslash
5883 is required. Any space or known delimiter must be escaped using
5884 a backslash ('\'). The pattern applies to a full line at a time.
5885 The "rspdeny" keyword strictly matches case while "rspideny"
5886 ignores case.
5887
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005888 <cond> is an optional matching condition built from ACLs. It makes it
5889 possible to ignore this rule when other conditions are not met.
5890
Willy Tarreau303c0352008-01-17 19:01:39 +01005891 A response containing any line which matches extended regular expression
5892 <search> will mark the request as denied. The test applies both to the
5893 response line and to response headers. Keep in mind that header names are not
5894 case-sensitive.
5895
5896 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01005897 block the response before it reaches the client. If a response is denied, it
5898 will be replaced with an HTTP 502 error so that the client never retrieves
5899 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01005900
5901 It is easier, faster and more powerful to use ACLs to write access policies.
5902 Rspdeny should be avoided in new designs.
5903
5904 Example :
5905 # Ensure that no content type matching ms-word will leak
5906 rspideny ^Content-type:\.*/ms-word
5907
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005908 See also: "reqdeny", "acl", "block", section 6 about HTTP header manipulation
5909 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005910
5911
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005912rsprep <search> <string> [{if | unless} <cond>]
5913rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005914 Replace a regular expression with a string in an HTTP response line
5915 May be used in sections : defaults | frontend | listen | backend
5916 no | yes | yes | yes
5917 Arguments :
5918 <search> is the regular expression applied to HTTP headers and to the
5919 response line. This is an extended regular expression, so
5920 parenthesis grouping is supported and no preliminary backslash
5921 is required. Any space or known delimiter must be escaped using
5922 a backslash ('\'). The pattern applies to a full line at a time.
5923 The "rsprep" keyword strictly matches case while "rspirep"
5924 ignores case.
5925
5926 <string> is the complete line to be added. Any space or known delimiter
5927 must be escaped using a backslash ('\'). References to matched
5928 pattern groups are possible using the common \N form, with N
5929 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005930 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005931
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005932 <cond> is an optional matching condition built from ACLs. It makes it
5933 possible to ignore this rule when other conditions are not met.
5934
Willy Tarreau303c0352008-01-17 19:01:39 +01005935 Any line matching extended regular expression <search> in the response (both
5936 the response line and header lines) will be completely replaced with
5937 <string>. Most common use of this is to rewrite Location headers.
5938
5939 Header transformations only apply to traffic which passes through HAProxy,
5940 and not to traffic generated by HAProxy, such as health-checks or error
5941 responses. Note that for increased readability, it is suggested to add enough
5942 spaces between the request and the response. Keep in mind that header names
5943 are not case-sensitive.
5944
5945 Example :
5946 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
5947 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
5948
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005949 See also: "rspadd", "rspdel", "reqrep", section 6 about HTTP header
5950 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005951
5952
David du Colombier486df472011-03-17 10:40:26 +01005953server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005954 Declare a server in a backend
5955 May be used in sections : defaults | frontend | listen | backend
5956 no | no | yes | yes
5957 Arguments :
5958 <name> is the internal name assigned to this server. This name will
Cyril Bonté941a0c62012-10-15 19:44:24 +02005959 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05005960 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005961
David du Colombier486df472011-03-17 10:40:26 +01005962 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
5963 resolvable hostname is supported, but this name will be resolved
5964 during start-up. Address "0.0.0.0" or "*" has a special meaning.
5965 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02005966 address as the one from the client connection. This is useful in
5967 transparent proxy architectures where the client's connection is
5968 intercepted and haproxy must forward to the original destination
5969 address. This is more or less what the "transparent" keyword does
5970 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01005971 to report statistics. Optionally, an address family prefix may be
5972 used before the address to force the family regardless of the
5973 address format, which can be useful to specify a path to a unix
5974 socket with no slash ('/'). Currently supported prefixes are :
5975 - 'ipv4@' -> address is always IPv4
5976 - 'ipv6@' -> address is always IPv6
5977 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02005978 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01005979 Any part of the address string may reference any number of
5980 environment variables by preceding their name with a dollar
5981 sign ('$') and optionally enclosing them with braces ('{}'),
5982 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005983
Willy Tarreaub6205fd2012-09-24 12:27:33 +02005984 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005985 be sent to this port. If unset, the same port the client
5986 connected to will be used. The port may also be prefixed by a "+"
5987 or a "-". In this case, the server's port will be determined by
5988 adding this value to the client's port.
5989
5990 <param*> is a list of parameters for this server. The "server" keywords
5991 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005992 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005993
5994 Examples :
5995 server first 10.1.1.1:1080 cookie first check inter 1000
5996 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01005997 server transp ipv4@
Willy Tarreaudad36a32013-03-11 01:20:04 +01005998 server backup ${SRV_BACKUP}:1080 backup
5999 server www1_dc1 ${LAN_DC1}.101:80
6000 server www1_dc2 ${LAN_DC2}.101:80
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006001
Willy Tarreauf0c95fc2015-09-27 15:03:15 +02006002 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
6003 sun_path length is used for the address length. Some other programs
6004 such as socat use the string length only by default. Pass the option
6005 ",unix-tightsocklen=0" to any abstract socket definition in socat to
6006 make it compatible with HAProxy's.
6007
Mark Lamourinec2247f02012-01-04 13:02:01 -05006008 See also: "default-server", "http-send-name-header" and section 5 about
6009 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006010
6011
6012source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02006013source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006014source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006015 Set the source address for outgoing connections
6016 May be used in sections : defaults | frontend | listen | backend
6017 yes | no | yes | yes
6018 Arguments :
6019 <addr> is the IPv4 address HAProxy will bind to before connecting to a
6020 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01006021
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006022 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01006023 the most appropriate address to reach its destination. Optionally
6024 an address family prefix may be used before the address to force
6025 the family regardless of the address format, which can be useful
6026 to specify a path to a unix socket with no slash ('/'). Currently
6027 supported prefixes are :
6028 - 'ipv4@' -> address is always IPv4
6029 - 'ipv6@' -> address is always IPv6
6030 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02006031 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01006032 Any part of the address string may reference any number of
6033 environment variables by preceding their name with a dollar
6034 sign ('$') and optionally enclosing them with braces ('{}'),
6035 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006036
6037 <port> is an optional port. It is normally not needed but may be useful
6038 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02006039 the system will select a free port. Note that port ranges are not
6040 supported in the backend. If you want to force port ranges, you
6041 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006042
6043 <addr2> is the IP address to present to the server when connections are
6044 forwarded in full transparent proxy mode. This is currently only
6045 supported on some patched Linux kernels. When this address is
6046 specified, clients connecting to the server will be presented
6047 with this address, while health checks will still use the address
6048 <addr>.
6049
6050 <port2> is the optional port to present to the server when connections
6051 are forwarded in full transparent proxy mode (see <addr2> above).
6052 The default value of zero means the system will select a free
6053 port.
6054
Willy Tarreaubce70882009-09-07 11:51:47 +02006055 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
6056 This is the name of a comma-separated header list which can
6057 contain multiple IP addresses. By default, the last occurrence is
6058 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01006059 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02006060 by previous proxy, typically Stunnel. In order to use another
6061 occurrence from the last one, please see the <occ> parameter
6062 below. When the header (or occurrence) is not found, no binding
6063 is performed so that the proxy's default IP address is used. Also
6064 keep in mind that the header name is case insensitive, as for any
6065 HTTP header.
6066
6067 <occ> is the occurrence number of a value to be used in a multi-value
6068 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006069 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02006070 address. Positive values indicate a position from the first
6071 occurrence, 1 being the first one. Negative values indicate
6072 positions relative to the last one, -1 being the last one. This
6073 is helpful for situations where an X-Forwarded-For header is set
6074 at the entry point of an infrastructure and must be used several
6075 proxy layers away. When this value is not specified, -1 is
6076 assumed. Passing a zero here disables the feature.
6077
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006078 <name> is an optional interface name to which to bind to for outgoing
6079 traffic. On systems supporting this features (currently, only
6080 Linux), this allows one to bind all traffic to the server to
6081 this interface even if it is not the one the system would select
6082 based on routing tables. This should be used with extreme care.
6083 Note that using this option requires root privileges.
6084
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006085 The "source" keyword is useful in complex environments where a specific
6086 address only is allowed to connect to the servers. It may be needed when a
6087 private address must be used through a public gateway for instance, and it is
6088 known that the system cannot determine the adequate source address by itself.
6089
6090 An extension which is available on certain patched Linux kernels may be used
6091 through the "usesrc" optional keyword. It makes it possible to connect to the
6092 servers with an IP address which does not belong to the system itself. This
6093 is called "full transparent proxy mode". For this to work, the destination
6094 servers have to route their traffic back to this address through the machine
6095 running HAProxy, and IP forwarding must generally be enabled on this machine.
6096
6097 In this "full transparent proxy" mode, it is possible to force a specific IP
6098 address to be presented to the servers. This is not much used in fact. A more
6099 common use is to tell HAProxy to present the client's IP address. For this,
6100 there are two methods :
6101
6102 - present the client's IP and port addresses. This is the most transparent
6103 mode, but it can cause problems when IP connection tracking is enabled on
6104 the machine, because a same connection may be seen twice with different
6105 states. However, this solution presents the huge advantage of not
6106 limiting the system to the 64k outgoing address+port couples, because all
6107 of the client ranges may be used.
6108
6109 - present only the client's IP address and select a spare port. This
6110 solution is still quite elegant but slightly less transparent (downstream
6111 firewalls logs will not match upstream's). It also presents the downside
6112 of limiting the number of concurrent connections to the usual 64k ports.
6113 However, since the upstream and downstream ports are different, local IP
6114 connection tracking on the machine will not be upset by the reuse of the
6115 same session.
6116
6117 Note that depending on the transparent proxy technology used, it may be
6118 required to force the source address. In fact, cttproxy version 2 requires an
6119 IP address in <addr> above, and does not support setting of "0.0.0.0" as the
6120 IP address because it creates NAT entries which much match the exact outgoing
6121 address. Tproxy version 4 and some other kernel patches which work in pure
6122 forwarding mode generally will not have this limitation.
6123
6124 This option sets the default source for all servers in the backend. It may
6125 also be specified in a "defaults" section. Finer source address specification
6126 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006127 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006128
Baptiste Assmannea31f222015-07-17 21:59:42 +02006129 In order to work, "usesrc" requires root privileges.
6130
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006131 Examples :
6132 backend private
6133 # Connect to the servers using our 192.168.1.200 source address
6134 source 192.168.1.200
6135
6136 backend transparent_ssl1
6137 # Connect to the SSL farm from the client's source address
6138 source 192.168.1.200 usesrc clientip
6139
6140 backend transparent_ssl2
6141 # Connect to the SSL farm from the client's source address and port
6142 # not recommended if IP conntrack is present on the local machine.
6143 source 192.168.1.200 usesrc client
6144
6145 backend transparent_ssl3
6146 # Connect to the SSL farm from the client's source address. It
6147 # is more conntrack-friendly.
6148 source 192.168.1.200 usesrc clientip
6149
6150 backend transparent_smtp
6151 # Connect to the SMTP farm from the client's source address/port
6152 # with Tproxy version 4.
6153 source 0.0.0.0 usesrc clientip
6154
Willy Tarreaubce70882009-09-07 11:51:47 +02006155 backend transparent_http
6156 # Connect to the servers using the client's IP as seen by previous
6157 # proxy.
6158 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
6159
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006160 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006161 the Linux kernel on www.balabit.com, the "bind" keyword.
6162
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006163
Willy Tarreau844e3c52008-01-11 16:28:18 +01006164srvtimeout <timeout> (deprecated)
6165 Set the maximum inactivity time on the server side.
6166 May be used in sections : defaults | frontend | listen | backend
6167 yes | no | yes | yes
6168 Arguments :
6169 <timeout> is the timeout value specified in milliseconds by default, but
6170 can be in any other unit if the number is suffixed by the unit,
6171 as explained at the top of this document.
6172
6173 The inactivity timeout applies when the server is expected to acknowledge or
6174 send data. In HTTP mode, this timeout is particularly important to consider
6175 during the first phase of the server's response, when it has to send the
6176 headers, as it directly represents the server's processing time for the
6177 request. To find out what value to put there, it's often good to start with
6178 what would be considered as unacceptable response times, then check the logs
6179 to observe the response time distribution, and adjust the value accordingly.
6180
6181 The value is specified in milliseconds by default, but can be in any other
6182 unit if the number is suffixed by the unit, as specified at the top of this
6183 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
6184 recommended that the client timeout remains equal to the server timeout in
6185 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006186 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01006187 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreaud72758d2010-01-12 10:42:19 +01006188 seconds (eg: 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01006189
6190 This parameter is specific to backends, but can be specified once for all in
6191 "defaults" sections. This is in fact one of the easiest solutions not to
6192 forget about it. An unspecified timeout results in an infinite timeout, which
6193 is not recommended. Such a usage is accepted and works but reports a warning
6194 during startup because it may results in accumulation of expired sessions in
6195 the system if the system's timeouts are not configured either.
6196
6197 This parameter is provided for compatibility but is currently deprecated.
6198 Please use "timeout server" instead.
6199
Willy Tarreauce887fd2012-05-12 12:50:00 +02006200 See also : "timeout server", "timeout tunnel", "timeout client" and
6201 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01006202
6203
Cyril Bonté66c327d2010-10-12 00:14:37 +02006204stats admin { if | unless } <cond>
6205 Enable statistics admin level if/unless a condition is matched
6206 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006207 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02006208
6209 This statement enables the statistics admin level if/unless a condition is
6210 matched.
6211
6212 The admin level allows to enable/disable servers from the web interface. By
6213 default, statistics page is read-only for security reasons.
6214
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006215 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6216 unless you know what you do : memory is not shared between the
6217 processes, which can result in random behaviours.
6218
Cyril Bonté23b39d92011-02-10 22:54:44 +01006219 Currently, the POST request is limited to the buffer size minus the reserved
6220 buffer space, which means that if the list of servers is too long, the
6221 request won't be processed. It is recommended to alter few servers at a
6222 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006223
6224 Example :
6225 # statistics admin level only for localhost
6226 backend stats_localhost
6227 stats enable
6228 stats admin if LOCALHOST
6229
6230 Example :
6231 # statistics admin level always enabled because of the authentication
6232 backend stats_auth
6233 stats enable
6234 stats auth admin:AdMiN123
6235 stats admin if TRUE
6236
6237 Example :
6238 # statistics admin level depends on the authenticated user
6239 userlist stats-auth
6240 group admin users admin
6241 user admin insecure-password AdMiN123
6242 group readonly users haproxy
6243 user haproxy insecure-password haproxy
6244
6245 backend stats_auth
6246 stats enable
6247 acl AUTH http_auth(stats-auth)
6248 acl AUTH_ADMIN http_auth_group(stats-auth) admin
6249 stats http-request auth unless AUTH
6250 stats admin if AUTH_ADMIN
6251
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006252 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
6253 "bind-process", section 3.4 about userlists and section 7 about
6254 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006255
6256
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006257stats auth <user>:<passwd>
6258 Enable statistics with authentication and grant access to an account
6259 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006260 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006261 Arguments :
6262 <user> is a user name to grant access to
6263
6264 <passwd> is the cleartext password associated to this user
6265
6266 This statement enables statistics with default settings, and restricts access
6267 to declared users only. It may be repeated as many times as necessary to
6268 allow as many users as desired. When a user tries to access the statistics
6269 without a valid account, a "401 Forbidden" response will be returned so that
6270 the browser asks the user to provide a valid user and password. The real
6271 which will be returned to the browser is configurable using "stats realm".
6272
6273 Since the authentication method is HTTP Basic Authentication, the passwords
6274 circulate in cleartext on the network. Thus, it was decided that the
6275 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006276 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006277
6278 It is also possible to reduce the scope of the proxies which appear in the
6279 report using "stats scope".
6280
6281 Though this statement alone is enough to enable statistics reporting, it is
6282 recommended to set all other settings in order to avoid relying on default
6283 unobvious parameters.
6284
6285 Example :
6286 # public access (limited to this backend only)
6287 backend public_www
6288 server srv1 192.168.0.1:80
6289 stats enable
6290 stats hide-version
6291 stats scope .
6292 stats uri /admin?stats
6293 stats realm Haproxy\ Statistics
6294 stats auth admin1:AdMiN123
6295 stats auth admin2:AdMiN321
6296
6297 # internal monitoring access (unlimited)
6298 backend private_monitoring
6299 stats enable
6300 stats uri /admin?stats
6301 stats refresh 5s
6302
6303 See also : "stats enable", "stats realm", "stats scope", "stats uri"
6304
6305
6306stats enable
6307 Enable statistics reporting with default settings
6308 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006309 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006310 Arguments : none
6311
6312 This statement enables statistics reporting with default settings defined
6313 at build time. Unless stated otherwise, these settings are used :
6314 - stats uri : /haproxy?stats
6315 - stats realm : "HAProxy Statistics"
6316 - stats auth : no authentication
6317 - stats scope : no restriction
6318
6319 Though this statement alone is enough to enable statistics reporting, it is
6320 recommended to set all other settings in order to avoid relying on default
6321 unobvious parameters.
6322
6323 Example :
6324 # public access (limited to this backend only)
6325 backend public_www
6326 server srv1 192.168.0.1:80
6327 stats enable
6328 stats hide-version
6329 stats scope .
6330 stats uri /admin?stats
6331 stats realm Haproxy\ Statistics
6332 stats auth admin1:AdMiN123
6333 stats auth admin2:AdMiN321
6334
6335 # internal monitoring access (unlimited)
6336 backend private_monitoring
6337 stats enable
6338 stats uri /admin?stats
6339 stats refresh 5s
6340
6341 See also : "stats auth", "stats realm", "stats uri"
6342
6343
Willy Tarreaud63335a2010-02-26 12:56:52 +01006344stats hide-version
6345 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006346 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006347 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006348 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006349
Willy Tarreaud63335a2010-02-26 12:56:52 +01006350 By default, the stats page reports some useful status information along with
6351 the statistics. Among them is HAProxy's version. However, it is generally
6352 considered dangerous to report precise version to anyone, as it can help them
6353 target known weaknesses with specific attacks. The "stats hide-version"
6354 statement removes the version from the statistics report. This is recommended
6355 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006356
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006357 Though this statement alone is enough to enable statistics reporting, it is
6358 recommended to set all other settings in order to avoid relying on default
6359 unobvious parameters.
6360
Willy Tarreaud63335a2010-02-26 12:56:52 +01006361 Example :
6362 # public access (limited to this backend only)
6363 backend public_www
6364 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006365 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006366 stats hide-version
6367 stats scope .
6368 stats uri /admin?stats
6369 stats realm Haproxy\ Statistics
6370 stats auth admin1:AdMiN123
6371 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006372
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006373 # internal monitoring access (unlimited)
6374 backend private_monitoring
6375 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006376 stats uri /admin?stats
6377 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01006378
Willy Tarreaud63335a2010-02-26 12:56:52 +01006379 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006380
Willy Tarreau983e01e2010-01-11 18:42:06 +01006381
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02006382stats http-request { allow | deny | auth [realm <realm>] }
6383 [ { if | unless } <condition> ]
6384 Access control for statistics
6385
6386 May be used in sections: defaults | frontend | listen | backend
6387 no | no | yes | yes
6388
6389 As "http-request", these set of options allow to fine control access to
6390 statistics. Each option may be followed by if/unless and acl.
6391 First option with matched condition (or option without condition) is final.
6392 For "deny" a 403 error will be returned, for "allow" normal processing is
6393 performed, for "auth" a 401/407 error code is returned so the client
6394 should be asked to enter a username and password.
6395
6396 There is no fixed limit to the number of http-request statements per
6397 instance.
6398
6399 See also : "http-request", section 3.4 about userlists and section 7
6400 about ACL usage.
6401
6402
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006403stats realm <realm>
6404 Enable statistics and set authentication realm
6405 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006406 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006407 Arguments :
6408 <realm> is the name of the HTTP Basic Authentication realm reported to
6409 the browser. The browser uses it to display it in the pop-up
6410 inviting the user to enter a valid username and password.
6411
6412 The realm is read as a single word, so any spaces in it should be escaped
6413 using a backslash ('\').
6414
6415 This statement is useful only in conjunction with "stats auth" since it is
6416 only related to authentication.
6417
6418 Though this statement alone is enough to enable statistics reporting, it is
6419 recommended to set all other settings in order to avoid relying on default
6420 unobvious parameters.
6421
6422 Example :
6423 # public access (limited to this backend only)
6424 backend public_www
6425 server srv1 192.168.0.1:80
6426 stats enable
6427 stats hide-version
6428 stats scope .
6429 stats uri /admin?stats
6430 stats realm Haproxy\ Statistics
6431 stats auth admin1:AdMiN123
6432 stats auth admin2:AdMiN321
6433
6434 # internal monitoring access (unlimited)
6435 backend private_monitoring
6436 stats enable
6437 stats uri /admin?stats
6438 stats refresh 5s
6439
6440 See also : "stats auth", "stats enable", "stats uri"
6441
6442
6443stats refresh <delay>
6444 Enable statistics with automatic refresh
6445 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006446 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006447 Arguments :
6448 <delay> is the suggested refresh delay, specified in seconds, which will
6449 be returned to the browser consulting the report page. While the
6450 browser is free to apply any delay, it will generally respect it
6451 and refresh the page this every seconds. The refresh interval may
6452 be specified in any other non-default time unit, by suffixing the
6453 unit after the value, as explained at the top of this document.
6454
6455 This statement is useful on monitoring displays with a permanent page
6456 reporting the load balancer's activity. When set, the HTML report page will
6457 include a link "refresh"/"stop refresh" so that the user can select whether
6458 he wants automatic refresh of the page or not.
6459
6460 Though this statement alone is enough to enable statistics reporting, it is
6461 recommended to set all other settings in order to avoid relying on default
6462 unobvious parameters.
6463
6464 Example :
6465 # public access (limited to this backend only)
6466 backend public_www
6467 server srv1 192.168.0.1:80
6468 stats enable
6469 stats hide-version
6470 stats scope .
6471 stats uri /admin?stats
6472 stats realm Haproxy\ Statistics
6473 stats auth admin1:AdMiN123
6474 stats auth admin2:AdMiN321
6475
6476 # internal monitoring access (unlimited)
6477 backend private_monitoring
6478 stats enable
6479 stats uri /admin?stats
6480 stats refresh 5s
6481
6482 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6483
6484
6485stats scope { <name> | "." }
6486 Enable statistics and limit access scope
6487 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006488 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006489 Arguments :
6490 <name> is the name of a listen, frontend or backend section to be
6491 reported. The special name "." (a single dot) designates the
6492 section in which the statement appears.
6493
6494 When this statement is specified, only the sections enumerated with this
6495 statement will appear in the report. All other ones will be hidden. This
6496 statement may appear as many times as needed if multiple sections need to be
6497 reported. Please note that the name checking is performed as simple string
6498 comparisons, and that it is never checked that a give section name really
6499 exists.
6500
6501 Though this statement alone is enough to enable statistics reporting, it is
6502 recommended to set all other settings in order to avoid relying on default
6503 unobvious parameters.
6504
6505 Example :
6506 # public access (limited to this backend only)
6507 backend public_www
6508 server srv1 192.168.0.1:80
6509 stats enable
6510 stats hide-version
6511 stats scope .
6512 stats uri /admin?stats
6513 stats realm Haproxy\ Statistics
6514 stats auth admin1:AdMiN123
6515 stats auth admin2:AdMiN321
6516
6517 # internal monitoring access (unlimited)
6518 backend private_monitoring
6519 stats enable
6520 stats uri /admin?stats
6521 stats refresh 5s
6522
6523 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6524
Willy Tarreaud63335a2010-02-26 12:56:52 +01006525
Willy Tarreauc9705a12010-07-27 20:05:50 +02006526stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01006527 Enable reporting of a description on the statistics page.
6528 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006529 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006530
Willy Tarreauc9705a12010-07-27 20:05:50 +02006531 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01006532 description from global section is automatically used instead.
6533
6534 This statement is useful for users that offer shared services to their
6535 customers, where node or description should be different for each customer.
6536
6537 Though this statement alone is enough to enable statistics reporting, it is
6538 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006539 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006540
6541 Example :
6542 # internal monitoring access (unlimited)
6543 backend private_monitoring
6544 stats enable
6545 stats show-desc Master node for Europe, Asia, Africa
6546 stats uri /admin?stats
6547 stats refresh 5s
6548
6549 See also: "show-node", "stats enable", "stats uri" and "description" in
6550 global section.
6551
6552
6553stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02006554 Enable reporting additional information on the statistics page
6555 May be used in sections : defaults | frontend | listen | backend
6556 yes | yes | yes | yes
6557 Arguments : none
6558
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006559 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01006560 - cap: capabilities (proxy)
6561 - mode: one of tcp, http or health (proxy)
6562 - id: SNMP ID (proxy, socket, server)
6563 - IP (socket, server)
6564 - cookie (backend, server)
6565
6566 Though this statement alone is enough to enable statistics reporting, it is
6567 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006568 unobvious parameters. Default behaviour is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006569
6570 See also: "stats enable", "stats uri".
6571
6572
6573stats show-node [ <name> ]
6574 Enable reporting of a host name on the statistics page.
6575 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006576 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006577 Arguments:
6578 <name> is an optional name to be reported. If unspecified, the
6579 node name from global section is automatically used instead.
6580
6581 This statement is useful for users that offer shared services to their
6582 customers, where node or description might be different on a stats page
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006583 provided for each customer. Default behaviour is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006584
6585 Though this statement alone is enough to enable statistics reporting, it is
6586 recommended to set all other settings in order to avoid relying on default
6587 unobvious parameters.
6588
6589 Example:
6590 # internal monitoring access (unlimited)
6591 backend private_monitoring
6592 stats enable
6593 stats show-node Europe-1
6594 stats uri /admin?stats
6595 stats refresh 5s
6596
6597 See also: "show-desc", "stats enable", "stats uri", and "node" in global
6598 section.
6599
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006600
6601stats uri <prefix>
6602 Enable statistics and define the URI prefix to access them
6603 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006604 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006605 Arguments :
6606 <prefix> is the prefix of any URI which will be redirected to stats. This
6607 prefix may contain a question mark ('?') to indicate part of a
6608 query string.
6609
6610 The statistics URI is intercepted on the relayed traffic, so it appears as a
6611 page within the normal application. It is strongly advised to ensure that the
6612 selected URI will never appear in the application, otherwise it will never be
6613 possible to reach it in the application.
6614
6615 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006616 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006617 It is generally a good idea to include a question mark in the URI so that
6618 intermediate proxies refrain from caching the results. Also, since any string
6619 beginning with the prefix will be accepted as a stats request, the question
6620 mark helps ensuring that no valid URI will begin with the same words.
6621
6622 It is sometimes very convenient to use "/" as the URI prefix, and put that
6623 statement in a "listen" instance of its own. That makes it easy to dedicate
6624 an address or a port to statistics only.
6625
6626 Though this statement alone is enough to enable statistics reporting, it is
6627 recommended to set all other settings in order to avoid relying on default
6628 unobvious parameters.
6629
6630 Example :
6631 # public access (limited to this backend only)
6632 backend public_www
6633 server srv1 192.168.0.1:80
6634 stats enable
6635 stats hide-version
6636 stats scope .
6637 stats uri /admin?stats
6638 stats realm Haproxy\ Statistics
6639 stats auth admin1:AdMiN123
6640 stats auth admin2:AdMiN321
6641
6642 # internal monitoring access (unlimited)
6643 backend private_monitoring
6644 stats enable
6645 stats uri /admin?stats
6646 stats refresh 5s
6647
6648 See also : "stats auth", "stats enable", "stats realm"
6649
6650
Willy Tarreaud63335a2010-02-26 12:56:52 +01006651stick match <pattern> [table <table>] [{if | unless} <cond>]
6652 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006653 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01006654 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006655
6656 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006657 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006658 describes what elements of the incoming request or connection
6659 will be analysed in the hope to find a matching entry in a
6660 stickiness table. This rule is mandatory.
6661
6662 <table> is an optional stickiness table name. If unspecified, the same
6663 backend's table is used. A stickiness table is declared using
6664 the "stick-table" statement.
6665
6666 <cond> is an optional matching condition. It makes it possible to match
6667 on a certain criterion only when other conditions are met (or
6668 not met). For instance, it could be used to match on a source IP
6669 address except when a request passes through a known proxy, in
6670 which case we'd match on a header containing that IP address.
6671
6672 Some protocols or applications require complex stickiness rules and cannot
6673 always simply rely on cookies nor hashing. The "stick match" statement
6674 describes a rule to extract the stickiness criterion from an incoming request
6675 or connection. See section 7 for a complete list of possible patterns and
6676 transformation rules.
6677
6678 The table has to be declared using the "stick-table" statement. It must be of
6679 a type compatible with the pattern. By default it is the one which is present
6680 in the same backend. It is possible to share a table with other backends by
6681 referencing it using the "table" keyword. If another table is referenced,
6682 the server's ID inside the backends are used. By default, all server IDs
6683 start at 1 in each backend, so the server ordering is enough. But in case of
6684 doubt, it is highly recommended to force server IDs using their "id" setting.
6685
6686 It is possible to restrict the conditions where a "stick match" statement
6687 will apply, using "if" or "unless" followed by a condition. See section 7 for
6688 ACL based conditions.
6689
6690 There is no limit on the number of "stick match" statements. The first that
6691 applies and matches will cause the request to be directed to the same server
6692 as was used for the request which created the entry. That way, multiple
6693 matches can be used as fallbacks.
6694
6695 The stick rules are checked after the persistence cookies, so they will not
6696 affect stickiness if a cookie has already been used to select a server. That
6697 way, it becomes very easy to insert cookies and match on IP addresses in
6698 order to maintain stickiness between HTTP and HTTPS.
6699
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006700 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6701 unless you know what you do : memory is not shared between the
6702 processes, which can result in random behaviours.
6703
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006704 Example :
6705 # forward SMTP users to the same server they just used for POP in the
6706 # last 30 minutes
6707 backend pop
6708 mode tcp
6709 balance roundrobin
6710 stick store-request src
6711 stick-table type ip size 200k expire 30m
6712 server s1 192.168.1.1:110
6713 server s2 192.168.1.1:110
6714
6715 backend smtp
6716 mode tcp
6717 balance roundrobin
6718 stick match src table pop
6719 server s1 192.168.1.1:25
6720 server s2 192.168.1.1:25
6721
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006722 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006723 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006724
6725
6726stick on <pattern> [table <table>] [{if | unless} <condition>]
6727 Define a request pattern to associate a user to a server
6728 May be used in sections : defaults | frontend | listen | backend
6729 no | no | yes | yes
6730
6731 Note : This form is exactly equivalent to "stick match" followed by
6732 "stick store-request", all with the same arguments. Please refer
6733 to both keywords for details. It is only provided as a convenience
6734 for writing more maintainable configurations.
6735
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006736 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6737 unless you know what you do : memory is not shared between the
6738 processes, which can result in random behaviours.
6739
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006740 Examples :
6741 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01006742 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006743
6744 # ...is strictly equivalent to this one :
6745 stick match src table pop if !localhost
6746 stick store-request src table pop if !localhost
6747
6748
6749 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
6750 # well as HTTP without cookie. Share the same table between both accesses.
6751 backend http
6752 mode http
6753 balance roundrobin
6754 stick on src table https
6755 cookie SRV insert indirect nocache
6756 server s1 192.168.1.1:80 cookie s1
6757 server s2 192.168.1.1:80 cookie s2
6758
6759 backend https
6760 mode tcp
6761 balance roundrobin
6762 stick-table type ip size 200k expire 30m
6763 stick on src
6764 server s1 192.168.1.1:443
6765 server s2 192.168.1.1:443
6766
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006767 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006768
6769
6770stick store-request <pattern> [table <table>] [{if | unless} <condition>]
6771 Define a request pattern used to create an entry in a stickiness table
6772 May be used in sections : defaults | frontend | listen | backend
6773 no | no | yes | yes
6774
6775 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006776 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006777 describes what elements of the incoming request or connection
6778 will be analysed, extracted and stored in the table once a
6779 server is selected.
6780
6781 <table> is an optional stickiness table name. If unspecified, the same
6782 backend's table is used. A stickiness table is declared using
6783 the "stick-table" statement.
6784
6785 <cond> is an optional storage condition. It makes it possible to store
6786 certain criteria only when some conditions are met (or not met).
6787 For instance, it could be used to store the source IP address
6788 except when the request passes through a known proxy, in which
6789 case we'd store a converted form of a header containing that IP
6790 address.
6791
6792 Some protocols or applications require complex stickiness rules and cannot
6793 always simply rely on cookies nor hashing. The "stick store-request" statement
6794 describes a rule to decide what to extract from the request and when to do
6795 it, in order to store it into a stickiness table for further requests to
6796 match it using the "stick match" statement. Obviously the extracted part must
6797 make sense and have a chance to be matched in a further request. Storing a
6798 client's IP address for instance often makes sense. Storing an ID found in a
6799 URL parameter also makes sense. Storing a source port will almost never make
6800 any sense because it will be randomly matched. See section 7 for a complete
6801 list of possible patterns and transformation rules.
6802
6803 The table has to be declared using the "stick-table" statement. It must be of
6804 a type compatible with the pattern. By default it is the one which is present
6805 in the same backend. It is possible to share a table with other backends by
6806 referencing it using the "table" keyword. If another table is referenced,
6807 the server's ID inside the backends are used. By default, all server IDs
6808 start at 1 in each backend, so the server ordering is enough. But in case of
6809 doubt, it is highly recommended to force server IDs using their "id" setting.
6810
6811 It is possible to restrict the conditions where a "stick store-request"
6812 statement will apply, using "if" or "unless" followed by a condition. This
6813 condition will be evaluated while parsing the request, so any criteria can be
6814 used. See section 7 for ACL based conditions.
6815
6816 There is no limit on the number of "stick store-request" statements, but
6817 there is a limit of 8 simultaneous stores per request or response. This
6818 makes it possible to store up to 8 criteria, all extracted from either the
6819 request or the response, regardless of the number of rules. Only the 8 first
6820 ones which match will be kept. Using this, it is possible to feed multiple
6821 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01006822 another protocol or access method. Using multiple store-request rules with
6823 the same table is possible and may be used to find the best criterion to rely
6824 on, by arranging the rules by decreasing preference order. Only the first
6825 extracted criterion for a given table will be stored. All subsequent store-
6826 request rules referencing the same table will be skipped and their ACLs will
6827 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006828
6829 The "store-request" rules are evaluated once the server connection has been
6830 established, so that the table will contain the real server that processed
6831 the request.
6832
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006833 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6834 unless you know what you do : memory is not shared between the
6835 processes, which can result in random behaviours.
6836
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006837 Example :
6838 # forward SMTP users to the same server they just used for POP in the
6839 # last 30 minutes
6840 backend pop
6841 mode tcp
6842 balance roundrobin
6843 stick store-request src
6844 stick-table type ip size 200k expire 30m
6845 server s1 192.168.1.1:110
6846 server s2 192.168.1.1:110
6847
6848 backend smtp
6849 mode tcp
6850 balance roundrobin
6851 stick match src table pop
6852 server s1 192.168.1.1:25
6853 server s2 192.168.1.1:25
6854
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006855 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006856 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006857
6858
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006859stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02006860 size <size> [expire <expire>] [nopurge] [peers <peersect>]
6861 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08006862 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006863 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02006864 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006865
6866 Arguments :
6867 ip a table declared with "type ip" will only store IPv4 addresses.
6868 This form is very compact (about 50 bytes per entry) and allows
6869 very fast entry lookup and stores with almost no overhead. This
6870 is mainly used to store client source IP addresses.
6871
David du Colombier9a6d3c92011-03-17 10:40:24 +01006872 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
6873 This form is very compact (about 60 bytes per entry) and allows
6874 very fast entry lookup and stores with almost no overhead. This
6875 is mainly used to store client source IP addresses.
6876
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006877 integer a table declared with "type integer" will store 32bit integers
6878 which can represent a client identifier found in a request for
6879 instance.
6880
6881 string a table declared with "type string" will store substrings of up
6882 to <len> characters. If the string provided by the pattern
6883 extractor is larger than <len>, it will be truncated before
6884 being stored. During matching, at most <len> characters will be
6885 compared between the string in the table and the extracted
6886 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006887 to 32 characters.
6888
6889 binary a table declared with "type binary" will store binary blocks
6890 of <len> bytes. If the block provided by the pattern
6891 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02006892 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006893 is shorter than <len>, it will be padded by 0. When not
6894 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006895
6896 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006897 "string" type table (See type "string" above). Or the number
6898 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006899 changing this parameter as memory usage will proportionally
6900 increase.
6901
6902 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01006903 value directly impacts memory usage. Count approximately
6904 50 bytes per entry, plus the size of a string if any. The size
6905 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006906
6907 [nopurge] indicates that we refuse to purge older entries when the table
6908 is full. When not specified and the table is full when haproxy
6909 wants to store an entry in it, it will flush a few of the oldest
6910 entries in order to release some space for the new ones. This is
6911 most often the desired behaviour. In some specific cases, it
6912 be desirable to refuse new entries instead of purging the older
6913 ones. That may be the case when the amount of data to store is
6914 far above the hardware limits and we prefer not to offer access
6915 to new clients than to reject the ones already connected. When
6916 using this parameter, be sure to properly set the "expire"
6917 parameter (see below).
6918
Emeric Brunf099e792010-09-27 12:05:28 +02006919 <peersect> is the name of the peers section to use for replication. Entries
6920 which associate keys to server IDs are kept synchronized with
6921 the remote peers declared in this section. All entries are also
6922 automatically learned from the local peer (old process) during a
6923 soft restart.
6924
Willy Tarreauc85ad792015-05-01 19:21:02 +02006925 NOTE : each peers section may be referenced only by tables
6926 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006927
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006928 <expire> defines the maximum duration of an entry in the table since it
6929 was last created, refreshed or matched. The expiration delay is
6930 defined using the standard time format, similarly as the various
6931 timeouts. The maximum duration is slightly above 24 days. See
6932 section 2.2 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02006933 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006934 be removed once full. Be sure not to use the "nopurge" parameter
6935 if not expiration delay is specified.
6936
Willy Tarreau08d5f982010-06-06 13:34:54 +02006937 <data_type> is used to store additional information in the stick-table. This
6938 may be used by ACLs in order to control various criteria related
6939 to the activity of the client matching the stick-table. For each
6940 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02006941 that the additional data can fit. Several data types may be
6942 stored with an entry. Multiple data types may be specified after
6943 the "store" keyword, as a comma-separated list. Alternatively,
6944 it is possible to repeat the "store" keyword followed by one or
6945 several data types. Except for the "server_id" type which is
6946 automatically detected and enabled, all data types must be
6947 explicitly declared to be stored. If an ACL references a data
6948 type which is not stored, the ACL will simply not match. Some
6949 data types require an argument which must be passed just after
6950 the type between parenthesis. See below for the supported data
6951 types and their arguments.
6952
6953 The data types that can be stored with an entry are the following :
6954 - server_id : this is an integer which holds the numeric ID of the server a
6955 request was assigned to. It is used by the "stick match", "stick store",
6956 and "stick on" rules. It is automatically enabled when referenced.
6957
6958 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
6959 integer which may be used for anything. Most of the time it will be used
6960 to put a special tag on some entries, for instance to note that a
6961 specific behaviour was detected and must be known for future matches.
6962
Willy Tarreauba2ffd12013-05-29 15:54:14 +02006963 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
6964 over a period. It is a positive 32-bit integer integer which may be used
6965 for anything. Just like <gpc0>, it counts events, but instead of keeping
6966 a cumulative count, it maintains the rate at which the counter is
6967 incremented. Most of the time it will be used to measure the frequency of
6968 occurrence of certain events (eg: requests to a specific URL).
6969
Willy Tarreauc9705a12010-07-27 20:05:50 +02006970 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
6971 the absolute number of connections received from clients which matched
6972 this entry. It does not mean the connections were accepted, just that
6973 they were received.
6974
6975 - conn_cur : Current Connections. It is a positive 32-bit integer which
6976 stores the concurrent connection counts for the entry. It is incremented
6977 once an incoming connection matches the entry, and decremented once the
6978 connection leaves. That way it is possible to know at any time the exact
6979 number of concurrent connections for an entry.
6980
6981 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6982 integer parameter <period> which indicates in milliseconds the length
6983 of the period over which the average is measured. It reports the average
6984 incoming connection rate over that period, in connections per period. The
6985 result is an integer which can be matched using ACLs.
6986
6987 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
6988 the absolute number of sessions received from clients which matched this
6989 entry. A session is a connection that was accepted by the layer 4 rules.
6990
6991 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6992 integer parameter <period> which indicates in milliseconds the length
6993 of the period over which the average is measured. It reports the average
6994 incoming session rate over that period, in sessions per period. The
6995 result is an integer which can be matched using ACLs.
6996
6997 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
6998 counts the absolute number of HTTP requests received from clients which
6999 matched this entry. It does not matter whether they are valid requests or
7000 not. Note that this is different from sessions when keep-alive is used on
7001 the client side.
7002
7003 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7004 integer parameter <period> which indicates in milliseconds the length
7005 of the period over which the average is measured. It reports the average
7006 HTTP request rate over that period, in requests per period. The result is
7007 an integer which can be matched using ACLs. It does not matter whether
7008 they are valid requests or not. Note that this is different from sessions
7009 when keep-alive is used on the client side.
7010
7011 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
7012 counts the absolute number of HTTP requests errors induced by clients
7013 which matched this entry. Errors are counted on invalid and truncated
7014 requests, as well as on denied or tarpitted requests, and on failed
7015 authentications. If the server responds with 4xx, then the request is
7016 also counted as an error since it's an error triggered by the client
7017 (eg: vulnerability scan).
7018
7019 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7020 integer parameter <period> which indicates in milliseconds the length
7021 of the period over which the average is measured. It reports the average
7022 HTTP request error rate over that period, in requests per period (see
7023 http_err_cnt above for what is accounted as an error). The result is an
7024 integer which can be matched using ACLs.
7025
7026 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
7027 integer which counts the cumulated amount of bytes received from clients
7028 which matched this entry. Headers are included in the count. This may be
7029 used to limit abuse of upload features on photo or video servers.
