blob: 1d95b5bdfdfaa45e9fb0e3e3c8705413a61e7b05 [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 Tarreau16f863f2015-07-03 17:35:11 +02005 version 1.5.14
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau16f863f2015-07-03 17:35:11 +02007 2015/07/02
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
Rauf Kuliyev38b41562011-01-04 15:14:13 +01001370option pgsql-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001371option nolinger (*) X X X X
1372option originalto X X X X
1373option persist (*) X - X X
1374option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02001375option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001376option smtpchk X - X X
1377option socket-stats (*) X X X -
1378option splice-auto (*) X X X X
1379option splice-request (*) X X X X
1380option splice-response (*) X X X X
1381option srvtcpka (*) X - X X
1382option ssl-hello-chk X - X X
1383-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01001384option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001385option tcp-smart-accept (*) X X X -
1386option tcp-smart-connect (*) X - X X
1387option tcpka X X X X
1388option tcplog X X X X
1389option transparent (*) X - X X
1390persist rdp-cookie X - X X
1391rate-limit sessions X X X -
1392redirect - X X X
1393redisp (deprecated) X - X X
1394redispatch (deprecated) X - X X
1395reqadd - X X X
1396reqallow - X X X
1397reqdel - X X X
1398reqdeny - X X X
1399reqiallow - X X X
1400reqidel - X X X
1401reqideny - X X X
1402reqipass - X X X
1403reqirep - X X X
1404reqisetbe - X X X
1405reqitarpit - X X X
1406reqpass - X X X
1407reqrep - X X X
1408-- keyword -------------------------- defaults - frontend - listen -- backend -
1409reqsetbe - X X X
1410reqtarpit - X X X
1411retries X - X X
1412rspadd - X X X
1413rspdel - X X X
1414rspdeny - X X X
1415rspidel - X X X
1416rspideny - X X X
1417rspirep - X X X
1418rsprep - X X X
1419server - - X X
1420source X - X X
1421srvtimeout (deprecated) X - X X
Cyril Bonté66c327d2010-10-12 00:14:37 +02001422stats admin - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001423stats auth X - X X
1424stats enable X - X X
1425stats hide-version X - X X
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02001426stats http-request - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001427stats realm X - X X
1428stats refresh X - X X
1429stats scope X - X X
1430stats show-desc X - X X
1431stats show-legends X - X X
1432stats show-node X - X X
1433stats uri X - X X
1434-- keyword -------------------------- defaults - frontend - listen -- backend -
1435stick match - - X X
1436stick on - - X X
1437stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02001438stick store-response - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001439stick-table - - X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02001440tcp-check connect - - X X
1441tcp-check expect - - X X
1442tcp-check send - - X X
1443tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02001444tcp-request connection - X X -
1445tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02001446tcp-request inspect-delay - X X X
Emeric Brun0a3b67f2010-09-24 15:34:53 +02001447tcp-response content - - X X
1448tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001449timeout check X - X X
1450timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02001451timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001452timeout clitimeout (deprecated) X X X -
1453timeout connect X - X X
1454timeout contimeout (deprecated) X - X X
1455timeout http-keep-alive X X X X
1456timeout http-request X X X X
1457timeout queue X - X X
1458timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02001459timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001460timeout srvtimeout (deprecated) X - X X
1461timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02001462timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001463transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01001464unique-id-format X X X -
1465unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001466use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02001467use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001468------------------------------------+----------+----------+---------+---------
1469 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02001470
Willy Tarreau0ba27502007-12-24 16:55:16 +01001471
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014724.2. Alphabetically sorted keywords reference
1473---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001474
1475This section provides a description of each keyword and its usage.
1476
1477
1478acl <aclname> <criterion> [flags] [operator] <value> ...