7030
7031 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7032 integer parameter <period> which indicates in milliseconds the length
7033 of the period over which the average is measured. It reports the average
7034 incoming bytes rate over that period, in bytes per period. It may be used
7035 to detect users which upload too much and too fast. Warning: with large
7036 uploads, it is possible that the amount of uploaded data will be counted
7037 once upon termination, thus causing spikes in the average transfer speed
7038 instead of having a smooth one. This may partially be smoothed with
7039 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
7040 recommended for better fairness.
7041
7042 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
7043 integer which counts the cumulated amount of bytes sent to clients which
7044 matched this entry. Headers are included in the count. This may be used
7045 to limit abuse of bots sucking the whole site.
7046
7047 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
7048 an integer parameter <period> which indicates in milliseconds the length
7049 of the period over which the average is measured. It reports the average
7050 outgoing bytes rate over that period, in bytes per period. It may be used
7051 to detect users which download too much and too fast. Warning: with large
7052 transfers, it is possible that the amount of transferred data will be
7053 counted once upon termination, thus causing spikes in the average
7054 transfer speed instead of having a smooth one. This may partially be
7055 smoothed with "option contstats" though this is not perfect yet. Use of
7056 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02007057
Willy Tarreauc00cdc22010-06-06 16:48:26 +02007058 There is only one stick-table per proxy. At the moment of writing this doc,
7059 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007060 to be required, simply create a dummy backend with a stick-table in it and
7061 reference it.
7062
7063 It is important to understand that stickiness based on learning information
7064 has some limitations, including the fact that all learned associations are
7065 lost upon restart. In general it can be good as a complement but not always
7066 as an exclusive stickiness.
7067
Willy Tarreauc9705a12010-07-27 20:05:50 +02007068 Last, memory requirements may be important when storing many data types.
7069 Indeed, storing all indicators above at once in each entry requires 116 bytes
7070 per entry, or 116 MB for a 1-million entries table. This is definitely not
7071 something that can be ignored.
7072
7073 Example:
7074 # Keep track of counters of up to 1 million IP addresses over 5 minutes
7075 # and store a general purpose counter and the average connection rate
7076 # computed over a sliding window of 30 seconds.
7077 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
7078
7079 See also : "stick match", "stick on", "stick store-request", section 2.2
David du Colombiera13d1b92011-03-17 10:40:22 +01007080 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007081
7082
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007083stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmannfd9aad02016-03-06 23:27:24 +01007084 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007085 May be used in sections : defaults | frontend | listen | backend
7086 no | no | yes | yes
7087
7088 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007089 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007090 describes what elements of the response or connection will
7091 be analysed, extracted and stored in the table once a
7092 server is selected.
7093
7094 <table> is an optional stickiness table name. If unspecified, the same
7095 backend's table is used. A stickiness table is declared using
7096 the "stick-table" statement.
7097
7098 <cond> is an optional storage condition. It makes it possible to store
7099 certain criteria only when some conditions are met (or not met).
7100 For instance, it could be used to store the SSL session ID only
7101 when the response is a SSL server hello.
7102
7103 Some protocols or applications require complex stickiness rules and cannot
7104 always simply rely on cookies nor hashing. The "stick store-response"
7105 statement describes a rule to decide what to extract from the response and
7106 when to do it, in order to store it into a stickiness table for further
7107 requests to match it using the "stick match" statement. Obviously the
7108 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007109 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007110 See section 7 for a complete list of possible patterns and transformation
7111 rules.
7112
7113 The table has to be declared using the "stick-table" statement. It must be of
7114 a type compatible with the pattern. By default it is the one which is present
7115 in the same backend. It is possible to share a table with other backends by
7116 referencing it using the "table" keyword. If another table is referenced,
7117 the server's ID inside the backends are used. By default, all server IDs
7118 start at 1 in each backend, so the server ordering is enough. But in case of
7119 doubt, it is highly recommended to force server IDs using their "id" setting.
7120
7121 It is possible to restrict the conditions where a "stick store-response"
7122 statement will apply, using "if" or "unless" followed by a condition. This
7123 condition will be evaluated while parsing the response, so any criteria can
7124 be used. See section 7 for ACL based conditions.
7125
7126 There is no limit on the number of "stick store-response" statements, but
7127 there is a limit of 8 simultaneous stores per request or response. This
7128 makes it possible to store up to 8 criteria, all extracted from either the
7129 request or the response, regardless of the number of rules. Only the 8 first
7130 ones which match will be kept. Using this, it is possible to feed multiple
7131 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01007132 another protocol or access method. Using multiple store-response rules with
7133 the same table is possible and may be used to find the best criterion to rely
7134 on, by arranging the rules by decreasing preference order. Only the first
7135 extracted criterion for a given table will be stored. All subsequent store-
7136 response rules referencing the same table will be skipped and their ACLs will
7137 not be evaluated. However, even if a store-request rule references a table, a
7138 store-response rule may also use the same table. This means that each table
7139 may learn exactly one element from the request and one element from the
7140 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007141
7142 The table will contain the real server that processed the request.
7143
7144 Example :
7145 # Learn SSL session ID from both request and response and create affinity.
7146 backend https
7147 mode tcp
7148 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02007149 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007150 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007151
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007152 acl clienthello req_ssl_hello_type 1
7153 acl serverhello rep_ssl_hello_type 2
7154
7155 # use tcp content accepts to detects ssl client and server hello.
7156 tcp-request inspect-delay 5s
7157 tcp-request content accept if clienthello
7158
7159 # no timeout on response inspect delay by default.
7160 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007161
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007162 # SSL session ID (SSLID) may be present on a client or server hello.
7163 # Its length is coded on 1 byte at offset 43 and its value starts
7164 # at offset 44.
7165
7166 # Match and learn on request if client hello.
7167 stick on payload_lv(43,1) if clienthello
7168
7169 # Learn on response if server hello.
7170 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02007171
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007172 server s1 192.168.1.1:443
7173 server s2 192.168.1.1:443
7174
7175 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
7176 extraction.
7177
7178
Willy Tarreau938c7fe2014-04-25 14:21:39 +02007179tcp-check connect [params*]
7180 Opens a new connection
7181 May be used in sections: defaults | frontend | listen | backend
7182 no | no | yes | yes
7183
7184 When an application lies on more than a single TCP port or when HAProxy
7185 load-balance many services in a single backend, it makes sense to probe all
7186 the services individually before considering a server as operational.
7187
7188 When there are no TCP port configured on the server line neither server port
7189 directive, then the 'tcp-check connect port <port>' must be the first step
7190 of the sequence.
7191
7192 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
7193 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
7194 do.
7195
7196 Parameters :
7197 They are optional and can be used to describe how HAProxy should open and
7198 use the TCP connection.
7199
7200 port if not set, check port or server port is used.
7201 It tells HAProxy where to open the connection to.
7202 <port> must be a valid TCP port source integer, from 1 to 65535.
7203
7204 send-proxy send a PROXY protocol string
7205
7206 ssl opens a ciphered connection
7207
7208 Examples:
7209 # check HTTP and HTTPs services on a server.
7210 # first open port 80 thanks to server line port directive, then
7211 # tcp-check opens port 443, ciphered and run a request on it:
7212 option tcp-check
7213 tcp-check connect
7214 tcp-check send GET\ /\ HTTP/1.0\r\n
7215 tcp-check send Host:\ haproxy.1wt.eu\r\n
7216 tcp-check send \r\n
7217 tcp-check expect rstring (2..|3..)
7218 tcp-check connect port 443 ssl
7219 tcp-check send GET\ /\ HTTP/1.0\r\n
7220 tcp-check send Host:\ haproxy.1wt.eu\r\n
7221 tcp-check send \r\n
7222 tcp-check expect rstring (2..|3..)
7223 server www 10.0.0.1 check port 80
7224
7225 # check both POP and IMAP from a single server:
7226 option tcp-check
7227 tcp-check connect port 110
7228 tcp-check expect string +OK\ POP3\ ready
7229 tcp-check connect port 143
7230 tcp-check expect string *\ OK\ IMAP4\ ready
7231 server mail 10.0.0.1 check
7232
7233 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
7234
7235
7236tcp-check expect [!] <match> <pattern>
7237 Specify data to be collected and analysed during a generic health check
7238 May be used in sections: defaults | frontend | listen | backend
7239 no | no | yes | yes
7240
7241 Arguments :
7242 <match> is a keyword indicating how to look for a specific pattern in the
7243 response. The keyword may be one of "string", "rstring" or
7244 binary.
7245 The keyword may be preceded by an exclamation mark ("!") to negate
7246 the match. Spaces are allowed between the exclamation mark and the
7247 keyword. See below for more details on the supported keywords.
7248
7249 <pattern> is the pattern to look for. It may be a string or a regular
7250 expression. If the pattern contains spaces, they must be escaped
7251 with the usual backslash ('\').
7252 If the match is set to binary, then the pattern must be passed as
7253 a serie of hexadecimal digits in an even number. Each sequence of
7254 two digits will represent a byte. The hexadecimal digits may be
7255 used upper or lower case.
7256
7257
7258 The available matches are intentionally similar to their http-check cousins :
7259
7260 string <string> : test the exact string matches in the response buffer.
7261 A health check response will be considered valid if the
7262 response's buffer contains this exact string. If the
7263 "string" keyword is prefixed with "!", then the response
7264 will be considered invalid if the body contains this
7265 string. This can be used to look for a mandatory pattern
7266 in a protocol response, or to detect a failure when a
7267 specific error appears in a protocol banner.
7268
7269 rstring <regex> : test a regular expression on the response buffer.
7270 A health check response will be considered valid if the
7271 response's buffer matches this expression. If the
7272 "rstring" keyword is prefixed with "!", then the response
7273 will be considered invalid if the body matches the
7274 expression.
7275
7276 binary <hexstring> : test the exact string in its hexadecimal form matches
7277 in the response buffer. A health check response will
7278 be considered valid if the response's buffer contains
7279 this exact hexadecimal string.
7280 Purpose is to match data on binary protocols.
7281
7282 It is important to note that the responses will be limited to a certain size
7283 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
7284 Thus, too large responses may not contain the mandatory pattern when using
7285 "string", "rstring" or binary. If a large response is absolutely required, it
7286 is possible to change the default max size by setting the global variable.
7287 However, it is worth keeping in mind that parsing very large responses can
7288 waste some CPU cycles, especially when regular expressions are used, and that
7289 it is always better to focus the checks on smaller resources. Also, in its
7290 current state, the check will not find any string nor regex past a null
7291 character in the response. Similarly it is not possible to request matching
7292 the null character.
7293
7294 Examples :
7295 # perform a POP check
7296 option tcp-check
7297 tcp-check expect string +OK\ POP3\ ready
7298
7299 # perform an IMAP check
7300 option tcp-check
7301 tcp-check expect string *\ OK\ IMAP4\ ready
7302
7303 # look for the redis master server
7304 option tcp-check
7305 tcp-check send PING\r\n
7306 tcp-check expect +PONG
7307 tcp-check send info\ replication\r\n
7308 tcp-check expect string role:master
7309 tcp-check send QUIT\r\n
7310 tcp-check expect string +OK
7311
7312
7313 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
7314 "tcp-check send-binary", "http-check expect", tune.chksize
7315
7316
7317tcp-check send <data>
7318 Specify a string to be sent as a question during a generic health check
7319 May be used in sections: defaults | frontend | listen | backend
7320 no | no | yes | yes
7321
7322 <data> : the data to be sent as a question during a generic health check
7323 session. For now, <data> must be a string.
7324
7325 Examples :
7326 # look for the redis master server
7327 option tcp-check
7328 tcp-check send info\ replication\r\n
7329 tcp-check expect string role:master
7330
7331 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7332 "tcp-check send-binary", tune.chksize
7333
7334
7335tcp-check send-binary <hexastring>
7336 Specify an hexa digits string to be sent as a binary question during a raw
7337 tcp health check
7338 May be used in sections: defaults | frontend | listen | backend
7339 no | no | yes | yes
7340
7341 <data> : the data to be sent as a question during a generic health check
7342 session. For now, <data> must be a string.
7343 <hexastring> : test the exact string in its hexadecimal form matches in the
7344 response buffer. A health check response will be considered
7345 valid if the response's buffer contains this exact
7346 hexadecimal string.
7347 Purpose is to send binary data to ask on binary protocols.
7348
7349 Examples :
7350 # redis check in binary
7351 option tcp-check
7352 tcp-check send-binary 50494e470d0a # PING\r\n
7353 tcp-check expect binary 2b504F4e47 # +PONG
7354
7355
7356 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7357 "tcp-check send", tune.chksize
7358
7359
Willy Tarreaue9656522010-08-17 15:40:09 +02007360tcp-request connection <action> [{if | unless} <condition>]
7361 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02007362 May be used in sections : defaults | frontend | listen | backend
7363 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02007364 Arguments :
7365 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007366 actions include : "accept", "reject", "track-sc0", "track-sc1",
7367 "track-sc2", and "expect-proxy". See below for more details.
Willy Tarreau1a687942010-05-23 22:40:30 +02007368
Willy Tarreaue9656522010-08-17 15:40:09 +02007369 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007370
7371 Immediately after acceptance of a new incoming connection, it is possible to
7372 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02007373 or dropped or have its counters tracked. Those conditions cannot make use of
7374 any data contents because the connection has not been read from yet, and the
7375 buffers are not yet allocated. This is used to selectively and very quickly
7376 accept or drop connections from various sources with a very low overhead. If
7377 some contents need to be inspected in order to take the decision, the
7378 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007379
Willy Tarreaue9656522010-08-17 15:40:09 +02007380 The "tcp-request connection" rules are evaluated in their exact declaration
7381 order. If no rule matches or if there is no rule, the default action is to
7382 accept the incoming connection. There is no specific limit to the number of
7383 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007384
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007385 Five types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02007386 - accept :
7387 accepts the connection if the condition is true (when used with "if")
7388 or false (when used with "unless"). The first such rule executed ends
7389 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007390
Willy Tarreaue9656522010-08-17 15:40:09 +02007391 - reject :
7392 rejects the connection if the condition is true (when used with "if")
7393 or false (when used with "unless"). The first such rule executed ends
7394 the rules evaluation. Rejected connections do not even become a
7395 session, which is why they are accounted separately for in the stats,
7396 as "denied connections". They are not considered for the session
7397 rate-limit and are not logged either. The reason is that these rules
7398 should only be used to filter extremely high connection rates such as
7399 the ones encountered during a massive DDoS attack. Under these extreme
7400 conditions, the simple action of logging each event would make the
7401 system collapse and would considerably lower the filtering capacity. If
7402 logging is absolutely desired, then "tcp-request content" rules should
7403 be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007404
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007405 - expect-proxy layer4 :
7406 configures the client-facing connection to receive a PROXY protocol
7407 header before any byte is read from the socket. This is equivalent to
7408 having the "accept-proxy" keyword on the "bind" line, except that using
7409 the TCP rule allows the PROXY protocol to be accepted only for certain
7410 IP address ranges using an ACL. This is convenient when multiple layers
7411 of load balancers are passed through by traffic coming from public
7412 hosts.
7413
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007414 - capture <sample> len <length> :
7415 This only applies to "tcp-request content" rules. It captures sample
7416 expression <sample> from the request buffer, and converts it to a
7417 string of at most <len> characters. The resulting string is stored into
7418 the next request "capture" slot, so it will possibly appear next to
7419 some captured HTTP headers. It will then automatically appear in the
7420 logs, and it will be possible to extract it using sample fetch rules to
7421 feed it into headers or anything. The length should be limited given
7422 that this size will be allocated for each capture during the whole
7423 session life. Since it applies to Please check section 7.3 (Fetching
7424 samples) and "capture request header" for more information.
7425
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007426 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02007427 enables tracking of sticky counters from current connection. These
7428 rules do not stop evaluation and do not change default action. Two sets
7429 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007430 first "track-sc0" rule executed enables tracking of the counters of the
7431 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02007432 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007433 set. The first "track-sc2" rule executed enables tracking of the
7434 counters of the specified table as the third set. It is a recommended
7435 practice to use the first set of counters for the per-frontend counters
7436 and the second set for the per-backend ones. But this is just a
7437 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007438
Willy Tarreaue9656522010-08-17 15:40:09 +02007439 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007440 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02007441 in section 7.3. It describes what elements of the incoming
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007442 request or connection will be analysed, extracted, combined,
7443 and used to select which table entry to update the counters.
7444 Note that "tcp-request connection" cannot use content-based
7445 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007446
Willy Tarreaue9656522010-08-17 15:40:09 +02007447 <table> is an optional table to be used instead of the default one,
7448 which is the stick-table declared in the current proxy. All
7449 the counters for the matches and updates for the key will
7450 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007451
Willy Tarreaue9656522010-08-17 15:40:09 +02007452 Once a "track-sc*" rule is executed, the key is looked up in the table
7453 and if it is not found, an entry is allocated for it. Then a pointer to
7454 that entry is kept during all the session's life, and this entry's
7455 counters are updated as often as possible, every time the session's
7456 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007457 Counters are only updated for events that happen after the tracking has
7458 been started. For example, connection counters will not be updated when
7459 tracking layer 7 information, since the connection event happens before
7460 layer7 information is extracted.
7461
Willy Tarreaue9656522010-08-17 15:40:09 +02007462 If the entry tracks concurrent connection counters, one connection is
7463 counted for as long as the entry is tracked, and the entry will not
7464 expire during that time. Tracking counters also provides a performance
7465 advantage over just checking the keys, because only one table lookup is
7466 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007467
Willy Tarreaue9656522010-08-17 15:40:09 +02007468 Note that the "if/unless" condition is optional. If no condition is set on
7469 the action, it is simply performed unconditionally. That can be useful for
7470 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007471
Willy Tarreaue9656522010-08-17 15:40:09 +02007472 Example: accept all connections from white-listed hosts, reject too fast
7473 connection without counting them, and track accepted connections.
7474 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007475
Willy Tarreaue9656522010-08-17 15:40:09 +02007476 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007477 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007478 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007479
Willy Tarreaue9656522010-08-17 15:40:09 +02007480 Example: accept all connections from white-listed hosts, count all other
7481 connections and reject too fast ones. This results in abusive ones
7482 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007483
Willy Tarreaue9656522010-08-17 15:40:09 +02007484 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007485 tcp-request connection track-sc0 src
7486 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007487
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007488 Example: enable the PROXY protocol for traffic coming from all known proxies.
7489
7490 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
7491
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007492 See section 7 about ACL usage.
7493
Willy Tarreaue9656522010-08-17 15:40:09 +02007494 See also : "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007495
7496
Willy Tarreaue9656522010-08-17 15:40:09 +02007497tcp-request content <action> [{if | unless} <condition>]
7498 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02007499 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007500 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02007501 Arguments :
7502 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007503 actions include : "accept", "reject", "track-sc0", "track-sc1",
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007504 "track-sc2" and "capture". See "tcp-request connection" above
7505 for their signification.
Willy Tarreau62644772008-07-16 18:36:06 +02007506
Willy Tarreaue9656522010-08-17 15:40:09 +02007507 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02007508
Willy Tarreaue9656522010-08-17 15:40:09 +02007509 A request's contents can be analysed at an early stage of request processing
7510 called "TCP content inspection". During this stage, ACL-based rules are
7511 evaluated every time the request contents are updated, until either an
7512 "accept" or a "reject" rule matches, or the TCP request inspection delay
7513 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02007514
Willy Tarreaue9656522010-08-17 15:40:09 +02007515 The first difference between these rules and "tcp-request connection" rules
7516 is that "tcp-request content" rules can make use of contents to take a
7517 decision. Most often, these decisions will consider a protocol recognition or
7518 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01007519 both frontends and backends. In case of HTTP keep-alive with the client, all
7520 tcp-request content rules are evaluated again, so haproxy keeps a record of
7521 what sticky counters were assigned by a "tcp-request connection" versus a
7522 "tcp-request content" rule, and flushes all the content-related ones after
7523 processing an HTTP request, so that they may be evaluated again by the rules
7524 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007525 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01007526 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02007527
Willy Tarreaue9656522010-08-17 15:40:09 +02007528 Content-based rules are evaluated in their exact declaration order. If no
7529 rule matches or if there is no rule, the default action is to accept the
7530 contents. There is no specific limit to the number of rules which may be
7531 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02007532
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007533 Four types of actions are supported :
7534 - accept : the request is accepted
7535 - reject : the request is rejected and the connection is closed
7536 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007537 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Willy Tarreau62644772008-07-16 18:36:06 +02007538
Willy Tarreaue9656522010-08-17 15:40:09 +02007539 They have the same meaning as their counter-parts in "tcp-request connection"
7540 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02007541
Willy Tarreauf3338342014-01-28 21:40:28 +01007542 While there is nothing mandatory about it, it is recommended to use the
7543 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
7544 content" rules in the frontend, and track-sc2 for "tcp-request content"
7545 rules in the backend, because that makes the configuration more readable
7546 and easier to troubleshoot, but this is just a guideline and all counters
7547 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02007548
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007549 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02007550 the action, it is simply performed unconditionally. That can be useful for
7551 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02007552
Willy Tarreaue9656522010-08-17 15:40:09 +02007553 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02007554 rules, since HTTP-specific ACL matches are able to preliminarily parse the
7555 contents of a buffer before extracting the required data. If the buffered
7556 contents do not parse as a valid HTTP message, then the ACL does not match.
7557 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01007558 processing, so there is no risk of parsing something differently. In an HTTP
7559 backend connected to from an HTTP frontend, it is guaranteed that HTTP
7560 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02007561
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007562 Tracking layer7 information is also possible provided that the information
Willy Tarreau531485c2014-09-16 15:48:15 +02007563 are present when the rule is processed. The rule processing engine is able to
7564 wait until the inspect delay expires when the data to be tracked is not yet
7565 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007566
Willy Tarreau62644772008-07-16 18:36:06 +02007567 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02007568 # Accept HTTP requests containing a Host header saying "example.com"
7569 # and reject everything else.
7570 acl is_host_com hdr(Host) -i example.com
7571 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02007572 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02007573 tcp-request content reject
7574
7575 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02007576 # reject SMTP connection if client speaks first
7577 tcp-request inspect-delay 30s
7578 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007579 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02007580
7581 # Forward HTTPS connection only if client speaks
7582 tcp-request inspect-delay 30s
7583 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007584 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02007585 tcp-request content reject
7586
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007587 Example:
7588 # Track the last IP from X-Forwarded-For
7589 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007590 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007591
7592 Example:
7593 # track request counts per "base" (concatenation of Host+URL)
7594 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007595 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007596
Willy Tarreaue9656522010-08-17 15:40:09 +02007597 Example: track per-frontend and per-backend counters, block abusers at the
7598 frontend when the backend detects abuse.
7599
7600 frontend http
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007601 # Use General Purpose Couter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02007602 # protecting all our sites
7603 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007604 tcp-request connection track-sc0 src
7605 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02007606 ...
7607 use_backend http_dynamic if { path_end .php }
7608
7609 backend http_dynamic
7610 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007611 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02007612 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007613 acl click_too_fast sc1_http_req_rate gt 10
7614 acl mark_as_abuser sc0_inc_gpc0 gt 0
7615 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02007616 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02007617
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007618 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02007619
Willy Tarreaue9656522010-08-17 15:40:09 +02007620 See also : "tcp-request connection", "tcp-request inspect-delay"
Willy Tarreau62644772008-07-16 18:36:06 +02007621
7622
7623tcp-request inspect-delay <timeout>
7624 Set the maximum allowed time to wait for data during content inspection
7625 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007626 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02007627 Arguments :
7628 <timeout> is the timeout value specified in milliseconds by default, but
7629 can be in any other unit if the number is suffixed by the unit,
7630 as explained at the top of this document.
7631
7632 People using haproxy primarily as a TCP relay are often worried about the
7633 risk of passing any type of protocol to a server without any analysis. In
7634 order to be able to analyze the request contents, we must first withhold
7635 the data then analyze them. This statement simply enables withholding of
7636 data for at most the specified amount of time.
7637
Willy Tarreaufb356202010-08-03 14:02:05 +02007638 TCP content inspection applies very early when a connection reaches a
7639 frontend, then very early when the connection is forwarded to a backend. This
7640 means that a connection may experience a first delay in the frontend and a
7641 second delay in the backend if both have tcp-request rules.
7642
Willy Tarreau62644772008-07-16 18:36:06 +02007643 Note that when performing content inspection, haproxy will evaluate the whole
7644 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007645 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02007646 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01007647 contents are definitive. If no delay is set, haproxy will not wait at all
7648 and will immediately apply a verdict based on the available information.
7649 Obviously this is unlikely to be very useful and might even be racy, so such
7650 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02007651
7652 As soon as a rule matches, the request is released and continues as usual. If
7653 the timeout is reached and no rule matches, the default policy will be to let
7654 it pass through unaffected.
7655
7656 For most protocols, it is enough to set it to a few seconds, as most clients
7657 send the full request immediately upon connection. Add 3 or more seconds to
7658 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01007659 to use large values, for instance to ensure that the client never talks
Willy Tarreau62644772008-07-16 18:36:06 +02007660 before the server (eg: SMTP), or to wait for a client to talk before passing
7661 data to the server (eg: SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02007662 least the inspection delay, otherwise it will expire first. If the client
7663 closes the connection or if the buffer is full, the delay immediately expires
7664 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02007665
Willy Tarreau55165fe2009-05-10 12:02:55 +02007666 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02007667 "timeout client".
7668
7669
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007670tcp-response content <action> [{if | unless} <condition>]
7671 Perform an action on a session response depending on a layer 4-7 condition
7672 May be used in sections : defaults | frontend | listen | backend
7673 no | no | yes | yes
7674 Arguments :
7675 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007676 actions include : "accept", "close", "reject".
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007677
7678 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
7679
7680 Response contents can be analysed at an early stage of response processing
7681 called "TCP content inspection". During this stage, ACL-based rules are
7682 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007683 "accept", "close" or a "reject" rule matches, or a TCP response inspection
7684 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007685
7686 Most often, these decisions will consider a protocol recognition or validity.
7687
7688 Content-based rules are evaluated in their exact declaration order. If no
7689 rule matches or if there is no rule, the default action is to accept the
7690 contents. There is no specific limit to the number of rules which may be
7691 inserted.
7692
7693 Two types of actions are supported :
7694 - accept :
7695 accepts the response if the condition is true (when used with "if")
7696 or false (when used with "unless"). The first such rule executed ends
7697 the rules evaluation.
7698
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007699 - close :
7700 immediately closes the connection with the server if the condition is
7701 true (when used with "if"), or false (when used with "unless"). The
7702 first such rule executed ends the rules evaluation. The main purpose of
7703 this action is to force a connection to be finished between a client
7704 and a server after an exchange when the application protocol expects
7705 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007706 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007707 protocols.
7708
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007709 - reject :
7710 rejects the response if the condition is true (when used with "if")
7711 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007712 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007713
7714 Note that the "if/unless" condition is optional. If no condition is set on
7715 the action, it is simply performed unconditionally. That can be useful for
7716 for changing the default action to a reject.
7717
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007718 It is perfectly possible to match layer 7 contents with "tcp-response
7719 content" rules, but then it is important to ensure that a full response has
7720 been buffered, otherwise no contents will match. In order to achieve this,
7721 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007722 period.
7723
7724 See section 7 about ACL usage.
7725
7726 See also : "tcp-request content", "tcp-response inspect-delay"
7727
7728
7729tcp-response inspect-delay <timeout>
7730 Set the maximum allowed time to wait for a response during content inspection
7731 May be used in sections : defaults | frontend | listen | backend
7732 no | no | yes | yes
7733 Arguments :
7734 <timeout> is the timeout value specified in milliseconds by default, but
7735 can be in any other unit if the number is suffixed by the unit,
7736 as explained at the top of this document.
7737
7738 See also : "tcp-response content", "tcp-request inspect-delay".
7739
7740
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007741timeout check <timeout>
7742 Set additional check timeout, but only after a connection has been already
7743 established.
7744
7745 May be used in sections: defaults | frontend | listen | backend
7746 yes | no | yes | yes
7747 Arguments:
7748 <timeout> is the timeout value specified in milliseconds by default, but
7749 can be in any other unit if the number is suffixed by the unit,
7750 as explained at the top of this document.
7751
7752 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
7753 for check and "timeout check" as an additional read timeout. The "min" is
7754 used so that people running with *very* long "timeout connect" (eg. those
7755 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01007756 (Please also note that there is no valid reason to have such long connect
7757 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
7758 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007759
7760 If "timeout check" is not set haproxy uses "inter" for complete check
7761 timeout (connect + read) exactly like all <1.3.15 version.
7762
7763 In most cases check request is much simpler and faster to handle than normal
7764 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01007765 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007766
7767 This parameter is specific to backends, but can be specified once for all in
7768 "defaults" sections. This is in fact one of the easiest solutions not to
7769 forget about it.
7770
Willy Tarreau41a340d2008-01-22 12:25:31 +01007771 See also: "timeout connect", "timeout queue", "timeout server",
7772 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007773
7774
Willy Tarreau0ba27502007-12-24 16:55:16 +01007775timeout client <timeout>
7776timeout clitimeout <timeout> (deprecated)
7777 Set the maximum inactivity time on the client side.
7778 May be used in sections : defaults | frontend | listen | backend
7779 yes | yes | yes | no
7780 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007781 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007782 can be in any other unit if the number is suffixed by the unit,
7783 as explained at the top of this document.
7784
7785 The inactivity timeout applies when the client is expected to acknowledge or
7786 send data. In HTTP mode, this timeout is particularly important to consider
7787 during the first phase, when the client sends the request, and during the
Baptiste Assmannfdecd032016-03-06 23:24:12 +01007788 response while it is reading data sent by the server. That said, for the
7789 first phase, it is preferable to set the "timeout http-request" to better
7790 protect HAProxy from Slowloris like attacks. The value is specified in
7791 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +01007792 suffixed by the unit, as specified at the top of this document. In TCP mode
7793 (and to a lesser extent, in HTTP mode), it is highly recommended that the
7794 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007795 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01007796 losses by specifying timeouts that are slightly above multiples of 3 seconds
Willy Tarreauce887fd2012-05-12 12:50:00 +02007797 (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
7798 sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02007799 which overrides "timeout client" and "timeout server" for tunnels, as well as
7800 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007801
7802 This parameter is specific to frontends, but can be specified once for all in
7803 "defaults" sections. This is in fact one of the easiest solutions not to
7804 forget about it. An unspecified timeout results in an infinite timeout, which
7805 is not recommended. Such a usage is accepted and works but reports a warning
7806 during startup because it may results in accumulation of expired sessions in
7807 the system if the system's timeouts are not configured either.
7808
7809 This parameter replaces the old, deprecated "clitimeout". It is recommended
7810 to use it to write new configurations. The form "timeout clitimeout" is
7811 provided only by backwards compatibility but its use is strongly discouraged.
7812
Baptiste Assmannfdecd032016-03-06 23:24:12 +01007813 See also : "clitimeout", "timeout server", "timeout tunnel",
7814 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007815
7816
Willy Tarreau05cdd962014-05-10 14:30:07 +02007817timeout client-fin <timeout>
7818 Set the inactivity timeout on the client side for half-closed connections.
7819 May be used in sections : defaults | frontend | listen | backend
7820 yes | yes | yes | no
7821 Arguments :
7822 <timeout> is the timeout value specified in milliseconds by default, but
7823 can be in any other unit if the number is suffixed by the unit,
7824 as explained at the top of this document.
7825
7826 The inactivity timeout applies when the client is expected to acknowledge or
7827 send data while one direction is already shut down. This timeout is different
7828 from "timeout client" in that it only applies to connections which are closed
7829 in one direction. This is particularly useful to avoid keeping connections in
7830 FIN_WAIT state for too long when clients do not disconnect cleanly. This
7831 problem is particularly common long connections such as RDP or WebSocket.
7832 Note that this timeout can override "timeout tunnel" when a connection shuts
7833 down in one direction.
7834
7835 This parameter is specific to frontends, but can be specified once for all in
7836 "defaults" sections. By default it is not set, so half-closed connections
7837 will use the other timeouts (timeout.client or timeout.tunnel).
7838
7839 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
7840
7841
Willy Tarreau0ba27502007-12-24 16:55:16 +01007842timeout connect <timeout>
7843timeout contimeout <timeout> (deprecated)
7844 Set the maximum time to wait for a connection attempt to a server to succeed.
7845 May be used in sections : defaults | frontend | listen | backend
7846 yes | no | yes | yes
7847 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007848 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007849 can be in any other unit if the number is suffixed by the unit,
7850 as explained at the top of this document.
7851
7852 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007853 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01007854 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01007855 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007856 connect timeout also presets both queue and tarpit timeouts to the same value
7857 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007858
7859 This parameter is specific to backends, but can be specified once for all in
7860 "defaults" sections. This is in fact one of the easiest solutions not to
7861 forget about it. An unspecified timeout results in an infinite timeout, which
7862 is not recommended. Such a usage is accepted and works but reports a warning
7863 during startup because it may results in accumulation of failed sessions in
7864 the system if the system's timeouts are not configured either.
7865
7866 This parameter replaces the old, deprecated "contimeout". It is recommended
7867 to use it to write new configurations. The form "timeout contimeout" is
7868 provided only by backwards compatibility but its use is strongly discouraged.
7869
Willy Tarreau41a340d2008-01-22 12:25:31 +01007870 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
7871 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007872
7873
Willy Tarreaub16a5742010-01-10 14:46:16 +01007874timeout http-keep-alive <timeout>
7875 Set the maximum allowed time to wait for a new HTTP request to appear
7876 May be used in sections : defaults | frontend | listen | backend
7877 yes | yes | yes | yes
7878 Arguments :
7879 <timeout> is the timeout value specified in milliseconds by default, but
7880 can be in any other unit if the number is suffixed by the unit,
7881 as explained at the top of this document.
7882
7883 By default, the time to wait for a new request in case of keep-alive is set
7884 by "timeout http-request". However this is not always convenient because some
7885 people want very short keep-alive timeouts in order to release connections
7886 faster, and others prefer to have larger ones but still have short timeouts
7887 once the request has started to present itself.
7888
7889 The "http-keep-alive" timeout covers these needs. It will define how long to
7890 wait for a new HTTP request to start coming after a response was sent. Once
7891 the first byte of request has been seen, the "http-request" timeout is used
7892 to wait for the complete request to come. Note that empty lines prior to a
7893 new request do not refresh the timeout and are not counted as a new request.
7894
7895 There is also another difference between the two timeouts : when a connection
7896 expires during timeout http-keep-alive, no error is returned, the connection
7897 just closes. If the connection expires in "http-request" while waiting for a
7898 connection to complete, a HTTP 408 error is returned.
7899
7900 In general it is optimal to set this value to a few tens to hundreds of
7901 milliseconds, to allow users to fetch all objects of a page at once but
7902 without waiting for further clicks. Also, if set to a very small value (eg:
7903 1 millisecond) it will probably only accept pipelined requests but not the
7904 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +02007905 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +01007906
7907 If this parameter is not set, the "http-request" timeout applies, and if both
7908 are not set, "timeout client" still applies at the lower level. It should be
7909 set in the frontend to take effect, unless the frontend is in TCP mode, in
7910 which case the HTTP backend's timeout will be used.
7911
7912 See also : "timeout http-request", "timeout client".
7913
7914
Willy Tarreau036fae02008-01-06 13:24:40 +01007915timeout http-request <timeout>
7916 Set the maximum allowed time to wait for a complete HTTP request
7917 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007918 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +01007919 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007920 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +01007921 can be in any other unit if the number is suffixed by the unit,
7922 as explained at the top of this document.
7923
7924 In order to offer DoS protection, it may be required to lower the maximum
7925 accepted time to receive a complete HTTP request without affecting the client
7926 timeout. This helps protecting against established connections on which
7927 nothing is sent. The client timeout cannot offer a good protection against
7928 this abuse because it is an inactivity timeout, which means that if the
7929 attacker sends one character every now and then, the timeout will not
7930 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +02007931 types, the request will be aborted if it does not complete in time. When the
7932 timeout expires, an HTTP 408 response is sent to the client to inform it
7933 about the problem, and the connection is closed. The logs will report
7934 termination codes "cR". Some recent browsers are having problems with this
7935 standard, well-documented behaviour, so it might be needed to hide the 408
Willy Tarreau1c3a6122015-05-01 15:37:53 +02007936 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
7937 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +01007938
7939 Note that this timeout only applies to the header part of the request, and
7940 not to any data. As soon as the empty line is received, this timeout is not
Willy Tarreaub16a5742010-01-10 14:46:16 +01007941 used anymore. It is used again on keep-alive connections to wait for a second
7942 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +01007943
7944 Generally it is enough to set it to a few seconds, as most clients send the
7945 full request immediately upon connection. Add 3 or more seconds to cover TCP
7946 retransmits but that's all. Setting it to very low values (eg: 50 ms) will
7947 generally work on local networks as long as there are no packet losses. This
7948 will prevent people from sending bare HTTP requests using telnet.
7949
7950 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007951 chunk of the incoming request. It should be set in the frontend to take
7952 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
7953 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +01007954
Willy Tarreau1c3a6122015-05-01 15:37:53 +02007955 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
7956 "timeout client".
Willy Tarreau036fae02008-01-06 13:24:40 +01007957
Willy Tarreau844e3c52008-01-11 16:28:18 +01007958
7959timeout queue <timeout>
7960 Set the maximum time to wait in the queue for a connection slot to be free
7961 May be used in sections : defaults | frontend | listen | backend
7962 yes | no | yes | yes
7963 Arguments :
7964 <timeout> is the timeout value specified in milliseconds by default, but
7965 can be in any other unit if the number is suffixed by the unit,
7966 as explained at the top of this document.
7967
7968 When a server's maxconn is reached, connections are left pending in a queue
7969 which may be server-specific or global to the backend. In order not to wait
7970 indefinitely, a timeout is applied to requests pending in the queue. If the
7971 timeout is reached, it is considered that the request will almost never be
7972 served, so it is dropped and a 503 error is returned to the client.
7973
7974 The "timeout queue" statement allows to fix the maximum time for a request to
7975 be left pending in a queue. If unspecified, the same value as the backend's
7976 connection timeout ("timeout connect") is used, for backwards compatibility
7977 with older versions with no "timeout queue" parameter.
7978
7979 See also : "timeout connect", "contimeout".
7980
7981
7982timeout server <timeout>
7983timeout srvtimeout <timeout> (deprecated)
7984 Set the maximum inactivity time on the server side.
7985 May be used in sections : defaults | frontend | listen | backend
7986 yes | no | yes | yes
7987 Arguments :
7988 <timeout> is the timeout value specified in milliseconds by default, but
7989 can be in any other unit if the number is suffixed by the unit,
7990 as explained at the top of this document.
7991
7992 The inactivity timeout applies when the server is expected to acknowledge or
7993 send data. In HTTP mode, this timeout is particularly important to consider
7994 during the first phase of the server's response, when it has to send the
7995 headers, as it directly represents the server's processing time for the
7996 request. To find out what value to put there, it's often good to start with
7997 what would be considered as unacceptable response times, then check the logs
7998 to observe the response time distribution, and adjust the value accordingly.
7999
8000 The value is specified in milliseconds by default, but can be in any other
8001 unit if the number is suffixed by the unit, as specified at the top of this
8002 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8003 recommended that the client timeout remains equal to the server timeout in
8004 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008005 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008006 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreauce887fd2012-05-12 12:50:00 +02008007 seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed
8008 with short-lived sessions (eg: WebSocket and HTTP), it's worth considering
8009 "timeout tunnel", which overrides "timeout client" and "timeout server" for
8010 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008011
8012 This parameter is specific to backends, but can be specified once for all in
8013 "defaults" sections. This is in fact one of the easiest solutions not to
8014 forget about it. An unspecified timeout results in an infinite timeout, which
8015 is not recommended. Such a usage is accepted and works but reports a warning
8016 during startup because it may results in accumulation of expired sessions in
8017 the system if the system's timeouts are not configured either.
8018
8019 This parameter replaces the old, deprecated "srvtimeout". It is recommended
8020 to use it to write new configurations. The form "timeout srvtimeout" is
8021 provided only by backwards compatibility but its use is strongly discouraged.
8022
Willy Tarreauce887fd2012-05-12 12:50:00 +02008023 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008024
Willy Tarreau05cdd962014-05-10 14:30:07 +02008025
8026timeout server-fin <timeout>
8027 Set the inactivity timeout on the server side for half-closed connections.
8028 May be used in sections : defaults | frontend | listen | backend
8029 yes | no | yes | yes
8030 Arguments :
8031 <timeout> is the timeout value specified in milliseconds by default, but
8032 can be in any other unit if the number is suffixed by the unit,
8033 as explained at the top of this document.
8034
8035 The inactivity timeout applies when the server is expected to acknowledge or
8036 send data while one direction is already shut down. This timeout is different
8037 from "timeout server" in that it only applies to connections which are closed
8038 in one direction. This is particularly useful to avoid keeping connections in
8039 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
8040 This problem is particularly common long connections such as RDP or WebSocket.
8041 Note that this timeout can override "timeout tunnel" when a connection shuts
8042 down in one direction. This setting was provided for completeness, but in most
8043 situations, it should not be needed.
8044
8045 This parameter is specific to backends, but can be specified once for all in
8046 "defaults" sections. By default it is not set, so half-closed connections
8047 will use the other timeouts (timeout.server or timeout.tunnel).
8048
8049 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
8050
Willy Tarreau844e3c52008-01-11 16:28:18 +01008051
8052timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +01008053 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +01008054 May be used in sections : defaults | frontend | listen | backend
8055 yes | yes | yes | yes
8056 Arguments :
8057 <timeout> is the tarpit duration specified in milliseconds by default, but
8058 can be in any other unit if the number is suffixed by the unit,
8059 as explained at the top of this document.
8060
8061 When a connection is tarpitted using "reqtarpit", it is maintained open with
8062 no activity for a certain amount of time, then closed. "timeout tarpit"
8063 defines how long it will be maintained open.
8064
8065 The value is specified in milliseconds by default, but can be in any other
8066 unit if the number is suffixed by the unit, as specified at the top of this
8067 document. If unspecified, the same value as the backend's connection timeout
8068 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +01008069 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008070
8071 See also : "timeout connect", "contimeout".
8072
8073
Willy Tarreauce887fd2012-05-12 12:50:00 +02008074timeout tunnel <timeout>
8075 Set the maximum inactivity time on the client and server side for tunnels.
8076 May be used in sections : defaults | frontend | listen | backend
8077 yes | no | yes | yes
8078 Arguments :
8079 <timeout> is the timeout value specified in milliseconds by default, but
8080 can be in any other unit if the number is suffixed by the unit,
8081 as explained at the top of this document.
8082
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008083 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +02008084 between a client and a server, and the connection remains inactive in both
8085 directions. This timeout supersedes both the client and server timeouts once
8086 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
8087 analyser remains attached to either connection (eg: tcp content rules are
8088 accepted). In HTTP, this timeout is used when a connection is upgraded (eg:
8089 when switching to the WebSocket protocol, or forwarding a CONNECT request
8090 to a proxy), or after the first response when no keepalive/close option is
8091 specified.
8092
Willy Tarreau05cdd962014-05-10 14:30:07 +02008093 Since this timeout is usually used in conjunction with long-lived connections,
8094 it usually is a good idea to also set "timeout client-fin" to handle the
8095 situation where a client suddenly disappears from the net and does not
8096 acknowledge a close, or sends a shutdown and does not acknowledge pending
8097 data anymore. This can happen in lossy networks where firewalls are present,
8098 and is detected by the presence of large amounts of sessions in a FIN_WAIT
8099 state.
8100
Willy Tarreauce887fd2012-05-12 12:50:00 +02008101 The value is specified in milliseconds by default, but can be in any other
8102 unit if the number is suffixed by the unit, as specified at the top of this
8103 document. Whatever the expected normal idle time, it is a good practice to
8104 cover at least one or several TCP packet losses by specifying timeouts that
8105 are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum).
8106
8107 This parameter is specific to backends, but can be specified once for all in
8108 "defaults" sections. This is in fact one of the easiest solutions not to
8109 forget about it.
8110
8111 Example :
8112 defaults http
8113 option http-server-close
8114 timeout connect 5s
8115 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +02008116 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +02008117 timeout server 30s
8118 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
8119
Willy Tarreau05cdd962014-05-10 14:30:07 +02008120 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +02008121
8122
Willy Tarreau844e3c52008-01-11 16:28:18 +01008123transparent (deprecated)
8124 Enable client-side transparent proxying
8125 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008126 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008127 Arguments : none
8128
8129 This keyword was introduced in order to provide layer 7 persistence to layer
8130 3 load balancers. The idea is to use the OS's ability to redirect an incoming
8131 connection for a remote address to a local process (here HAProxy), and let
8132 this process know what address was initially requested. When this option is
8133 used, sessions without cookies will be forwarded to the original destination
8134 IP address of the incoming request (which should match that of another
8135 equipment), while requests with cookies will still be forwarded to the
8136 appropriate server.
8137
8138 The "transparent" keyword is deprecated, use "option transparent" instead.
8139
8140 Note that contrary to a common belief, this option does NOT make HAProxy
8141 present the client's IP to the server when establishing the connection.
8142
Willy Tarreau844e3c52008-01-11 16:28:18 +01008143 See also: "option transparent"
8144
William Lallemanda73203e2012-03-12 12:48:57 +01008145unique-id-format <string>
8146 Generate a unique ID for each request.
8147 May be used in sections : defaults | frontend | listen | backend
8148 yes | yes | yes | no
8149 Arguments :
8150 <string> is a log-format string.
8151
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008152 This keyword creates a ID for each request using the custom log format. A
8153 unique ID is useful to trace a request passing through many components of
8154 a complex infrastructure. The newly created ID may also be logged using the
8155 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +01008156
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008157 The format should be composed from elements that are guaranteed to be
8158 unique when combined together. For instance, if multiple haproxy instances
8159 are involved, it might be important to include the node name. It is often
8160 needed to log the incoming connection's source and destination addresses
8161 and ports. Note that since multiple requests may be performed over the same
8162 connection, including a request counter may help differentiate them.
8163 Similarly, a timestamp may protect against a rollover of the counter.
8164 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +01008165
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008166 It is recommended to use hexadecimal notation for many fields since it
8167 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +01008168
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008169 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008170
Julien Vehentf21be322014-03-07 08:27:34 -05008171 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008172
8173 will generate:
8174
8175 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8176
8177 See also: "unique-id-header"
8178
8179unique-id-header <name>
8180 Add a unique ID header in the HTTP request.
8181 May be used in sections : defaults | frontend | listen | backend
8182 yes | yes | yes | no
8183 Arguments :
8184 <name> is the name of the header.
8185
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008186 Add a unique-id header in the HTTP request sent to the server, using the
8187 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +01008188
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008189 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008190
Julien Vehentf21be322014-03-07 08:27:34 -05008191 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008192 unique-id-header X-Unique-ID
8193
8194 will generate:
8195
8196 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8197
8198 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +01008199
Willy Tarreauf51658d2014-04-23 01:21:56 +02008200use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008201 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008202 May be used in sections : defaults | frontend | listen | backend
8203 no | yes | yes | no
8204 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008205 <backend> is the name of a valid backend or "listen" section, or a
8206 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008207
Willy Tarreauf51658d2014-04-23 01:21:56 +02008208 <condition> is a condition composed of ACLs, as described in section 7. If
8209 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008210
8211 When doing content-switching, connections arrive on a frontend and are then
8212 dispatched to various backends depending on a number of conditions. The
8213 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008214 "use_backend" keyword. While it is normally used with HTTP processing, it can
8215 also be used in pure TCP, either without content using stateless ACLs (eg:
8216 source address validation) or combined with a "tcp-request" rule to wait for
8217 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008218
8219 There may be as many "use_backend" rules as desired. All of these rules are
8220 evaluated in their declaration order, and the first one which matches will
8221 assign the backend.
8222
8223 In the first form, the backend will be used if the condition is met. In the
8224 second form, the backend will be used if the condition is not met. If no
8225 condition is valid, the backend defined with "default_backend" will be used.
8226 If no default backend is defined, either the servers in the same section are
8227 used (in case of a "listen" section) or, in case of a frontend, no server is
8228 used and a 503 service unavailable response is returned.
8229
Willy Tarreau51aecc72009-07-12 09:47:04 +02008230 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008231 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +02008232 and backend processing will immediately follow, or the backend will wait for
8233 a complete HTTP request to get in. This feature is useful when a frontend
8234 must decode several protocols on a unique port, one of them being HTTP.
8235
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008236 When <backend> is a simple name, it is resolved at configuration time, and an
8237 error is reported if the specified backend does not exist. If <backend> is
8238 a log-format string instead, no check may be done at configuration time, so
8239 the backend name is resolved dynamically at run time. If the resulting
8240 backend name does not correspond to any valid backend, no other rule is
8241 evaluated, and the default_backend directive is applied instead. Note that
8242 when using dynamic backend names, it is highly recommended to use a prefix
8243 that no other backend uses in order to ensure that an unauthorized backend
8244 cannot be forced from the request.
8245
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008246 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008247 used to detect the association between frontends and backends to compute the
8248 backend's "fullconn" setting. This cannot be done for dynamic names.
8249
8250 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
8251 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008252
Willy Tarreau036fae02008-01-06 13:24:40 +01008253
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008254use-server <server> if <condition>
8255use-server <server> unless <condition>
8256 Only use a specific server if/unless an ACL-based condition is matched.
8257 May be used in sections : defaults | frontend | listen | backend
8258 no | no | yes | yes
8259 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008260 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008261
8262 <condition> is a condition composed of ACLs, as described in section 7.
8263
8264 By default, connections which arrive to a backend are load-balanced across
8265 the available servers according to the configured algorithm, unless a
8266 persistence mechanism such as a cookie is used and found in the request.
8267
8268 Sometimes it is desirable to forward a particular request to a specific
8269 server without having to declare a dedicated backend for this server. This
8270 can be achieved using the "use-server" rules. These rules are evaluated after
8271 the "redirect" rules and before evaluating cookies, and they have precedence
8272 on them. There may be as many "use-server" rules as desired. All of these
8273 rules are evaluated in their declaration order, and the first one which
8274 matches will assign the server.
8275
8276 If a rule designates a server which is down, and "option persist" is not used
8277 and no force-persist rule was validated, it is ignored and evaluation goes on
8278 with the next rules until one matches.
8279
8280 In the first form, the server will be used if the condition is met. In the
8281 second form, the server will be used if the condition is not met. If no
8282 condition is valid, the processing continues and the server will be assigned
8283 according to other persistence mechanisms.
8284
8285 Note that even if a rule is matched, cookie processing is still performed but
8286 does not assign the server. This allows prefixed cookies to have their prefix
8287 stripped.
8288
8289 The "use-server" statement works both in HTTP and TCP mode. This makes it
8290 suitable for use with content-based inspection. For instance, a server could
8291 be selected in a farm according to the TLS SNI field. And if these servers
8292 have their weight set to zero, they will not be used for other traffic.
8293
8294 Example :
8295 # intercept incoming TLS requests based on the SNI field
8296 use-server www if { req_ssl_sni -i www.example.com }
8297 server www 192.168.0.1:443 weight 0
8298 use-server mail if { req_ssl_sni -i mail.example.com }
8299 server mail 192.168.0.1:587 weight 0
8300 use-server imap if { req_ssl_sni -i imap.example.com }
8301 server mail 192.168.0.1:993 weight 0
8302 # all the rest is forwarded to this server
8303 server default 192.168.0.2:443 check
8304
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008305 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008306
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008307
83085. Bind and Server options
8309--------------------------
8310
8311The "bind", "server" and "default-server" keywords support a number of settings
8312depending on some build options and on the system HAProxy was built on. These
8313settings generally each consist in one word sometimes followed by a value,
8314written on the same line as the "bind" or "server" line. All these options are
8315described in this section.
8316
8317
83185.1. Bind options
8319-----------------
8320
8321The "bind" keyword supports a certain number of settings which are all passed
8322as arguments on the same line. The order in which those arguments appear makes
8323no importance, provided that they appear after the bind address. All of these
8324parameters are optional. Some of them consist in a single words (booleans),
8325while other ones expect a value after them. In this case, the value must be
8326provided immediately after the setting name.
8327
8328The currently supported settings are the following ones.
8329
8330accept-proxy
8331 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +02008332 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
8333 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008334 3/4 addresses of the incoming connection to be used everywhere an address is
8335 used, with the only exception of "tcp-request connection" rules which will
8336 only see the real connection address. Logs will reflect the addresses
8337 indicated in the protocol, unless it is violated, in which case the real
8338 address will still be used. This keyword combined with support from external
8339 components can be used as an efficient and reliable alternative to the
8340 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +02008341 usable. See also "tcp-request connection expect-proxy" for a finer-grained
8342 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008343
Willy Tarreauab861d32013-04-02 02:30:41 +02008344alpn <protocols>
8345 This enables the TLS ALPN extension and advertises the specified protocol
8346 list as supported on top of ALPN. The protocol list consists in a comma-
8347 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
8348 quotes). This requires that the SSL library is build with support for TLS
8349 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
8350 initial NPN extension.
8351
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008352backlog <backlog>
8353 Sets the socket's backlog to this value. If unspecified, the frontend's
8354 backlog is used instead, which generally defaults to the maxconn value.
8355
Emeric Brun7fb34422012-09-28 15:26:15 +02008356ecdhe <named curve>
8357 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +01008358 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
8359 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +02008360
Emeric Brunfd33a262012-10-11 16:28:27 +02008361ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008362 This setting is only available when support for OpenSSL was built in. It
8363 designates a PEM file from which to load CA certificates used to verify
8364 client's certificate.
8365
Emeric Brunb6dc9342012-09-28 17:55:37 +02008366ca-ignore-err [all|<errorID>,...]
8367 This setting is only available when support for OpenSSL was built in.
8368 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
8369 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
8370 error is ignored.
8371
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008372ciphers <ciphers>
8373 This setting is only available when support for OpenSSL was built in. It sets
8374 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008375 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008376 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
8377 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
8378
Emeric Brunfd33a262012-10-11 16:28:27 +02008379crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008380 This setting is only available when support for OpenSSL was built in. It
8381 designates a PEM file from which to load certificate revocation list used
8382 to verify client's certificate.
8383
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008384crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +00008385 This setting is only available when support for OpenSSL was built in. It
8386 designates a PEM file containing both the required certificates and any
8387 associated private keys. This file can be built by concatenating multiple
8388 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
8389 requires an intermediate certificate, this can also be concatenated into this
8390 file.
8391
8392 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
8393 are loaded.
8394
8395 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté62043c92015-01-25 00:16:08 +01008396 that directory will be loaded in alphabetic order unless their name ends with
8397 '.issuer' or '.ocsp' (reserved extensions). This directive may be specified
8398 multiple times in order to load certificates from multiple files or
8399 directories. The certificates will be presented to clients who provide a valid
8400 TLS Server Name Indication field matching one of their CN or alt subjects.
8401 Wildcards are supported, where a wildcard character '*' is used instead of the
8402 first hostname component (eg: *.example.org matches www.example.org but not
Alex Davies0fbf0162013-03-02 16:04:50 +00008403 www.sub.example.org).
8404
8405 If no SNI is provided by the client or if the SSL library does not support
8406 TLS extensions, or if the client provides an SNI hostname which does not
8407 match any certificate, then the first loaded certificate will be presented.
8408 This means that when loading certificates from a directory, it is highly
Cyril Bonté62043c92015-01-25 00:16:08 +01008409 recommended to load the default one first as a file or to ensure that it will
8410 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +00008411
Emeric Brune032bfa2012-09-28 13:01:45 +02008412 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008413
Alex Davies0fbf0162013-03-02 16:04:50 +00008414 Some CAs (such as Godaddy) offer a drop down list of server types that do not
8415 include HAProxy when obtaining a certificate. If this happens be sure to
Godbach8bf60a12014-04-21 21:42:41 +08008416 choose a webserver that the CA believes requires an intermediate CA (for
Alex Davies0fbf0162013-03-02 16:04:50 +00008417 Godaddy, selection Apache Tomcat will get the correct bundle, but many
8418 others, e.g. nginx, result in a wrong bundle that will not work for some
8419 clients).
8420
Emeric Brun4147b2e2014-06-16 18:36:30 +02008421 For each PEM file, haproxy checks for the presence of file at the same path
8422 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
8423 Status Request extension (also known as "OCSP stapling") is automatically
8424 enabled. The content of this file is optional. If not empty, it must contain
8425 a valid OCSP Response in DER format. In order to be valid an OCSP Response
8426 must comply with the following rules: it has to indicate a good status,
8427 it has to be a single response for the certificate of the PEM file, and it
8428 has to be valid at the moment of addition. If these rules are not respected
8429 the OCSP Response is ignored and a warning is emitted. In order to identify
8430 which certificate an OCSP Response applies to, the issuer's certificate is
8431 necessary. If the issuer's certificate is not found in the PEM file, it will
8432 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
8433 if it exists otherwise it will fail with an error.
8434
Emeric Brunb6dc9342012-09-28 17:55:37 +02008435crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +00008436 This setting is only available when support for OpenSSL was built in. Sets a
8437 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008438 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +00008439 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +02008440
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008441crt-list <file>
8442 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008443 designates a list of PEM file with an optional list of SNI filter per
8444 certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008445
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008446 <crtfile> [[!]<snifilter> ...]
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008447
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008448 Wildcards are supported in the SNI filter. Negative filter are also supported,
8449 only useful in combination with a wildcard filter to exclude a particular SNI.
8450 The certificates will be presented to clients who provide a valid TLS Server
8451 Name Indication field matching one of the SNI filters. If no SNI filter is
8452 specified, the CN and alt subjects are used. This directive may be specified
8453 multiple times. See the "crt" option for more information. The default
8454 certificate is still needed to meet OpenSSL expectations. If it is not used,
8455 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008456
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008457defer-accept
8458 Is an optional keyword which is supported only on certain Linux kernels. It
8459 states that a connection will only be accepted once some data arrive on it,
8460 or at worst after the first retransmit. This should be used only on protocols
8461 for which the client talks first (eg: HTTP). It can slightly improve
8462 performance by ensuring that most of the request is already available when
8463 the connection is accepted. On the other hand, it will not be able to detect
8464 connections which don't talk. It is important to note that this option is
8465 broken in all kernels up to 2.6.31, as the connection is never accepted until
8466 the client talks. This can cause issues with front firewalls which would see
8467 an established connection while the proxy will only see it in SYN_RECV. This
8468 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
8469
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008470force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008471 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008472 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun42a3e202014-10-30 15:56:50 +01008473 for high connection rates. This option is also available on global statement
8474 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008475
8476force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008477 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008478 this listener. This option is also available on global statement
8479 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008480
8481force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008482 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008483 this listener. This option is also available on global statement
8484 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008485
8486force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008487 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008488 this listener. This option is also available on global statement
8489 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008490
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008491gid <gid>
8492 Sets the group of the UNIX sockets to the designated system gid. It can also
8493 be set by default in the global section's "unix-bind" statement. Note that
8494 some platforms simply ignore this. This setting is equivalent to the "group"
8495 setting except that the group ID is used instead of its name. This setting is
8496 ignored by non UNIX sockets.
8497
8498group <group>
8499 Sets the group of the UNIX sockets to the designated system group. It can
8500 also be set by default in the global section's "unix-bind" statement. Note
8501 that some platforms simply ignore this. This setting is equivalent to the
8502 "gid" setting except that the group name is used instead of its gid. This
8503 setting is ignored by non UNIX sockets.
8504
8505id <id>
8506 Fixes the socket ID. By default, socket IDs are automatically assigned, but
8507 sometimes it is more convenient to fix them to ease monitoring. This value
8508 must be strictly positive and unique within the listener/frontend. This
8509 option can only be used when defining only a single socket.
8510
8511interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +01008512 Restricts the socket to a specific interface. When specified, only packets
8513 received from that particular interface are processed by the socket. This is
8514 currently only supported on Linux. The interface must be a primary system
8515 interface, not an aliased interface. It is also possible to bind multiple
8516 frontends to the same address if they are bound to different interfaces. Note
8517 that binding to a network interface requires root privileges. This parameter
8518 is only compatible with TCPv4/TCPv6 sockets.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008519
Willy Tarreauabb175f2012-09-24 12:43:26 +02008520level <level>
8521 This setting is used with the stats sockets only to restrict the nature of
8522 the commands that can be issued on the socket. It is ignored by other
8523 sockets. <level> can be one of :
8524 - "user" is the least privileged level ; only non-sensitive stats can be
8525 read, and no change is allowed. It would make sense on systems where it
8526 is not easy to restrict access to the socket.
8527 - "operator" is the default level and fits most common uses. All data can
8528 be read, and only non-sensitive changes are permitted (eg: clear max
8529 counters).
8530 - "admin" should be used with care, as everything is permitted (eg: clear
8531 all counters).
8532
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008533maxconn <maxconn>
8534 Limits the sockets to this number of concurrent connections. Extraneous
8535 connections will remain in the system's backlog until a connection is
8536 released. If unspecified, the limit will be the same as the frontend's
8537 maxconn. Note that in case of port ranges or multiple addresses, the same
8538 value will be applied to each socket. This setting enables different
8539 limitations on expensive sockets, for instance SSL entries which may easily
8540 eat all memory.
8541
8542mode <mode>
8543 Sets the octal mode used to define access permissions on the UNIX socket. It
8544 can also be set by default in the global section's "unix-bind" statement.
8545 Note that some platforms simply ignore this. This setting is ignored by non
8546 UNIX sockets.
8547
8548mss <maxseg>
8549 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
8550 connections. This can be used to force a lower MSS for certain specific
8551 ports, for instance for connections passing through a VPN. Note that this
8552 relies on a kernel feature which is theoretically supported under Linux but
8553 was buggy in all versions prior to 2.6.28. It may or may not work on other
8554 operating systems. It may also not change the advertised value but change the
8555 effective size of outgoing segments. The commonly advertised value for TCPv4
8556 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
8557 positive, it will be used as the advertised MSS. If it is negative, it will
8558 indicate by how much to reduce the incoming connection's advertised MSS for
8559 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
8560
8561name <name>
8562 Sets an optional name for these sockets, which will be reported on the stats
8563 page.
8564
8565nice <nice>
8566 Sets the 'niceness' of connections initiated from the socket. Value must be
8567 in the range -1024..1024 inclusive, and defaults to zero. Positive values
8568 means that such connections are more friendly to others and easily offer
8569 their place in the scheduler. On the opposite, negative values mean that
8570 connections want to run with a higher priority than others. The difference
8571 only happens under high loads when the system is close to saturation.
8572 Negative values are appropriate for low-latency or administration services,
8573 and high values are generally recommended for CPU intensive tasks such as SSL
8574 processing or bulk transfers which are less sensible to latency. For example,
8575 it may make sense to use a positive value for an SMTP socket and a negative
8576 one for an RDP socket.
8577
Emeric Brun9b3009b2012-10-05 11:55:06 +02008578no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008579 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008580 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008581 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun42a3e202014-10-30 15:56:50 +01008582 be enabled using any configuration option. This option is also available on
8583 global statement "ssl-default-bind-options". See also "force-tls*",
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008584 and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008585
Emeric Brun90ad8722012-10-02 14:00:59 +02008586no-tls-tickets
8587 This setting is only available when support for OpenSSL was built in. It
8588 disables the stateless session resumption (RFC 5077 TLS Ticket
8589 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01008590 session resumption is more expensive in CPU usage. This option is also
8591 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +02008592
Emeric Brun9b3009b2012-10-05 11:55:06 +02008593no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008594 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008595 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008596 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008597 cannot be enabled using any configuration option. This option is also
8598 available on global statement "ssl-default-bind-options". See also
8599 "force-tlsv*", and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008600
Emeric Brun9b3009b2012-10-05 11:55:06 +02008601no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +02008602 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008603 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008604 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008605 cannot be enabled using any configuration option. This option is also
8606 available on global statement "ssl-default-bind-options". See also
8607 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008608
Emeric Brun9b3009b2012-10-05 11:55:06 +02008609no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +02008610 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008611 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008612 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008613 cannot be enabled using any configuration option. This option is also
8614 available on global statement "ssl-default-bind-options". See also
8615 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008616
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008617npn <protocols>
8618 This enables the NPN TLS extension and advertises the specified protocol list
8619 as supported on top of NPN. The protocol list consists in a comma-delimited
8620 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
8621 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +02008622 enabled (check with haproxy -vv). Note that the NPN extension has been
8623 replaced with the ALPN extension (see the "alpn" keyword).
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008624
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008625process [ all | odd | even | <number 1-64>[-<number 1-64>] ]
8626 This restricts the list of processes on which this listener is allowed to
8627 run. It does not enforce any process but eliminates those which do not match.
8628 If the frontend uses a "bind-process" setting, the intersection between the
8629 two is applied. If in the end the listener is not allowed to run on any
8630 remaining process, a warning is emitted, and the listener will either run on
8631 the first process of the listener if a single process was specified, or on
8632 all of its processes if multiple processes were specified. For the unlikely
Willy Tarreauae302532014-05-07 19:22:24 +02008633 case where several ranges are needed, this directive may be repeated. The
8634 main purpose of this directive is to be used with the stats sockets and have
8635 one different socket per process. The second purpose is to have multiple bind
8636 lines sharing the same IP:port but not the same process in a listener, so
8637 that the system can distribute the incoming connections into multiple queues
8638 and allow a smoother inter-process load balancing. Currently Linux 3.9 and
8639 above is known for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008640
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008641ssl
8642 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008643 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008644 certificate is necessary (see "crt" above). All contents in the buffers will
8645 appear in clear text, so that ACLs and HTTP processing will only have access
8646 to deciphered contents.
8647
Emmanuel Hocdet65623372013-01-24 17:17:15 +01008648strict-sni
8649 This setting is only available when support for OpenSSL was built in. The
8650 SSL/TLS negotiation is allow only if the client provided an SNI which match
8651 a certificate. The default certificate is not used.
8652 See the "crt" option for more information.
8653
Willy Tarreau1c862c52012-10-05 16:21:00 +02008654tfo
Lukas Tribus0defb902013-02-13 23:35:39 +01008655 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +02008656 enables TCP Fast Open on the listening socket, which means that clients which
8657 support this feature will be able to send a request and receive a response
8658 during the 3-way handshake starting from second connection, thus saving one
8659 round-trip after the first connection. This only makes sense with protocols
8660 that use high connection rates and where each round trip matters. This can
8661 possibly cause issues with many firewalls which do not accept data on SYN
8662 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +02008663 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
8664 need to build HAProxy with USE_TFO=1 if your libc doesn't define
8665 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +02008666
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008667transparent
8668 Is an optional keyword which is supported only on certain Linux kernels. It
8669 indicates that the addresses will be bound even if they do not belong to the
8670 local machine, and that packets targeting any of these addresses will be
8671 intercepted just as if the addresses were locally configured. This normally
8672 requires that IP forwarding is enabled. Caution! do not use this with the
8673 default address '*', as it would redirect any traffic for the specified port.
8674 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
8675 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
8676 kernel version. Some distribution kernels include backports of the feature,
8677 so check for support with your vendor.
8678
Willy Tarreau77e3af92012-11-24 15:07:23 +01008679v4v6
8680 Is an optional keyword which is supported only on most recent systems
8681 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
8682 and IPv6 when it uses the default address. Doing so is sometimes necessary
8683 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008684 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +01008685
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008686v6only
8687 Is an optional keyword which is supported only on most recent systems
8688 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
8689 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +01008690 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
8691 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008692
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008693uid <uid>
8694 Sets the owner of the UNIX sockets to the designated system uid. It can also
8695 be set by default in the global section's "unix-bind" statement. Note that
8696 some platforms simply ignore this. This setting is equivalent to the "user"
8697 setting except that the user numeric ID is used instead of its name. This
8698 setting is ignored by non UNIX sockets.
8699
8700user <user>
8701 Sets the owner of the UNIX sockets to the designated system user. It can also
8702 be set by default in the global section's "unix-bind" statement. Note that
8703 some platforms simply ignore this. This setting is equivalent to the "uid"
8704 setting except that the user name is used instead of its uid. This setting is
8705 ignored by non UNIX sockets.
8706
Emeric Brun1a073b42012-09-28 17:07:34 +02008707verify [none|optional|required]
8708 This setting is only available when support for OpenSSL was built in. If set
8709 to 'none', client certificate is not requested. This is the default. In other
8710 cases, a client certificate is requested. If the client does not provide a
8711 certificate after the request and if 'verify' is set to 'required', then the
8712 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +02008713 certificate provided by the client is always verified using CAs from
8714 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
8715 is aborted, regardless of the 'verify' option, unless the error code exactly
8716 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008717
Willy Tarreaub6205fd2012-09-24 12:27:33 +020087185.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +01008719------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02008720
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008721The "server" and "default-server" keywords support a certain number of settings
8722which are all passed as arguments on the server line. The order in which those
8723arguments appear does not count, and they are all optional. Some of those
8724settings are single words (booleans) while others expect one or several values
8725after them. In this case, the values must immediately follow the setting name.
8726Except default-server, all those settings must be specified after the server's
8727address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +02008728
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008729 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008730 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +02008731
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008732The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008733
Willy Tarreauceb4ac92012-04-28 00:41:46 +02008734addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008735 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann611dfcc2016-03-06 23:14:36 +01008736 to send health-checks or to probe the agent-check. On some servers, it may be
8737 desirable to dedicate an IP address to specific component able to perform
8738 complex tests which are more suitable to health-checks than the application.
8739 This parameter is ignored if the "check" parameter is not set. See also the
8740 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008741
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008742 Supported in default-server: No
8743
Simon Hormand60d6912013-11-25 10:46:36 +09008744agent-check
8745 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +01008746 health check. An agent health check is performed by making a TCP connection
8747 to the port set by the "agent-port" parameter and reading an ASCII string.
8748 The string is made of a series of words delimited by spaces, tabs or commas
8749 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +09008750
Willy Tarreau81f5d942013-12-09 20:51:51 +01008751 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +09008752 Values in this format will set the weight proportional to the initial
Willy Tarreau47067ca2014-10-07 15:27:33 +02008753 weight of a server as configured when haproxy starts. Note that a zero
8754 weight is reported on the stats page as "DRAIN" since it has the same
8755 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +09008756
Willy Tarreau81f5d942013-12-09 20:51:51 +01008757 - The word "ready". This will turn the server's administrative state to the
8758 READY mode, thus cancelling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +09008759
Willy Tarreau81f5d942013-12-09 20:51:51 +01008760 - The word "drain". This will turn the server's administrative state to the
8761 DRAIN mode, thus it will not accept any new connections other than those
8762 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +09008763
Willy Tarreau81f5d942013-12-09 20:51:51 +01008764 - The word "maint". This will turn the server's administrative state to the
8765 MAINT mode, thus it will not accept any new connections at all, and health
8766 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +09008767
Willy Tarreau81f5d942013-12-09 20:51:51 +01008768 - The words "down", "failed", or "stopped", optionally followed by a
8769 description string after a sharp ('#'). All of these mark the server's
8770 operating state as DOWN, but since the word itself is reported on the stats
8771 page, the difference allows an administrator to know if the situation was
8772 expected or not : the service may intentionally be stopped, may appear up
8773 but fail some validity tests, or may be seen as down (eg: missing process,
8774 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +09008775
Willy Tarreau81f5d942013-12-09 20:51:51 +01008776 - The word "up" sets back the server's operating state as UP if health checks
8777 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +09008778
Willy Tarreau81f5d942013-12-09 20:51:51 +01008779 Parameters which are not advertised by the agent are not changed. For
8780 example, an agent might be designed to monitor CPU usage and only report a
8781 relative weight and never interact with the operating status. Similarly, an
8782 agent could be designed as an end-user interface with 3 radio buttons
8783 allowing an administrator to change only the administrative state. However,
8784 it is important to consider that only the agent may revert its own actions,
8785 so if a server is set to DRAIN mode or to DOWN state using the agent, the
8786 agent must implement the other equivalent actions to bring the service into
8787 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +09008788
Simon Horman2f1f9552013-11-25 10:46:37 +09008789 Failure to connect to the agent is not considered an error as connectivity
8790 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +01008791 parameter. Warning though, it is not a good idea to stop an agent after it
8792 reports "down", since only an agent reporting "up" will be able to turn the
8793 server up again. Note that the CLI on the Unix stats socket is also able to
8794 force an agent's result in order to workaround a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +09008795
Willy Tarreau81f5d942013-12-09 20:51:51 +01008796 Requires the "agent-port" parameter to be set. See also the "agent-inter"
8797 parameter.
Simon Hormand60d6912013-11-25 10:46:36 +09008798
8799 Supported in default-server: No
8800
8801agent-inter <delay>
8802 The "agent-inter" parameter sets the interval between two agent checks
8803 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8804
8805 Just as with every other time-based parameter, it may be entered in any
8806 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
8807 parameter also serves as a timeout for agent checks "timeout check" is
8808 not set. In order to reduce "resonance" effects when multiple servers are
8809 hosted on the same hardware, the agent and health checks of all servers
8810 are started with a small time offset between them. It is also possible to
8811 add some random noise in the agent and health checks interval using the
8812 global "spread-checks" keyword. This makes sense for instance when a lot
8813 of backends use the same servers.
8814
8815 See also the "agent-check" and "agent-port" parameters.
8816
8817 Supported in default-server: Yes
8818
8819agent-port <port>
8820 The "agent-port" parameter sets the TCP port used for agent checks.
8821
8822 See also the "agent-check" and "agent-inter" parameters.
8823
8824 Supported in default-server: Yes
8825
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008826backup
8827 When "backup" is present on a server line, the server is only used in load
8828 balancing when all other non-backup servers are unavailable. Requests coming
8829 with a persistence cookie referencing the server will always be served
8830 though. By default, only the first operational backup server is used, unless
8831 the "allbackups" option is set in the backend. See also the "allbackups"
8832 option.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008833
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008834 Supported in default-server: No
8835
Emeric Brunef42d922012-10-11 16:11:36 +02008836ca-file <cafile>
8837 This setting is only available when support for OpenSSL was built in. It
8838 designates a PEM file from which to load CA certificates used to verify
8839 server's certificate.
8840
8841 Supported in default-server: No
8842
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008843check
8844 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +01008845 always considered available. If "check" is set, the server is available when
8846 accepting periodic TCP connections, to ensure that it is really able to serve
8847 requests. The default address and port to send the tests to are those of the
8848 server, and the default source is the same as the one defined in the
8849 backend. It is possible to change the address using the "addr" parameter, the
8850 port using the "port" parameter, the source address using the "source"
8851 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +09008852 parameters. The request method is define in the backend using the "httpchk",
8853 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
8854 refer to those options and parameters for more information.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008855
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008856 Supported in default-server: No
8857
Willy Tarreau6c16adc2012-10-05 00:04:16 +02008858check-send-proxy
8859 This option forces emission of a PROXY protocol line with outgoing health
8860 checks, regardless of whether the server uses send-proxy or not for the
8861 normal traffic. By default, the PROXY protocol is enabled for health checks
8862 if it is already enabled for normal traffic and if no "port" nor "addr"
8863 directive is present. However, if such a directive is present, the
8864 "check-send-proxy" option needs to be used to force the use of the
8865 protocol. See also the "send-proxy" option for more information.
8866
8867 Supported in default-server: No
8868
Willy Tarreau763a95b2012-10-04 23:15:39 +02008869check-ssl
8870 This option forces encryption of all health checks over SSL, regardless of
8871 whether the server uses SSL or not for the normal traffic. This is generally
8872 used when an explicit "port" or "addr" directive is specified and SSL health
8873 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008874 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +02008875 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
8876 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
8877 All SSL settings are common to health checks and traffic (eg: ciphers).
8878 See the "ssl" option for more information.
8879
8880 Supported in default-server: No
8881
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008882ciphers <ciphers>
8883 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008884 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008885 string is defined in "man 1 ciphers". When SSL is used to communicate with
8886 servers on the local network, it is common to see a weaker set of algorithms
8887 than what is used over the internet. Doing so reduces CPU usage on both the
8888 server and haproxy while still keeping it compatible with deployed software.
8889 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
8890 is needed and just connectivity, using DES can be appropriate.
8891
Willy Tarreau763a95b2012-10-04 23:15:39 +02008892 Supported in default-server: No
8893
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008894cookie <value>
8895 The "cookie" parameter sets the cookie value assigned to the server to
8896 <value>. This value will be checked in incoming requests, and the first
8897 operational server possessing the same value will be selected. In return, in
8898 cookie insertion or rewrite modes, this value will be assigned to the cookie
8899 sent to the client. There is nothing wrong in having several servers sharing
8900 the same cookie value, and it is in fact somewhat common between normal and
8901 backup servers. See also the "cookie" keyword in backend section.
8902
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008903 Supported in default-server: No
8904
Emeric Brunef42d922012-10-11 16:11:36 +02008905crl-file <crlfile>
8906 This setting is only available when support for OpenSSL was built in. It
8907 designates a PEM file from which to load certificate revocation list used
8908 to verify server's certificate.
8909
8910 Supported in default-server: No
8911
Emeric Bruna7aa3092012-10-26 12:58:00 +02008912crt <cert>
8913 This setting is only available when support for OpenSSL was built in.
8914 It designates a PEM file from which to load both a certificate and the
8915 associated private key. This file can be built by concatenating both PEM
8916 files into one. This certificate will be sent if the server send a client
8917 certificate request.
8918
8919 Supported in default-server: No
8920
Willy Tarreau96839092010-03-29 10:02:24 +02008921disabled
8922 The "disabled" keyword starts the server in the "disabled" state. That means
8923 that it is marked down in maintenance mode, and no connection other than the
8924 ones allowed by persist mode will reach it. It is very well suited to setup
8925 new servers, because normal traffic will never reach them, while it is still
8926 possible to test the service by making use of the force-persist mechanism.
8927
8928 Supported in default-server: No
8929
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008930error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +01008931 If health observing is enabled, the "error-limit" parameter specifies the
8932 number of consecutive errors that triggers event selected by the "on-error"
8933 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008934
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008935 Supported in default-server: Yes
8936
8937 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008938
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008939fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008940 The "fall" parameter states that a server will be considered as dead after
8941 <count> consecutive unsuccessful health checks. This value defaults to 3 if
8942 unspecified. See also the "check", "inter" and "rise" parameters.
8943
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008944 Supported in default-server: Yes
8945
Emeric Brun8694b9a2012-10-05 14:39:07 +02008946force-sslv3
8947 This option enforces use of SSLv3 only when SSL is used to communicate with
8948 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun42a3e202014-10-30 15:56:50 +01008949 high connection rates. This option is also available on global statement
8950 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008951
8952 Supported in default-server: No
8953
8954force-tlsv10
8955 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008956 the server. This option is also available on global statement
8957 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008958
8959 Supported in default-server: No
8960
8961force-tlsv11
8962 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008963 the server. This option is also available on global statement
8964 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008965
8966 Supported in default-server: No
8967
8968force-tlsv12
8969 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008970 the server. This option is also available on global statement
8971 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008972
8973 Supported in default-server: No
8974
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008975id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02008976 Set a persistent ID for the server. This ID must be positive and unique for
8977 the proxy. An unused ID will automatically be assigned if unset. The first
8978 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008979
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008980 Supported in default-server: No
8981
8982inter <delay>
8983fastinter <delay>
8984downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008985 The "inter" parameter sets the interval between two consecutive health checks
8986 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8987 It is also possible to use "fastinter" and "downinter" to optimize delays
8988 between checks depending on the server state :
8989
8990 Server state | Interval used
8991 ---------------------------------+-----------------------------------------
8992 UP 100% (non-transitional) | "inter"
8993 ---------------------------------+-----------------------------------------
8994 Transitionally UP (going down), |
8995 Transitionally DOWN (going up), | "fastinter" if set, "inter" otherwise.
8996 or yet unchecked. |
8997 ---------------------------------+-----------------------------------------
8998 DOWN 100% (non-transitional) | "downinter" if set, "inter" otherwise.
8999 ---------------------------------+-----------------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +01009000
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009001 Just as with every other time-based parameter, they can be entered in any
9002 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
9003 serves as a timeout for health checks sent to servers if "timeout check" is
9004 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +09009005 hosted on the same hardware, the agent and health checks of all servers
9006 are started with a small time offset between them. It is also possible to
9007 add some random noise in the agent and health checks interval using the
9008 global "spread-checks" keyword. This makes sense for instance when a lot
9009 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009010
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009011 Supported in default-server: Yes
9012
9013maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009014 The "maxconn" parameter specifies the maximal number of concurrent
9015 connections that will be sent to this server. If the number of incoming
9016 concurrent requests goes higher than this value, they will be queued, waiting
9017 for a connection to be released. This parameter is very important as it can
9018 save fragile servers from going down under extreme loads. If a "minconn"
9019 parameter is specified, the limit becomes dynamic. The default value is "0"
9020 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
9021 the backend's "fullconn" keyword.
9022
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009023 Supported in default-server: Yes
9024
9025maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009026 The "maxqueue" parameter specifies the maximal number of connections which
9027 will wait in the queue for this server. If this limit is reached, next
9028 requests will be redispatched to other servers instead of indefinitely
9029 waiting to be served. This will break persistence but may allow people to
9030 quickly re-log in when the server they try to connect to is dying. The
9031 default value is "0" which means the queue is unlimited. See also the
9032 "maxconn" and "minconn" parameters.
9033
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009034 Supported in default-server: Yes
9035
9036minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009037 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
9038 limit following the backend's load. The server will always accept at least
9039 <minconn> connections, never more than <maxconn>, and the limit will be on
9040 the ramp between both values when the backend has less than <fullconn>
9041 concurrent connections. This makes it possible to limit the load on the
9042 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009043 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009044 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009045
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009046 Supported in default-server: Yes
9047
Emeric Brun9b3009b2012-10-05 11:55:06 +02009048no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009049 This option disables support for SSLv3 when SSL is used to communicate with
9050 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emeric Brun8694b9a2012-10-05 14:39:07 +02009051 using any configuration option. See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009052
Willy Tarreau763a95b2012-10-04 23:15:39 +02009053 Supported in default-server: No
9054
Emeric Brunf9c5c472012-10-11 15:28:34 +02009055no-tls-tickets
9056 This setting is only available when support for OpenSSL was built in. It
9057 disables the stateless session resumption (RFC 5077 TLS Ticket
9058 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01009059 session resumption is more expensive in CPU usage for servers. This option
9060 is also available on global statement "ssl-default-server-options".
Emeric Brunf9c5c472012-10-11 15:28:34 +02009061
9062 Supported in default-server: No
9063
Emeric Brun9b3009b2012-10-05 11:55:06 +02009064no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +02009065 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009066 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9067 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009068 often makes sense to disable it when communicating with local servers. This
9069 option is also available on global statement "ssl-default-server-options".
9070 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009071
Willy Tarreau763a95b2012-10-04 23:15:39 +02009072 Supported in default-server: No
9073
Emeric Brun9b3009b2012-10-05 11:55:06 +02009074no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +02009075 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009076 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9077 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009078 often makes sense to disable it when communicating with local servers. This
9079 option is also available on global statement "ssl-default-server-options".
9080 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009081
Willy Tarreau763a95b2012-10-04 23:15:39 +02009082 Supported in default-server: No
9083
Emeric Brun9b3009b2012-10-05 11:55:06 +02009084no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +02009085 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009086 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9087 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009088 often makes sense to disable it when communicating with local servers. This
9089 option is also available on global statement "ssl-default-server-options".
9090 See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009091
Willy Tarreau763a95b2012-10-04 23:15:39 +02009092 Supported in default-server: No
9093
Simon Hormanfa461682011-06-25 09:39:49 +09009094non-stick
9095 Never add connections allocated to this sever to a stick-table.
9096 This may be used in conjunction with backup to ensure that
9097 stick-table persistence is disabled for backup servers.
9098
Willy Tarreau763a95b2012-10-04 23:15:39 +02009099 Supported in default-server: No
9100
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009101observe <mode>
9102 This option enables health adjusting based on observing communication with
9103 the server. By default this functionality is disabled and enabling it also
9104 requires to enable health checks. There are two supported modes: "layer4" and
9105 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
9106 significant. In layer7, which is only allowed for http proxies, responses
9107 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +01009108 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009109
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009110 Supported in default-server: No
9111
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009112 See also the "check", "on-error" and "error-limit".
9113
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009114on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009115 Select what should happen when enough consecutive errors are detected.
9116 Currently, four modes are available:
9117 - fastinter: force fastinter
9118 - fail-check: simulate a failed check, also forces fastinter (default)
9119 - sudden-death: simulate a pre-fatal failed health check, one more failed
9120 check will mark a server down, forces fastinter
9121 - mark-down: mark the server immediately down and force fastinter
9122
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009123 Supported in default-server: Yes
9124
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009125 See also the "check", "observe" and "error-limit".