1479 Declare or complete an access list.
1480 May be used in sections : defaults | frontend | listen | backend
1481 no | yes | yes | yes
1482 Example:
1483 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1484 acl invalid_src src_port 0:1023
1485 acl local_dst hdr(host) -i localhost
1486
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001487 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001488
1489
Cyril Bontéb21570a2009-11-29 20:04:48 +01001490appsession <cookie> len <length> timeout <holdtime>
1491 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001492 Define session stickiness on an existing application cookie.
1493 May be used in sections : defaults | frontend | listen | backend
1494 no | no | yes | yes
1495 Arguments :
1496 <cookie> this is the name of the cookie used by the application and which
1497 HAProxy will have to learn for each new session.
1498
Cyril Bontéb21570a2009-11-29 20:04:48 +01001499 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01001500 checked in each cookie value.
1501
1502 <holdtime> this is the time after which the cookie will be removed from
1503 memory if unused. If no unit is specified, this time is in
1504 milliseconds.
1505
Cyril Bontébf47aeb2009-10-15 00:15:40 +02001506 request-learn
1507 If this option is specified, then haproxy will be able to learn
1508 the cookie found in the request in case the server does not
1509 specify any in response. This is typically what happens with
1510 PHPSESSID cookies, or when haproxy's session expires before
1511 the application's session and the correct server is selected.
1512 It is recommended to specify this option to improve reliability.
1513
Cyril Bontéb21570a2009-11-29 20:04:48 +01001514 prefix When this option is specified, haproxy will match on the cookie
1515 prefix (or URL parameter prefix). The appsession value is the
1516 data following this prefix.
1517
1518 Example :
1519 appsession ASPSESSIONID len 64 timeout 3h prefix
1520
1521 This will match the cookie ASPSESSIONIDXXXX=XXXXX,
1522 the appsession value will be XXXX=XXXXX.
1523
1524 mode This option allows to change the URL parser mode.
1525 2 modes are currently supported :
1526 - path-parameters :
1527 The parser looks for the appsession in the path parameters
1528 part (each parameter is separated by a semi-colon), which is
1529 convenient for JSESSIONID for example.
1530 This is the default mode if the option is not set.
1531 - query-string :
1532 In this mode, the parser will look for the appsession in the
1533 query string.
1534
Willy Tarreau0ba27502007-12-24 16:55:16 +01001535 When an application cookie is defined in a backend, HAProxy will check when
1536 the server sets such a cookie, and will store its value in a table, and
1537 associate it with the server's identifier. Up to <length> characters from
1538 the value will be retained. On each connection, haproxy will look for this
Cyril Bontéb21570a2009-11-29 20:04:48 +01001539 cookie both in the "Cookie:" headers, and as a URL parameter (depending on
1540 the mode used). If a known value is found, the client will be directed to the
1541 server associated with this value. Otherwise, the load balancing algorithm is
Willy Tarreau0ba27502007-12-24 16:55:16 +01001542 applied. Cookies are automatically removed from memory when they have been
1543 unused for a duration longer than <holdtime>.
1544
1545 The definition of an application cookie is limited to one per backend.
1546
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001547 Note : Consider not using this feature in multi-process mode (nbproc > 1)
1548 unless you know what you do : memory is not shared between the
1549 processes, which can result in random behaviours.
1550
Willy Tarreau0ba27502007-12-24 16:55:16 +01001551 Example :
1552 appsession JSESSIONID len 52 timeout 3h
1553
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001554 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
1555 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001556
1557
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001558backlog <conns>
1559 Give hints to the system about the approximate listen backlog desired size
1560 May be used in sections : defaults | frontend | listen | backend
1561 yes | yes | yes | no
1562 Arguments :
1563 <conns> is the number of pending connections. Depending on the operating
1564 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02001565 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001566
1567 In order to protect against SYN flood attacks, one solution is to increase
1568 the system's SYN backlog size. Depending on the system, sometimes it is just
1569 tunable via a system parameter, sometimes it is not adjustable at all, and
1570 sometimes the system relies on hints given by the application at the time of
1571 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
1572 to the listen() syscall. On systems which can make use of this value, it can
1573 sometimes be useful to be able to specify a different value, hence this
1574 backlog parameter.