9126
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009127on-marked-down <action>
9128 Modify what occurs when a server is marked down.
9129 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009130 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
9131 all connections to the server are immediately terminated when the server
9132 goes down. It might be used if the health check detects more complex cases
9133 than a simple connection status, and long timeouts would cause the service
9134 to remain unresponsive for too long a time. For instance, a health check
9135 might detect that a database is stuck and that there's no chance to reuse
9136 existing connections anymore. Connections killed this way are logged with
9137 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009138
9139 Actions are disabled by default
9140
9141 Supported in default-server: Yes
9142
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009143on-marked-up <action>
9144 Modify what occurs when a server is marked up.
9145 Currently one action is available:
9146 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
9147 done only if the server is not in backup state and if it is not disabled
9148 (it must have an effective weight > 0). This can be used sometimes to force
9149 an active server to take all the traffic back after recovery when dealing
9150 with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble
9151 than it tries to solve (eg: incomplete transactions), so use this feature
9152 with extreme care. Sessions killed because a server comes up are logged
9153 with an 'U' termination code (for "Up").
9154
9155 Actions are disabled by default
9156
9157 Supported in default-server: Yes
9158
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009159port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009160 Using the "port" parameter, it becomes possible to use a different port to
9161 send health-checks. On some servers, it may be desirable to dedicate a port
9162 to a specific component able to perform complex tests which are more suitable
9163 to health-checks than the application. It is common to run a simple script in
9164 inetd for instance. This parameter is ignored if the "check" parameter is not
9165 set. See also the "addr" parameter.
9166
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009167 Supported in default-server: Yes
9168
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009169redir <prefix>
9170 The "redir" parameter enables the redirection mode for all GET and HEAD
9171 requests addressing this server. This means that instead of having HAProxy
9172 forward the request to the server, it will send an "HTTP 302" response with
9173 the "Location" header composed of this prefix immediately followed by the
9174 requested URI beginning at the leading '/' of the path component. That means
9175 that no trailing slash should be used after <prefix>. All invalid requests
9176 will be rejected, and all non-GET or HEAD requests will be normally served by
9177 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009178 mangling nor cookie insertion is possible in the response. However, cookies in
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009179 requests are still analysed, making this solution completely usable to direct
9180 users to a remote location in case of local disaster. Main use consists in
9181 increasing bandwidth for static servers by having the clients directly
9182 connect to them. Note: never use a relative location here, it would cause a
9183 loop between the client and HAProxy!
9184
9185 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
9186
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009187 Supported in default-server: No
9188
9189rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009190 The "rise" parameter states that a server will be considered as operational
9191 after <count> consecutive successful health checks. This value defaults to 2
9192 if unspecified. See also the "check", "inter" and "fall" parameters.
9193
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009194 Supported in default-server: Yes
9195
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009196send-proxy
9197 The "send-proxy" parameter enforces use of the PROXY protocol over any
9198 connection established to this server. The PROXY protocol informs the other
9199 end about the layer 3/4 addresses of the incoming connection, so that it can
9200 know the client's address or the public address it accessed to, whatever the
9201 upper layer protocol. For connections accepted by an "accept-proxy" listener,
9202 the advertised address will be used. Only TCPv4 and TCPv6 address families
9203 are supported. Other families such as Unix sockets, will report an UNKNOWN
9204 family. Servers using this option can fully be chained to another instance of
9205 haproxy listening with an "accept-proxy" setting. This setting must not be
Willy Tarreau6c16adc2012-10-05 00:04:16 +02009206 used if the server isn't aware of the protocol. When health checks are sent
9207 to the server, the PROXY protocol is automatically used when this option is
9208 set, unless there is an explicit "port" or "addr" directive, in which case an
9209 explicit "check-send-proxy" directive would also be needed to use the PROXY
9210 protocol. See also the "accept-proxy" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009211
9212 Supported in default-server: No
9213
David Safb76832014-05-08 23:42:08 -04009214send-proxy-v2
9215 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
9216 over any connection established to this server. The PROXY protocol informs
9217 the other end about the layer 3/4 addresses of the incoming connection, so
9218 that it can know the client's address or the public address it accessed to,
9219 whatever the upper layer protocol. This setting must not be used if the
9220 server isn't aware of this version of the protocol. See also the "send-proxy"
9221 option of the "bind" keyword.
9222
9223 Supported in default-server: No
9224
9225send-proxy-v2-ssl
9226 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9227 2 over any connection established to this server. The PROXY protocol informs
9228 the other end about the layer 3/4 addresses of the incoming connection, so
9229 that it can know the client's address or the public address it accessed to,
9230 whatever the upper layer protocol. In addition, the SSL information extension
9231 of the PROXY protocol is added to the PROXY protocol header. This setting
9232 must not be used if the server isn't aware of this version of the protocol.
9233 See also the "send-proxy-v2" option of the "bind" keyword.
9234
9235 Supported in default-server: No
9236
9237send-proxy-v2-ssl-cn
9238 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9239 2 over any connection established to this server. The PROXY protocol informs
9240 the other end about the layer 3/4 addresses of the incoming connection, so
9241 that it can know the client's address or the public address it accessed to,
9242 whatever the upper layer protocol. In addition, the SSL information extension
9243 of the PROXY protocol, along along with the Common Name from the subject of
9244 the client certificate (if any), is added to the PROXY protocol header. This
9245 setting must not be used if the server isn't aware of this version of the
9246 protocol. See also the "send-proxy-v2" option of the "bind" keyword.
9247
9248 Supported in default-server: No
9249
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009250slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009251 The "slowstart" parameter for a server accepts a value in milliseconds which
9252 indicates after how long a server which has just come back up will run at
9253 full speed. Just as with every other time-based parameter, it can be entered
9254 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
9255 linearly from 0 to 100% during this time. The limitation applies to two
9256 parameters :
9257
9258 - maxconn: the number of connections accepted by the server will grow from 1
9259 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
9260
9261 - weight: when the backend uses a dynamic weighted algorithm, the weight
9262 grows linearly from 1 to 100%. In this case, the weight is updated at every
9263 health-check. For this reason, it is important that the "inter" parameter
9264 is smaller than the "slowstart", in order to maximize the number of steps.
9265
9266 The slowstart never applies when haproxy starts, otherwise it would cause
9267 trouble to running servers. It only applies when a server has been previously
9268 seen as failed.
9269
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009270 Supported in default-server: Yes
9271
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009272source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009273source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009274source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009275 The "source" parameter sets the source address which will be used when
9276 connecting to the server. It follows the exact same parameters and principle
9277 as the backend "source" keyword, except that it only applies to the server
9278 referencing it. Please consult the "source" keyword for details.
9279
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009280 Additionally, the "source" statement on a server line allows one to specify a
9281 source port range by indicating the lower and higher bounds delimited by a
9282 dash ('-'). Some operating systems might require a valid IP address when a
9283 source port range is specified. It is permitted to have the same IP/range for
9284 several servers. Doing so makes it possible to bypass the maximum of 64k
9285 total concurrent connections. The limit will then reach 64k connections per
9286 server.
9287
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009288 Supported in default-server: No
9289
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009290ssl
Willy Tarreau44f65392013-06-25 07:56:20 +02009291 This option enables SSL ciphering on outgoing connections to the server. It
9292 is critical to verify server certificates using "verify" when using SSL to
9293 connect to servers, otherwise the communication is prone to trivial man in
9294 the-middle attacks rendering SSL useless. When this option is used, health
9295 checks are automatically sent in SSL too unless there is a "port" or an
9296 "addr" directive indicating the check should be sent to a different location.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009297 See the "check-ssl" option to force SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +02009298
9299 Supported in default-server: No
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009300
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009301track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +02009302 This option enables ability to set the current state of the server by tracking
9303 another one. It is possible to track a server which itself tracks another
9304 server, provided that at the end of the chain, a server has health checks
9305 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009306 used, it has to be enabled on both proxies.
9307
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009308 Supported in default-server: No
9309
Emeric Brunef42d922012-10-11 16:11:36 +02009310verify [none|required]
9311 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +01009312 to 'none', server certificate is not verified. In the other case, The
9313 certificate provided by the server is verified using CAs from 'ca-file'
9314 and optional CRLs from 'crl-file'. If 'ssl_server_verify' is not specified
9315 in global section, this is the default. On verify failure the handshake
Willy Tarreau44f65392013-06-25 07:56:20 +02009316 is aborted. It is critically important to verify server certificates when
9317 using SSL to connect to servers, otherwise the communication is prone to
9318 trivial man-in-the-middle attacks rendering SSL totally useless.
Emeric Brunef42d922012-10-11 16:11:36 +02009319
9320 Supported in default-server: No
9321
Evan Broderbe554312013-06-27 00:05:25 -07009322verifyhost <hostname>
9323 This setting is only available when support for OpenSSL was built in, and
9324 only takes effect if 'verify required' is also specified. When set, the
9325 hostnames in the subject and subjectAlternateNames of the certificate
9326 provided by the server are checked. If none of the hostnames in the
9327 certificate match the specified hostname, the handshake is aborted. The
9328 hostnames in the server-provided certificate may include wildcards.
9329
9330 Supported in default-server: No
9331
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009332weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009333 The "weight" parameter is used to adjust the server's weight relative to
9334 other servers. All servers will receive a load proportional to their weight
9335 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +02009336 load. The default weight is 1, and the maximal value is 256. A value of 0
9337 means the server will not participate in load-balancing but will still accept
9338 persistent connections. If this parameter is used to distribute the load
9339 according to server's capacity, it is recommended to start with values which
9340 can both grow and shrink, for instance between 10 and 100 to leave enough
9341 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009342
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009343 Supported in default-server: Yes
9344
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009345
93466. HTTP header manipulation
9347---------------------------
9348
9349In HTTP mode, it is possible to rewrite, add or delete some of the request and
9350response headers based on regular expressions. It is also possible to block a
9351request or a response if a particular header matches a regular expression,
9352which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +01009353against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009354
Willy Tarreau70dffda2014-01-30 03:07:23 +01009355If HAProxy encounters an "Informational Response" (status code 1xx), it is able
9356to process all rsp* rules which can allow, deny, rewrite or delete a header,
9357but it will refuse to add a header to any such messages as this is not
9358HTTP-compliant. The reason for still processing headers in such responses is to
9359stop and/or fix any possible information leak which may happen, for instance
9360because another downstream equipment would unconditionally add a header, or if
9361a server name appears there. When such messages are seen, normal processing
9362still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +02009363
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009364This section covers common usage of the following keywords, described in detail
9365in section 4.2 :
9366
9367 - reqadd <string>
9368 - reqallow <search>
9369 - reqiallow <search>
9370 - reqdel <search>
9371 - reqidel <search>
9372 - reqdeny <search>
9373 - reqideny <search>
9374 - reqpass <search>
9375 - reqipass <search>
9376 - reqrep <search> <replace>
9377 - reqirep <search> <replace>
9378 - reqtarpit <search>
9379 - reqitarpit <search>
9380 - rspadd <string>
9381 - rspdel <search>
9382 - rspidel <search>
9383 - rspdeny <search>
9384 - rspideny <search>
9385 - rsprep <search> <replace>
9386 - rspirep <search> <replace>
9387
9388With all these keywords, the same conventions are used. The <search> parameter
9389is a POSIX extended regular expression (regex) which supports grouping through
9390parenthesis (without the backslash). Spaces and other delimiters must be
9391prefixed with a backslash ('\') to avoid confusion with a field delimiter.
9392Other characters may be prefixed with a backslash to change their meaning :
9393
9394 \t for a tab
9395 \r for a carriage return (CR)
9396 \n for a new line (LF)
9397 \ to mark a space and differentiate it from a delimiter
9398 \# to mark a sharp and differentiate it from a comment
9399 \\ to use a backslash in a regex
9400 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
9401 \xXX to write the ASCII hex code XX as in the C language
9402
9403The <replace> parameter contains the string to be used to replace the largest
9404portion of text matching the regex. It can make use of the special characters
9405above, and can reference a substring which is delimited by parenthesis in the
9406regex, by writing a backslash ('\') immediately followed by one digit from 0 to
94079 indicating the group position (0 designating the entire line). This practice
9408is very common to users of the "sed" program.
9409
9410The <string> parameter represents the string which will systematically be added
9411after the last header line. It can also use special character sequences above.
9412
9413Notes related to these keywords :
9414---------------------------------
9415 - these keywords are not always convenient to allow/deny based on header
9416 contents. It is strongly recommended to use ACLs with the "block" keyword
9417 instead, resulting in far more flexible and manageable rules.
9418
9419 - lines are always considered as a whole. It is not possible to reference
9420 a header name only or a value only. This is important because of the way
9421 headers are written (notably the number of spaces after the colon).
9422
9423 - the first line is always considered as a header, which makes it possible to
9424 rewrite or filter HTTP requests URIs or response codes, but in turn makes
9425 it harder to distinguish between headers and request line. The regex prefix
9426 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
9427 ^[^ \t:]*: matches any header name followed by a colon.
9428
9429 - for performances reasons, the number of characters added to a request or to
9430 a response is limited at build time to values between 1 and 4 kB. This
9431 should normally be far more than enough for most usages. If it is too short
9432 on occasional usages, it is possible to gain some space by removing some
9433 useless headers before adding new ones.
9434
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009435 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009436 without the 'i' letter except that they ignore case when matching patterns.
9437
9438 - when a request passes through a frontend then a backend, all req* rules
9439 from the frontend will be evaluated, then all req* rules from the backend
9440 will be evaluated. The reverse path is applied to responses.
9441
9442 - req* statements are applied after "block" statements, so that "block" is
9443 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +01009444 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009445
9446
Willy Tarreau74ca5042013-06-11 23:12:07 +020094477. Using ACLs and fetching samples
9448----------------------------------
9449
9450Haproxy is capable of extracting data from request or response streams, from
9451client or server information, from tables, environmental information etc...
9452The action of extracting such data is called fetching a sample. Once retrieved,
9453these samples may be used for various purposes such as a key to a stick-table,
9454but most common usages consist in matching them against predefined constant
9455data called patterns.
9456
9457
94587.1. ACL basics
9459---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009460
9461The use of Access Control Lists (ACL) provides a flexible solution to perform
9462content switching and generally to take decisions based on content extracted
9463from the request, the response or any environmental status. The principle is
9464simple :
9465
Willy Tarreau74ca5042013-06-11 23:12:07 +02009466 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009467 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +02009468 - apply one or multiple pattern matching methods on this sample
9469 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009470
Willy Tarreau74ca5042013-06-11 23:12:07 +02009471The actions generally consist in blocking a request, selecting a backend, or
9472adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009473
9474In order to define a test, the "acl" keyword is used. The syntax is :
9475
Willy Tarreau74ca5042013-06-11 23:12:07 +02009476 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009477
9478This creates a new ACL <aclname> or completes an existing one with new tests.
9479Those tests apply to the portion of request/response specified in <criterion>
9480and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009481an operator which may be specified before the set of values. Optionally some
9482conversion operators may be applied to the sample, and they will be specified
9483as a comma-delimited list of keywords just after the first keyword. The values
9484are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009485
9486ACL names must be formed from upper and lower case letters, digits, '-' (dash),
9487'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
9488which means that "my_acl" and "My_Acl" are two different ACLs.
9489
9490There is no enforced limit to the number of ACLs. The unused ones do not affect
9491performance, they just consume a small amount of memory.
9492
Willy Tarreau74ca5042013-06-11 23:12:07 +02009493The criterion generally is the name of a sample fetch method, or one of its ACL
9494specific declinations. The default test method is implied by the output type of
9495this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009496methods of a same sample fetch method. The sample fetch methods are the only
9497ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009498
9499Sample fetch methods return data which can be of the following types :
9500 - boolean
9501 - integer (signed or unsigned)
9502 - IPv4 or IPv6 address
9503 - string
9504 - data block
9505
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009506Converters transform any of these data into any of these. For example, some
9507converters might convert a string to a lower-case string while other ones
9508would turn a string to an IPv4 address, or apply a netmask to an IP address.
9509The resulting sample is of the type of the last converter applied to the list,
9510which defaults to the type of the sample fetch method.
9511
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009512Each sample or converter returns data of a specific type, specified with its
9513keyword in this documentation. When an ACL is declared using a standard sample
9514fetch method, certain types automatically involved a default matching method
9515which are summarized in the table below :
9516
9517 +---------------------+-----------------+
9518 | Sample or converter | Default |
9519 | output type | matching method |
9520 +---------------------+-----------------+
9521 | boolean | bool |
9522 +---------------------+-----------------+
9523 | integer | int |
9524 +---------------------+-----------------+
9525 | ip | ip |
9526 +---------------------+-----------------+
9527 | string | str |
9528 +---------------------+-----------------+
9529 | binary | none, use "-m" |
9530 +---------------------+-----------------+
9531
9532Note that in order to match a binary samples, it is mandatory to specify a
9533matching method, see below.
9534
Willy Tarreau74ca5042013-06-11 23:12:07 +02009535The ACL engine can match these types against patterns of the following types :
9536 - boolean
9537 - integer or integer range
9538 - IP address / network
9539 - string (exact, substring, suffix, prefix, subdir, domain)
9540 - regular expression
9541 - hex block
9542
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009543The following ACL flags are currently supported :
9544
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009545 -i : ignore case during matching of all subsequent patterns.
9546 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009547 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009548 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009549 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009550 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +02009551 -- : force end of flags. Useful when a string looks like one of the flags.
9552
Willy Tarreau74ca5042013-06-11 23:12:07 +02009553The "-f" flag is followed by the name of a file from which all lines will be
9554read as individual values. It is even possible to pass multiple "-f" arguments
9555if the patterns are to be loaded from multiple files. Empty lines as well as
9556lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
9557will be stripped. If it is absolutely necessary to insert a valid pattern
9558beginning with a sharp, just prefix it with a space so that it is not taken for
9559a comment. Depending on the data type and match method, haproxy may load the
9560lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
9561exact string matching. In this case, duplicates will automatically be removed.
9562
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009563The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
9564parsed as two column file. The first column contains the patterns used by the
9565ACL, and the second column contain the samples. The sample can be used later by
9566a map. This can be useful in some rare cases where an ACL would just be used to
9567check for the existence of a pattern in a map before a mapping is applied.
9568
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009569The "-u" flag forces the unique id of the ACL. This unique id is used with the
9570socket interface to identify ACL and dynamically change its values. Note that a
9571file is always identified by its name even if an id is set.
9572
Willy Tarreau74ca5042013-06-11 23:12:07 +02009573Also, note that the "-i" flag applies to subsequent entries and not to entries
9574loaded from files preceding it. For instance :
9575
9576 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
9577
9578In this example, each line of "exact-ua.lst" will be exactly matched against
9579the "user-agent" header of the request. Then each line of "generic-ua" will be
9580case-insensitively matched. Then the word "test" will be insensitively matched
9581as well.
9582
9583The "-m" flag is used to select a specific pattern matching method on the input
9584sample. All ACL-specific criteria imply a pattern matching method and generally
9585do not need this flag. However, this flag is useful with generic sample fetch
9586methods to describe how they're going to be matched against the patterns. This
9587is required for sample fetches which return data type for which there is no
9588obvious matching method (eg: string or binary). When "-m" is specified and
9589followed by a pattern matching method name, this method is used instead of the
9590default one for the criterion. This makes it possible to match contents in ways
9591that were not initially planned, or with sample fetch methods which return a
9592string. The matching method also affects the way the patterns are parsed.
9593
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009594The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
9595By default, if the parser cannot parse ip address it considers that the parsed
9596string is maybe a domain name and try dns resolution. The flag "-n" disable this
9597resolution. It is useful for detecting malformed ip lists. Note that if the DNS
9598server is not reachable, the haproxy configuration parsing may last many minutes
9599waiting fir the timeout. During this time no error messages are displayed. The
9600flag "-n" disable this behavior. Note also that during the runtime, this
9601function is disabled for the dynamic acl modifications.
9602
Willy Tarreau74ca5042013-06-11 23:12:07 +02009603There are some restrictions however. Not all methods can be used with all
9604sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
9605be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +02009606
9607 - "found" : only check if the requested sample could be found in the stream,
9608 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +02009609 to pass any pattern to avoid confusion. This matching method is
9610 particularly useful to detect presence of certain contents such
9611 as headers, cookies, etc... even if they are empty and without
9612 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009613
9614 - "bool" : check the value as a boolean. It can only be applied to fetches
9615 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009616 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009617
9618 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +02009619 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009620
9621 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +02009622 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009623
9624 - "bin" : match the contents against an hexadecimal string representing a
9625 binary sequence. This may be used with binary or string samples.
9626
9627 - "len" : match the sample's length as an integer. This may be used with
9628 binary or string samples.
9629
Willy Tarreau74ca5042013-06-11 23:12:07 +02009630 - "str" : exact match : match the contents against a string. This may be
9631 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009632
Willy Tarreau74ca5042013-06-11 23:12:07 +02009633 - "sub" : substring match : check that the contents contain at least one of
9634 the provided string patterns. This may be used with binary or
9635 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009636
Willy Tarreau74ca5042013-06-11 23:12:07 +02009637 - "reg" : regex match : match the contents against a list of regular
9638 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009639
Willy Tarreau74ca5042013-06-11 23:12:07 +02009640 - "beg" : prefix match : check that the contents begin like the provided
9641 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009642
Willy Tarreau74ca5042013-06-11 23:12:07 +02009643 - "end" : suffix match : check that the contents end like the provided
9644 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009645
Willy Tarreau74ca5042013-06-11 23:12:07 +02009646 - "dir" : subdir match : check that a slash-delimited portion of the
9647 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009648 This may be used with binary or string samples.
9649
Willy Tarreau74ca5042013-06-11 23:12:07 +02009650 - "dom" : domain match : check that a dot-delimited portion of the contents
9651 exactly match one of the provided string patterns. This may be
9652 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009653
9654For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
9655request, it is possible to do :
9656
9657 acl jsess_present cook(JSESSIONID) -m found
9658
9659In order to apply a regular expression on the 500 first bytes of data in the
9660buffer, one would use the following acl :
9661
9662 acl script_tag payload(0,500) -m reg -i <script>
9663
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009664On systems where the regex library is much slower when using "-i", it is
9665possible to convert the sample to lowercase before matching, like this :
9666
9667 acl script_tag payload(0,500),lower -m reg <script>
9668
Willy Tarreau74ca5042013-06-11 23:12:07 +02009669All ACL-specific criteria imply a default matching method. Most often, these
9670criteria are composed by concatenating the name of the original sample fetch
9671method and the matching method. For example, "hdr_beg" applies the "beg" match
9672to samples retrieved using the "hdr" fetch method. Since all ACL-specific
9673criteria rely on a sample fetch method, it is always possible instead to use
9674the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009675
Willy Tarreau74ca5042013-06-11 23:12:07 +02009676If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009677the matching method is simply applied to the underlying sample fetch method.
9678For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009679
Willy Tarreau74ca5042013-06-11 23:12:07 +02009680 acl short_form hdr_beg(host) www.
9681 acl alternate1 hdr_beg(host) -m beg www.
9682 acl alternate2 hdr_dom(host) -m beg www.
9683 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009684
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009685
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009686The table below summarizes the compatibility matrix between sample or converter
9687types and the pattern types to fetch against. It indicates for each compatible
9688combination the name of the matching method to be used, surrounded with angle
9689brackets ">" and "<" when the method is the default one and will work by
9690default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009691
Willy Tarreau74ca5042013-06-11 23:12:07 +02009692 +-------------------------------------------------+
9693 | Input sample type |
9694 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009695 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009696 +----------------------+---------+---------+---------+---------+---------+
9697 | none (presence only) | found | found | found | found | found |
9698 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009699 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009700 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009701 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009702 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009703 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009704 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009705 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009706 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009707 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009708 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009709 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009710 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009711 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009712 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009713 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009714 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009715 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009716 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009717 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009718 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009719 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009720 +----------------------+---------+---------+---------+---------+---------+
9721 | hex block | | | | bin | bin |
9722 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +02009723
9724
Willy Tarreau74ca5042013-06-11 23:12:07 +020097257.1.1. Matching booleans
9726------------------------
9727
9728In order to match a boolean, no value is needed and all values are ignored.
9729Boolean matching is used by default for all fetch methods of type "boolean".
9730When boolean matching is used, the fetched value is returned as-is, which means
9731that a boolean "true" will always match and a boolean "false" will never match.
9732
9733Boolean matching may also be enforced using "-m bool" on fetch methods which
9734return an integer value. Then, integer value 0 is converted to the boolean
9735"false" and all other values are converted to "true".
9736
Willy Tarreau6a06a402007-07-15 20:15:28 +02009737
Willy Tarreau74ca5042013-06-11 23:12:07 +020097387.1.2. Matching integers
9739------------------------
9740
9741Integer matching applies by default to integer fetch methods. It can also be
9742enforced on boolean fetches using "-m int". In this case, "false" is converted
9743to the integer 0, and "true" is converted to the integer 1.
9744
9745Integer matching also supports integer ranges and operators. Note that integer
9746matching only applies to positive values. A range is a value expressed with a
9747lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009748
9749For instance, "1024:65535" is a valid range to represent a range of
9750unprivileged ports, and "1024:" would also work. "0:1023" is a valid
9751representation of privileged ports, and ":1023" would also work.
9752
Willy Tarreau62644772008-07-16 18:36:06 +02009753As a special case, some ACL functions support decimal numbers which are in fact
9754two integers separated by a dot. This is used with some version checks for
9755instance. All integer properties apply to those decimal numbers, including
9756ranges and operators.
9757
Willy Tarreau6a06a402007-07-15 20:15:28 +02009758For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +01009759operators with ranges does not make much sense and is strongly discouraged.
9760Similarly, it does not make much sense to perform order comparisons with a set
9761of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009762
Willy Tarreau0ba27502007-12-24 16:55:16 +01009763Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009764
9765 eq : true if the tested value equals at least one value
9766 ge : true if the tested value is greater than or equal to at least one value
9767 gt : true if the tested value is greater than at least one value
9768 le : true if the tested value is less than or equal to at least one value
9769 lt : true if the tested value is less than at least one value
9770
Willy Tarreau0ba27502007-12-24 16:55:16 +01009771For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009772
9773 acl negative-length hdr_val(content-length) lt 0
9774
Willy Tarreau62644772008-07-16 18:36:06 +02009775This one matches SSL versions between 3.0 and 3.1 (inclusive) :
9776
9777 acl sslv3 req_ssl_ver 3:3.1
9778
Willy Tarreau6a06a402007-07-15 20:15:28 +02009779
Willy Tarreau74ca5042013-06-11 23:12:07 +020097807.1.3. Matching strings
9781-----------------------
9782
9783String matching applies to string or binary fetch methods, and exists in 6
9784different forms :
9785
9786 - exact match (-m str) : the extracted string must exactly match the
9787 patterns ;
9788
9789 - substring match (-m sub) : the patterns are looked up inside the
9790 extracted string, and the ACL matches if any of them is found inside ;
9791
9792 - prefix match (-m beg) : the patterns are compared with the beginning of
9793 the extracted string, and the ACL matches if any of them matches.
9794
9795 - suffix match (-m end) : the patterns are compared with the end of the
9796 extracted string, and the ACL matches if any of them matches.
9797
Baptiste Assmanna2fae942016-03-06 23:32:10 +01009798 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +02009799 string, delimited with slashes ("/"), and the ACL matches if any of them
9800 matches.
9801
9802 - domain match (-m dom) : the patterns are looked up inside the extracted
9803 string, delimited with dots ("."), and the ACL matches if any of them
9804 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009805
9806String matching applies to verbatim strings as they are passed, with the
9807exception of the backslash ("\") which makes it possible to escape some
9808characters such as the space. If the "-i" flag is passed before the first
9809string, then the matching will be performed ignoring the case. In order
9810to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +01009811before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009812
9813
Willy Tarreau74ca5042013-06-11 23:12:07 +020098147.1.4. Matching regular expressions (regexes)
9815---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009816
9817Just like with string matching, regex matching applies to verbatim strings as
9818they are passed, with the exception of the backslash ("\") which makes it
9819possible to escape some characters such as the space. If the "-i" flag is
9820passed before the first regex, then the matching will be performed ignoring
9821the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +01009822the "--" flag before the first string. Same principle applies of course to
9823match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009824
9825
Willy Tarreau74ca5042013-06-11 23:12:07 +020098267.1.5. Matching arbitrary data blocks
9827-------------------------------------
9828
9829It is possible to match some extracted samples against a binary block which may
9830not safely be represented as a string. For this, the patterns must be passed as
9831a series of hexadecimal digits in an even number, when the match method is set
9832to binary. Each sequence of two digits will represent a byte. The hexadecimal
9833digits may be used upper or lower case.
9834
9835Example :
9836 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
9837 acl hello payload(0,6) -m bin 48656c6c6f0a
9838
9839
98407.1.6. Matching IPv4 and IPv6 addresses
9841---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009842
9843IPv4 addresses values can be specified either as plain addresses or with a
9844netmask appended, in which case the IPv4 address matches whenever it is
9845within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009846host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +01009847difficult to read and debug configurations. If hostnames are used, you should
9848at least ensure that they are present in /etc/hosts so that the configuration
9849does not depend on any random DNS match at the moment the configuration is
9850parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009851
Willy Tarreauceb4ac92012-04-28 00:41:46 +02009852IPv6 may be entered in their usual form, with or without a netmask appended.
9853Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
9854trouble with randomly resolved IP addresses, host names are never allowed in
9855IPv6 patterns.
9856
9857HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
9858following situations :
9859 - tested address is IPv4, pattern address is IPv4, the match applies
9860 in IPv4 using the supplied mask if any.
9861 - tested address is IPv6, pattern address is IPv6, the match applies
9862 in IPv6 using the supplied mask if any.
9863 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
9864 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
9865 ::IPV4 or ::ffff:IPV4, otherwise it fails.
9866 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
9867 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
9868 applied in IPv6 using the supplied IPv6 mask.
9869
Willy Tarreau74ca5042013-06-11 23:12:07 +02009870
98717.2. Using ACLs to form conditions
9872----------------------------------
9873
9874Some actions are only performed upon a valid condition. A condition is a
9875combination of ACLs with operators. 3 operators are supported :
9876
9877 - AND (implicit)
9878 - OR (explicit with the "or" keyword or the "||" operator)
9879 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +02009880
Willy Tarreau74ca5042013-06-11 23:12:07 +02009881A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +02009882
Willy Tarreau74ca5042013-06-11 23:12:07 +02009883 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +02009884
Willy Tarreau74ca5042013-06-11 23:12:07 +02009885Such conditions are generally used after an "if" or "unless" statement,
9886indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +02009887
Willy Tarreau74ca5042013-06-11 23:12:07 +02009888For instance, to block HTTP requests to the "*" URL with methods other than
9889"OPTIONS", as well as POST requests without content-length, and GET or HEAD
9890requests with a content-length greater than 0, and finally every request which
9891is not either GET/HEAD/POST/OPTIONS !
9892
9893 acl missing_cl hdr_cnt(Content-length) eq 0
9894 block if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
9895 block if METH_GET HTTP_CONTENT
9896 block unless METH_GET or METH_POST or METH_OPTIONS
9897
9898To select a different backend for requests to static contents on the "www" site
9899and to every request on the "img", "video", "download" and "ftp" hosts :
9900
9901 acl url_static path_beg /static /images /img /css
9902 acl url_static path_end .gif .png .jpg .css .js
9903 acl host_www hdr_beg(host) -i www
9904 acl host_static hdr_beg(host) -i img. video. download. ftp.
9905
9906 # now use backend "static" for all static-only hosts, and for static urls
9907 # of host "www". Use backend "www" for the rest.
9908 use_backend static if host_static or host_www url_static
9909 use_backend www if host_www
9910
9911It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
9912expressions that are built on the fly without needing to be declared. They must
9913be enclosed between braces, with a space before and after each brace (because
9914the braces must be seen as independent words). Example :
9915
9916 The following rule :
9917
9918 acl missing_cl hdr_cnt(Content-length) eq 0
9919 block if METH_POST missing_cl
9920
9921 Can also be written that way :
9922
9923 block if METH_POST { hdr_cnt(Content-length) eq 0 }
9924
9925It is generally not recommended to use this construct because it's a lot easier
9926to leave errors in the configuration when written that way. However, for very
9927simple rules matching only one source IP address for instance, it can make more
9928sense to use them than to declare ACLs with random names. Another example of
9929good use is the following :
9930
9931 With named ACLs :
9932
9933 acl site_dead nbsrv(dynamic) lt 2
9934 acl site_dead nbsrv(static) lt 2
9935 monitor fail if site_dead
9936
9937 With anonymous ACLs :
9938
9939 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
9940
9941See section 4.2 for detailed help on the "block" and "use_backend" keywords.
9942
9943
99447.3. Fetching samples
9945---------------------
9946
9947Historically, sample fetch methods were only used to retrieve data to match
9948against patterns using ACLs. With the arrival of stick-tables, a new class of
9949sample fetch methods was created, most often sharing the same syntax as their
9950ACL counterpart. These sample fetch methods are also known as "fetches". As
9951of now, ACLs and fetches have converged. All ACL fetch methods have been made
9952available as fetch methods, and ACLs may use any sample fetch method as well.
9953
9954This section details all available sample fetch methods and their output type.
9955Some sample fetch methods have deprecated aliases that are used to maintain
9956compatibility with existing configurations. They are then explicitly marked as
9957deprecated and should not be used in new setups.
9958
9959The ACL derivatives are also indicated when available, with their respective
9960matching methods. These ones all have a well defined default pattern matching
9961method, so it is never necessary (though allowed) to pass the "-m" option to
9962indicate how the sample will be matched using ACLs.
9963
9964As indicated in the sample type versus matching compatibility matrix above,
9965when using a generic sample fetch method in an ACL, the "-m" option is
9966mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
9967the same keyword exists as an ACL keyword and as a standard fetch method, the
9968ACL engine will automatically pick the ACL-only one by default.
9969
9970Some of these keywords support one or multiple mandatory arguments, and one or
9971multiple optional arguments. These arguments are strongly typed and are checked
9972when the configuration is parsed so that there is no risk of running with an
9973incorrect argument (eg: an unresolved backend name). Fetch function arguments
9974are passed between parenthesis and are delimited by commas. When an argument
9975is optional, it will be indicated below between square brackets ('[ ]'). When
9976all arguments are optional, the parenthesis may be omitted.
9977
9978Thus, the syntax of a standard sample fetch method is one of the following :
9979 - name
9980 - name(arg1)
9981 - name(arg1,arg2)
9982
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009983
99847.3.1. Converters
9985-----------------
9986
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009987Sample fetch methods may be combined with transformations to be applied on top
9988of the fetched sample (also called "converters"). These combinations form what
9989is called "sample expressions" and the result is a "sample". Initially this
9990was only supported by "stick on" and "stick store-request" directives but this
9991has now be extended to all places where samples may be used (acls, log-format,
9992unique-id-format, add-header, ...).
9993
9994These transformations are enumerated as a series of specific keywords after the
9995sample fetch method. These keywords may equally be appended immediately after
9996the fetch keyword's argument, delimited by a comma. These keywords can also
9997support some arguments (eg: a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009998
Willy Tarreau74ca5042013-06-11 23:12:07 +02009999The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010010000
Emeric Brun53d1a982014-04-30 18:21:37 +020010001base64
10002 Converts a binary input sample to a base64 string. It is used to log or
10003 transfer binary content in a way that can be reliably transferred (eg:
10004 an SSL ID can be copied in a header).
10005
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010006hex
10007 Converts a binary input sample to an hex string containing two hex digits per
10008 input byte. It is used to log or transfer hex dumps of some binary input data
10009 in a way that can be reliably transferred (eg: an SSL ID can be copied in a
10010 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010010011
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010012http_date([<offset>])
10013 Converts an integer supposed to contain a date since epoch to a string
10014 representing this date in a format suitable for use in HTTP header fields. If
10015 an offset value is specified, then it is a number of seconds that is added to
10016 the date before the conversion is operated. This is particularly useful to
10017 emit Date header fields, Expires values in responses when combined with a
10018 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010019
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010020ipmask(<mask>)
10021 Apply a mask to an IPv4 address, and use the result for lookups and storage.
10022 This can be used to make all hosts within a certain mask to share the same
10023 table entries and as such use the same server. The mask can be passed in
10024 dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24).
10025
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010026language(<value>[,<default>])
10027 Returns the value with the highest q-factor from a list as extracted from the
10028 "accept-language" header using "req.fhdr". Values with no q-factor have a
10029 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
10030 belong to the list of semi-colon delimited <values> will be considered. The
10031 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
10032 given list and a default value is provided, it is returned. Note that language
10033 names may have a variant after a dash ('-'). If this variant is present in the
10034 list, it will be matched, but if it is not, only the base language is checked.
10035 The match is case-sensitive, and the output string is always one of those
10036 provided in arguments. The ordering of arguments is meaningless, only the
10037 ordering of the values in the request counts, as the first value among
10038 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020010039
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010040 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010041
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010042 # this configuration switches to the backend matching a
10043 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010044
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010045 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
10046 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
10047 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
10048 use_backend spanish if es
10049 use_backend french if fr
10050 use_backend english if en
10051 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020010052
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010053lower
10054 Convert a string sample to lower case. This can only be placed after a string
10055 sample fetch function or after a transformation keyword returning a string
10056 type. The result is of type string.
10057
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010058map(<map_file>[,<default_value>])
10059map_<match_type>(<map_file>[,<default_value>])
10060map_<match_type>_<output_type>(<map_file>[,<default_value>])
10061 Search the input value from <map_file> using the <match_type> matching method,
10062 and return the associated value converted to the type <output_type>. If the
10063 input value cannot be found in the <map_file>, the converter returns the
10064 <default_value>. If the <default_value> is not set, the converter fails and
10065 acts as if no input value could be fetched. If the <match_type> is not set, it
10066 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
10067 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
10068 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010069
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010070 It is important to avoid overlapping between the keys : IP addresses and
10071 strings are stored in trees, so the first of the finest match will be used.
10072 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010073
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010074 The following array contains the list of all map functions avalaible sorted by
10075 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010076
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010077 input type | match method | output type str | output type int | output type ip
10078 -----------+--------------+-----------------+-----------------+---------------
10079 str | str | map_str | map_str_int | map_str_ip
10080 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020010081 str | beg | map_beg | map_beg_int | map_end_ip
10082 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010083 str | sub | map_sub | map_sub_int | map_sub_ip
10084 -----------+--------------+-----------------+-----------------+---------------
10085 str | dir | map_dir | map_dir_int | map_dir_ip
10086 -----------+--------------+-----------------+-----------------+---------------
10087 str | dom | map_dom | map_dom_int | map_dom_ip
10088 -----------+--------------+-----------------+-----------------+---------------
10089 str | end | map_end | map_end_int | map_end_ip
10090 -----------+--------------+-----------------+-----------------+---------------
10091 str | reg | map_reg | map_reg_int | map_reg_ip
10092 -----------+--------------+-----------------+-----------------+---------------
10093 int | int | map_int | map_int_int | map_int_ip
10094 -----------+--------------+-----------------+-----------------+---------------
10095 ip | ip | map_ip | map_ip_int | map_ip_ip
10096 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010097
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010098 The file contains one key + value per line. Lines which start with '#' are
10099 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
10100 is then the first "word" (series of non-space/tabs characters), and the value
10101 is what follows this series of space/tab till the end of the line excluding
10102 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010103
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010104 Example :
10105
10106 # this is a comment and is ignored
10107 2.22.246.0/23 United Kingdom \n
10108 <-><-----------><--><------------><---->
10109 | | | | `- trailing spaces ignored
10110 | | | `---------- value
10111 | | `-------------------- middle spaces ignored
10112 | `---------------------------- key
10113 `------------------------------------ leading spaces ignored
10114
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010115upper
10116 Convert a string sample to upper case. This can only be placed after a string
10117 sample fetch function or after a transformation keyword returning a string
10118 type. The result is of type string.
10119
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010120
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200101217.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020010122--------------------------------------------
10123
10124A first set of sample fetch methods applies to internal information which does
10125not even relate to any client information. These ones are sometimes used with
10126"monitor-fail" directives to report an internal status to external watchers.
10127The sample fetch methods described in this section are usable anywhere.
10128
10129always_false : boolean
10130 Always returns the boolean "false" value. It may be used with ACLs as a
10131 temporary replacement for another one when adjusting configurations.
10132
10133always_true : boolean
10134 Always returns the boolean "true" value. It may be used with ACLs as a
10135 temporary replacement for another one when adjusting configurations.
10136
10137avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010138 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020010139 divided by the number of active servers. The current backend is used if no
10140 backend is specified. This is very similar to "queue" except that the size of
10141 the farm is considered, in order to give a more accurate measurement of the
10142 time it may take for a new connection to be processed. The main usage is with
10143 ACL to return a sorry page to new users when it becomes certain they will get
10144 a degraded service, or to pass to the backend servers in a header so that
10145 they decide to work in degraded mode or to disable some functions to speed up
10146 the processing a bit. Note that in the event there would not be any active
10147 server anymore, twice the number of queued connections would be considered as
10148 the measured value. This is a fair estimate, as we expect one server to get
10149 back soon anyway, but we still prefer to send new traffic to another backend
10150 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
10151 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010010152
Willy Tarreau74ca5042013-06-11 23:12:07 +020010153be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020010154 Applies to the number of currently established connections on the backend,
10155 possibly including the connection being evaluated. If no backend name is
10156 specified, the current one is used. But it is also possible to check another
10157 backend. It can be used to use a specific farm when the nominal one is full.
10158 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020010159
Willy Tarreau74ca5042013-06-11 23:12:07 +020010160be_sess_rate([<backend>]) : integer
10161 Returns an integer value corresponding to the sessions creation rate on the
10162 backend, in number of new sessions per second. This is used with ACLs to
10163 switch to an alternate backend when an expensive or fragile one reaches too
10164 high a session rate, or to limit abuse of service (eg. prevent sucking of an
10165 online dictionary). It can also be useful to add this element to logs using a
10166 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010167
10168 Example :
10169 # Redirect to an error page if the dictionary is requested too often
10170 backend dynamic
10171 mode http
10172 acl being_scanned be_sess_rate gt 100
10173 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010010174
Willy Tarreau74ca5042013-06-11 23:12:07 +020010175connslots([<backend>]) : integer
10176 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010177 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020010178 connections on all servers and the maximum queue size. This is probably only
10179 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050010180
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010181 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020010182 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010183 usage; see "use_backend" keyword) can be redirected to a different backend.