1575
1576 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
1577 used as a hint and the system accepts up to the smallest greater power of
1578 two, and never more than some limits (usually 32768).
1579
1580 See also : "maxconn" and the target operating system's tuning guide.
1581
1582
Willy Tarreau0ba27502007-12-24 16:55:16 +01001583balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02001584balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001585 Define the load balancing algorithm to be used in a backend.
1586 May be used in sections : defaults | frontend | listen | backend
1587 yes | no | yes | yes
1588 Arguments :
1589 <algorithm> is the algorithm used to select a server when doing load
1590 balancing. This only applies when no persistence information
1591 is available, or when a connection is redispatched to another
1592 server. <algorithm> may be one of the following :
1593
1594 roundrobin Each server is used in turns, according to their weights.
1595 This is the smoothest and fairest algorithm when the server's
1596 processing time remains equally distributed. This algorithm
1597 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02001598 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08001599 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02001600 large farms, when a server becomes up after having been down
1601 for a very short time, it may sometimes take a few hundreds
1602 requests for it to be re-integrated into the farm and start
1603 receiving traffic. This is normal, though very rare. It is
1604 indicated here in case you would have the chance to observe
1605 it, so that you don't worry.
1606
1607 static-rr Each server is used in turns, according to their weights.
1608 This algorithm is as similar to roundrobin except that it is
1609 static, which means that changing a server's weight on the
1610 fly will have no effect. On the other hand, it has no design
1611 limitation on the number of servers, and when a server goes
1612 up, it is always immediately reintroduced into the farm, once
1613 the full map is recomputed. It also uses slightly less CPU to
1614 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01001615
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01001616 leastconn The server with the lowest number of connections receives the
1617 connection. Round-robin is performed within groups of servers
1618 of the same load to ensure that all servers will be used. Use
1619 of this algorithm is recommended where very long sessions are
1620 expected, such as LDAP, SQL, TSE, etc... but is not very well
1621 suited for protocols using short sessions such as HTTP. This
1622 algorithm is dynamic, which means that server weights may be
1623 adjusted on the fly for slow starts for instance.
1624
Willy Tarreauf09c6602012-02-13 17:12:08 +01001625 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001626 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01001627 identifier to the highest (see server parameter "id"), which
1628 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02001629 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01001630 not make sense to use this algorithm without setting maxconn.
1631 The purpose of this algorithm is to always use the smallest
1632 number of servers so that extra servers can be powered off
1633 during non-intensive hours. This algorithm ignores the server
1634 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02001635 or IMAP than HTTP, though it can be useful there too. In
1636 order to use this algorithm efficiently, it is recommended
1637 that a cloud controller regularly checks server usage to turn
1638 them off when unused, and regularly checks backend queue to
1639 turn new servers on when the queue inflates. Alternatively,
1640 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01001641
Willy Tarreau0ba27502007-12-24 16:55:16 +01001642 source The source IP address is hashed and divided by the total
1643 weight of the running servers to designate which server will
1644 receive the request. This ensures that the same client IP
1645 address will always reach the same server as long as no
1646 server goes down or up. If the hash result changes due to the
1647 number of running servers changing, many clients will be
1648 directed to a different server. This algorithm is generally
1649 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001650 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01001651 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001652 static by default, which means that changing a server's
1653 weight on the fly will have no effect, but this can be
1654 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001655
Oskar Stolc8dc41842012-05-19 10:19:54 +01001656 uri This algorithm hashes either the left part of the URI (before
1657 the question mark) or the whole URI (if the "whole" parameter
1658 is present) and divides the hash value by the total weight of
1659 the running servers. The result designates which server will
1660 receive the request. This ensures that the same URI will
1661 always be directed to the same server as long as no server
1662 goes up or down. This is used with proxy caches and
1663 anti-virus proxies in order to maximize the cache hit rate.
1664 Note that this algorithm may only be used in an HTTP backend.
1665 This algorithm is static by default, which means that
1666 changing a server's weight on the fly will have no effect,
1667 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001668
Oskar Stolc8dc41842012-05-19 10:19:54 +01001669 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001670 "depth", both followed by a positive integer number. These
1671 options may be helpful when it is needed to balance servers
1672 based on the beginning of the URI only. The "len" parameter
1673 indicates that the algorithm should only consider that many
1674 characters at the beginning of the URI to compute the hash.
1675 Note that having "len" set to 1 rarely makes sense since most
1676 URIs start with a leading "/".
1677
1678 The "depth" parameter indicates the maximum directory depth
1679 to be used to compute the hash. One level is counted for each
1680 slash in the request. If both parameters are specified, the
1681 evaluation stops when either is reached.
1682
Willy Tarreau0ba27502007-12-24 16:55:16 +01001683 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001684 the query string of each HTTP GET request.
1685
1686 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02001687 request entity will be searched for the parameter argument,
1688 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02001689 ('?') in the URL. The message body will only start to be
1690 analyzed once either the advertised amount of data has been
1691 received or the request buffer is full. In the unlikely event
1692 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02001693 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02001694 be randomly balanced if at all. This keyword used to support
1695 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001696
1697 If the parameter is found followed by an equal sign ('=') and
1698 a value, then the value is hashed and divided by the total
1699 weight of the running servers. The result designates which
1700 server will receive the request.
1701
1702 This is used to track user identifiers in requests and ensure
1703 that a same user ID will always be sent to the same server as
1704 long as no server goes up or down. If no value is found or if
1705 the parameter is not found, then a round robin algorithm is
1706 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001707 backend. This algorithm is static by default, which means
1708 that changing a server's weight on the fly will have no
1709 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001710
Cyril Bontédc4d9032012-04-08 21:57:39 +02001711 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
1712 request. Just as with the equivalent ACL 'hdr()' function,
1713 the header name in parenthesis is not case sensitive. If the
1714 header is absent or if it does not contain any value, the
1715 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01001716
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001717 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01001718 reducing the hash algorithm to the main domain part with some
1719 specific headers such as 'Host'. For instance, in the Host
1720 value "haproxy.1wt.eu", only "1wt" will be considered.
1721
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001722 This algorithm is static by default, which means that
1723 changing a server's weight on the fly will have no effect,
1724 but this can be changed using "hash-type".
1725
Emeric Brun736aa232009-06-30 17:56:00 +02001726 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02001727 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02001728 The RDP cookie <name> (or "mstshash" if omitted) will be
1729 looked up and hashed for each incoming TCP request. Just as
1730 with the equivalent ACL 'req_rdp_cookie()' function, the name
1731 is not case-sensitive. This mechanism is useful as a degraded
1732 persistence mode, as it makes it possible to always send the
1733 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001734 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02001735 used instead.
1736
1737 Note that for this to work, the frontend must ensure that an
1738 RDP cookie is already present in the request buffer. For this
1739 you must use 'tcp-request content accept' rule combined with
1740 a 'req_rdp_cookie_cnt' ACL.
1741
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001742 This algorithm is static by default, which means that
1743 changing a server's weight on the fly will have no effect,
1744 but this can be changed using "hash-type".
1745
Cyril Bontédc4d9032012-04-08 21:57:39 +02001746 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09001747
Willy Tarreau0ba27502007-12-24 16:55:16 +01001748 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001749 algorithms. Right now, only "url_param" and "uri" support an
1750 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001751
Willy Tarreau3cd9af22009-03-15 14:06:41 +01001752 The load balancing algorithm of a backend is set to roundrobin when no other
1753 algorithm, mode nor option have been set. The algorithm may only be set once
1754 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001755
1756 Examples :
1757 balance roundrobin
1758 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001759 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01001760 balance hdr(User-Agent)
1761 balance hdr(host)
1762 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001763
1764 Note: the following caveats and limitations on using the "check_post"
1765 extension with "url_param" must be considered :
1766
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001767 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001768 to determine if the parameters will be found in the body or entity which
1769 may contain binary data. Therefore another method may be required to
1770 restrict consideration of POST requests that have no URL parameters in
1771 the body. (see acl reqideny http_end)
1772
1773 - using a <max_wait> value larger than the request buffer size does not
1774 make sense and is useless. The buffer size is set at build time, and
1775 defaults to 16 kB.