10184
Willy Tarreau55165fe2009-05-10 12:02:55 +020010185 'connslots' = number of available server connection slots, + number of
10186 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010187
Willy Tarreaua36af912009-10-10 12:02:45 +020010188 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020010189 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020010190 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020010191 you want to be able to differentiate between different backends, and their
10192 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020010193 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020010194 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010195
Willy Tarreau55165fe2009-05-10 12:02:55 +020010196 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
10197 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020010198 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020010199 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010200
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010201date([<offset>]) : integer
10202 Returns the current date as the epoch (number of seconds since 01/01/1970).
10203 If an offset value is specified, then it is a number of seconds that is added
10204 to the current date before returning the value. This is particularly useful
10205 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020010206 It is useful combined with the http_date converter.
10207
10208 Example :
10209
10210 # set an expires header to now+1 hour in every response
10211 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010212
Willy Tarreau595ec542013-06-12 21:34:28 +020010213env(<name>) : string
10214 Returns a string containing the value of environment variable <name>. As a
10215 reminder, environment variables are per-process and are sampled when the
10216 process starts. This can be useful to pass some information to a next hop
10217 server, or with ACLs to take specific action when the process is started a
10218 certain way.
10219
10220 Examples :
10221 # Pass the Via header to next hop with the local hostname in it
10222 http-request add-header Via 1.1\ %[env(HOSTNAME)]
10223
10224 # reject cookie-less requests when the STOP environment variable is set
10225 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
10226
Willy Tarreau74ca5042013-06-11 23:12:07 +020010227fe_conn([<frontend>]) : integer
10228 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010010229 possibly including the connection being evaluated. If no frontend name is
10230 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010231 frontend. It can be used to return a sorry page before hard-blocking, or to
10232 use a specific backend to drain new requests when the farm is considered
10233 full. This is mostly used with ACLs but can also be used to pass some
10234 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
10235 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020010236
Willy Tarreau74ca5042013-06-11 23:12:07 +020010237fe_sess_rate([<frontend>]) : integer
10238 Returns an integer value corresponding to the sessions creation rate on the
10239 frontend, in number of new sessions per second. This is used with ACLs to
10240 limit the incoming session rate to an acceptable range in order to prevent
10241 abuse of service at the earliest moment, for example when combined with other
10242 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
10243 down below the limit. It can also be useful to add this element to logs using
10244 a log-format directive. See also the "rate-limit sessions" directive for use
10245 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010246
10247 Example :
10248 # This frontend limits incoming mails to 10/s with a max of 100
10249 # concurrent connections. We accept any connection below 10/s, and
10250 # force excess clients to wait for 100 ms. Since clients are limited to
10251 # 100 max, there cannot be more than 10 incoming mails per second.
10252 frontend mail
10253 bind :25
10254 mode tcp
10255 maxconn 100
10256 acl too_fast fe_sess_rate ge 10
10257 tcp-request inspect-delay 100ms
10258 tcp-request content accept if ! too_fast
10259 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010010260
Willy Tarreaufad7f072014-11-24 16:02:05 +010010261nbproc : integer
10262 Returns an integer value corresponding to the number of processes that were
10263 started (it equals the global "nbproc" setting). This is useful for logging
10264 and debugging purposes.
10265
Willy Tarreau74ca5042013-06-11 23:12:07 +020010266nbsrv([<backend>]) : integer
10267 Returns an integer value corresponding to the number of usable servers of
10268 either the current backend or the named backend. This is mostly used with
10269 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010010270 switch to an alternate backend when the number of servers is too low to
10271 to handle some load. It is useful to report a failure when combined with
10272 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010273
Willy Tarreaufad7f072014-11-24 16:02:05 +010010274proc : integer
10275 Returns an integer value corresponding to the position of the process calling
10276 the function, between 1 and global.nbproc. This is useful for logging and
10277 debugging purposes.
10278
Willy Tarreau74ca5042013-06-11 23:12:07 +020010279queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010280 Returns the total number of queued connections of the designated backend,
10281 including all the connections in server queues. If no backend name is
10282 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010283 one. This is useful with ACLs or to pass statistics to backend servers. This
10284 can be used to take actions when queuing goes above a known level, generally
10285 indicating a surge of traffic or a massive slowdown on the servers. One
10286 possible action could be to reject new users but still accept old ones. See
10287 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
10288
Willy Tarreau84310e22014-02-14 11:59:04 +010010289rand([<range>]) : integer
10290 Returns a random integer value within a range of <range> possible values,
10291 starting at zero. If the range is not specified, it defaults to 2^32, which
10292 gives numbers between 0 and 4294967295. It can be useful to pass some values
10293 needed to take some routing decisions for example, or just for debugging
10294 purposes. This random must not be used for security purposes.
10295
Willy Tarreau74ca5042013-06-11 23:12:07 +020010296srv_conn([<backend>/]<server>) : integer
10297 Returns an integer value corresponding to the number of currently established
10298 connections on the designated server, possibly including the connection being
10299 evaluated. If <backend> is omitted, then the server is looked up in the
10300 current backend. It can be used to use a specific farm when one server is
10301 full, or to inform the server about our view of the number of active
10302 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
10303 methods.
10304
10305srv_is_up([<backend>/]<server>) : boolean
10306 Returns true when the designated server is UP, and false when it is either
10307 DOWN or in maintenance mode. If <backend> is omitted, then the server is
10308 looked up in the current backend. It is mainly used to take action based on
10309 an external status reported via a health check (eg: a geographical site's
10310 availability). Another possible use which is more of a hack consists in
10311 using dummy servers as boolean variables that can be enabled or disabled from
10312 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
10313
10314srv_sess_rate([<backend>/]<server>) : integer
10315 Returns an integer corresponding to the sessions creation rate on the
10316 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010317 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020010318 used with ACLs but can make sense with logs too. This is used to switch to an
10319 alternate backend when an expensive or fragile one reaches too high a session
10320 rate, or to limit abuse of service (eg. prevent latent requests from
10321 overloading servers).
10322
10323 Example :
10324 # Redirect to a separate back
10325 acl srv1_full srv_sess_rate(be1/srv1) gt 50
10326 acl srv2_full srv_sess_rate(be1/srv2) gt 50
10327 use_backend be2 if srv1_full or srv2_full
10328
Willy Tarreaufad7f072014-11-24 16:02:05 +010010329stopping : boolean
10330 Returns TRUE if the process calling the function is currently stopping. This
10331 can be useful for logging, or for relaxing certain checks or helping close
10332 certain connections upon graceful shutdown.
10333
Willy Tarreau74ca5042013-06-11 23:12:07 +020010334table_avl([<table>]) : integer
10335 Returns the total number of available entries in the current proxy's
10336 stick-table or in the designated stick-table. See also table_cnt.
10337
10338table_cnt([<table>]) : integer
10339 Returns the total number of entries currently in use in the current proxy's
10340 stick-table or in the designated stick-table. See also src_conn_cnt and
10341 table_avl for other entry counting methods.
10342
10343
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200103447.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020010345----------------------------------
10346
10347The layer 4 usually describes just the transport layer which in haproxy is
10348closest to the connection, where no content is yet made available. The fetch
10349methods described here are usable as low as the "tcp-request connection" rule
10350sets unless they require some future information. Those generally include
10351TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010352the incoming connection. For retrieving a value from a sticky counters, the
10353counter number can be explicitly set as 0, 1, or 2 using the pre-defined
10354"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010355argument when using the "sc_" prefix. An optional table may be specified with
10356the "sc*" form, in which case the currently tracked key will be looked up into
10357this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010358
10359be_id : integer
10360 Returns an integer containing the current backend's id. It can be used in
10361 frontends with responses to check which backend processed the request.
10362
10363dst : ip
10364 This is the destination IPv4 address of the connection on the client side,
10365 which is the address the client connected to. It can be useful when running
10366 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
10367 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
10368 RFC 4291.
10369
10370dst_conn : integer
10371 Returns an integer value corresponding to the number of currently established
10372 connections on the same socket including the one being evaluated. It is
10373 normally used with ACLs but can as well be used to pass the information to
10374 servers in an HTTP header or in logs. It can be used to either return a sorry
10375 page before hard-blocking, or to use a specific backend to drain new requests
10376 when the socket is considered saturated. This offers the ability to assign
10377 different limits to different listening ports or addresses. See also the
10378 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010379
Willy Tarreau74ca5042013-06-11 23:12:07 +020010380dst_port : integer
10381 Returns an integer value corresponding to the destination TCP port of the
10382 connection on the client side, which is the port the client connected to.
10383 This might be used when running in transparent mode, when assigning dynamic
10384 ports to some clients for a whole application session, to stick all users to
10385 a same server, or to pass the destination port information to a server using
10386 an HTTP header.
10387
10388fe_id : integer
10389 Returns an integer containing the current frontend's id. It can be used in
10390 backends to check from which backend it was called, or to stick all users
10391 coming via a same frontend to the same server.
10392
Cyril Bonté62ba8702014-04-22 23:52:25 +020010393sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010394sc0_bytes_in_rate([<table>]) : integer
10395sc1_bytes_in_rate([<table>]) : integer
10396sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010397 Returns the average client-to-server bytes rate from the currently tracked
10398 counters, measured in amount of bytes over the period configured in the
10399 table. See also src_bytes_in_rate.
10400
Cyril Bonté62ba8702014-04-22 23:52:25 +020010401sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010402sc0_bytes_out_rate([<table>]) : integer
10403sc1_bytes_out_rate([<table>]) : integer
10404sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010405 Returns the average server-to-client bytes rate from the currently tracked
10406 counters, measured in amount of bytes over the period configured in the
10407 table. See also src_bytes_out_rate.
10408
Cyril Bonté62ba8702014-04-22 23:52:25 +020010409sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010410sc0_clr_gpc0([<table>]) : integer
10411sc1_clr_gpc0([<table>]) : integer
10412sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020010413 Clears the first General Purpose Counter associated to the currently tracked
10414 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010010415 stored value is zero, so first invocation will always return zero. This is
10416 typically used as a second ACL in an expression in order to mark a connection
10417 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010418
10419 # block if 5 consecutive requests continue to come faster than 10 sess
10420 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010421 acl abuse sc0_http_req_rate gt 10
10422 acl kill sc0_inc_gpc0 gt 5
10423 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010424 tcp-request connection accept if !abuse save
10425 tcp-request connection reject if abuse kill
10426
Cyril Bonté62ba8702014-04-22 23:52:25 +020010427sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010428sc0_conn_cnt([<table>]) : integer
10429sc1_conn_cnt([<table>]) : integer
10430sc2_conn_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010431 Returns the cumulated number of incoming connections from currently tracked
10432 counters. See also src_conn_cnt.
10433
Cyril Bonté62ba8702014-04-22 23:52:25 +020010434sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010435sc0_conn_cur([<table>]) : integer
10436sc1_conn_cur([<table>]) : integer
10437sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010438 Returns the current amount of concurrent connections tracking the same
10439 tracked counters. This number is automatically incremented when tracking
10440 begins and decremented when tracking stops. See also src_conn_cur.
10441
Cyril Bonté62ba8702014-04-22 23:52:25 +020010442sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010443sc0_conn_rate([<table>]) : integer
10444sc1_conn_rate([<table>]) : integer
10445sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010446 Returns the average connection rate from the currently tracked counters,
10447 measured in amount of connections over the period configured in the table.
10448 See also src_conn_rate.
10449
Cyril Bonté62ba8702014-04-22 23:52:25 +020010450sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010451sc0_get_gpc0([<table>]) : integer
10452sc1_get_gpc0([<table>]) : integer
10453sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010454 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010455 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010456
Cyril Bonté62ba8702014-04-22 23:52:25 +020010457sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010458sc0_gpc0_rate([<table>]) : integer
10459sc1_gpc0_rate([<table>]) : integer
10460sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010461 Returns the average increment rate of the first General Purpose Counter
10462 associated to the currently tracked counters. It reports the frequency
10463 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010464 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10465 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10466 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020010467
Cyril Bonté62ba8702014-04-22 23:52:25 +020010468sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010469sc0_http_err_cnt([<table>]) : integer
10470sc1_http_err_cnt([<table>]) : integer
10471sc2_http_err_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010472 Returns the cumulated number of HTTP errors from the currently tracked
10473 counters. This includes the both request errors and 4xx error responses.
10474 See also src_http_err_cnt.
10475
Cyril Bonté62ba8702014-04-22 23:52:25 +020010476sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010477sc0_http_err_rate([<table>]) : integer
10478sc1_http_err_rate([<table>]) : integer
10479sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010480 Returns the average rate of HTTP errors from the currently tracked counters,
10481 measured in amount of errors over the period configured in the table. This
10482 includes the both request errors and 4xx error responses. See also
10483 src_http_err_rate.
10484
Cyril Bonté62ba8702014-04-22 23:52:25 +020010485sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010486sc0_http_req_cnt([<table>]) : integer
10487sc1_http_req_cnt([<table>]) : integer
10488sc2_http_req_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010489 Returns the cumulated number of HTTP requests from the currently tracked
10490 counters. This includes every started request, valid or not. See also
10491 src_http_req_cnt.
10492
Cyril Bonté62ba8702014-04-22 23:52:25 +020010493sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010494sc0_http_req_rate([<table>]) : integer
10495sc1_http_req_rate([<table>]) : integer
10496sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010497 Returns the average rate of HTTP requests from the currently tracked
10498 counters, measured in amount of requests over the period configured in
10499 the table. This includes every started request, valid or not. See also
10500 src_http_req_rate.
10501
Cyril Bonté62ba8702014-04-22 23:52:25 +020010502sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010503sc0_inc_gpc0([<table>]) : integer
10504sc1_inc_gpc0([<table>]) : integer
10505sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010506 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010010507 tracked counters, and returns its new value. Before the first invocation,
10508 the stored value is zero, so first invocation will increase it to 1 and will
10509 return 1. This is typically used as a second ACL in an expression in order
10510 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020010511
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010512 acl abuse sc0_http_req_rate gt 10
10513 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010514 tcp-request connection reject if abuse kill
10515
Cyril Bonté62ba8702014-04-22 23:52:25 +020010516sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010517sc0_kbytes_in([<table>]) : integer
10518sc1_kbytes_in([<table>]) : integer
10519sc2_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010520 Returns the total amount of client-to-server data from the currently tracked
10521 counters, measured in kilobytes. The test is currently performed on 32-bit
10522 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020010523
Cyril Bonté62ba8702014-04-22 23:52:25 +020010524sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010525sc0_kbytes_out([<table>]) : integer
10526sc1_kbytes_out([<table>]) : integer
10527sc2_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010528 Returns the total amount of server-to-client data from the currently tracked
10529 counters, measured in kilobytes. The test is currently performed on 32-bit
10530 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020010531
Cyril Bonté62ba8702014-04-22 23:52:25 +020010532sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010533sc0_sess_cnt([<table>]) : integer
10534sc1_sess_cnt([<table>]) : integer
10535sc2_sess_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010536 Returns the cumulated number of incoming connections that were transformed
10537 into sessions, which means that they were accepted by a "tcp-request
10538 connection" rule, from the currently tracked counters. A backend may count
10539 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010540 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020010541 with the client. See also src_sess_cnt.
10542
Cyril Bonté62ba8702014-04-22 23:52:25 +020010543sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010544sc0_sess_rate([<table>]) : integer
10545sc1_sess_rate([<table>]) : integer
10546sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010547 Returns the average session rate from the currently tracked counters,
10548 measured in amount of sessions over the period configured in the table. A
10549 session is a connection that got past the early "tcp-request connection"
10550 rules. A backend may count more sessions than connections because each
10551 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010552 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020010553
Cyril Bonté62ba8702014-04-22 23:52:25 +020010554sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020010555sc0_tracked([<table>]) : boolean
10556sc1_tracked([<table>]) : boolean
10557sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020010558 Returns true if the designated session counter is currently being tracked by
10559 the current session. This can be useful when deciding whether or not we want
10560 to set some values in a header passed to the server.
10561
Cyril Bonté62ba8702014-04-22 23:52:25 +020010562sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010563sc0_trackers([<table>]) : integer
10564sc1_trackers([<table>]) : integer
10565sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010010566 Returns the current amount of concurrent connections tracking the same
10567 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010568 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010010569 that it does not rely on any stored information but on the table's reference
10570 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020010571 may sometimes be more suited for layer7 tracking. It can be used to tell a
10572 server how many concurrent connections there are from a given address for
10573 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010010574
Willy Tarreau74ca5042013-06-11 23:12:07 +020010575so_id : integer
10576 Returns an integer containing the current listening socket's id. It is useful
10577 in frontends involving many "bind" lines, or to stick all users coming via a
10578 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010579
Willy Tarreau74ca5042013-06-11 23:12:07 +020010580src : ip
10581 This is the source IPv4 address of the client of the session. It is of type
10582 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
10583 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
10584 TCP-level source address which is used, and not the address of a client
10585 behind a proxy. However if the "accept-proxy" bind directive is used, it can
10586 be the address of a client behind another PROXY-protocol compatible component
10587 for all rule sets except "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010588
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010589 Example:
10590 # add an HTTP header in requests with the originating address' country
10591 http-request set-header X-Country %[src,map_ip(geoip.lst)]
10592
Willy Tarreau74ca5042013-06-11 23:12:07 +020010593src_bytes_in_rate([<table>]) : integer
10594 Returns the average bytes rate from the incoming connection's source address
10595 in the current proxy's stick-table or in the designated stick-table, measured
10596 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010597 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010598
Willy Tarreau74ca5042013-06-11 23:12:07 +020010599src_bytes_out_rate([<table>]) : integer
10600 Returns the average bytes rate to the incoming connection's source address in
10601 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010602 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010603 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010604
Willy Tarreau74ca5042013-06-11 23:12:07 +020010605src_clr_gpc0([<table>]) : integer
10606 Clears the first General Purpose Counter associated to the incoming
10607 connection's source address in the current proxy's stick-table or in the
10608 designated stick-table, and returns its previous value. If the address is not
10609 found, an entry is created and 0 is returned. This is typically used as a
10610 second ACL in an expression in order to mark a connection when a first ACL
10611 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010612
10613 # block if 5 consecutive requests continue to come faster than 10 sess
10614 # per second, and reset the counter as soon as the traffic slows down.
10615 acl abuse src_http_req_rate gt 10
10616 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010010617 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010618 tcp-request connection accept if !abuse save
10619 tcp-request connection reject if abuse kill
10620
Willy Tarreau74ca5042013-06-11 23:12:07 +020010621src_conn_cnt([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010622 Returns the cumulated number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020010623 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010624 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010625 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010626
Willy Tarreau74ca5042013-06-11 23:12:07 +020010627src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010628 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010629 current incoming connection's source address in the current proxy's
10630 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010631 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010632
Willy Tarreau74ca5042013-06-11 23:12:07 +020010633src_conn_rate([<table>]) : integer
10634 Returns the average connection rate from the incoming connection's source
10635 address in the current proxy's stick-table or in the designated stick-table,
10636 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010637 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010638
Willy Tarreau74ca5042013-06-11 23:12:07 +020010639src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010640 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010641 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010642 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010643 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010644
Willy Tarreau74ca5042013-06-11 23:12:07 +020010645src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010646 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020010647 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010648 stick-table or in the designated stick-table. It reports the frequency
10649 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010650 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10651 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10652 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010653
Willy Tarreau74ca5042013-06-11 23:12:07 +020010654src_http_err_cnt([<table>]) : integer
10655 Returns the cumulated number of HTTP errors from the incoming connection's
10656 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020010657 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010658 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020010659 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010660
Willy Tarreau74ca5042013-06-11 23:12:07 +020010661src_http_err_rate([<table>]) : integer
10662 Returns the average rate of HTTP errors from the incoming connection's source
10663 address in the current proxy's stick-table or in the designated stick-table,
10664 measured in amount of errors over the period configured in the table. This
10665 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010666 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010667
Willy Tarreau74ca5042013-06-11 23:12:07 +020010668src_http_req_cnt([<table>]) : integer
10669 Returns the cumulated number of HTTP requests from the incoming connection's
10670 source address in the current proxy's stick-table or in the designated stick-
10671 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010672 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010673
Willy Tarreau74ca5042013-06-11 23:12:07 +020010674src_http_req_rate([<table>]) : integer
10675 Returns the average rate of HTTP requests from the incoming connection's
10676 source address in the current proxy's stick-table or in the designated stick-
10677 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020010678 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010679 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010680
Willy Tarreau74ca5042013-06-11 23:12:07 +020010681src_inc_gpc0([<table>]) : integer
10682 Increments the first General Purpose Counter associated to the incoming
10683 connection's source address in the current proxy's stick-table or in the
10684 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010685 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010686 This is typically used as a second ACL in an expression in order to mark a
10687 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020010688
10689 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010010690 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010691 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020010692
Willy Tarreau74ca5042013-06-11 23:12:07 +020010693src_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010694 Returns the total amount of data received from the incoming connection's
10695 source address in the current proxy's stick-table or in the designated
10696 stick-table, measured in kilobytes. If the address is not found, zero is
10697 returned. The test is currently performed on 32-bit integers, which limits
10698 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010699
Willy Tarreau74ca5042013-06-11 23:12:07 +020010700src_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010701 Returns the total amount of data sent to the incoming connection's source
10702 address in the current proxy's stick-table or in the designated stick-table,
10703 measured in kilobytes. If the address is not found, zero is returned. The
10704 test is currently performed on 32-bit integers, which limits values to 4
10705 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010706
Willy Tarreau74ca5042013-06-11 23:12:07 +020010707src_port : integer
10708 Returns an integer value corresponding to the TCP source port of the
10709 connection on the client side, which is the port the client connected from.
10710 Usage of this function is very limited as modern protocols do not care much
10711 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010712
Willy Tarreau74ca5042013-06-11 23:12:07 +020010713src_sess_cnt([<table>]) : integer
10714 Returns the cumulated number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020010715 connection's source IPv4 address in the current proxy's stick-table or in the
10716 designated stick-table, that were transformed into sessions, which means that
10717 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010718 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010719
Willy Tarreau74ca5042013-06-11 23:12:07 +020010720src_sess_rate([<table>]) : integer
10721 Returns the average session rate from the incoming connection's source
10722 address in the current proxy's stick-table or in the designated stick-table,
10723 measured in amount of sessions over the period configured in the table. A
10724 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010725 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010726
Willy Tarreau74ca5042013-06-11 23:12:07 +020010727src_updt_conn_cnt([<table>]) : integer
10728 Creates or updates the entry associated to the incoming connection's source
10729 address in the current proxy's stick-table or in the designated stick-table.
10730 This table must be configured to store the "conn_cnt" data type, otherwise
10731 the match will be ignored. The current count is incremented by one, and the
10732 expiration timer refreshed. The updated count is returned, so this match
10733 can't return zero. This was used to reject service abusers based on their
10734 source address. Note: it is recommended to use the more complete "track-sc*"
10735 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010736
10737 Example :
10738 # This frontend limits incoming SSH connections to 3 per 10 second for
10739 # each source address, and rejects excess connections until a 10 second
10740 # silence is observed. At most 20 addresses are tracked.
10741 listen ssh
10742 bind :22
10743 mode tcp
10744 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020010745 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020010746 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010747 server local 127.0.0.1:22
10748
Willy Tarreau74ca5042013-06-11 23:12:07 +020010749srv_id : integer
10750 Returns an integer containing the server's id when processing the response.
10751 While it's almost only used with ACLs, it may be used for logging or
10752 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020010753
Hervé COMMOWICK35ed8012010-12-15 14:04:51 +010010754
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200107557.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020010756----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020010757
Willy Tarreau74ca5042013-06-11 23:12:07 +020010758The layer 5 usually describes just the session layer which in haproxy is
10759closest to the session once all the connection handshakes are finished, but
10760when no content is yet made available. The fetch methods described here are
10761usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010762future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020010763
Emeric Brun645ae792014-04-30 14:21:06 +020010764ssl_bc : boolean
10765 Returns true when the back connection was made via an SSL/TLS transport
10766 layer and is locally deciphered. This means the outgoing connection was made
10767 other a server with the "ssl" option.
10768
10769ssl_bc_alg_keysize : integer
10770 Returns the symmetric cipher key size supported in bits when the outgoing
10771 connection was made over an SSL/TLS transport layer.
10772
10773ssl_bc_cipher : string
10774 Returns the name of the used cipher when the outgoing connection was made
10775 over an SSL/TLS transport layer.
10776
10777ssl_bc_protocol : string
10778 Returns the name of the used protocol when the outgoing connection was made
10779 over an SSL/TLS transport layer.
10780
Emeric Brunb73a9b02014-04-30 18:49:19 +020010781ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020010782 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010783 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10784 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020010785
10786ssl_bc_session_id : binary
10787 Returns the SSL ID of the back connection when the outgoing connection was
10788 made over an SSL/TLS transport layer. It is useful to log if we want to know
10789 if session was reused or not.
10790
10791ssl_bc_use_keysize : integer
10792 Returns the symmetric cipher key size used in bits when the outgoing
10793 connection was made over an SSL/TLS transport layer.
10794
Willy Tarreau74ca5042013-06-11 23:12:07 +020010795ssl_c_ca_err : integer
10796 When the incoming connection was made over an SSL/TLS transport layer,
10797 returns the ID of the first error detected during verification of the client
10798 certificate at depth > 0, or 0 if no error was encountered during this
10799 verification process. Please refer to your SSL library's documentation to
10800 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020010801
Willy Tarreau74ca5042013-06-11 23:12:07 +020010802ssl_c_ca_err_depth : integer
10803 When the incoming connection was made over an SSL/TLS transport layer,
10804 returns the depth in the CA chain of the first error detected during the
10805 verification of the client certificate. If no error is encountered, 0 is
10806 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010807
Emeric Brunb3cc4252014-10-29 19:03:26 +010010808ssl_c_der : binary
10809 Returns the DER formatted certificate presented by the client when the
10810 incoming connection was made over an SSL/TLS transport layer. When used for
10811 an ACL, the value(s) to match against can be passed in hexadecimal form.
10812
Willy Tarreau74ca5042013-06-11 23:12:07 +020010813ssl_c_err : integer
10814 When the incoming connection was made over an SSL/TLS transport layer,
10815 returns the ID of the first error detected during verification at depth 0, or
10816 0 if no error was encountered during this verification process. Please refer
10817 to your SSL library's documentation to find the exhaustive list of error
10818 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020010819
Willy Tarreau74ca5042013-06-11 23:12:07 +020010820ssl_c_i_dn([<entry>[,<occ>]]) : string
10821 When the incoming connection was made over an SSL/TLS transport layer,
10822 returns the full distinguished name of the issuer of the certificate
10823 presented by the client when no <entry> is specified, or the value of the
10824 first given entry found from the beginning of the DN. If a positive/negative
10825 occurrence number is specified as the optional second argument, it returns
10826 the value of the nth given entry value from the beginning/end of the DN.
10827 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
10828 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020010829
Willy Tarreau74ca5042013-06-11 23:12:07 +020010830ssl_c_key_alg : string
10831 Returns the name of the algorithm used to generate the key of the certificate
10832 presented by the client when the incoming connection was made over an SSL/TLS
10833 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020010834
Willy Tarreau74ca5042013-06-11 23:12:07 +020010835ssl_c_notafter : string
10836 Returns the end date presented by the client as a formatted string
10837 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10838 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020010839
Willy Tarreau74ca5042013-06-11 23:12:07 +020010840ssl_c_notbefore : string
10841 Returns the start date presented by the client as a formatted string
10842 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10843 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010844
Willy Tarreau74ca5042013-06-11 23:12:07 +020010845ssl_c_s_dn([<entry>[,<occ>]]) : string
10846 When the incoming connection was made over an SSL/TLS transport layer,
10847 returns the full distinguished name of the subject of the certificate
10848 presented by the client when no <entry> is specified, or the value of the
10849 first given entry found from the beginning of the DN. If a positive/negative
10850 occurrence number is specified as the optional second argument, it returns
10851 the value of the nth given entry value from the beginning/end of the DN.
10852 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
10853 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010854
Willy Tarreau74ca5042013-06-11 23:12:07 +020010855ssl_c_serial : binary
10856 Returns the serial of the certificate presented by the client when the
10857 incoming connection was made over an SSL/TLS transport layer. When used for
10858 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010859
Willy Tarreau74ca5042013-06-11 23:12:07 +020010860ssl_c_sha1 : binary
10861 Returns the SHA-1 fingerprint of the certificate presented by the client when
10862 the incoming connection was made over an SSL/TLS transport layer. This can be
10863 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau9fe4cb62014-07-02 19:01:22 +020010864 Note that the output is binary, so if you want to pass that signature to the
10865 server, you need to encode it in hex or base64, such as in the example below:
10866
10867 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020010868
Willy Tarreau74ca5042013-06-11 23:12:07 +020010869ssl_c_sig_alg : string
10870 Returns the name of the algorithm used to sign the certificate presented by
10871 the client when the incoming connection was made over an SSL/TLS transport
10872 layer.
Emeric Brun87855892012-10-17 17:39:35 +020010873
Willy Tarreau74ca5042013-06-11 23:12:07 +020010874ssl_c_used : boolean
10875 Returns true if current SSL session uses a client certificate even if current
10876 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020010877
Willy Tarreau74ca5042013-06-11 23:12:07 +020010878ssl_c_verify : integer
10879 Returns the verify result error ID when the incoming connection was made over
10880 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
10881 refer to your SSL library's documentation for an exhaustive list of error
10882 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020010883
Willy Tarreau74ca5042013-06-11 23:12:07 +020010884ssl_c_version : integer
10885 Returns the version of the certificate presented by the client when the
10886 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020010887
Emeric Brunb3cc4252014-10-29 19:03:26 +010010888ssl_f_der : binary
10889 Returns the DER formatted certificate presented by the frontend when the
10890 incoming connection was made over an SSL/TLS transport layer. When used for
10891 an ACL, the value(s) to match against can be passed in hexadecimal form.
10892
Willy Tarreau74ca5042013-06-11 23:12:07 +020010893ssl_f_i_dn([<entry>[,<occ>]]) : string
10894 When the incoming connection was made over an SSL/TLS transport layer,
10895 returns the full distinguished name of the issuer of the certificate
10896 presented by the frontend when no <entry> is specified, or the value of the
10897 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020010898 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020010899 the value of the nth given entry value from the beginning/end of the DN.
10900 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
10901 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020010902
Willy Tarreau74ca5042013-06-11 23:12:07 +020010903ssl_f_key_alg : string
10904 Returns the name of the algorithm used to generate the key of the certificate
10905 presented by the frontend when the incoming connection was made over an
10906 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010907
Willy Tarreau74ca5042013-06-11 23:12:07 +020010908ssl_f_notafter : string
10909 Returns the end date presented by the frontend as a formatted string
10910 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10911 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010912
Willy Tarreau74ca5042013-06-11 23:12:07 +020010913ssl_f_notbefore : string
10914 Returns the start date presented by the frontend as a formatted string
10915 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10916 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020010917
Willy Tarreau74ca5042013-06-11 23:12:07 +020010918ssl_f_s_dn([<entry>[,<occ>]]) : string
10919 When the incoming connection was made over an SSL/TLS transport layer,
10920 returns the full distinguished name of the subject of the certificate
10921 presented by the frontend when no <entry> is specified, or the value of the
10922 first given entry found from the beginning of the DN. If a positive/negative
10923 occurrence number is specified as the optional second argument, it returns
10924 the value of the nth given entry value from the beginning/end of the DN.
10925 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
10926 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020010927
Willy Tarreau74ca5042013-06-11 23:12:07 +020010928ssl_f_serial : binary
10929 Returns the serial of the certificate presented by the frontend when the
10930 incoming connection was made over an SSL/TLS transport layer. When used for
10931 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020010932
Emeric Brun55f4fa82014-04-30 17:11:25 +020010933ssl_f_sha1 : binary
10934 Returns the SHA-1 fingerprint of the certificate presented by the frontend
10935 when the incoming connection was made over an SSL/TLS transport layer. This
10936 can be used to know which certificate was chosen using SNI.
10937
Willy Tarreau74ca5042013-06-11 23:12:07 +020010938ssl_f_sig_alg : string
10939 Returns the name of the algorithm used to sign the certificate presented by
10940 the frontend when the incoming connection was made over an SSL/TLS transport
10941 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010942
Willy Tarreau74ca5042013-06-11 23:12:07 +020010943ssl_f_version : integer
10944 Returns the version of the certificate presented by the frontend when the
10945 incoming connection was made over an SSL/TLS transport layer.
10946
10947ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010948 Returns true when the front connection was made via an SSL/TLS transport
10949 layer and is locally deciphered. This means it has matched a socket declared
10950 with a "bind" line having the "ssl" option.
10951
Willy Tarreau74ca5042013-06-11 23:12:07 +020010952 Example :
10953 # This passes "X-Proto: https" to servers when client connects over SSL
10954 listen http-https
10955 bind :80
10956 bind :443 ssl crt /etc/haproxy.pem
10957 http-request add-header X-Proto https if { ssl_fc }
10958
10959ssl_fc_alg_keysize : integer
10960 Returns the symmetric cipher key size supported in bits when the incoming
10961 connection was made over an SSL/TLS transport layer.
10962
10963ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010964 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020010965 incoming connection made via a TLS transport layer and locally deciphered by
10966 haproxy. The result is a string containing the protocol name advertised by
10967 the client. The SSL library must have been built with support for TLS
10968 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
10969 not advertised unless the "alpn" keyword on the "bind" line specifies a
10970 protocol list. Also, nothing forces the client to pick a protocol from this
10971 list, any other one may be requested. The TLS ALPN extension is meant to
10972 replace the TLS NPN extension. See also "ssl_fc_npn".
10973
Willy Tarreau74ca5042013-06-11 23:12:07 +020010974ssl_fc_cipher : string
10975 Returns the name of the used cipher when the incoming connection was made
10976 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020010977
Willy Tarreau74ca5042013-06-11 23:12:07 +020010978ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010979 Returns true if a client certificate is present in an incoming connection over
10980 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010010981 Note: on SSL session resumption with Session ID or TLS ticket, client
10982 certificate is not present in the current connection but may be retrieved
10983 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
10984 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010985
Willy Tarreau74ca5042013-06-11 23:12:07 +020010986ssl_fc_has_sni : boolean
10987 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020010988 in an incoming connection was made over an SSL/TLS transport layer. Returns
10989 true when the incoming connection presents a TLS SNI field. This requires
10990 that the SSL library is build with support for TLS extensions enabled (check
10991 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020010992
Willy Tarreau74ca5042013-06-11 23:12:07 +020010993ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010994 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020010995 made via a TLS transport layer and locally deciphered by haproxy. The result
10996 is a string containing the protocol name advertised by the client. The SSL
10997 library must have been built with support for TLS extensions enabled (check
10998 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
10999 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
11000 forces the client to pick a protocol from this list, any other one may be
11001 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020011002
Willy Tarreau74ca5042013-06-11 23:12:07 +020011003ssl_fc_protocol : string
11004 Returns the name of the used protocol when the incoming connection was made
11005 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020011006
Emeric Brunb73a9b02014-04-30 18:49:19 +020011007ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040011008 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020011009 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
11010 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040011011
Willy Tarreau74ca5042013-06-11 23:12:07 +020011012ssl_fc_session_id : binary
11013 Returns the SSL ID of the front connection when the incoming connection was
11014 made over an SSL/TLS transport layer. It is useful to stick a given client to
11015 a server. It is important to note that some browsers refresh their session ID
11016 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020011017
Willy Tarreau74ca5042013-06-11 23:12:07 +020011018ssl_fc_sni : string
11019 This extracts the Server Name Indication TLS extension (SNI) field from an
11020 incoming connection made via an SSL/TLS transport layer and locally
11021 deciphered by haproxy. The result (when present) typically is a string
11022 matching the HTTPS host name (253 chars or less). The SSL library must have
11023 been built with support for TLS extensions enabled (check haproxy -vv).
11024
11025 This fetch is different from "req_ssl_sni" above in that it applies to the
11026 connection being deciphered by haproxy and not to SSL contents being blindly
11027 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020011028 requires that the SSL library is build with support for TLS extensions
11029 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020011030
Willy Tarreau74ca5042013-06-11 23:12:07 +020011031 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020011032 ssl_fc_sni_end : suffix match
11033 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020011034
Willy Tarreau74ca5042013-06-11 23:12:07 +020011035ssl_fc_use_keysize : integer
11036 Returns the symmetric cipher key size used in bits when the incoming
11037 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011038
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011039
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200110407.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011041------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011042
Willy Tarreau74ca5042013-06-11 23:12:07 +020011043Fetching samples from buffer contents is a bit different from the previous
11044sample fetches above because the sampled data are ephemeral. These data can
11045only be used when they're available and will be lost when they're forwarded.
11046For this reason, samples fetched from buffer contents during a request cannot
11047be used in a response for example. Even while the data are being fetched, they
11048can change. Sometimes it is necessary to set some delays or combine multiple
11049sample fetch methods to ensure that the expected data are complete and usable,
11050for example through TCP request content inspection. Please see the "tcp-request
11051content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020011052
Willy Tarreau74ca5042013-06-11 23:12:07 +020011053payload(<offset>,<length>) : binary (deprecated)
11054 This is an alias for "req.payload" when used in the context of a request (eg:
11055 "stick on", "stick match"), and for "res.payload" when used in the context of
11056 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011057
Willy Tarreau74ca5042013-06-11 23:12:07 +020011058payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
11059 This is an alias for "req.payload_lv" when used in the context of a request
11060 (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the
11061 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011062
Willy Tarreau74ca5042013-06-11 23:12:07 +020011063req.len : integer
11064req_len : integer (deprecated)
11065 Returns an integer value corresponding to the number of bytes present in the
11066 request buffer. This is mostly used in ACL. It is important to understand
11067 that this test does not return false as long as the buffer is changing. This
11068 means that a check with equality to zero will almost always immediately match
11069 at the beginning of the session, while a test for more data will wait for
11070 that data to come in and return false only when haproxy is certain that no
11071 more data will come in. This test was designed to be used with TCP request
11072 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011073
Willy Tarreau74ca5042013-06-11 23:12:07 +020011074req.payload(<offset>,<length>) : binary
11075 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011076 in the request buffer. As a special case, if the <length> argument is zero,
11077 the the whole buffer from <offset> to the end is extracted. This can be used
11078 with ACLs in order to check for the presence of some content in a buffer at
11079 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011080
Willy Tarreau74ca5042013-06-11 23:12:07 +020011081 ACL alternatives :
11082 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011083
Willy Tarreau74ca5042013-06-11 23:12:07 +020011084req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11085 This extracts a binary block whose size is specified at <offset1> for <length>
11086 bytes, and which starts at <offset2> if specified or just after the length in
11087 the request buffer. The <offset2> parameter also supports relative offsets if
11088 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011089
Willy Tarreau74ca5042013-06-11 23:12:07 +020011090 ACL alternatives :
11091 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011092
Willy Tarreau74ca5042013-06-11 23:12:07 +020011093 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011094
Willy Tarreau74ca5042013-06-11 23:12:07 +020011095req.proto_http : boolean
11096req_proto_http : boolean (deprecated)
11097 Returns true when data in the request buffer look like HTTP and correctly
11098 parses as such. It is the same parser as the common HTTP request parser which
11099 is used so there should be no surprises. The test does not match until the
11100 request is complete, failed or timed out. This test may be used to report the
11101 protocol in TCP logs, but the biggest use is to block TCP request analysis
11102 until a complete HTTP request is present in the buffer, for example to track
11103 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011104
Willy Tarreau74ca5042013-06-11 23:12:07 +020011105 Example:
11106 # track request counts per "base" (concatenation of Host+URL)
11107 tcp-request inspect-delay 10s
11108 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011109 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011110
Willy Tarreau74ca5042013-06-11 23:12:07 +020011111req.rdp_cookie([<name>]) : string
11112rdp_cookie([<name>]) : string (deprecated)
11113 When the request buffer looks like the RDP protocol, extracts the RDP cookie
11114 <name>, or any cookie if unspecified. The parser only checks for the first
11115 cookie, as illustrated in the RDP protocol specification. The cookie name is
11116 case insensitive. Generally the "MSTS" cookie name will be used, as it can
11117 contain the user name of the client connecting to the server if properly
11118 configured on the client. The "MSTSHASH" cookie is often used as well for
11119 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011120
Willy Tarreau74ca5042013-06-11 23:12:07 +020011121 This differs from "balance rdp-cookie" in that any balancing algorithm may be
11122 used and thus the distribution of clients to backend servers is not linked to
11123 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
11124 such as "balance roundrobin" or "balance leastconn" will lead to a more even
11125 distribution of clients to backend servers than the hash used by "balance
11126 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011127
Willy Tarreau74ca5042013-06-11 23:12:07 +020011128 ACL derivatives :
11129 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011130
Willy Tarreau74ca5042013-06-11 23:12:07 +020011131 Example :
11132 listen tse-farm
11133 bind 0.0.0.0:3389
11134 # wait up to 5s for an RDP cookie in the request
11135 tcp-request inspect-delay 5s
11136 tcp-request content accept if RDP_COOKIE
11137 # apply RDP cookie persistence
11138 persist rdp-cookie
11139 # Persist based on the mstshash cookie
11140 # This is only useful makes sense if
11141 # balance rdp-cookie is not used
11142 stick-table type string size 204800
11143 stick on req.rdp_cookie(mstshash)
11144 server srv1 1.1.1.1:3389
11145 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011146
Willy Tarreau74ca5042013-06-11 23:12:07 +020011147 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
11148 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011149
Willy Tarreau74ca5042013-06-11 23:12:07 +020011150req.rdp_cookie_cnt([name]) : integer
11151rdp_cookie_cnt([name]) : integer (deprecated)
11152 Tries to parse the request buffer as RDP protocol, then returns an integer
11153 corresponding to the number of RDP cookies found. If an optional cookie name
11154 is passed, only cookies matching this name are considered. This is mostly
11155 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011156
Willy Tarreau74ca5042013-06-11 23:12:07 +020011157 ACL derivatives :
11158 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011159
Willy Tarreau74ca5042013-06-11 23:12:07 +020011160req.ssl_hello_type : integer
11161req_ssl_hello_type : integer (deprecated)
11162 Returns an integer value containing the type of the SSL hello message found
11163 in the request buffer if the buffer contains data that parse as a complete
11164 SSL (v3 or superior) client hello message. Note that this only applies to raw
11165 contents found in the request buffer and not to contents deciphered via an
11166 SSL data layer, so this will not work with "bind" lines having the "ssl"
11167 option. This is mostly used in ACL to detect presence of an SSL hello message
11168 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011169
Willy Tarreau74ca5042013-06-11 23:12:07 +020011170req.ssl_sni : string
11171req_ssl_sni : string (deprecated)
11172 Returns a string containing the value of the Server Name TLS extension sent
11173 by a client in a TLS stream passing through the request buffer if the buffer
11174 contains data that parse as a complete SSL (v3 or superior) client hello
11175 message. Note that this only applies to raw contents found in the request
11176 buffer and not to contents deciphered via an SSL data layer, so this will not
11177 work with "bind" lines having the "ssl" option. SNI normally contains the
11178 name of the host the client tries to connect to (for recent browsers). SNI is
11179 useful for allowing or denying access to certain hosts when SSL/TLS is used
11180 by the client. This test was designed to be used with TCP request content
11181 inspection. If content switching is needed, it is recommended to first wait
11182 for a complete client hello (type 1), like in the example below. See also
11183 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011184
Willy Tarreau74ca5042013-06-11 23:12:07 +020011185 ACL derivatives :
11186 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011187
Willy Tarreau74ca5042013-06-11 23:12:07 +020011188 Examples :
11189 # Wait for a client hello for at most 5 seconds
11190 tcp-request inspect-delay 5s
11191 tcp-request content accept if { req_ssl_hello_type 1 }
11192 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
11193 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011194
Willy Tarreau74ca5042013-06-11 23:12:07 +020011195res.ssl_hello_type : integer
11196rep_ssl_hello_type : integer (deprecated)
11197 Returns an integer value containing the type of the SSL hello message found
11198 in the response buffer if the buffer contains data that parses as a complete
11199 SSL (v3 or superior) hello message. Note that this only applies to raw
11200 contents found in the response buffer and not to contents deciphered via an
11201 SSL data layer, so this will not work with "server" lines having the "ssl"
11202 option. This is mostly used in ACL to detect presence of an SSL hello message
11203 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau51539362012-05-08 12:46:28 +020011204
Willy Tarreau74ca5042013-06-11 23:12:07 +020011205req.ssl_ver : integer
11206req_ssl_ver : integer (deprecated)
11207 Returns an integer value containing the version of the SSL/TLS protocol of a
11208 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
11209 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
11210 composed of the major version multiplied by 65536, added to the minor
11211 version. Note that this only applies to raw contents found in the request
11212 buffer and not to contents deciphered via an SSL data layer, so this will not
11213 work with "bind" lines having the "ssl" option. The ACL version of the test
11214 matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This
11215 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011216
Willy Tarreau74ca5042013-06-11 23:12:07 +020011217 ACL derivatives :
11218 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011219
Willy Tarreau47e8eba2013-09-11 23:28:46 +020011220res.len : integer
11221 Returns an integer value corresponding to the number of bytes present in the
11222 response buffer. This is mostly used in ACL. It is important to understand
11223 that this test does not return false as long as the buffer is changing. This
11224 means that a check with equality to zero will almost always immediately match
11225 at the beginning of the session, while a test for more data will wait for
11226 that data to come in and return false only when haproxy is certain that no
11227 more data will come in. This test was designed to be used with TCP response
11228 content inspection.