1776
1777 - Content-Encoding is not supported, the parameter search will probably
1778 fail; and load balancing will fall back to Round Robin.
1779
1780 - Expect: 100-continue is not supported, load balancing will fall back to
1781 Round Robin.
1782
1783 - Transfer-Encoding (RFC2616 3.6.1) is only supported in the first chunk.
1784 If the entire parameter value is not present in the first chunk, the
1785 selection of server is undefined (actually, defined by how little
1786 actually appeared in the first chunk).
1787
1788 - This feature does not support generation of a 100, 411 or 501 response.
1789
1790 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001791 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001792 white space or control characters are found, indicating the end of what
1793 might be a URL parameter list. This is probably not a concern with SGML
1794 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001795
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001796 See also : "dispatch", "cookie", "appsession", "transparent", "hash-type" and
1797 "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001798
1799
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001800bind [<address>]:<port_range> [, ...] [param*]
1801bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001802 Define one or several listening addresses and/or ports in a frontend.
1803 May be used in sections : defaults | frontend | listen | backend
1804 no | yes | yes | no
1805 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001806 <address> is optional and can be a host name, an IPv4 address, an IPv6
1807 address, or '*'. It designates the address the frontend will
1808 listen on. If unset, all IPv4 addresses of the system will be
1809 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01001810 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01001811 Optionally, an address family prefix may be used before the
1812 address to force the family regardless of the address format,
1813 which can be useful to specify a path to a unix socket with
1814 no slash ('/'). Currently supported prefixes are :
1815 - 'ipv4@' -> address is always IPv4
1816 - 'ipv6@' -> address is always IPv6
1817 - 'unix@' -> address is a path to a local unix socket
Willy Tarreaub4fca5d2014-07-08 00:37:50 +02001818 - 'abns@' -> address is in abstract namespace (Linux only).
1819 Note: since abstract sockets are not "rebindable", they
1820 do not cope well with multi-process mode during
1821 soft-restart, so it is better to avoid them if
1822 nbproc is greater than 1. The effect is that if the
1823 new process fails to start, only one of the old ones
1824 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01001825 - 'fd@<n>' -> use file descriptor <n> inherited from the
1826 parent. The fd must be bound and may or may not already
1827 be listening.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001828 Any part of the address string may reference any number of
1829 environment variables by preceding their name with a dollar
1830 sign ('$') and optionally enclosing them with braces ('{}'),
1831 similarly to what is done in Bourne shell.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001832
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001833 <port_range> is either a unique TCP port, or a port range for which the
1834 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001835 above. The port is mandatory for TCP listeners. Note that in
1836 the case of an IPv6 address, the port is always the number
1837 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001838 - a numerical port (ex: '80')
1839 - a dash-delimited ports range explicitly stating the lower
1840 and upper bounds (ex: '2000-2100') which are included in
1841 the range.
1842
1843 Particular care must be taken against port ranges, because
1844 every <address:port> couple consumes one socket (= a file
1845 descriptor), so it's easy to consume lots of descriptors
1846 with a simple range, and to run out of sockets. Also, each
1847 <address:port> couple must be used only once among all
1848 instances running on a same system. Please note that binding
1849 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001850 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001851 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001852
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001853 <path> is a UNIX socket path beginning with a slash ('/'). This is
1854 alternative to the TCP listening port. Haproxy will then
1855 receive UNIX connections on the socket located at this place.
1856 The path must begin with a slash and by default is absolute.
1857 It can be relative to the prefix defined by "unix-bind" in
1858 the global section. Note that the total length of the prefix
1859 followed by the socket path cannot exceed some system limits
1860 for UNIX sockets, which commonly are set to 107 characters.