11229
Willy Tarreau74ca5042013-06-11 23:12:07 +020011230res.payload(<offset>,<length>) : binary
11231 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011232 in the response buffer. As a special case, if the <length> argument is zero,
11233 the the whole buffer from <offset> to the end is extracted. This can be used
11234 with ACLs in order to check for the presence of some content in a buffer at
11235 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011236
Willy Tarreau74ca5042013-06-11 23:12:07 +020011237res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11238 This extracts a binary block whose size is specified at <offset1> for <length>
11239 bytes, and which starts at <offset2> if specified or just after the length in
11240 the response buffer. The <offset2> parameter also supports relative offsets
11241 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011242
Willy Tarreau74ca5042013-06-11 23:12:07 +020011243 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011244
Willy Tarreau74ca5042013-06-11 23:12:07 +020011245wait_end : boolean
11246 This fetch either returns true when the inspection period is over, or does
11247 not fetch. It is only used in ACLs, in conjunction with content analysis to
11248 avoid returning a wrong verdict early. It may also be used to delay some
11249 actions, such as a delayed reject for some special addresses. Since it either
11250 stops the rules evaluation or immediately returns true, it is recommended to
11251 use this acl as the last one in a rule. Please note that the default ACL
11252 "WAIT_END" is always usable without prior declaration. This test was designed
11253 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011254
Willy Tarreau74ca5042013-06-11 23:12:07 +020011255 Examples :
11256 # delay every incoming request by 2 seconds
11257 tcp-request inspect-delay 2s
11258 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010011259
Willy Tarreau74ca5042013-06-11 23:12:07 +020011260 # don't immediately tell bad guys they are rejected
11261 tcp-request inspect-delay 10s
11262 acl goodguys src 10.0.0.0/24
11263 acl badguys src 10.0.1.0/24
11264 tcp-request content accept if goodguys
11265 tcp-request content reject if badguys WAIT_END
11266 tcp-request content reject
11267
11268
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200112697.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011270--------------------------------------
11271
11272It is possible to fetch samples from HTTP contents, requests and responses.
11273This application layer is also called layer 7. It is only possible to fetch the
11274data in this section when a full HTTP request or response has been parsed from
11275its respective request or response buffer. This is always the case with all
11276HTTP specific rules and for sections running with "mode http". When using TCP
11277content inspection, it may be necessary to support an inspection delay in order
11278to let the request or response come in first. These fetches may require a bit
11279more CPU resources than the layer 4 ones, but not much since the request and
11280response are indexed.
11281
11282base : string
11283 This returns the concatenation of the first Host header and the path part of
11284 the request, which starts at the first slash and ends before the question
11285 mark. It can be useful in virtual hosted environments to detect URL abuses as
11286 well as to improve shared caches efficiency. Using this with a limited size
11287 stick table also allows one to collect statistics about most commonly
11288 requested objects by host/path. With ACLs it can allow simple content
11289 switching rules involving the host and the path at the same time, such as
11290 "www.example.com/favicon.ico". See also "path" and "uri".
11291
11292 ACL derivatives :
11293 base : exact string match
11294 base_beg : prefix match
11295 base_dir : subdir match
11296 base_dom : domain match
11297 base_end : suffix match
11298 base_len : length match
11299 base_reg : regex match
11300 base_sub : substring match
11301
11302base32 : integer
11303 This returns a 32-bit hash of the value returned by the "base" fetch method
11304 above. This is useful to track per-URL activity on high traffic sites without
11305 having to store all URLs. Instead a shorter hash is stored, saving a lot of
11306 memory. The output type is an unsigned integer.
11307
11308base32+src : binary
11309 This returns the concatenation of the base32 fetch above and the src fetch
11310 below. The resulting type is of type binary, with a size of 8 or 20 bytes
11311 depending on the source address family. This can be used to track per-IP,
11312 per-URL counters.
11313
William Lallemand65ad6e12014-01-31 15:08:02 +010011314capture.req.hdr(<idx>) : string
11315 This extracts the content of the header captured by the "capture request
11316 header", idx is the position of the capture keyword in the configuration.
11317 The first entry is an index of 0. See also: "capture request header".
11318
11319capture.req.method : string
11320 This extracts the METHOD of an HTTP request. It can be used in both request
11321 and response. Unlike "method", it can be used in both request and response
11322 because it's allocated.
11323
11324capture.req.uri : string
11325 This extracts the request's URI, which starts at the first slash and ends
11326 before the first space in the request (without the host part). Unlike "path"
11327 and "url", it can be used in both request and response because it's
11328 allocated.
11329
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011330capture.req.ver : string
11331 This extracts the request's HTTP version and returns either "HTTP/1.0" or
11332 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
11333 logs because it relies on a persistent flag.
11334
William Lallemand65ad6e12014-01-31 15:08:02 +010011335capture.res.hdr(<idx>) : string
11336 This extracts the content of the header captured by the "capture response
11337 header", idx is the position of the capture keyword in the configuration.
11338 The first entry is an index of 0.
11339 See also: "capture response header"
11340
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011341capture.res.ver : string
11342 This extracts the response's HTTP version and returns either "HTTP/1.0" or
11343 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
11344 persistent flag.
11345
Willy Tarreau74ca5042013-06-11 23:12:07 +020011346req.cook([<name>]) : string
11347cook([<name>]) : string (deprecated)
11348 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11349 header line from the request, and returns its value as string. If no name is
11350 specified, the first cookie value is returned. When used with ACLs, all
11351 matching cookies are evaluated. Spaces around the name and the value are
11352 ignored as requested by the Cookie header specification (RFC6265). The cookie
11353 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
11354 well return an empty value if it is present. Use the "found" match to detect
11355 presence. Use the res.cook() variant for response cookies sent by the server.
11356
11357 ACL derivatives :
11358 cook([<name>]) : exact string match
11359 cook_beg([<name>]) : prefix match
11360 cook_dir([<name>]) : subdir match
11361 cook_dom([<name>]) : domain match
11362 cook_end([<name>]) : suffix match
11363 cook_len([<name>]) : length match
11364 cook_reg([<name>]) : regex match
11365 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011366
Willy Tarreau74ca5042013-06-11 23:12:07 +020011367req.cook_cnt([<name>]) : integer
11368cook_cnt([<name>]) : integer (deprecated)
11369 Returns an integer value representing the number of occurrences of the cookie
11370 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011371
Willy Tarreau74ca5042013-06-11 23:12:07 +020011372req.cook_val([<name>]) : integer
11373cook_val([<name>]) : integer (deprecated)
11374 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11375 header line from the request, and converts its value to an integer which is
11376 returned. If no name is specified, the first cookie value is returned. When
11377 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020011378
Willy Tarreau74ca5042013-06-11 23:12:07 +020011379cookie([<name>]) : string (deprecated)
11380 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11381 header line from the request, or a "Set-Cookie" header from the response, and
11382 returns its value as a string. A typical use is to get multiple clients
11383 sharing a same profile use the same server. This can be similar to what
11384 "appsession" does with the "request-learn" statement, but with support for
11385 multi-peer synchronization and state keeping across restarts. If no name is
11386 specified, the first cookie value is returned. This fetch should not be used
11387 anymore and should be replaced by req.cook() or res.cook() instead as it
11388 ambiguously uses the direction based on the context where it is used.
11389 See also : "appsession".
Willy Tarreaud63335a2010-02-26 12:56:52 +010011390
Willy Tarreau74ca5042013-06-11 23:12:07 +020011391hdr([<name>[,<occ>]]) : string
11392 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
11393 used on responses. Please refer to these respective fetches for more details.
11394 In case of doubt about the fetch direction, please use the explicit ones.
11395 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011396 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011397
Willy Tarreau74ca5042013-06-11 23:12:07 +020011398req.fhdr(<name>[,<occ>]) : string
11399 This extracts the last occurrence of header <name> in an HTTP request. When
11400 used from an ACL, all occurrences are iterated over until a match is found.
11401 Optionally, a specific occurrence might be specified as a position number.
11402 Positive values indicate a position from the first occurrence, with 1 being
11403 the first one. Negative values indicate positions relative to the last one,
11404 with -1 being the last one. It differs from req.hdr() in that any commas
11405 present in the value are returned and are not used as delimiters. This is
11406 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011407
Willy Tarreau74ca5042013-06-11 23:12:07 +020011408req.fhdr_cnt([<name>]) : integer
11409 Returns an integer value representing the number of occurrences of request
11410 header field name <name>, or the total number of header fields if <name> is
11411 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
11412 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011413
Willy Tarreau74ca5042013-06-11 23:12:07 +020011414req.hdr([<name>[,<occ>]]) : string
11415 This extracts the last occurrence of header <name> in an HTTP request. When
11416 used from an ACL, all occurrences are iterated over until a match is found.
11417 Optionally, a specific occurrence might be specified as a position number.
11418 Positive values indicate a position from the first occurrence, with 1 being
11419 the first one. Negative values indicate positions relative to the last one,
11420 with -1 being the last one. A typical use is with the X-Forwarded-For header
11421 once converted to IP, associated with an IP stick-table. The function
11422 considers any comma as a delimiter for distinct values. If full-line headers
11423 are desired instead, use req.fhdr(). Please carefully check RFC2616 to know
11424 how certain headers are supposed to be parsed. Also, some of them are case
11425 insensitive (eg: Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010011426
Willy Tarreau74ca5042013-06-11 23:12:07 +020011427 ACL derivatives :
11428 hdr([<name>[,<occ>]]) : exact string match
11429 hdr_beg([<name>[,<occ>]]) : prefix match
11430 hdr_dir([<name>[,<occ>]]) : subdir match
11431 hdr_dom([<name>[,<occ>]]) : domain match
11432 hdr_end([<name>[,<occ>]]) : suffix match
11433 hdr_len([<name>[,<occ>]]) : length match
11434 hdr_reg([<name>[,<occ>]]) : regex match
11435 hdr_sub([<name>[,<occ>]]) : substring match
11436
11437req.hdr_cnt([<name>]) : integer
11438hdr_cnt([<header>]) : integer (deprecated)
11439 Returns an integer value representing the number of occurrences of request
11440 header field name <name>, or the total number of header field values if
11441 <name> is not specified. It is important to remember that one header line may
11442 count as several headers if it has several values. The function considers any
11443 comma as a delimiter for distinct values. If full-line headers are desired
11444 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
11445 detect presence, absence or abuse of a specific header, as well as to block
11446 request smuggling attacks by rejecting requests which contain more than one
11447 of certain headers. See "req.hdr" for more information on header matching.
11448
11449req.hdr_ip([<name>[,<occ>]]) : ip
11450hdr_ip([<name>[,<occ>]]) : ip (deprecated)
11451 This extracts the last occurrence of header <name> in an HTTP request,
11452 converts it to an IPv4 or IPv6 address and returns this address. When used
11453 with ACLs, all occurrences are checked, and if <name> is omitted, every value
11454 of every header is checked. Optionally, a specific occurrence might be
11455 specified as a position number. Positive values indicate a position from the
11456 first occurrence, with 1 being the first one. Negative values indicate
11457 positions relative to the last one, with -1 being the last one. A typical use
11458 is with the X-Forwarded-For and X-Client-IP headers.
11459
11460req.hdr_val([<name>[,<occ>]]) : integer
11461hdr_val([<name>[,<occ>]]) : integer (deprecated)
11462 This extracts the last occurrence of header <name> in an HTTP request, and
11463 converts it to an integer value. When used with ACLs, all occurrences are
11464 checked, and if <name> is omitted, every value of every header is checked.
11465 Optionally, a specific occurrence might be specified as a position number.
11466 Positive values indicate a position from the first occurrence, with 1 being
11467 the first one. Negative values indicate positions relative to the last one,
11468 with -1 being the last one. A typical use is with the X-Forwarded-For header.
11469
11470http_auth(<userlist>) : boolean
11471 Returns a boolean indicating whether the authentication data received from
11472 the client match a username & password stored in the specified userlist. This
11473 fetch function is not really useful outside of ACLs. Currently only http
11474 basic auth is supported.
11475
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011476http_auth_group(<userlist>) : string
11477 Returns a string corresponding to the user name found in the authentication
11478 data received from the client if both the user name and password are valid
11479 according to the specified userlist. The main purpose is to use it in ACLs
11480 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011481 This fetch function is not really useful outside of ACLs. Currently only http
11482 basic auth is supported.
11483
11484 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011485 http_auth_group(<userlist>) : group ...
11486 Returns true when the user extracted from the request and whose password is
11487 valid according to the specified userlist belongs to at least one of the
11488 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011489
11490http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020011491 Returns true when the request being processed is the first one of the
11492 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020011493 from some requests when a request is not the first one, or to help grouping
11494 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020011495
Willy Tarreau74ca5042013-06-11 23:12:07 +020011496method : integer + string
11497 Returns an integer value corresponding to the method in the HTTP request. For
11498 example, "GET" equals 1 (check sources to establish the matching). Value 9
11499 means "other method" and may be converted to a string extracted from the
11500 stream. This should not be used directly as a sample, this is only meant to
11501 be used from ACLs, which transparently convert methods from patterns to these
11502 integer + string values. Some predefined ACL already check for most common
11503 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011504
Willy Tarreau74ca5042013-06-11 23:12:07 +020011505 ACL derivatives :
11506 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011507
Willy Tarreau74ca5042013-06-11 23:12:07 +020011508 Example :
11509 # only accept GET and HEAD requests
11510 acl valid_method method GET HEAD
11511 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020011512
Willy Tarreau74ca5042013-06-11 23:12:07 +020011513path : string
11514 This extracts the request's URL path, which starts at the first slash and
11515 ends before the question mark (without the host part). A typical use is with
11516 prefetch-capable caches, and with portals which need to aggregate multiple
11517 information from databases and keep them in caches. Note that with outgoing
11518 caches, it would be wiser to use "url" instead. With ACLs, it's typically
11519 used to match exact file names (eg: "/login.php"), or directory parts using
11520 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011521
Willy Tarreau74ca5042013-06-11 23:12:07 +020011522 ACL derivatives :
11523 path : exact string match
11524 path_beg : prefix match
11525 path_dir : subdir match
11526 path_dom : domain match
11527 path_end : suffix match
11528 path_len : length match
11529 path_reg : regex match
11530 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011531
Willy Tarreau74ca5042013-06-11 23:12:07 +020011532req.ver : string
11533req_ver : string (deprecated)
11534 Returns the version string from the HTTP request, for example "1.1". This can
11535 be useful for logs, but is mostly there for ACL. Some predefined ACL already
11536 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011537
Willy Tarreau74ca5042013-06-11 23:12:07 +020011538 ACL derivatives :
11539 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020011540
Willy Tarreau74ca5042013-06-11 23:12:07 +020011541res.comp : boolean
11542 Returns the boolean "true" value if the response has been compressed by
11543 HAProxy, otherwise returns boolean "false". This may be used to add
11544 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011545
Willy Tarreau74ca5042013-06-11 23:12:07 +020011546res.comp_algo : string
11547 Returns a string containing the name of the algorithm used if the response
11548 was compressed by HAProxy, for example : "deflate". This may be used to add
11549 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011550
Willy Tarreau74ca5042013-06-11 23:12:07 +020011551res.cook([<name>]) : string
11552scook([<name>]) : string (deprecated)
11553 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11554 header line from the response, and returns its value as string. If no name is
11555 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011556
Willy Tarreau74ca5042013-06-11 23:12:07 +020011557 ACL derivatives :
11558 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011559
Willy Tarreau74ca5042013-06-11 23:12:07 +020011560res.cook_cnt([<name>]) : integer
11561scook_cnt([<name>]) : integer (deprecated)
11562 Returns an integer value representing the number of occurrences of the cookie
11563 <name> in the response, or all cookies if <name> is not specified. This is
11564 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011565
Willy Tarreau74ca5042013-06-11 23:12:07 +020011566res.cook_val([<name>]) : integer
11567scook_val([<name>]) : integer (deprecated)
11568 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11569 header line from the response, and converts its value to an integer which is
11570 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011571
Willy Tarreau74ca5042013-06-11 23:12:07 +020011572res.fhdr([<name>[,<occ>]]) : string
11573 This extracts the last occurrence of header <name> in an HTTP response, or of
11574 the last header if no <name> is specified. Optionally, a specific occurrence
11575 might be specified as a position number. Positive values indicate a position
11576 from the first occurrence, with 1 being the first one. Negative values
11577 indicate positions relative to the last one, with -1 being the last one. It
11578 differs from res.hdr() in that any commas present in the value are returned
11579 and are not used as delimiters. If this is not desired, the res.hdr() fetch
11580 should be used instead. This is sometimes useful with headers such as Date or
11581 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011582
Willy Tarreau74ca5042013-06-11 23:12:07 +020011583res.fhdr_cnt([<name>]) : integer
11584 Returns an integer value representing the number of occurrences of response
11585 header field name <name>, or the total number of header fields if <name> is
11586 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
11587 the number of full line headers and does not stop on commas. If this is not
11588 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011589
Willy Tarreau74ca5042013-06-11 23:12:07 +020011590res.hdr([<name>[,<occ>]]) : string
11591shdr([<name>[,<occ>]]) : string (deprecated)
11592 This extracts the last occurrence of header <name> in an HTTP response, or of
11593 the last header if no <name> is specified. Optionally, a specific occurrence
11594 might be specified as a position number. Positive values indicate a position
11595 from the first occurrence, with 1 being the first one. Negative values
11596 indicate positions relative to the last one, with -1 being the last one. This
11597 can be useful to learn some data into a stick-table. The function considers
11598 any comma as a delimiter for distinct values. If this is not desired, the
11599 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011600
Willy Tarreau74ca5042013-06-11 23:12:07 +020011601 ACL derivatives :
11602 shdr([<name>[,<occ>]]) : exact string match
11603 shdr_beg([<name>[,<occ>]]) : prefix match
11604 shdr_dir([<name>[,<occ>]]) : subdir match
11605 shdr_dom([<name>[,<occ>]]) : domain match
11606 shdr_end([<name>[,<occ>]]) : suffix match
11607 shdr_len([<name>[,<occ>]]) : length match
11608 shdr_reg([<name>[,<occ>]]) : regex match
11609 shdr_sub([<name>[,<occ>]]) : substring match
11610
11611res.hdr_cnt([<name>]) : integer
11612shdr_cnt([<name>]) : integer (deprecated)
11613 Returns an integer value representing the number of occurrences of response
11614 header field name <name>, or the total number of header fields if <name> is
11615 not specified. The function considers any comma as a delimiter for distinct
11616 values. If this is not desired, the res.fhdr_cnt() fetch should be used
11617 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011618
Willy Tarreau74ca5042013-06-11 23:12:07 +020011619res.hdr_ip([<name>[,<occ>]]) : ip
11620shdr_ip([<name>[,<occ>]]) : ip (deprecated)
11621 This extracts the last occurrence of header <name> in an HTTP response,
11622 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
11623 specific occurrence might be specified as a position number. Positive values
11624 indicate a position from the first occurrence, with 1 being the first one.
11625 Negative values indicate positions relative to the last one, with -1 being
11626 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011627
Willy Tarreau74ca5042013-06-11 23:12:07 +020011628res.hdr_val([<name>[,<occ>]]) : integer
11629shdr_val([<name>[,<occ>]]) : integer (deprecated)
11630 This extracts the last occurrence of header <name> in an HTTP response, and
11631 converts it to an integer value. Optionally, a specific occurrence might be
11632 specified as a position number. Positive values indicate a position from the
11633 first occurrence, with 1 being the first one. Negative values indicate
11634 positions relative to the last one, with -1 being the last one. This can be
11635 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011636
Willy Tarreau74ca5042013-06-11 23:12:07 +020011637res.ver : string
11638resp_ver : string (deprecated)
11639 Returns the version string from the HTTP response, for example "1.1". This
11640 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020011641
Willy Tarreau74ca5042013-06-11 23:12:07 +020011642 ACL derivatives :
11643 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011644
Willy Tarreau74ca5042013-06-11 23:12:07 +020011645set-cookie([<name>]) : string (deprecated)
11646 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11647 header line from the response and uses the corresponding value to match. This
11648 can be comparable to what "appsession" does with default options, but with
11649 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011650
Willy Tarreau74ca5042013-06-11 23:12:07 +020011651 This fetch function is deprecated and has been superseded by the "res.cook"
11652 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011653
Willy Tarreau74ca5042013-06-11 23:12:07 +020011654 See also : "appsession"
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011655
Willy Tarreau74ca5042013-06-11 23:12:07 +020011656status : integer
11657 Returns an integer containing the HTTP status code in the HTTP response, for
11658 example, 302. It is mostly used within ACLs and integer ranges, for example,
11659 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011660
Willy Tarreau74ca5042013-06-11 23:12:07 +020011661url : string
11662 This extracts the request's URL as presented in the request. A typical use is
11663 with prefetch-capable caches, and with portals which need to aggregate
11664 multiple information from databases and keep them in caches. With ACLs, using
11665 "path" is preferred over using "url", because clients may send a full URL as
11666 is normally done with proxies. The only real use is to match "*" which does
11667 not match in "path", and for which there is already a predefined ACL. See
11668 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011669
Willy Tarreau74ca5042013-06-11 23:12:07 +020011670 ACL derivatives :
11671 url : exact string match
11672 url_beg : prefix match
11673 url_dir : subdir match
11674 url_dom : domain match
11675 url_end : suffix match
11676 url_len : length match
11677 url_reg : regex match
11678 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011679
Willy Tarreau74ca5042013-06-11 23:12:07 +020011680url_ip : ip
11681 This extracts the IP address from the request's URL when the host part is
11682 presented as an IP address. Its use is very limited. For instance, a
11683 monitoring system might use this field as an alternative for the source IP in
11684 order to test what path a given source address would follow, or to force an
11685 entry in a table for a given source address. With ACLs it can be used to
11686 restrict access to certain systems through a proxy, for example when combined
11687 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011688
Willy Tarreau74ca5042013-06-11 23:12:07 +020011689url_port : integer
11690 This extracts the port part from the request's URL. Note that if the port is
11691 not specified in the request, port 80 is assumed. With ACLs it can be used to
11692 restrict access to certain systems through a proxy, for example when combined
11693 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011694
Willy Tarreau74ca5042013-06-11 23:12:07 +020011695urlp(<name>[,<delim>]) : string
11696url_param(<name>[,<delim>]) : string
11697 This extracts the first occurrence of the parameter <name> in the query
11698 string, which begins after either '?' or <delim>, and which ends before '&',
11699 ';' or <delim>. The parameter name is case-sensitive. The result is a string
11700 corresponding to the value of the parameter <name> as presented in the
11701 request (no URL decoding is performed). This can be used for session
11702 stickiness based on a client ID, to extract an application cookie passed as a
11703 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
11704 this fetch do not iterate over multiple parameters and stop at the first one
11705 as well.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011706
Willy Tarreau74ca5042013-06-11 23:12:07 +020011707 ACL derivatives :
11708 urlp(<name>[,<delim>]) : exact string match
11709 urlp_beg(<name>[,<delim>]) : prefix match
11710 urlp_dir(<name>[,<delim>]) : subdir match
11711 urlp_dom(<name>[,<delim>]) : domain match
11712 urlp_end(<name>[,<delim>]) : suffix match
11713 urlp_len(<name>[,<delim>]) : length match
11714 urlp_reg(<name>[,<delim>]) : regex match
11715 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011716
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011717
Willy Tarreau74ca5042013-06-11 23:12:07 +020011718 Example :
11719 # match http://example.com/foo?PHPSESSIONID=some_id
11720 stick on urlp(PHPSESSIONID)
11721 # match http://example.com/foo;JSESSIONID=some_id
11722 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011723
Willy Tarreau74ca5042013-06-11 23:12:07 +020011724urlp_val(<name>[,<delim>]) : integer
11725 See "urlp" above. This one extracts the URL parameter <name> in the request
11726 and converts it to an integer value. This can be used for session stickiness
11727 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020011728
Willy Tarreau198a7442008-01-17 12:05:32 +010011729
Willy Tarreau74ca5042013-06-11 23:12:07 +0200117307.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011731---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011732
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011733Some predefined ACLs are hard-coded so that they do not have to be declared in
11734every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020011735order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010011736
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011737ACL name Equivalent to Usage
11738---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011739FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020011740HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011741HTTP_1.0 req_ver 1.0 match HTTP version 1.0
11742HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010011743HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
11744HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
11745HTTP_URL_SLASH url_beg / match URL beginning with "/"
11746HTTP_URL_STAR url * match URL equal to "*"
11747LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011748METH_CONNECT method CONNECT match HTTP CONNECT method
11749METH_GET method GET HEAD match HTTP GET or HEAD method
11750METH_HEAD method HEAD match HTTP HEAD method
11751METH_OPTIONS method OPTIONS match HTTP OPTIONS method
11752METH_POST method POST match HTTP POST method
11753METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020011754RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011755REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010011756TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011757WAIT_END wait_end wait for end of content analysis
11758---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011759
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011760
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117618. Logging
11762----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010011763
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011764One of HAProxy's strong points certainly lies is its precise logs. It probably
11765provides the finest level of information available for such a product, which is
11766very important for troubleshooting complex environments. Standard information
11767provided in logs include client ports, TCP/HTTP state timers, precise session
11768state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011769to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011770headers.
11771
11772In order to improve administrators reactivity, it offers a great transparency
11773about encountered problems, both internal and external, and it is possible to
11774send logs to different sources at the same time with different level filters :
11775
11776 - global process-level logs (system errors, start/stop, etc..)
11777 - per-instance system and internal errors (lack of resource, bugs, ...)
11778 - per-instance external troubles (servers up/down, max connections)
11779 - per-instance activity (client connections), either at the establishment or
11780 at the termination.
11781
11782The ability to distribute different levels of logs to different log servers
11783allow several production teams to interact and to fix their problems as soon
11784as possible. For example, the system team might monitor system-wide errors,
11785while the application team might be monitoring the up/down for their servers in
11786real time, and the security team might analyze the activity logs with one hour
11787delay.
11788
11789
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117908.1. Log levels
11791---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011792
Simon Hormandf791f52011-05-29 15:01:10 +090011793TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011794source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090011795HTTP request, HTTP return code, number of bytes transmitted, conditions
11796in which the session ended, and even exchanged cookies values. For example
11797track a particular user's problems. All messages may be sent to up to two
11798syslog servers. Check the "log" keyword in section 4.2 for more information
11799about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011800
11801
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118028.2. Log formats
11803----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011804
William Lallemand48940402012-01-30 16:47:22 +010011805HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090011806and will be detailed in the following sections. A few of them may vary
11807slightly with the configuration, due to indicators specific to certain
11808options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011809
11810 - the default format, which is very basic and very rarely used. It only
11811 provides very basic information about the incoming connection at the moment
11812 it is accepted : source IP:port, destination IP:port, and frontend-name.
11813 This mode will eventually disappear so it will not be described to great
11814 extents.
11815
11816 - the TCP format, which is more advanced. This format is enabled when "option
11817 tcplog" is set on the frontend. HAProxy will then usually wait for the
11818 connection to terminate before logging. This format provides much richer
11819 information, such as timers, connection counts, queue size, etc... This
11820 format is recommended for pure TCP proxies.
11821
11822 - the HTTP format, which is the most advanced for HTTP proxying. This format
11823 is enabled when "option httplog" is set on the frontend. It provides the
11824 same information as the TCP format with some HTTP-specific fields such as
11825 the request, the status code, and captures of headers and cookies. This
11826 format is recommended for HTTP proxies.
11827
Emeric Brun3a058f32009-06-30 18:26:00 +020011828 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
11829 fields arranged in the same order as the CLF format. In this mode, all
11830 timers, captures, flags, etc... appear one per field after the end of the
11831 common fields, in the same order they appear in the standard HTTP format.
11832
William Lallemand48940402012-01-30 16:47:22 +010011833 - the custom log format, allows you to make your own log line.
11834
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011835Next sections will go deeper into details for each of these formats. Format
11836specification will be performed on a "field" basis. Unless stated otherwise, a
11837field is a portion of text delimited by any number of spaces. Since syslog
11838servers are susceptible of inserting fields at the beginning of a line, it is
11839always assumed that the first field is the one containing the process name and
11840identifier.
11841
11842Note : Since log lines may be quite long, the log examples in sections below
11843 might be broken into multiple lines. The example log lines will be
11844 prefixed with 3 closing angle brackets ('>>>') and each time a log is
11845 broken into multiple lines, each non-final line will end with a
11846 backslash ('\') and the next line will start indented by two characters.
11847
11848
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118498.2.1. Default log format
11850-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011851
11852This format is used when no specific option is set. The log is emitted as soon
11853as the connection is accepted. One should note that this currently is the only
11854format which logs the request's destination IP and ports.
11855
11856 Example :
11857 listen www
11858 mode http
11859 log global
11860 server srv1 127.0.0.1:8000
11861
11862 >>> Feb 6 12:12:09 localhost \
11863 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
11864 (www/HTTP)
11865
11866 Field Format Extract from the example above
11867 1 process_name '[' pid ']:' haproxy[14385]:
11868 2 'Connect from' Connect from
11869 3 source_ip ':' source_port 10.0.1.2:33312
11870 4 'to' to
11871 5 destination_ip ':' destination_port 10.0.3.31:8012
11872 6 '(' frontend_name '/' mode ')' (www/HTTP)
11873
11874Detailed fields description :
11875 - "source_ip" is the IP address of the client which initiated the connection.
11876 - "source_port" is the TCP port of the client which initiated the connection.
11877 - "destination_ip" is the IP address the client connected to.
11878 - "destination_port" is the TCP port the client connected to.
11879 - "frontend_name" is the name of the frontend (or listener) which received
11880 and processed the connection.
11881 - "mode is the mode the frontend is operating (TCP or HTTP).
11882
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011883In case of a UNIX socket, the source and destination addresses are marked as
11884"unix:" and the ports reflect the internal ID of the socket which accepted the
11885connection (the same ID as reported in the stats).
11886
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011887It is advised not to use this deprecated format for newer installations as it
11888will eventually disappear.
11889
11890
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118918.2.2. TCP log format
11892---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011893
11894The TCP format is used when "option tcplog" is specified in the frontend, and
11895is the recommended format for pure TCP proxies. It provides a lot of precious
11896information for troubleshooting. Since this format includes timers and byte
11897counts, the log is normally emitted at the end of the session. It can be
11898emitted earlier if "option logasap" is specified, which makes sense in most
11899environments with long sessions such as remote terminals. Sessions which match
11900the "monitor" rules are never logged. It is also possible not to emit logs for
11901sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020011902specifying "option dontlognull" in the frontend. Successful connections will
11903not be logged if "option dontlog-normal" is specified in the frontend. A few
11904fields may slightly vary depending on some configuration options, those are
11905marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011906
11907 Example :
11908 frontend fnt
11909 mode tcp
11910 option tcplog
11911 log global
11912 default_backend bck
11913
11914 backend bck
11915 server srv1 127.0.0.1:8000
11916
11917 >>> Feb 6 12:12:56 localhost \
11918 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
11919 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
11920
11921 Field Format Extract from the example above
11922 1 process_name '[' pid ']:' haproxy[14387]:
11923 2 client_ip ':' client_port 10.0.1.2:33313
11924 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
11925 4 frontend_name fnt
11926 5 backend_name '/' server_name bck/srv1
11927 6 Tw '/' Tc '/' Tt* 0/0/5007
11928 7 bytes_read* 212
11929 8 termination_state --
11930 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
11931 10 srv_queue '/' backend_queue 0/0
11932
11933Detailed fields description :
11934 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011935 connection to haproxy. If the connection was accepted on a UNIX socket
11936 instead, the IP address would be replaced with the word "unix". Note that
11937 when the connection is accepted on a socket configured with "accept-proxy"
11938 and the PROXY protocol is correctly used, then the logs will reflect the
11939 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011940
11941 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011942 If the connection was accepted on a UNIX socket instead, the port would be
11943 replaced with the ID of the accepting socket, which is also reported in the
11944 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011945
11946 - "accept_date" is the exact date when the connection was received by haproxy
11947 (which might be very slightly different from the date observed on the
11948 network if there was some queuing in the system's backlog). This is usually
11949 the same date which may appear in any upstream firewall's log.
11950
11951 - "frontend_name" is the name of the frontend (or listener) which received
11952 and processed the connection.
11953
11954 - "backend_name" is the name of the backend (or listener) which was selected
11955 to manage the connection to the server. This will be the same as the
11956 frontend if no switching rule has been applied, which is common for TCP
11957 applications.
11958
11959 - "server_name" is the name of the last server to which the connection was
11960 sent, which might differ from the first one if there were connection errors
11961 and a redispatch occurred. Note that this server belongs to the backend
11962 which processed the request. If the connection was aborted before reaching
11963 a server, "<NOSRV>" is indicated instead of a server name.
11964
11965 - "Tw" is the total time in milliseconds spent waiting in the various queues.
11966 It can be "-1" if the connection was aborted before reaching the queue.
11967 See "Timers" below for more details.
11968
11969 - "Tc" is the total time in milliseconds spent waiting for the connection to
11970 establish to the final server, including retries. It can be "-1" if the
11971 connection was aborted before a connection could be established. See
11972 "Timers" below for more details.
11973
11974 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011975 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011976 "option logasap" was specified, then the time counting stops at the moment
11977 the log is emitted. In this case, a '+' sign is prepended before the value,
11978 indicating that the final one will be larger. See "Timers" below for more
11979 details.
11980
11981 - "bytes_read" is the total number of bytes transmitted from the server to
11982 the client when the log is emitted. If "option logasap" is specified, the
11983 this value will be prefixed with a '+' sign indicating that the final one
11984 may be larger. Please note that this value is a 64-bit counter, so log
11985 analysis tools must be able to handle it without overflowing.
11986
11987 - "termination_state" is the condition the session was in when the session
11988 ended. This indicates the session state, which side caused the end of
11989 session to happen, and for what reason (timeout, error, ...). The normal
11990 flags should be "--", indicating the session was closed by either end with
11991 no data remaining in buffers. See below "Session state at disconnection"
11992 for more details.
11993
11994 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011995 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011996 limits have been reached. For instance, if actconn is close to 512 when
11997 multiple connection errors occur, chances are high that the system limits
11998 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011999 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012000
12001 - "feconn" is the total number of concurrent connections on the frontend when
12002 the session was logged. It is useful to estimate the amount of resource
12003 required to sustain high loads, and to detect when the frontend's "maxconn"
12004 has been reached. Most often when this value increases by huge jumps, it is
12005 because there is congestion on the backend servers, but sometimes it can be
12006 caused by a denial of service attack.
12007
12008 - "beconn" is the total number of concurrent connections handled by the
12009 backend when the session was logged. It includes the total number of
12010 concurrent connections active on servers as well as the number of
12011 connections pending in queues. It is useful to estimate the amount of
12012 additional servers needed to support high loads for a given application.
12013 Most often when this value increases by huge jumps, it is because there is
12014 congestion on the backend servers, but sometimes it can be caused by a
12015 denial of service attack.
12016
12017 - "srv_conn" is the total number of concurrent connections still active on
12018 the server when the session was logged. It can never exceed the server's
12019 configured "maxconn" parameter. If this value is very often close or equal
12020 to the server's "maxconn", it means that traffic regulation is involved a
12021 lot, meaning that either the server's maxconn value is too low, or that
12022 there aren't enough servers to process the load with an optimal response
12023 time. When only one of the server's "srv_conn" is high, it usually means
12024 that this server has some trouble causing the connections to take longer to
12025 be processed than on other servers.
12026
12027 - "retries" is the number of connection retries experienced by this session
12028 when trying to connect to the server. It must normally be zero, unless a
12029 server is being stopped at the same moment the connection was attempted.
12030 Frequent retries generally indicate either a network problem between
12031 haproxy and the server, or a misconfigured system backlog on the server
12032 preventing new connections from being queued. This field may optionally be
12033 prefixed with a '+' sign, indicating that the session has experienced a
12034 redispatch after the maximal retry count has been reached on the initial
12035 server. In this case, the server name appearing in the log is the one the
12036 connection was redispatched to, and not the first one, though both may
12037 sometimes be the same in case of hashing for instance. So as a general rule
12038 of thumb, when a '+' is present in front of the retry count, this count
12039 should not be attributed to the logged server.
12040
12041 - "srv_queue" is the total number of requests which were processed before
12042 this one in the server queue. It is zero when the request has not gone
12043 through the server queue. It makes it possible to estimate the approximate
12044 server's response time by dividing the time spent in queue by the number of
12045 requests in the queue. It is worth noting that if a session experiences a
12046 redispatch and passes through two server queues, their positions will be
12047 cumulated. A request should not pass through both the server queue and the
12048 backend queue unless a redispatch occurs.
12049
12050 - "backend_queue" is the total number of requests which were processed before
12051 this one in the backend's global queue. It is zero when the request has not
12052 gone through the global queue. It makes it possible to estimate the average
12053 queue length, which easily translates into a number of missing servers when
12054 divided by a server's "maxconn" parameter. It is worth noting that if a
12055 session experiences a redispatch, it may pass twice in the backend's queue,
12056 and then both positions will be cumulated. A request should not pass
12057 through both the server queue and the backend queue unless a redispatch
12058 occurs.
12059
12060
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200120618.2.3. HTTP log format
12062----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012063
12064The HTTP format is the most complete and the best suited for HTTP proxies. It
12065is enabled by when "option httplog" is specified in the frontend. It provides
12066the same level of information as the TCP format with additional features which
12067are specific to the HTTP protocol. Just like the TCP format, the log is usually
12068emitted at the end of the session, unless "option logasap" is specified, which
12069generally only makes sense for download sites. A session which matches the
12070"monitor" rules will never logged. It is also possible not to log sessions for
12071which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012072frontend. Successful connections will not be logged if "option dontlog-normal"
12073is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012074
12075Most fields are shared with the TCP log, some being different. A few fields may
12076slightly vary depending on some configuration options. Those ones are marked
12077with a star ('*') after the field name below.
12078
12079 Example :
12080 frontend http-in
12081 mode http
12082 option httplog
12083 log global
12084 default_backend bck
12085
12086 backend static
12087 server srv1 127.0.0.1:8000
12088
12089 >>> Feb 6 12:14:14 localhost \
12090 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
12091 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 +010012092 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012093
12094 Field Format Extract from the example above
12095 1 process_name '[' pid ']:' haproxy[14389]:
12096 2 client_ip ':' client_port 10.0.1.2:33317
12097 3 '[' accept_date ']' [06/Feb/2009:12:14:14.655]
12098 4 frontend_name http-in
12099 5 backend_name '/' server_name static/srv1
12100 6 Tq '/' Tw '/' Tc '/' Tr '/' Tt* 10/0/30/69/109
12101 7 status_code 200
12102 8 bytes_read* 2750
12103 9 captured_request_cookie -
12104 10 captured_response_cookie -
12105 11 termination_state ----
12106 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
12107 13 srv_queue '/' backend_queue 0/0
12108 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
12109 15 '{' captured_response_headers* '}' {}
12110 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010012111
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012112
12113Detailed fields description :
12114 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012115 connection to haproxy. If the connection was accepted on a UNIX socket
12116 instead, the IP address would be replaced with the word "unix". Note that
12117 when the connection is accepted on a socket configured with "accept-proxy"
12118 and the PROXY protocol is correctly used, then the logs will reflect the
12119 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012120
12121 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012122 If the connection was accepted on a UNIX socket instead, the port would be
12123 replaced with the ID of the accepting socket, which is also reported in the
12124 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012125
12126 - "accept_date" is the exact date when the TCP connection was received by
12127 haproxy (which might be very slightly different from the date observed on
12128 the network if there was some queuing in the system's backlog). This is
12129 usually the same date which may appear in any upstream firewall's log. This
12130 does not depend on the fact that the client has sent the request or not.
12131
12132 - "frontend_name" is the name of the frontend (or listener) which received
12133 and processed the connection.
12134
12135 - "backend_name" is the name of the backend (or listener) which was selected
12136 to manage the connection to the server. This will be the same as the
12137 frontend if no switching rule has been applied.
12138
12139 - "server_name" is the name of the last server to which the connection was
12140 sent, which might differ from the first one if there were connection errors
12141 and a redispatch occurred. Note that this server belongs to the backend
12142 which processed the request. If the request was aborted before reaching a
12143 server, "<NOSRV>" is indicated instead of a server name. If the request was
12144 intercepted by the stats subsystem, "<STATS>" is indicated instead.
12145
12146 - "Tq" is the total time in milliseconds spent waiting for the client to send
12147 a full HTTP request, not counting data. It can be "-1" if the connection
12148 was aborted before a complete request could be received. It should always
12149 be very small because a request generally fits in one single packet. Large
12150 times here generally indicate network trouble between the client and
12151 haproxy. See "Timers" below for more details.
12152
12153 - "Tw" is the total time in milliseconds spent waiting in the various queues.
12154 It can be "-1" if the connection was aborted before reaching the queue.
12155 See "Timers" below for more details.
12156
12157 - "Tc" is the total time in milliseconds spent waiting for the connection to
12158 establish to the final server, including retries. It can be "-1" if the
12159 request was aborted before a connection could be established. See "Timers"
12160 below for more details.
12161
12162 - "Tr" is the total time in milliseconds spent waiting for the server to send
12163 a full HTTP response, not counting data. It can be "-1" if the request was
12164 aborted before a complete response could be received. It generally matches
12165 the server's processing time for the request, though it may be altered by
12166 the amount of data sent by the client to the server. Large times here on
12167 "GET" requests generally indicate an overloaded server. See "Timers" below
12168 for more details.
12169
12170 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012171 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012172 "option logasap" was specified, then the time counting stops at the moment
12173 the log is emitted. In this case, a '+' sign is prepended before the value,
12174 indicating that the final one will be larger. See "Timers" below for more
12175 details.
12176
12177 - "status_code" is the HTTP status code returned to the client. This status
12178 is generally set by the server, but it might also be set by haproxy when
12179 the server cannot be reached or when its response is blocked by haproxy.
12180
12181 - "bytes_read" is the total number of bytes transmitted to the client when
12182 the log is emitted. This does include HTTP headers. If "option logasap" is
12183 specified, the this value will be prefixed with a '+' sign indicating that
12184 the final one may be larger. Please note that this value is a 64-bit
12185 counter, so log analysis tools must be able to handle it without
12186 overflowing.
12187
12188 - "captured_request_cookie" is an optional "name=value" entry indicating that
12189 the client had this cookie in the request. The cookie name and its maximum
12190 length are defined by the "capture cookie" statement in the frontend
12191 configuration. The field is a single dash ('-') when the option is not
12192 set. Only one cookie may be captured, it is generally used to track session
12193 ID exchanges between a client and a server to detect session crossing
12194 between clients due to application bugs. For more details, please consult
12195 the section "Capturing HTTP headers and cookies" below.
12196
12197 - "captured_response_cookie" is an optional "name=value" entry indicating
12198 that the server has returned a cookie with its response. The cookie name
12199 and its maximum length are defined by the "capture cookie" statement in the
12200 frontend configuration. The field is a single dash ('-') when the option is
12201 not set. Only one cookie may be captured, it is generally used to track
12202 session ID exchanges between a client and a server to detect session
12203 crossing between clients due to application bugs. For more details, please
12204 consult the section "Capturing HTTP headers and cookies" below.
12205
12206 - "termination_state" is the condition the session was in when the session
12207 ended. This indicates the session state, which side caused the end of
12208 session to happen, for what reason (timeout, error, ...), just like in TCP
12209 logs, and information about persistence operations on cookies in the last
12210 two characters. The normal flags should begin with "--", indicating the
12211 session was closed by either end with no data remaining in buffers. See
12212 below "Session state at disconnection" for more details.
12213
12214 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012215 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012216 limits have been reached. For instance, if actconn is close to 512 or 1024
12217 when multiple connection errors occur, chances are high that the system
12218 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012219 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012220 system.
12221
12222 - "feconn" is the total number of concurrent connections on the frontend when
12223 the session was logged. It is useful to estimate the amount of resource
12224 required to sustain high loads, and to detect when the frontend's "maxconn"
12225 has been reached. Most often when this value increases by huge jumps, it is
12226 because there is congestion on the backend servers, but sometimes it can be
12227 caused by a denial of service attack.
12228
12229 - "beconn" is the total number of concurrent connections handled by the
12230 backend when the session was logged. It includes the total number of
12231 concurrent connections active on servers as well as the number of
12232 connections pending in queues. It is useful to estimate the amount of
12233 additional servers needed to support high loads for a given application.
12234 Most often when this value increases by huge jumps, it is because there is
12235 congestion on the backend servers, but sometimes it can be caused by a
12236 denial of service attack.
12237
12238 - "srv_conn" is the total number of concurrent connections still active on
12239 the server when the session was logged. It can never exceed the server's
12240 configured "maxconn" parameter. If this value is very often close or equal
12241 to the server's "maxconn", it means that traffic regulation is involved a
12242 lot, meaning that either the server's maxconn value is too low, or that
12243 there aren't enough servers to process the load with an optimal response
12244 time. When only one of the server's "srv_conn" is high, it usually means
12245 that this server has some trouble causing the requests to take longer to be
12246 processed than on other servers.
12247
12248 - "retries" is the number of connection retries experienced by this session
12249 when trying to connect to the server. It must normally be zero, unless a
12250 server is being stopped at the same moment the connection was attempted.
12251 Frequent retries generally indicate either a network problem between
12252 haproxy and the server, or a misconfigured system backlog on the server
12253 preventing new connections from being queued. This field may optionally be
12254 prefixed with a '+' sign, indicating that the session has experienced a
12255 redispatch after the maximal retry count has been reached on the initial
12256 server. In this case, the server name appearing in the log is the one the
12257 connection was redispatched to, and not the first one, though both may
12258 sometimes be the same in case of hashing for instance. So as a general rule
12259 of thumb, when a '+' is present in front of the retry count, this count
12260 should not be attributed to the logged server.
12261
12262 - "srv_queue" is the total number of requests which were processed before
12263 this one in the server queue. It is zero when the request has not gone
12264 through the server queue. It makes it possible to estimate the approximate
12265 server's response time by dividing the time spent in queue by the number of
12266 requests in the queue. It is worth noting that if a session experiences a
12267 redispatch and passes through two server queues, their positions will be
12268 cumulated. A request should not pass through both the server queue and the
12269 backend queue unless a redispatch occurs.
12270
12271 - "backend_queue" is the total number of requests which were processed before
12272 this one in the backend's global queue. It is zero when the request has not
12273 gone through the global queue. It makes it possible to estimate the average
12274 queue length, which easily translates into a number of missing servers when
12275 divided by a server's "maxconn" parameter. It is worth noting that if a
12276 session experiences a redispatch, it may pass twice in the backend's queue,
12277 and then both positions will be cumulated. A request should not pass
12278 through both the server queue and the backend queue unless a redispatch
12279 occurs.
12280
12281 - "captured_request_headers" is a list of headers captured in the request due
12282 to the presence of the "capture request header" statement in the frontend.
12283 Multiple headers can be captured, they will be delimited by a vertical bar
12284 ('|'). When no capture is enabled, the braces do not appear, causing a
12285 shift of remaining fields. It is important to note that this field may
12286 contain spaces, and that using it requires a smarter log parser than when
12287 it's not used. Please consult the section "Capturing HTTP headers and
12288 cookies" below for more details.
12289
12290 - "captured_response_headers" is a list of headers captured in the response
12291 due to the presence of the "capture response header" statement in the
12292 frontend. Multiple headers can be captured, they will be delimited by a
12293 vertical bar ('|'). When no capture is enabled, the braces do not appear,
12294 causing a shift of remaining fields. It is important to note that this
12295 field may contain spaces, and that using it requires a smarter log parser
12296 than when it's not used. Please consult the section "Capturing HTTP headers
12297 and cookies" below for more details.
12298
12299 - "http_request" is the complete HTTP request line, including the method,
12300 request and HTTP version string. Non-printable characters are encoded (see
12301 below the section "Non-printable characters"). This is always the last
12302 field, and it is always delimited by quotes and is the only one which can
12303 contain quotes. If new fields are added to the log format, they will be
12304 added before this field. This field might be truncated if the request is
12305 huge and does not fit in the standard syslog buffer (1024 characters). This
12306 is the reason why this field must always remain the last one.
12307
12308
Cyril Bontédc4d9032012-04-08 21:57:39 +0200123098.2.4. Custom log format
12310------------------------
William Lallemand48940402012-01-30 16:47:22 +010012311
Willy Tarreau2beef582012-12-20 17:22:52 +010012312The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010012313mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010012314
12315HAproxy understands some log format variables. % precedes log format variables.
12316Variables can take arguments using braces ('{}'), and multiple arguments are
12317separated by commas within the braces. Flags may be added or removed by
12318prefixing them with a '+' or '-' sign.
12319
12320Special variable "%o" may be used to propagate its flags to all other
12321variables on the same format string. This is particularly handy with quoted
12322string formats ("Q").
12323
Willy Tarreauc8368452012-12-21 00:09:23 +010012324If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020012325as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010012326less common information such as the client's SSL certificate's DN, or to log
12327the key that would be used to store an entry into a stick table.
12328
William Lallemand48940402012-01-30 16:47:22 +010012329Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012330In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010012331in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010012332
12333Flags are :
12334 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012335 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
William Lallemand48940402012-01-30 16:47:22 +010012336
12337 Example:
12338
12339 log-format %T\ %t\ Some\ Text
12340 log-format %{+Q}o\ %t\ %s\ %{-Q}r
12341
12342At the moment, the default HTTP format is defined this way :
12343
Willy Tarreau2beef582012-12-20 17:22:52 +010012344 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tq/%Tw/%Tc/%Tr/%Tt\ %ST\ %B\ %CC\ \
12345 %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r
William Lallemand48940402012-01-30 16:47:22 +010012346
William Lallemandbddd4fd2012-02-27 11:23:10 +010012347the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010012348
Willy Tarreau2beef582012-12-20 17:22:52 +010012349 log-format %{+Q}o\ %{-Q}ci\ -\ -\ [%T]\ %r\ %ST\ %B\ \"\"\ \"\"\ %cp\ \
Willy Tarreau773d65f2012-10-12 14:56:11 +020012350 %ms\ %ft\ %b\ %s\ \%Tq\ %Tw\ %Tc\ %Tr\ %Tt\ %tsc\ %ac\ %fc\ \
Willy Tarreau2beef582012-12-20 17:22:52 +010012351 %bc\ %sc\ %rc\ %sq\ %bq\ %CC\ %CS\ \%hrl\ %hsl
William Lallemand48940402012-01-30 16:47:22 +010012352
William Lallemandbddd4fd2012-02-27 11:23:10 +010012353and the default TCP format is defined this way :
12354
Willy Tarreau2beef582012-12-20 17:22:52 +010012355 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ \
William Lallemandbddd4fd2012-02-27 11:23:10 +010012356 %ac/%fc/%bc/%sc/%rc\ %sq/%bq
12357
William Lallemand48940402012-01-30 16:47:22 +010012358Please refer to the table below for currently defined variables :
12359
William Lallemandbddd4fd2012-02-27 11:23:10 +010012360 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012361 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012362 +---+------+-----------------------------------------------+-------------+
12363 | | %o | special variable, apply flags on all next var | |
12364 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010012365 | | %B | bytes_read (from server to client) | numeric |
12366 | H | %CC | captured_request_cookie | string |
12367 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020012368 | | %H | hostname | string |
William Lallemanda73203e2012-03-12 12:48:57 +010012369 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020012370 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012371 | | %T | gmt_date_time | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012372 | | %Tc | Tc | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080012373 | | %Tl | local_date_time | date |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012374 | H | %Tq | Tq | numeric |
12375 | H | %Tr | Tr | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012376 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012377 | | %Tt | Tt | numeric |
12378 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012379 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012380 | | %ac | actconn | numeric |
12381 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012382 | | %bc | beconn (backend concurrent connections) | numeric |
12383 | | %bi | backend_source_ip (connecting address) | IP |
12384 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012385 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012386 | | %ci | client_ip (accepted address) | IP |
12387 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012388 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012389 | | %fc | feconn (frontend concurrent connections) | numeric |
12390 | | %fi | frontend_ip (accepting address) | IP |
12391 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020012392 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020012393 | | %hr | captured_request_headers default style | string |
12394 | | %hrl | captured_request_headers CLF style | string list |
12395 | | %hs | captured_response_headers default style | string |
12396 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreauc3453d52015-08-09 10:56:35 +020012397 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012398 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012399 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012400 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010012401 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012402 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012403 | | %sc | srv_conn (server concurrent connections) | numeric |
12404 | | %si | server_IP (target address) | IP |
12405 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012406 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012407 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
12408 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012409 | | %t | date_time (with millisecond resolution) | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012410 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012411 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012412 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010012413
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012414 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010012415
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010012416
124178.2.5. Error log format
12418-----------------------
12419
12420When an incoming connection fails due to an SSL handshake or an invalid PROXY
12421protocol header, haproxy will log the event using a shorter, fixed line format.
12422By default, logs are emitted at the LOG_INFO level, unless the option
12423"log-separate-errors" is set in the backend, in which case the LOG_ERR level
12424will be used. Connections on which no data are exchanged (eg: probes) are not
12425logged if the "dontlognull" option is set.
12426
12427The format looks like this :
12428
12429 >>> Dec 3 18:27:14 localhost \
12430 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
12431 Connection error during SSL handshake
12432
12433 Field Format Extract from the example above
12434 1 process_name '[' pid ']:' haproxy[6103]:
12435 2 client_ip ':' client_port 127.0.0.1:56059
12436 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
12437 4 frontend_name "/" bind_name ":" frt/f1:
12438 5 message Connection error during SSL handshake
12439
12440These fields just provide minimal information to help debugging connection
12441failures.
12442
12443
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124448.3. Advanced logging options
12445-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012446
12447Some advanced logging options are often looked for but are not easy to find out
12448just by looking at the various options. Here is an entry point for the few
12449options which can enable better logging. Please refer to the keywords reference
12450for more information about their usage.
12451
12452
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124538.3.1. Disabling logging of external tests
12454------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012455
12456It is quite common to have some monitoring tools perform health checks on
12457haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
12458commercial load-balancer, and sometimes it will simply be a more complete
12459monitoring system such as Nagios. When the tests are very frequent, users often
12460ask how to disable logging for those checks. There are three possibilities :
12461
12462 - if connections come from everywhere and are just TCP probes, it is often
12463 desired to simply disable logging of connections without data exchange, by
12464 setting "option dontlognull" in the frontend. It also disables logging of
12465 port scans, which may or may not be desired.
12466
12467 - if the connection come from a known source network, use "monitor-net" to
12468 declare this network as monitoring only. Any host in this network will then
12469 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012470 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012471 such as other load-balancers.
12472
12473 - if the tests are performed on a known URI, use "monitor-uri" to declare
12474 this URI as dedicated to monitoring. Any host sending this request will
12475 only get the result of a health-check, and the request will not be logged.
12476
12477
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124788.3.2. Logging before waiting for the session to terminate
12479----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012480
12481The problem with logging at end of connection is that you have no clue about
12482what is happening during very long sessions, such as remote terminal sessions
12483or large file downloads. This problem can be worked around by specifying
12484"option logasap" in the frontend. Haproxy will then log as soon as possible,
12485just before data transfer begins. This means that in case of TCP, it will still
12486log the connection status to the server, and in case of HTTP, it will log just
12487after processing the server headers. In this case, the number of bytes reported
12488is the number of header bytes sent to the client. In order to avoid confusion
12489with normal logs, the total time field and the number of bytes are prefixed
12490with a '+' sign which means that real numbers are certainly larger.
12491
12492
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124938.3.3. Raising log level upon errors
12494------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012495
12496Sometimes it is more convenient to separate normal traffic from errors logs,
12497for instance in order to ease error monitoring from log files. When the option
12498"log-separate-errors" is used, connections which experience errors, timeouts,
12499retries, redispatches or HTTP status codes 5xx will see their syslog level
12500raised from "info" to "err". This will help a syslog daemon store the log in
12501a separate file. It is very important to keep the errors in the normal traffic
12502file too, so that log ordering is not altered. You should also be careful if
12503you already have configured your syslog daemon to store all logs higher than
12504"notice" in an "admin" file, because the "err" level is higher than "notice".
12505
12506
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200125078.3.4. Disabling logging of successful connections
12508--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012509
12510Although this may sound strange at first, some large sites have to deal with
12511multiple thousands of logs per second and are experiencing difficulties keeping
12512them intact for a long time or detecting errors within them. If the option
12513"dontlog-normal" is set on the frontend, all normal connections will not be
12514logged. In this regard, a normal connection is defined as one without any
12515error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
12516and a response with a status 5xx is not considered normal and will be logged
12517too. Of course, doing is is really discouraged as it will remove most of the
12518useful information from the logs. Do this only if you have no other
12519alternative.
12520
12521
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200125228.4. Timing events
12523------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012524
12525Timers provide a great help in troubleshooting network problems. All values are
12526reported in milliseconds (ms). These timers should be used in conjunction with
12527the session termination flags. In TCP mode with "option tcplog" set on the
12528frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
12529mode, 5 control points are reported under the form "Tq/Tw/Tc/Tr/Tt" :
12530
12531 - Tq: total time to get the client request (HTTP mode only). It's the time
12532 elapsed between the moment the client connection was accepted and the
12533 moment the proxy received the last HTTP header. The value "-1" indicates
12534 that the end of headers (empty line) has never been seen. This happens when
12535 the client closes prematurely or times out.
12536
12537 - Tw: total time spent in the queues waiting for a connection slot. It
12538 accounts for backend queue as well as the server queues, and depends on the
12539 queue size, and the time needed for the server to complete previous
12540 requests. The value "-1" means that the request was killed before reaching
12541 the queue, which is generally what happens with invalid or denied requests.
12542
12543 - Tc: total time to establish the TCP connection to the server. It's the time
12544 elapsed between the moment the proxy sent the connection request, and the
12545 moment it was acknowledged by the server, or between the TCP SYN packet and
12546 the matching SYN/ACK packet in return. The value "-1" means that the
12547 connection never established.
12548
12549 - Tr: server response time (HTTP mode only). It's the time elapsed between
12550 the moment the TCP connection was established to the server and the moment
12551 the server sent its complete response headers. It purely shows its request
12552 processing time, without the network overhead due to the data transmission.
12553 It is worth noting that when the client has data to send to the server, for
12554 instance during a POST request, the time already runs, and this can distort
12555 apparent response time. For this reason, it's generally wise not to trust
12556 too much this field for POST requests initiated from clients behind an
12557 untrusted network. A value of "-1" here means that the last the response
12558 header (empty line) was never seen, most likely because the server timeout
12559 stroke before the server managed to process the request.
12560
12561 - Tt: total session duration time, between the moment the proxy accepted it
12562 and the moment both ends were closed. The exception is when the "logasap"
12563 option is specified. In this case, it only equals (Tq+Tw+Tc+Tr), and is
12564 prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012565 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012566
12567 Td = Tt - (Tq + Tw + Tc + Tr)
12568
12569 Timers with "-1" values have to be excluded from this equation. In TCP
12570 mode, "Tq" and "Tr" have to be excluded too. Note that "Tt" can never be
12571 negative.
12572
12573These timers provide precious indications on trouble causes. Since the TCP
12574protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
12575that timers close to multiples of 3s are nearly always related to lost packets
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012576due to network problems (wires, negotiation, congestion). Moreover, if "Tt" is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012577close to a timeout value specified in the configuration, it often means that a
12578session has been aborted on timeout.
12579
12580Most common cases :
12581
12582 - If "Tq" is close to 3000, a packet has probably been lost between the
12583 client and the proxy. This is very rare on local networks but might happen
12584 when clients are on far remote networks and send large requests. It may
12585 happen that values larger than usual appear here without any network cause.
12586 Sometimes, during an attack or just after a resource starvation has ended,
12587 haproxy may accept thousands of connections in a few milliseconds. The time
12588 spent accepting these connections will inevitably slightly delay processing
12589 of other connections, and it can happen that request times in the order of
12590 a few tens of milliseconds are measured after a few thousands of new
Lukas Tribus3082bde2016-02-03 18:09:37 +010012591 connections have been accepted at once. Using one of the keep-alive modes
Patrick Mezard105faca2010-06-12 17:02:46 +020012592 may display larger request times since "Tq" also measures the time spent
12593 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012594
12595 - If "Tc" is close to 3000, a packet has probably been lost between the
12596 server and the proxy during the server connection phase. This value should
12597 always be very low, such as 1 ms on local networks and less than a few tens
12598 of ms on remote networks.
12599
Willy Tarreau55165fe2009-05-10 12:02:55 +020012600 - If "Tr" is nearly always lower than 3000 except some rare values which seem
12601 to be the average majored by 3000, there are probably some packets lost
12602 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012603
12604 - If "Tt" is large even for small byte counts, it generally is because
12605 neither the client nor the server decides to close the connection, for
12606 instance because both have agreed on a keep-alive connection mode. In order
12607 to solve this issue, it will be needed to specify "option httpclose" on
12608 either the frontend or the backend. If the problem persists, it means that
12609 the server ignores the "close" connection mode and expects the client to
12610 close. Then it will be required to use "option forceclose". Having the
12611 smallest possible 'Tt' is important when connection regulation is used with
12612 the "maxconn" option on the servers, since no new connection will be sent
12613 to the server until another one is released.
12614
12615Other noticeable HTTP log cases ('xx' means any value to be ignored) :
12616
12617 Tq/Tw/Tc/Tr/+Tt The "option logasap" is present on the frontend and the log
12618 was emitted before the data phase. All the timers are valid
12619 except "Tt" which is shorter than reality.
12620
12621 -1/xx/xx/xx/Tt The client was not able to send a complete request in time
12622 or it aborted too early. Check the session termination flags
12623 then "timeout http-request" and "timeout client" settings.
12624
12625 Tq/-1/xx/xx/Tt It was not possible to process the request, maybe because
12626 servers were out of order, because the request was invalid
12627 or forbidden by ACL rules. Check the session termination
12628 flags.
12629
12630 Tq/Tw/-1/xx/Tt The connection could not establish on the server. Either it
12631 actively refused it or it timed out after Tt-(Tq+Tw) ms.
12632 Check the session termination flags, then check the
12633 "timeout connect" setting. Note that the tarpit action might
12634 return similar-looking patterns, with "Tw" equal to the time
12635 the client connection was maintained open.
12636
12637 Tq/Tw/Tc/-1/Tt The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012638 a complete response in time, or it closed its connection
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012639 unexpectedly after Tt-(Tq+Tw+Tc) ms. Check the session
12640 termination flags, then check the "timeout server" setting.
12641
12642
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126438.5. Session state at disconnection
12644-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012645
12646TCP and HTTP logs provide a session termination indicator in the
12647"termination_state" field, just before the number of active connections. It is
126482-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
12649each of which has a special meaning :
12650
12651 - On the first character, a code reporting the first event which caused the
12652 session to terminate :
12653
12654 C : the TCP session was unexpectedly aborted by the client.
12655
12656 S : the TCP session was unexpectedly aborted by the server, or the
12657 server explicitly refused it.
12658
12659 P : the session was prematurely aborted by the proxy, because of a
12660 connection limit enforcement, because a DENY filter was matched,
12661 because of a security check which detected and blocked a dangerous
12662 error in server response which might have caused information leak
Willy Tarreau570f2212013-06-10 16:42:09 +020012663 (eg: cacheable cookie).
12664
12665 L : the session was locally processed by haproxy and was not passed to
12666 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012667
12668 R : a resource on the proxy has been exhausted (memory, sockets, source
12669 ports, ...). Usually, this appears during the connection phase, and
12670 system logs should contain a copy of the precise error. If this
12671 happens, it must be considered as a very serious anomaly which
12672 should be fixed as soon as possible by any means.
12673
12674 I : an internal error was identified by the proxy during a self-check.
12675 This should NEVER happen, and you are encouraged to report any log
12676 containing this, because this would almost certainly be a bug. It
12677 would be wise to preventively restart the process after such an
12678 event too, in case it would be caused by memory corruption.
12679
Simon Horman752dc4a2011-06-21 14:34:59 +090012680 D : the session was killed by haproxy because the server was detected
12681 as down and was configured to kill all connections when going down.
12682
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012683 U : the session was killed by haproxy on this backup server because an
12684 active server was detected as up and was configured to kill all
12685 backup connections when going up.
12686
Willy Tarreaua2a64e92011-09-07 23:01:56 +020012687 K : the session was actively killed by an admin operating on haproxy.
12688
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012689 c : the client-side timeout expired while waiting for the client to
12690 send or receive data.
12691
12692 s : the server-side timeout expired while waiting for the server to
12693 send or receive data.
12694
12695 - : normal session completion, both the client and the server closed
12696 with nothing left in the buffers.
12697
12698 - on the second character, the TCP or HTTP session state when it was closed :
12699
Willy Tarreauf7b30a92010-12-06 22:59:17 +010012700 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012701 (HTTP mode only). Nothing was sent to any server.
12702
12703 Q : the proxy was waiting in the QUEUE for a connection slot. This can
12704 only happen when servers have a 'maxconn' parameter set. It can
12705 also happen in the global queue after a redispatch consecutive to
12706 a failed attempt to connect to a dying server. If no redispatch is
12707 reported, then no connection attempt was made to any server.
12708
12709 C : the proxy was waiting for the CONNECTION to establish on the
12710 server. The server might at most have noticed a connection attempt.
12711
12712 H : the proxy was waiting for complete, valid response HEADERS from the
12713 server (HTTP only).
12714
12715 D : the session was in the DATA phase.
12716
12717 L : the proxy was still transmitting LAST data to the client while the
12718 server had already finished. This one is very rare as it can only
12719 happen when the client dies while receiving the last packets.
12720
12721 T : the request was tarpitted. It has been held open with the client
12722 during the whole "timeout tarpit" duration or until the client
12723 closed, both of which will be reported in the "Tw" timer.
12724
12725 - : normal session completion after end of data transfer.
12726
12727 - the third character tells whether the persistence cookie was provided by
12728 the client (only in HTTP mode) :
12729
12730 N : the client provided NO cookie. This is usually the case for new
12731 visitors, so counting the number of occurrences of this flag in the
12732 logs generally indicate a valid trend for the site frequentation.
12733
12734 I : the client provided an INVALID cookie matching no known server.
12735 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020012736 cookies between HTTP/HTTPS sites, persistence conditionally
12737 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012738
12739 D : the client provided a cookie designating a server which was DOWN,
12740 so either "option persist" was used and the client was sent to
12741 this server, or it was not set and the client was redispatched to
12742 another server.
12743
Willy Tarreau996a92c2010-10-13 19:30:47 +020012744 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012745 server.
12746
Willy Tarreau996a92c2010-10-13 19:30:47 +020012747 E : the client provided a valid cookie, but with a last date which was
12748 older than what is allowed by the "maxidle" cookie parameter, so
12749 the cookie is consider EXPIRED and is ignored. The request will be
12750 redispatched just as if there was no cookie.
12751
12752 O : the client provided a valid cookie, but with a first date which was
12753 older than what is allowed by the "maxlife" cookie parameter, so
12754 the cookie is consider too OLD and is ignored. The request will be
12755 redispatched just as if there was no cookie.
12756
Willy Tarreauc89ccb62012-04-05 21:18:22 +020012757 U : a cookie was present but was not used to select the server because
12758 some other server selection mechanism was used instead (typically a
12759 "use-server" rule).
12760
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012761 - : does not apply (no cookie set in configuration).
12762
12763 - the last character reports what operations were performed on the persistence
12764 cookie returned by the server (only in HTTP mode) :
12765
12766 N : NO cookie was provided by the server, and none was inserted either.
12767
12768 I : no cookie was provided by the server, and the proxy INSERTED one.
12769 Note that in "cookie insert" mode, if the server provides a cookie,
12770 it will still be overwritten and reported as "I" here.
12771
Willy Tarreau996a92c2010-10-13 19:30:47 +020012772 U : the proxy UPDATED the last date in the cookie that was presented by
12773 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012774 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020012775 date indicated in the cookie. If any other change happens, such as
12776 a redispatch, then the cookie will be marked as inserted instead.
12777
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012778 P : a cookie was PROVIDED by the server and transmitted as-is.
12779
12780 R : the cookie provided by the server was REWRITTEN by the proxy, which
12781 happens in "cookie rewrite" or "cookie prefix" modes.
12782
12783 D : the cookie provided by the server was DELETED by the proxy.
12784
12785 - : does not apply (no cookie set in configuration).
12786
Willy Tarreau996a92c2010-10-13 19:30:47 +020012787The combination of the two first flags gives a lot of information about what
12788was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012789helpful to detect server saturation, network troubles, local system resource
12790starvation, attacks, etc...
12791
12792The most common termination flags combinations are indicated below. They are
12793alphabetically sorted, with the lowercase set just after the upper case for
12794easier finding and understanding.
12795
12796 Flags Reason
12797
12798 -- Normal termination.
12799
12800 CC The client aborted before the connection could be established to the
12801 server. This can happen when haproxy tries to connect to a recently
12802 dead (or unchecked) server, and the client aborts while haproxy is
12803 waiting for the server to respond or for "timeout connect" to expire.
12804
12805 CD The client unexpectedly aborted during data transfer. This can be
12806 caused by a browser crash, by an intermediate equipment between the
12807 client and haproxy which decided to actively break the connection,
12808 by network routing issues between the client and haproxy, or by a
12809 keep-alive session between the server and the client terminated first
12810 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012811
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012812 cD The client did not send nor acknowledge any data for as long as the
12813 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020012814 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012815
12816 CH The client aborted while waiting for the server to start responding.
12817 It might be the server taking too long to respond or the client
12818 clicking the 'Stop' button too fast.
12819
12820 cH The "timeout client" stroke while waiting for client data during a
12821 POST request. This is sometimes caused by too large TCP MSS values
12822 for PPPoE networks which cannot transport full-sized packets. It can
12823 also happen when client timeout is smaller than server timeout and
12824 the server takes too long to respond.
12825
12826 CQ The client aborted while its session was queued, waiting for a server
12827 with enough empty slots to accept it. It might be that either all the
12828 servers were saturated or that the assigned server was taking too
12829 long a time to respond.
12830
12831 CR The client aborted before sending a full HTTP request. Most likely
12832 the request was typed by hand using a telnet client, and aborted
12833 too early. The HTTP status code is likely a 400 here. Sometimes this
12834 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau1c3a6122015-05-01 15:37:53 +020012835 and the client. "option http-ignore-probes" can be used to ignore
12836 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012837
12838 cR The "timeout http-request" stroke before the client sent a full HTTP
12839 request. This is sometimes caused by too large TCP MSS values on the
12840 client side for PPPoE networks which cannot transport full-sized
12841 packets, or by clients sending requests by hand and not typing fast
12842 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020012843 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau1c3a6122015-05-01 15:37:53 +020012844 some browsers started to implement a "pre-connect" feature consisting
12845 in speculatively connecting to some recently visited web sites just
12846 in case the user would like to visit them. This results in many
12847 connections being established to web sites, which end up in 408
12848 Request Timeout if the timeout strikes first, or 400 Bad Request when
12849 the browser decides to close them first. These ones pollute the log
12850 and feed the error counters. Some versions of some browsers have even
12851 been reported to display the error code. It is possible to work
12852 around the undesirable effects of this behaviour by adding "option
12853 http-ignore-probes" in the frontend, resulting in connections with
12854 zero data transfer to be totally ignored. This will definitely hide
12855 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012856
12857 CT The client aborted while its session was tarpitted. It is important to
12858 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020012859 wrong tarpit rules have been written. If a lot of them happen, it
12860 might make sense to lower the "timeout tarpit" value to something
12861 closer to the average reported "Tw" timer, in order not to consume
12862 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012863
Willy Tarreau570f2212013-06-10 16:42:09 +020012864 LR The request was intercepted and locally handled by haproxy. Generally
12865 it means that this was a redirect or a stats request.
12866
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012867 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012868 the TCP connection (the proxy received a TCP RST or an ICMP message
12869 in return). Under some circumstances, it can also be the network
12870 stack telling the proxy that the server is unreachable (eg: no route,
12871 or no ARP response on local network). When this happens in HTTP mode,
12872 the status code is likely a 502 or 503 here.
12873
12874 sC The "timeout connect" stroke before a connection to the server could
12875 complete. When this happens in HTTP mode, the status code is likely a
12876 503 or 504 here.
12877
12878 SD The connection to the server died with an error during the data
12879 transfer. This usually means that haproxy has received an RST from
12880 the server or an ICMP message from an intermediate equipment while
12881 exchanging data with the server. This can be caused by a server crash
12882 or by a network issue on an intermediate equipment.
12883
12884 sD The server did not send nor acknowledge any data for as long as the
12885 "timeout server" setting during the data phase. This is often caused
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012886 by too short timeouts on L4 equipments before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012887 load-balancers, ...), as well as keep-alive sessions maintained
12888 between the client and the server expiring first on haproxy.
12889
12890 SH The server aborted before sending its full HTTP response headers, or
12891 it crashed while processing the request. Since a server aborting at
12892 this moment is very rare, it would be wise to inspect its logs to
12893 control whether it crashed and why. The logged request may indicate a
12894 small set of faulty requests, demonstrating bugs in the application.
12895 Sometimes this might also be caused by an IDS killing the connection
12896 between haproxy and the server.
12897
12898 sH The "timeout server" stroke before the server could return its
12899 response headers. This is the most common anomaly, indicating too
12900 long transactions, probably caused by server or database saturation.
12901 The immediate workaround consists in increasing the "timeout server"
12902 setting, but it is important to keep in mind that the user experience
12903 will suffer from these long response times. The only long term
12904 solution is to fix the application.
12905
12906 sQ The session spent too much time in queue and has been expired. See
12907 the "timeout queue" and "timeout connect" settings to find out how to
12908 fix this if it happens too often. If it often happens massively in
12909 short periods, it may indicate general problems on the affected
12910 servers due to I/O or database congestion, or saturation caused by
12911 external attacks.
12912
12913 PC The proxy refused to establish a connection to the server because the
12914 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020012915 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012916 so that it does not happen anymore. This status is very rare and
12917 might happen when the global "ulimit-n" parameter is forced by hand.
12918
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012919 PD The proxy blocked an incorrectly formatted chunked encoded message in
12920 a request or a response, after the server has emitted its headers. In
12921 most cases, this will indicate an invalid message from the server to
Willy Tarreauf3a3e132013-08-31 08:16:26 +020012922 the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647
12923 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012924
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012925 PH The proxy blocked the server's response, because it was invalid,
12926 incomplete, dangerous (cache control), or matched a security filter.
12927 In any case, an HTTP 502 error is sent to the client. One possible
12928 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012929 containing unauthorized characters. It is also possible but quite
12930 rare, that the proxy blocked a chunked-encoding request from the
12931 client due to an invalid syntax, before the server responded. In this
12932 case, an HTTP 400 error is sent to the client and reported in the
12933 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012934
12935 PR The proxy blocked the client's HTTP request, either because of an
12936 invalid HTTP syntax, in which case it returned an HTTP 400 error to
12937 the client, or because a deny filter matched, in which case it
12938 returned an HTTP 403 error.
12939
12940 PT The proxy blocked the client's request and has tarpitted its
12941 connection before returning it a 500 server error. Nothing was sent
12942 to the server. The connection was maintained open for as long as
12943 reported by the "Tw" timer field.
12944
12945 RC A local resource has been exhausted (memory, sockets, source ports)
12946 preventing the connection to the server from establishing. The error
12947 logs will tell precisely what was missing. This is very rare and can
12948 only be solved by proper system tuning.
12949
Willy Tarreau996a92c2010-10-13 19:30:47 +020012950The combination of the two last flags gives a lot of information about how
12951persistence was handled by the client, the server and by haproxy. This is very
12952important to troubleshoot disconnections, when users complain they have to
12953re-authenticate. The commonly encountered flags are :
12954
12955 -- Persistence cookie is not enabled.
12956
12957 NN No cookie was provided by the client, none was inserted in the
12958 response. For instance, this can be in insert mode with "postonly"
12959 set on a GET request.
12960
12961 II A cookie designating an invalid server was provided by the client,
12962 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012963 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020012964 value can be presented by a client when no other server knows it.
12965
12966 NI No cookie was provided by the client, one was inserted in the
12967 response. This typically happens for first requests from every user
12968 in "insert" mode, which makes it an easy way to count real users.
12969
12970 VN A cookie was provided by the client, none was inserted in the
12971 response. This happens for most responses for which the client has
12972 already got a cookie.
12973
12974 VU A cookie was provided by the client, with a last visit date which is
12975 not completely up-to-date, so an updated cookie was provided in
12976 response. This can also happen if there was no date at all, or if
12977 there was a date but the "maxidle" parameter was not set, so that the
12978 cookie can be switched to unlimited time.
12979
12980 EI A cookie was provided by the client, with a last visit date which is
12981 too old for the "maxidle" parameter, so the cookie was ignored and a
12982 new cookie was inserted in the response.
12983
12984 OI A cookie was provided by the client, with a first visit date which is
12985 too old for the "maxlife" parameter, so the cookie was ignored and a
12986 new cookie was inserted in the response.
12987
12988 DI The server designated by the cookie was down, a new server was
12989 selected and a new cookie was emitted in the response.
12990
12991 VI The server designated by the cookie was not marked dead but could not
12992 be reached. A redispatch happened and selected another one, which was
12993 then advertised in the response.
12994
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012995
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129968.6. Non-printable characters
12997-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012998
12999In order not to cause trouble to log analysis tools or terminals during log
13000consulting, non-printable characters are not sent as-is into log files, but are
13001converted to the two-digits hexadecimal representation of their ASCII code,
13002prefixed by the character '#'. The only characters that can be logged without
13003being escaped are comprised between 32 and 126 (inclusive). Obviously, the
13004escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
13005is the same for the character '"' which becomes "#22", as well as '{', '|' and
13006'}' when logging headers.
13007
13008Note that the space character (' ') is not encoded in headers, which can cause
13009issues for tools relying on space count to locate fields. A typical header
13010containing spaces is "User-Agent".
13011
13012Last, it has been observed that some syslog daemons such as syslog-ng escape
13013the quote ('"') with a backslash ('\'). The reverse operation can safely be
13014performed since no quote may appear anywhere else in the logs.
13015
13016
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130178.7. Capturing HTTP cookies
13018---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013019
13020Cookie capture simplifies the tracking a complete user session. This can be
13021achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013022section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013023cookie will simultaneously be checked in the request ("Cookie:" header) and in
13024the response ("Set-Cookie:" header). The respective values will be reported in
13025the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013026locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013027not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
13028user switches to a new session for example, because the server will reassign it
13029a new cookie. It is also possible to detect if a server unexpectedly sets a
13030wrong cookie to a client, leading to session crossing.
13031
13032 Examples :
13033 # capture the first cookie whose name starts with "ASPSESSION"
13034 capture cookie ASPSESSION len 32
13035
13036 # capture the first cookie whose name is exactly "vgnvisitor"
13037 capture cookie vgnvisitor= len 32
13038
13039
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130408.8. Capturing HTTP headers
13041---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013042
13043Header captures are useful to track unique request identifiers set by an upper
13044proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
13045the response, one can search for information about the response length, how the
13046server asked the cache to behave, or an object location during a redirection.
13047
13048Header captures are performed using the "capture request header" and "capture
13049response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013050section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013051
13052It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013053time. Non-existent headers are logged as empty strings, and if one header
13054appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013055are grouped within braces '{' and '}' in the same order as they were declared,
13056and delimited with a vertical bar '|' without any space. Response headers
13057follow the same representation, but are displayed after a space following the
13058request headers block. These blocks are displayed just before the HTTP request
13059in the logs.
13060
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020013061As a special case, it is possible to specify an HTTP header capture in a TCP
13062frontend. The purpose is to enable logging of headers which will be parsed in
13063an HTTP backend if the request is then switched to this HTTP backend.
13064
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013065 Example :
13066 # This instance chains to the outgoing proxy
13067 listen proxy-out
13068 mode http
13069 option httplog
13070 option logasap
13071 log global
13072 server cache1 192.168.1.1:3128
13073
13074 # log the name of the virtual server
13075 capture request header Host len 20
13076
13077 # log the amount of data uploaded during a POST
13078 capture request header Content-Length len 10
13079
13080 # log the beginning of the referrer
13081 capture request header Referer len 20
13082
13083 # server name (useful for outgoing proxies only)
13084 capture response header Server len 20
13085
13086 # logging the content-length is useful with "option logasap"
13087 capture response header Content-Length len 10
13088
13089 # log the expected cache behaviour on the response
13090 capture response header Cache-Control len 8
13091
13092 # the Via header will report the next proxy's name
13093 capture response header Via len 20
13094
13095 # log the URL location during a redirection
13096 capture response header Location len 20
13097
13098 >>> Aug 9 20:26:09 localhost \
13099 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
13100 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
13101 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
13102 "GET http://fr.adserver.yahoo.com/"
13103
13104 >>> Aug 9 20:30:46 localhost \
13105 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
13106 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
13107 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013108 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013109
13110 >>> Aug 9 20:30:46 localhost \
13111 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
13112 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
13113 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
13114 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013115 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013116
13117
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131188.9. Examples of logs
13119---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013120
13121These are real-world examples of logs accompanied with an explanation. Some of
13122them have been made up by hand. The syslog part has been removed for better
13123reading. Their sole purpose is to explain how to decipher them.
13124
13125 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
13126 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
13127 "HEAD / HTTP/1.0"
13128
13129 => long request (6.5s) entered by hand through 'telnet'. The server replied
13130 in 147 ms, and the session ended normally ('----')
13131
13132 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
13133 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
13134 0/9 "HEAD / HTTP/1.0"
13135
13136 => Idem, but the request was queued in the global queue behind 9 other
13137 requests, and waited there for 1230 ms.
13138
13139 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
13140 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
13141 "GET /image.iso HTTP/1.0"
13142
13143 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013144 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013145 14 ms, 243 bytes of headers were sent to the client, and total time from
13146 accept to first data byte is 30 ms.
13147
13148 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
13149 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
13150 "GET /cgi-bin/bug.cgi? HTTP/1.0"
13151
13152 => the proxy blocked a server response either because of an "rspdeny" or
13153 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020013154 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013155 risked being cached. In this case, the response is replaced with a "502
13156 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
13157 to return the 502 and not the server.
13158
13159 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013160 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 +010013161
13162 => the client never completed its request and aborted itself ("C---") after
13163 8.5s, while the proxy was waiting for the request headers ("-R--").
13164 Nothing was sent to any server.
13165
13166 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
13167 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
13168
13169 => The client never completed its request, which was aborted by the
13170 time-out ("c---") after 50s, while the proxy was waiting for the request
13171 headers ("-R--"). Nothing was sent to any server, but the proxy could
13172 send a 408 return code to the client.
13173
13174 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
13175 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
13176
13177 => This log was produced with "option tcplog". The client timed out after
13178 5 seconds ("c----").
13179
13180 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
13181 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013182 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013183
13184 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013185 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013186 (config says 'retries 3'), and no redispatch (otherwise we would have
13187 seen "/+3"). Status code 503 was returned to the client. There were 115
13188 connections on this server, 202 connections on this proxy, and 205 on
13189 the global process. It is possible that the server refused the
13190 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013191
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013192
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131939. Statistics and monitoring
13194----------------------------
13195
13196It is possible to query HAProxy about its status. The most commonly used
13197mechanism is the HTTP statistics page. This page also exposes an alternative
13198CSV output format for monitoring tools. The same format is provided on the
13199Unix socket.
13200
13201
132029.1. CSV format
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013203---------------
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +010013204
Willy Tarreau7f062c42009-03-05 18:43:00 +010013205The statistics may be consulted either from the unix socket or from the HTTP
Willy Tarreaua3310dc2014-06-16 15:43:21 +020013206page. Both means provide a CSV format whose fields follow. The first line
13207begins with a sharp ('#') and has one word per comma-delimited field which
13208represents the title of the column. All other lines starting at the second one
13209use a classical CSV format using a comma as the delimiter, and the double quote
13210('"') as an optional text delimiter, but only if the enclosed text is ambiguous
13211(if it contains a quote or a comma). The double-quote character ('"') in the
13212text is doubled ('""'), which is the format that most tools recognize. Please
13213do not insert any column before these ones in order not to break tools which
13214use hard-coded column positions.
Willy Tarreau7f062c42009-03-05 18:43:00 +010013215
James Westby75c98ad2014-07-08 10:14:57 -040013216In brackets after each field name are the types which may have a value for
13217that field. The types are L (Listeners), F (Frontends), B (Backends), and
13218S (Servers).
13219
13220 0. pxname [LFBS]: proxy name
13221 1. svname [LFBS]: service name (FRONTEND for frontend, BACKEND for backend,
13222 any name for server/listener)
13223 2. qcur [..BS]: current queued requests. For the backend this reports the
13224 number queued without a server assigned.
13225 3. qmax [..BS]: max value of qcur
13226 4. scur [LFBS]: current sessions
13227 5. smax [LFBS]: max sessions
13228 6. slim [LFBS]: configured session limit
13229 7. stot [LFBS]: cumulative number of connections
13230 8. bin [LFBS]: bytes in
13231 9. bout [LFBS]: bytes out
13232 10. dreq [LFB.]: requests denied because of security concerns.
13233 - For tcp this is because of a matched tcp-request content rule.
13234 - For http this is because of a matched http-request or tarpit rule.
13235 11. dresp [LFBS]: responses denied because of security concerns.
13236 - For http this is because of a matched http-request rule, or
13237 "option checkcache".
13238 12. ereq [LF..]: request errors. Some of the possible causes are:
13239 - early termination from the client, before the request has been sent.
13240 - read error from the client
13241 - client timeout
13242 - client closed connection
13243 - various bad requests from the client.
13244 - request was tarpitted.
13245 13. econ [..BS]: number of requests that encountered an error trying to
13246 connect to a backend server. The backend stat is the sum of the stat
13247 for all servers of that backend, plus any connection errors not
13248 associated with a particular server (such as the backend having no
13249 active servers).
13250 14. eresp [..BS]: response errors. srv_abrt will be counted here also.
13251 Some other errors are:
13252 - write error on the client socket (won't be counted for the server stat)
13253 - failure applying filters to the response.
13254 15. wretr [..BS]: number of times a connection to a server was retried.
13255 16. wredis [..BS]: number of times a request was redispatched to another
13256 server. The server value counts the number of times that server was
13257 switched away from.
13258 17. status [LFBS]: status (UP/DOWN/NOLB/MAINT/MAINT(via)...)
Pavlos Parissisac372e12015-05-02 20:30:44 +020013259 18. weight [..BS]: total weight (backend), server weight (server)
13260 19. act [..BS]: number of active servers (backend), server is active (server)
13261 20. bck [..BS]: number of backup servers (backend), server is backup (server)
James Westby75c98ad2014-07-08 10:14:57 -040013262 21. chkfail [...S]: number of failed checks. (Only counts checks failed when
13263 the server is up.)
13264 22. chkdown [..BS]: number of UP->DOWN transitions. The backend counter counts
13265 transitions to the whole backend being down, rather than the sum of the
13266 counters for each server.
13267 23. lastchg [..BS]: number of seconds since the last UP<->DOWN transition
13268 24. downtime [..BS]: total downtime (in seconds). The value for the backend
13269 is the downtime for the whole backend, not the sum of the server downtime.
13270 25. qlimit [...S]: configured maxqueue for the server, or nothing in the
13271 value is 0 (default, meaning no limit)
13272 26. pid [LFBS]: process id (0 for first instance, 1 for second, ...)
13273 27. iid [LFBS]: unique proxy id
13274 28. sid [L..S]: server id (unique inside a proxy)
13275 29. throttle [...S]: current throttle percentage for the server, when
13276 slowstart is active, or no value if not in slowstart.
13277 30. lbtot [..BS]: total number of times a server was selected, either for new
13278 sessions, or when re-dispatching. The server counter is the number
13279 of times that server was selected.
13280 31. tracked [...S]: id of proxy/server if tracking is enabled.
13281 32. type [LFBS]: (0=frontend, 1=backend, 2=server, 3=socket/listener)
13282 33. rate [.FBS]: number of sessions per second over last elapsed second
13283 34. rate_lim [.F..]: configured limit on new sessions per second
13284 35. rate_max [.FBS]: max number of new sessions per second
13285 36. check_status [...S]: status of last health check, one of:
Cyril Bontéf0c60612010-02-06 14:44:47 +010013286 UNK -> unknown
13287 INI -> initializing
13288 SOCKERR -> socket error
13289 L4OK -> check passed on layer 4, no upper layers testing enabled
Jason Harveycac41222015-04-16 11:13:21 -080013290 L4TOUT -> layer 1-4 timeout
Cyril Bontéf0c60612010-02-06 14:44:47 +010013291 L4CON -> layer 1-4 connection problem, for example
13292 "Connection refused" (tcp rst) or "No route to host" (icmp)
13293 L6OK -> check passed on layer 6
13294 L6TOUT -> layer 6 (SSL) timeout
13295 L6RSP -> layer 6 invalid response - protocol error
13296 L7OK -> check passed on layer 7
13297 L7OKC -> check conditionally passed on layer 7, for example 404 with
13298 disable-on-404
13299 L7TOUT -> layer 7 (HTTP/SMTP) timeout
13300 L7RSP -> layer 7 invalid response - protocol error
13301 L7STS -> layer 7 response error, for example HTTP 5xx
James Westby75c98ad2014-07-08 10:14:57 -040013302 37. check_code [...S]: layer5-7 code, if available
13303 38. check_duration [...S]: time in ms took to finish last health check
13304 39. hrsp_1xx [.FBS]: http responses with 1xx code
13305 40. hrsp_2xx [.FBS]: http responses with 2xx code
13306 41. hrsp_3xx [.FBS]: http responses with 3xx code
13307 42. hrsp_4xx [.FBS]: http responses with 4xx code
13308 43. hrsp_5xx [.FBS]: http responses with 5xx code
13309 44. hrsp_other [.FBS]: http responses with other codes (protocol error)
13310 45. hanafail [...S]: failed health checks details
13311 46. req_rate [.F..]: HTTP requests per second over last elapsed second
13312 47. req_rate_max [.F..]: max number of HTTP requests per second observed
13313 48. req_tot [.F..]: total number of HTTP requests received
13314 49. cli_abrt [..BS]: number of data transfers aborted by the client
13315 50. srv_abrt [..BS]: number of data transfers aborted by the server
13316 (inc. in eresp)
13317 51. comp_in [.FB.]: number of HTTP response bytes fed to the compressor
13318 52. comp_out [.FB.]: number of HTTP response bytes emitted by the compressor
13319 53. comp_byp [.FB.]: number of bytes that bypassed the HTTP compressor
13320 (CPU/BW limit)
13321 54. comp_rsp [.FB.]: number of HTTP responses that were compressed
13322 55. lastsess [..BS]: number of seconds since last session assigned to
13323 server/backend
13324 56. last_chk [...S]: last health check contents or textual error
13325 57. last_agt [...S]: last agent check contents or textual error
13326 58. qtime [..BS]: the average queue time in ms over the 1024 last requests
13327 59. ctime [..BS]: the average connect time in ms over the 1024 last requests
13328 60. rtime [..BS]: the average response time in ms over the 1024 last requests
13329 (0 for TCP)
13330 61. ttime [..BS]: the average total session time in ms over the 1024 last
13331 requests
Willy Tarreau844e3c52008-01-11 16:28:18 +010013332
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013333
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200133349.2. Unix Socket commands
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013335-------------------------
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013336
Willy Tarreau468f4932013-08-01 16:50:16 +020013337The stats socket is not enabled by default. In order to enable it, it is
13338necessary to add one line in the global section of the haproxy configuration.
13339A second line is recommended to set a larger timeout, always appreciated when
13340issuing commands by hand :
13341
13342 global
13343 stats socket /var/run/haproxy.sock mode 600 level admin
13344 stats timeout 2m
13345
13346It is also possible to add multiple instances of the stats socket by repeating
13347the line, and make them listen to a TCP port instead of a UNIX socket. This is
13348never done by default because this is dangerous, but can be handy in some
13349situations :
13350
13351 global
13352 stats socket /var/run/haproxy.sock mode 600 level admin
13353 stats socket ipv4@192.168.0.1:9999 level admin
13354 stats timeout 2m
13355
13356To access the socket, an external utility such as "socat" is required. Socat is a
13357swiss-army knife to connect anything to anything. We use it to connect terminals
13358to the socket, or a couple of stdin/stdout pipes to it for scripts. The two main
13359syntaxes we'll use are the following :
13360
13361 # socat /var/run/haproxy.sock stdio
13362 # socat /var/run/haproxy.sock readline
13363
13364The first one is used with scripts. It is possible to send the output of a
13365script to haproxy, and pass haproxy's output to another script. That's useful
13366for retrieving counters or attack traces for example.
13367
13368The second one is only useful for issuing commands by hand. It has the benefit
13369that the terminal is handled by the readline library which supports line
13370editing and history, which is very convenient when issuing repeated commands
13371(eg: watch a counter).
13372
13373The socket supports two operation modes :
13374 - interactive
13375 - non-interactive
13376
13377The non-interactive mode is the default when socat connects to the socket. In
13378this mode, a single line may be sent. It is processed as a whole, responses are
13379sent back, and the connection closes after the end of the response. This is the
13380mode that scripts and monitoring tools use. It is possible to send multiple
13381commands in this mode, they need to be delimited by a semi-colon (';'). For
13382example :
13383
13384 # echo "show info;show stat;show table" | socat /var/run/haproxy stdio
13385
13386The interactive mode displays a prompt ('>') and waits for commands to be
13387entered on the line, then processes them, and displays the prompt again to wait
13388for a new command. This mode is entered via the "prompt" command which must be
13389sent on the first line in non-interactive mode. The mode is a flip switch, if
13390"prompt" is sent in interactive mode, it is disabled and the connection closes
13391after processing the last command of the same line.
13392
13393For this reason, when debugging by hand, it's quite common to start with the
13394"prompt" command :
13395
13396 # socat /var/run/haproxy readline
13397 prompt
13398 > show info
13399 ...
13400 >
13401
13402Since multiple commands may be issued at once, haproxy uses the empty line as a
13403delimiter to mark an end of output for each command, and takes care of ensuring
13404that no command can emit an empty line on output. A script can thus easily
13405parse the output even when multiple commands were pipelined on a single line.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013406
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013407It is important to understand that when multiple haproxy processes are started
13408on the same sockets, any process may pick up the request and will output its
13409own stats.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013410
Willy Tarreau468f4932013-08-01 16:50:16 +020013411The list of commands currently supported on the stats socket is provided below.
13412If an unknown command is sent, haproxy displays the usage message which reminds
13413all supported commands. Some commands support a more complex syntax, generally
13414it will explain what part of the command is invalid when this happens.
13415
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013416add acl <acl> <pattern>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013417 Add an entry into the acl <acl>. <acl> is the #<id> or the <file> returned by
13418 "show acl". This command does not verify if the entry already exists. This
13419 command cannot be used if the reference <acl> is a file also used with a map.
13420 In this case, you must use the command "add map" in place of "add acl".
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013421
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013422add map <map> <key> <value>
13423 Add an entry into the map <map> to associate the value <value> to the key
13424 <key>. This command does not verify if the entry already exists. It is
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013425 mainly used to fill a map after a clear operation. Note that if the reference
13426 <map> is a file and is shared with a map, this map will contain also a new
13427 pattern entry.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013428
Willy Tarreaud63335a2010-02-26 12:56:52 +010013429clear counters
13430 Clear the max values of the statistics counters in each proxy (frontend &
13431 backend) and in each server. The cumulated counters are not affected. This
13432 can be used to get clean counters after an incident, without having to
13433 restart nor to clear traffic counters. This command is restricted and can
13434 only be issued on sockets configured for levels "operator" or "admin".
13435
13436clear counters all
13437 Clear all statistics counters in each proxy (frontend & backend) and in each
13438 server. This has the same effect as restarting. This command is restricted
13439 and can only be issued on sockets configured for level "admin".
13440
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013441clear acl <acl>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013442 Remove all entries from the acl <acl>. <acl> is the #<id> or the <file>
13443 returned by "show acl". Note that if the reference <acl> is a file and is
13444 shared with a map, this map will be also cleared.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013445
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013446clear map <map>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013447 Remove all entries from the map <map>. <map> is the #<id> or the <file>
13448 returned by "show map". Note that if the reference <map> is a file and is
13449 shared with a acl, this acl will be also cleared.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013450
Simon Hormanc88b8872011-06-15 15:18:49 +090013451clear table <table> [ data.<type> <operator> <value> ] | [ key <key> ]
13452 Remove entries from the stick-table <table>.
13453
13454 This is typically used to unblock some users complaining they have been
13455 abusively denied access to a service, but this can also be used to clear some
13456 stickiness entries matching a server that is going to be replaced (see "show
13457 table" below for details). Note that sometimes, removal of an entry will be
13458 refused because it is currently tracked by a session. Retrying a few seconds
13459 later after the session ends is usual enough.
13460
13461 In the case where no options arguments are given all entries will be removed.
13462
13463 When the "data." form is used entries matching a filter applied using the
13464 stored data (see "stick-table" in section 4.2) are removed. A stored data
13465 type must be specified in <type>, and this data type must be stored in the
13466 table otherwise an error is reported. The data is compared according to
13467 <operator> with the 64-bit integer <value>. Operators are the same as with
13468 the ACLs :
13469
13470 - eq : match entries whose data is equal to this value
13471 - ne : match entries whose data is not equal to this value
13472 - le : match entries whose data is less than or equal to this value
13473 - ge : match entries whose data is greater than or equal to this value
13474 - lt : match entries whose data is less than this value
13475 - gt : match entries whose data is greater than this value
13476
13477 When the key form is used the entry <key> is removed. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013478 same type as the table, which currently is limited to IPv4, IPv6, integer and
13479 string.
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013480
13481 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013482 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013483 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013484 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13485 bytes_out_rate(60000)=187
13486 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13487 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013488
13489 $ echo "clear table http_proxy key 127.0.0.1" | socat stdio /tmp/sock1
13490
13491 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013492 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau62a36c42010-08-17 15:53:10 +020013493 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13494 bytes_out_rate(60000)=191
Simon Hormanc88b8872011-06-15 15:18:49 +090013495 $ echo "clear table http_proxy data.gpc0 eq 1" | socat stdio /tmp/sock1
13496 $ echo "show table http_proxy" | socat stdio /tmp/sock1
13497 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013498
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013499del acl <acl> [<key>|#<ref>]
13500 Delete all the acl entries from the acl <acl> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013501 <acl> is the #<id> or the <file> returned by "show acl". If the <ref> is used,
13502 this command delete only the listed reference. The reference can be found with
13503 listing the content of the acl. Note that if the reference <acl> is a file and
13504 is shared with a map, the entry will be also deleted in the map.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013505
13506del map <map> [<key>|#<ref>]
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013507 Delete all the map entries from the map <map> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013508 <map> is the #<id> or the <file> returned by "show map". If the <ref> is used,
13509 this command delete only the listed reference. The reference can be found with
13510 listing the content of the map. Note that if the reference <map> is a file and
13511 is shared with a acl, the entry will be also deleted in the map.
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013512
13513disable agent <backend>/<server>
Simon Horman671b6f02013-11-25 10:46:39 +090013514 Mark the auxiliary agent check as temporarily stopped.
13515
13516 In the case where an agent check is being run as a auxiliary check, due
13517 to the agent-check parameter of a server directive, new checks are only
13518 initialised when the agent is in the enabled. Thus, disable agent will
13519 prevent any new agent checks from begin initiated until the agent
13520 re-enabled using enable agent.
13521
13522 When an agent is disabled the processing of an auxiliary agent check that
13523 was initiated while the agent was set as enabled is as follows: All
13524 results that would alter the weight, specifically "drain" or a weight
13525 returned by the agent, are ignored. The processing of agent check is
13526 otherwise unchanged.
13527
13528 The motivation for this feature is to allow the weight changing effects
13529 of the agent checks to be paused to allow the weight of a server to be
13530 configured using set weight without being overridden by the agent.
13531
13532 This command is restricted and can only be issued on sockets configured for
13533 level "admin".
13534
Willy Tarreau532a4502011-09-07 22:37:44 +020013535disable frontend <frontend>
13536 Mark the frontend as temporarily stopped. This corresponds to the mode which
13537 is used during a soft restart : the frontend releases the port but can be
13538 enabled again if needed. This should be used with care as some non-Linux OSes
13539 are unable to enable it back. This is intended to be used in environments
13540 where stopping a proxy is not even imaginable but a misconfigured proxy must
13541 be fixed. That way it's possible to release the port and bind it into another
13542 process to restore operations. The frontend will appear with status "STOP"
13543 on the stats page.
13544
13545 The frontend may be specified either by its name or by its numeric ID,
13546 prefixed with a sharp ('#').
13547
13548 This command is restricted and can only be issued on sockets configured for
13549 level "admin".
13550
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013551disable health <backend>/<server>
13552 Mark the primary health check as temporarily stopped. This will disable
13553 sending of health checks, and the last health check result will be ignored.
13554 The server will be in unchecked state and considered UP unless an auxiliary
13555 agent check forces it down.
13556
13557 This command is restricted and can only be issued on sockets configured for
13558 level "admin".
13559
Willy Tarreaud63335a2010-02-26 12:56:52 +010013560disable server <backend>/<server>
13561 Mark the server DOWN for maintenance. In this mode, no more checks will be
13562 performed on the server until it leaves maintenance.
13563 If the server is tracked by other servers, those servers will be set to DOWN
13564 during the maintenance.
13565
13566 In the statistics page, a server DOWN for maintenance will appear with a
13567 "MAINT" status, its tracking servers with the "MAINT(via)" one.
13568
13569 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013570 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013571
13572 This command is restricted and can only be issued on sockets configured for
13573 level "admin".
13574
Simon Horman671b6f02013-11-25 10:46:39 +090013575enable agent <backend>/<server>
13576 Resume auxiliary agent check that was temporarily stopped.
13577
13578 See "disable agent" for details of the effect of temporarily starting
13579 and stopping an auxiliary agent.
13580
13581 This command is restricted and can only be issued on sockets configured for
13582 level "admin".
13583
Willy Tarreau532a4502011-09-07 22:37:44 +020013584enable frontend <frontend>
13585 Resume a frontend which was temporarily stopped. It is possible that some of
13586 the listening ports won't be able to bind anymore (eg: if another process
13587 took them since the 'disable frontend' operation). If this happens, an error
13588 is displayed. Some operating systems might not be able to resume a frontend
13589 which was disabled.
13590
13591 The frontend may be specified either by its name or by its numeric ID,
13592 prefixed with a sharp ('#').
13593
13594 This command is restricted and can only be issued on sockets configured for
13595 level "admin".
13596
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013597enable health <backend>/<server>
13598 Resume a primary health check that was temporarily stopped. This will enable
13599 sending of health checks again. Please see "disable health" for details.
13600
13601 This command is restricted and can only be issued on sockets configured for
13602 level "admin".
13603
Willy Tarreaud63335a2010-02-26 12:56:52 +010013604enable server <backend>/<server>
13605 If the server was previously marked as DOWN for maintenance, this marks the
13606 server UP and checks are re-enabled.
13607
13608 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013609 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013610
13611 This command is restricted and can only be issued on sockets configured for
13612 level "admin".
13613
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013614get map <map> <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013615get acl <acl> <value>
13616 Lookup the value <value> in the map <map> or in the ACL <acl>. <map> or <acl>
13617 are the #<id> or the <file> returned by "show map" or "show acl". This command
13618 returns all the matching patterns associated with this map. This is useful for
13619 debugging maps and ACLs. The output format is composed by one line par
13620 matching type. Each line is composed by space-delimited series of words.
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013621
13622 The first two words are:
13623
13624 <match method>: The match method applied. It can be "found", "bool",
13625 "int", "ip", "bin", "len", "str", "beg", "sub", "dir",
13626 "dom", "end" or "reg".
13627
13628 <match result>: The result. Can be "match" or "no-match".
13629
13630 The following words are returned only if the pattern matches an entry.
13631
13632 <index type>: "tree" or "list". The internal lookup algorithm.
13633
13634 <case>: "case-insensitive" or "case-sensitive". The
13635 interpretation of the case.
13636
13637 <entry matched>: match="<entry>". Return the matched pattern. It is
13638 useful with regular expressions.
13639
13640 The two last word are used to show the returned value and its type. With the
13641 "acl" case, the pattern doesn't exist.
13642
13643 return=nothing: No return because there are no "map".
13644 return="<value>": The value returned in the string format.
13645 return=cannot-display: The value cannot be converted as string.
13646
13647 type="<type>": The type of the returned sample.
13648
Willy Tarreaud63335a2010-02-26 12:56:52 +010013649get weight <backend>/<server>
13650 Report the current weight and the initial weight of server <server> in
13651 backend <backend> or an error if either doesn't exist. The initial weight is
13652 the one that appears in the configuration file. Both are normally equal
13653 unless the current weight has been changed. Both the backend and the server
13654 may be specified either by their name or by their numeric ID, prefixed with a
Willy Tarreauf5f31922011-08-02 11:32:07 +020013655 sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013656
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013657help
13658 Print the list of known keywords and their basic usage. The same help screen
13659 is also displayed for unknown commands.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013660
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013661prompt
13662 Toggle the prompt at the beginning of the line and enter or leave interactive
13663 mode. In interactive mode, the connection is not closed after a command
13664 completes. Instead, the prompt will appear again, indicating the user that
13665 the interpreter is waiting for a new command. The prompt consists in a right
13666 angle bracket followed by a space "> ". This mode is particularly convenient
13667 when one wants to periodically check information such as stats or errors.
13668 It is also a good idea to enter interactive mode before issuing a "help"
13669 command.
13670
13671quit
13672 Close the connection when in interactive mode.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013673
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013674set map <map> [<key>|#<ref>] <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013675 Modify the value corresponding to each key <key> in a map <map>. <map> is the
13676 #<id> or <file> returned by "show map". If the <ref> is used in place of
13677 <key>, only the entry pointed by <ref> is changed. The new value is <value>.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013678
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013679set maxconn frontend <frontend> <value>
Willy Tarreau3c7a79d2012-09-26 21:07:15 +020013680 Dynamically change the specified frontend's maxconn setting. Any positive
13681 value is allowed including zero, but setting values larger than the global
13682 maxconn does not make much sense. If the limit is increased and connections
13683 were pending, they will immediately be accepted. If it is lowered to a value
13684 below the current number of connections, new connections acceptation will be
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013685 delayed until the threshold is reached. The frontend might be specified by
13686 either its name or its numeric ID prefixed with a sharp ('#').
13687
Willy Tarreau91886b62011-09-07 14:38:31 +020013688set maxconn global <maxconn>
13689 Dynamically change the global maxconn setting within the range defined by the
13690 initial global maxconn setting. If it is increased and connections were
13691 pending, they will immediately be accepted. If it is lowered to a value below
13692 the current number of connections, new connections acceptation will be
13693 delayed until the threshold is reached. A value of zero restores the initial
13694 setting.
13695
Willy Tarreauf5b22872011-09-07 16:13:44 +020013696set rate-limit connections global <value>
13697 Change the process-wide connection rate limit, which is set by the global
13698 'maxconnrate' setting. A value of zero disables the limitation. This limit
13699 applies to all frontends and the change has an immediate effect. The value
13700 is passed in number of connections per second.
13701
William Lallemandd85f9172012-11-09 17:05:39 +010013702set rate-limit http-compression global <value>
13703 Change the maximum input compression rate, which is set by the global
13704 'maxcomprate' setting. A value of zero disables the limitation. The value is
William Lallemand096f5542012-11-19 17:26:05 +010013705 passed in number of kilobytes per second. The value is available in the "show
13706 info" on the line "CompressBpsRateLim" in bytes.
William Lallemandd85f9172012-11-09 17:05:39 +010013707
Willy Tarreau93e7c002013-10-07 18:51:07 +020013708set rate-limit sessions global <value>
13709 Change the process-wide session rate limit, which is set by the global
13710 'maxsessrate' setting. A value of zero disables the limitation. This limit
13711 applies to all frontends and the change has an immediate effect. The value
13712 is passed in number of sessions per second.
13713
Willy Tarreaue43d5322013-10-07 20:01:52 +020013714set rate-limit ssl-sessions global <value>
13715 Change the process-wide SSL session rate limit, which is set by the global
13716 'maxsslrate' setting. A value of zero disables the limitation. This limit
13717 applies to all frontends and the change has an immediate effect. The value
13718 is passed in number of sessions per second sent to the SSL stack. It applies
13719 before the handshake in order to protect the stack against handshake abuses.
13720
Willy Tarreau2a4b70f2014-05-22 18:42:35 +020013721set server <backend>/<server> agent [ up | down ]
13722 Force a server's agent to a new state. This can be useful to immediately
13723 switch a server's state regardless of some slow agent checks for example.
13724 Note that the change is propagated to tracking servers if any.
13725
13726set server <backend>/<server> health [ up | stopping | down ]
13727 Force a server's health to a new state. This can be useful to immediately
13728 switch a server's state regardless of some slow health checks for example.
13729 Note that the change is propagated to tracking servers if any.
13730
13731set server <backend>/<server> state [ ready | drain | maint ]
13732 Force a server's administrative state to a new state. This can be useful to
13733 disable load balancing and/or any traffic to a server. Setting the state to
13734 "ready" puts the server in normal mode, and the command is the equivalent of
13735 the "enable server" command. Setting the state to "maint" disables any traffic
13736 to the server as well as any health checks. This is the equivalent of the
13737 "disable server" command. Setting the mode to "drain" only removes the server
13738 from load balancing but still allows it to be checked and to accept new
13739 persistent connections. Changes are propagated to tracking servers if any.
13740
13741set server <backend>/<server> weight <weight>[%]
13742 Change a server's weight to the value passed in argument. This is the exact
13743 equivalent of the "set weight" command below.
13744
Emeric Brun4147b2e2014-06-16 18:36:30 +020013745set ssl ocsp-response <response>
13746 This command is used to update an OCSP Response for a certificate (see "crt"
13747 on "bind" lines). Same controls are performed as during the initial loading of
13748 the response. The <response> must be passed as a base64 encoded string of the
13749 DER encoded response from the OCSP server.
13750
13751 Example:
13752 openssl ocsp -issuer issuer.pem -cert server.pem \
13753 -host ocsp.issuer.com:80 -respout resp.der
13754 echo "set ssl ocsp-response $(base64 -w 10000 resp.der)" | \
13755 socat stdio /var/run/haproxy.stat
13756
Willy Tarreau47060b62013-08-01 21:11:42 +020013757set table <table> key <key> [data.<data_type> <value>]*
Willy Tarreau654694e2012-06-07 01:03:16 +020013758 Create or update a stick-table entry in the table. If the key is not present,
13759 an entry is inserted. See stick-table in section 4.2 to find all possible
13760 values for <data_type>. The most likely use consists in dynamically entering
13761 entries for source IP addresses, with a flag in gpc0 to dynamically block an
Willy Tarreau47060b62013-08-01 21:11:42 +020013762 IP address or affect its quality of service. It is possible to pass multiple
13763 data_types in a single call.
Willy Tarreau654694e2012-06-07 01:03:16 +020013764
Willy Tarreaud63335a2010-02-26 12:56:52 +010013765set timeout cli <delay>
13766 Change the CLI interface timeout for current connection. This can be useful
13767 during long debugging sessions where the user needs to constantly inspect
13768 some indicators without being disconnected. The delay is passed in seconds.
13769
13770set weight <backend>/<server> <weight>[%]
13771 Change a server's weight to the value passed in argument. If the value ends
13772 with the '%' sign, then the new weight will be relative to the initially
Simon Horman58b5d292013-02-12 10:45:52 +090013773 configured weight. Absolute weights are permitted between 0 and 256.
13774 Relative weights must be positive with the resulting absolute weight is
13775 capped at 256. Servers which are part of a farm running a static
13776 load-balancing algorithm have stricter limitations because the weight
13777 cannot change once set. Thus for these servers, the only accepted values
13778 are 0 and 100% (or 0 and the initial weight). Changes take effect
13779 immediately, though certain LB algorithms require a certain amount of
13780 requests to consider changes. A typical usage of this command is to
13781 disable a server during an update by setting its weight to zero, then to
13782 enable it again after the update by setting it back to 100%. This command
13783 is restricted and can only be issued on sockets configured for level
13784 "admin". Both the backend and the server may be specified either by their
13785 name or by their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013786
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013787show errors [<iid>]
13788 Dump last known request and response errors collected by frontends and
13789 backends. If <iid> is specified, the limit the dump to errors concerning
Willy Tarreau6162db22009-10-10 17:13:00 +020013790 either frontend or backend whose ID is <iid>. This command is restricted
13791 and can only be issued on sockets configured for levels "operator" or
13792 "admin".
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013793
13794 The errors which may be collected are the last request and response errors
13795 caused by protocol violations, often due to invalid characters in header
13796 names. The report precisely indicates what exact character violated the
13797 protocol. Other important information such as the exact date the error was
13798 detected, frontend and backend names, the server name (when known), the
13799 internal session ID and the source address which has initiated the session
13800 are reported too.
13801
13802 All characters are returned, and non-printable characters are encoded. The
13803 most common ones (\t = 9, \n = 10, \r = 13 and \e = 27) are encoded as one
13804 letter following a backslash. The backslash itself is encoded as '\\' to
13805 avoid confusion. Other non-printable characters are encoded '\xNN' where
13806 NN is the two-digits hexadecimal representation of the character's ASCII
13807 code.
13808
13809 Lines are prefixed with the position of their first character, starting at 0
13810 for the beginning of the buffer. At most one input line is printed per line,
13811 and large lines will be broken into multiple consecutive output lines so that
13812 the output never goes beyond 79 characters wide. It is easy to detect if a
13813 line was broken, because it will not end with '\n' and the next line's offset
13814 will be followed by a '+' sign, indicating it is a continuation of previous
13815 line.
13816
13817 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013818 $ echo "show errors" | socat stdio /tmp/sock1
13819 >>> [04/Mar/2009:15:46:56.081] backend http-in (#2) : invalid response
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013820 src 127.0.0.1, session #54, frontend fe-eth0 (#1), server s2 (#1)
13821 response length 213 bytes, error at position 23:
13822
13823 00000 HTTP/1.0 200 OK\r\n
13824 00017 header/bizarre:blah\r\n
13825 00038 Location: blah\r\n
13826 00054 Long-line: this is a very long line which should b
13827 00104+ e broken into multiple lines on the output buffer,
13828 00154+ otherwise it would be too large to print in a ter
13829 00204+ minal\r\n
13830 00211 \r\n
13831
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013832 In the example above, we see that the backend "http-in" which has internal
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013833 ID 2 has blocked an invalid response from its server s2 which has internal
13834 ID 1. The request was on session 54 initiated by source 127.0.0.1 and
13835 received by frontend fe-eth0 whose ID is 1. The total response length was
13836 213 bytes when the error was detected, and the error was at byte 23. This
13837 is the slash ('/') in header name "header/bizarre", which is not a valid
13838 HTTP character for a header name.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013839
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013840show info
13841 Dump info about haproxy status on current process.
13842
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013843show map [<map>]
13844 Dump info about map converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013845 maps is returned. If a <map> is specified, its contents are dumped. <map> is
13846 the #<id> or <file>. The first column is a unique identifier. It can be used
13847 as reference for the operation "del map" and "set map". The second column is
13848 the pattern and the third column is the sample if available. The data returned
13849 are not directly a list of available maps, but are the list of all patterns
13850 composing any map. Many of these patterns can be shared with ACL.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013851
13852show acl [<acl>]
13853 Dump info about acl converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013854 acls is returned. If a <acl> is specified, its contents are dumped. <acl> if
13855 the #<id> or <file>. The dump format is the same than the map even for the
13856 sample value. The data returned are not a list of available ACL, but are the
13857 list of all patterns composing any ACL. Many of these patterns can be shared
13858 with maps.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013859
Willy Tarreau12833bb2014-01-28 16:49:56 +010013860show pools
13861 Dump the status of internal memory pools. This is useful to track memory
13862 usage when suspecting a memory leak for example. It does exactly the same
13863 as the SIGQUIT when running in foreground except that it does not flush
13864 the pools.
13865
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013866show sess
13867 Dump all known sessions. Avoid doing this on slow connections as this can
Willy Tarreau6162db22009-10-10 17:13:00 +020013868 be huge. This command is restricted and can only be issued on sockets
13869 configured for levels "operator" or "admin".
13870
Willy Tarreau66dc20a2010-03-05 17:53:32 +010013871show sess <id>
13872 Display a lot of internal information about the specified session identifier.
13873 This identifier is the first field at the beginning of the lines in the dumps
13874 of "show sess" (it corresponds to the session pointer). Those information are
13875 useless to most users but may be used by haproxy developers to troubleshoot a
13876 complex bug. The output format is intentionally not documented so that it can
Oliviere99d44d2014-09-05 18:49:10 +020013877 freely evolve depending on demands. You may find a description of all fields
13878 returned in src/dumpstats.c
13879
13880 The special id "all" dumps the states of all sessions, which must be avoided
13881 as much as possible as it is highly CPU intensive and can take a lot of time.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013882
13883show stat [<iid> <type> <sid>]
13884 Dump statistics in the CSV format. By passing <id>, <type> and <sid>, it is
13885 possible to dump only selected items :
13886 - <iid> is a proxy ID, -1 to dump everything
13887 - <type> selects the type of dumpable objects : 1 for frontends, 2 for
13888 backends, 4 for servers, -1 for everything. These values can be ORed,
13889 for example:
13890 1 + 2 = 3 -> frontend + backend.
13891 1 + 2 + 4 = 7 -> frontend + backend + server.
13892 - <sid> is a server ID, -1 to dump everything from the selected proxy.
13893
13894 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013895 $ echo "show info;show stat" | socat stdio unix-connect:/tmp/sock1
13896 >>> Name: HAProxy
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013897 Version: 1.4-dev2-49
13898 Release_date: 2009/09/23
13899 Nbproc: 1
13900 Process_num: 1
13901 (...)
13902
13903 # pxname,svname,qcur,qmax,scur,smax,slim,stot,bin,bout,dreq, (...)
13904 stats,FRONTEND,,,0,0,1000,0,0,0,0,0,0,,,,,OPEN,,,,,,,,,1,1,0, (...)
13905 stats,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,0,0,0,,0,250,(...)
13906 (...)
13907 www1,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,1,1,0,,0,250, (...)
13908
13909 $
13910
13911 Here, two commands have been issued at once. That way it's easy to find
13912 which process the stats apply to in multi-process mode. Notice the empty
13913 line after the information output which marks the end of the first block.
13914 A similar empty line appears at the end of the second block (stats) so that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013915 the reader knows the output has not been truncated.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013916
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013917show table
13918 Dump general information on all known stick-tables. Their name is returned
13919 (the name of the proxy which holds them), their type (currently zero, always
13920 IP), their size in maximum possible number of entries, and the number of
13921 entries currently in use.
13922
13923 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013924 $ echo "show table" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013925 >>> # table: front_pub, type: ip, size:204800, used:171454
13926 >>> # table: back_rdp, type: ip, size:204800, used:0
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013927
Simon Horman17bce342011-06-15 15:18:47 +090013928show table <name> [ data.<type> <operator> <value> ] | [ key <key> ]
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013929 Dump contents of stick-table <name>. In this mode, a first line of generic
13930 information about the table is reported as with "show table", then all
13931 entries are dumped. Since this can be quite heavy, it is possible to specify
Simon Horman17bce342011-06-15 15:18:47 +090013932 a filter in order to specify what entries to display.
13933
13934 When the "data." form is used the filter applies to the stored data (see
13935 "stick-table" in section 4.2). A stored data type must be specified
13936 in <type>, and this data type must be stored in the table otherwise an
13937 error is reported. The data is compared according to <operator> with the
13938 64-bit integer <value>. Operators are the same as with the ACLs :
13939
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013940 - eq : match entries whose data is equal to this value
13941 - ne : match entries whose data is not equal to this value
13942 - le : match entries whose data is less than or equal to this value
13943 - ge : match entries whose data is greater than or equal to this value
13944 - lt : match entries whose data is less than this value
13945 - gt : match entries whose data is greater than this value
13946
Simon Hormanc88b8872011-06-15 15:18:49 +090013947
13948 When the key form is used the entry <key> is shown. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013949 same type as the table, which currently is limited to IPv4, IPv6, integer,
13950 and string.
Simon Horman17bce342011-06-15 15:18:47 +090013951
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013952 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013953 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013954 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013955 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13956 bytes_out_rate(60000)=187
13957 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13958 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013959
Willy Tarreau62a36c42010-08-17 15:53:10 +020013960 $ echo "show table http_proxy data.gpc0 gt 0" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013961 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013962 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13963 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013964
Willy Tarreau62a36c42010-08-17 15:53:10 +020013965 $ echo "show table http_proxy data.conn_rate gt 5" | \
13966 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013967 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013968 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13969 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013970
Simon Horman17bce342011-06-15 15:18:47 +090013971 $ echo "show table http_proxy key 127.0.0.2" | \
13972 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013973 >>> # table: http_proxy, type: ip, size:204800, used:2
Simon Horman17bce342011-06-15 15:18:47 +090013974 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13975 bytes_out_rate(60000)=191
13976
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013977 When the data criterion applies to a dynamic value dependent on time such as
13978 a bytes rate, the value is dynamically computed during the evaluation of the
13979 entry in order to decide whether it has to be dumped or not. This means that
13980 such a filter could match for some time then not match anymore because as
13981 time goes, the average event rate drops.
13982
13983 It is possible to use this to extract lists of IP addresses abusing the
13984 service, in order to monitor them or even blacklist them in a firewall.
13985 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013986 $ echo "show table http_proxy data.gpc0 gt 0" \
13987 | socat stdio /tmp/sock1 \
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013988 | fgrep 'key=' | cut -d' ' -f2 | cut -d= -f2 > abusers-ip.txt
13989 ( or | awk '/key/{ print a[split($2,a,"=")]; }' )
Krzysztof Piotr Oledzki719e7262009-10-04 15:02:46 +020013990
Willy Tarreau532a4502011-09-07 22:37:44 +020013991shutdown frontend <frontend>
13992 Completely delete the specified frontend. All the ports it was bound to will
13993 be released. It will not be possible to enable the frontend anymore after
13994 this operation. This is intended to be used in environments where stopping a
13995 proxy is not even imaginable but a misconfigured proxy must be fixed. That
13996 way it's possible to release the port and bind it into another process to
13997 restore operations. The frontend will not appear at all on the stats page
13998 once it is terminated.
13999
14000 The frontend may be specified either by its name or by its numeric ID,
14001 prefixed with a sharp ('#').
14002
14003 This command is restricted and can only be issued on sockets configured for
14004 level "admin".
14005
Willy Tarreaua295edc2011-09-07 23:21:03 +020014006shutdown session <id>
14007 Immediately terminate the session matching the specified session identifier.
14008 This identifier is the first field at the beginning of the lines in the dumps
14009 of "show sess" (it corresponds to the session pointer). This can be used to
14010 terminate a long-running session without waiting for a timeout or when an
14011 endless transfer is ongoing. Such terminated sessions are reported with a 'K'
14012 flag in the logs.
14013
Cyril Bontéde9789b2014-07-12 18:22:42 +020014014shutdown sessions server <backend>/<server>
Willy Tarreau52b2d222011-09-07 23:48:48 +020014015 Immediately terminate all the sessions attached to the specified server. This
14016 can be used to terminate long-running sessions after a server is put into
14017 maintenance mode, for instance. Such terminated sessions are reported with a
14018 'K' flag in the logs.
14019
Willy Tarreau0ba27502007-12-24 16:55:16 +010014020/*
14021 * Local variables:
14022 * fill-column: 79
14023 * End:
14024 */