1861
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001862 <param*> is a list of parameters common to all sockets declared on the
1863 same line. These numerous parameters depend on OS and build
1864 options and have a complete section dedicated to them. Please
1865 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001866
Willy Tarreau0ba27502007-12-24 16:55:16 +01001867 It is possible to specify a list of address:port combinations delimited by
1868 commas. The frontend will then listen on all of these addresses. There is no
1869 fixed limit to the number of addresses and ports which can be listened on in
1870 a frontend, as well as there is no limit to the number of "bind" statements
1871 in a frontend.
1872
1873 Example :
1874 listen http_proxy
1875 bind :80,:443
1876 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001877 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01001878
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001879 listen http_https_proxy
1880 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02001881 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001882
Willy Tarreau24709282013-03-10 21:32:12 +01001883 listen http_https_proxy_explicit
1884 bind ipv6@:80
1885 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
1886 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
1887
Willy Tarreaudad36a32013-03-11 01:20:04 +01001888 listen external_bind_app1
1889 bind fd@${FD_APP1}
1890
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001891 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001892 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001893
1894
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001895bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001896 Limit visibility of an instance to a certain set of processes numbers.
1897 May be used in sections : defaults | frontend | listen | backend
1898 yes | yes | yes | yes
1899 Arguments :
1900 all All process will see this instance. This is the default. It
1901 may be used to override a default value.
1902
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001903 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001904 option may be combined with other numbers.
1905
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001906 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001907 option may be combined with other numbers. Do not use it
1908 with less than 2 processes otherwise some instances might be
1909 missing from all processes.
1910
Willy Tarreau110ecc12012-11-15 17:50:01 +01001911 number The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001912 whose values must all be between 1 and 32 or 64 depending on
Willy Tarreau102df612014-05-07 23:56:38 +02001913 the machine's word size. If a proxy is bound to process
1914 numbers greater than the configured global.nbproc, it will
1915 either be forced to process #1 if a single process was
1916 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001917
1918 This keyword limits binding of certain instances to certain processes. This
1919 is useful in order not to have too many processes listening to the same
1920 ports. For instance, on a dual-core machine, it might make sense to set
1921 'nbproc 2' in the global section, then distributes the listeners among 'odd'
1922 and 'even' instances.
1923
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001924 At the moment, it is not possible to reference more than 32 or 64 processes
1925 using this keyword, but this should be more than enough for most setups.
1926 Please note that 'all' really means all processes regardless of the machine's
1927 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001928
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001929 Each "bind" line may further be limited to a subset of the proxy's processes,
1930 please consult the "process" bind keyword in section 5.1.
1931
Willy Tarreaue56c4f12014-09-16 13:21:03 +02001932 When a frontend has no explicit "bind-process" line, it tries to bind to all
1933 the processes referenced by its "bind" lines. That means that frontends can
1934 easily adapt to their listeners' processes.
1935
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001936 If some backends are referenced by frontends bound to other processes, the
1937 backend automatically inherits the frontend's processes.
1938
1939 Example :
1940 listen app_ip1
1941 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001942 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001943
1944 listen app_ip2
1945 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001946 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001947
1948 listen management
1949 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001950 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001951
Willy Tarreau110ecc12012-11-15 17:50:01 +01001952 listen management
1953 bind 10.0.0.4:80
1954 bind-process 1-4
1955
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001956 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001957
1958
Willy Tarreau0ba27502007-12-24 16:55:16 +01001959block { if | unless } <condition>
1960 Block a layer 7 request if/unless a condition is matched
1961 May be used in sections : defaults | frontend | listen | backend
1962 no | yes | yes | yes
1963
1964 The HTTP request will be blocked very early in the layer 7 processing
1965 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001966 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02001967 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01001968 conditions are met or not met. There is no fixed limit to the number of
1969 "block" statements per instance.
1970
1971 Example:
1972 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1973 acl invalid_src src_port 0:1023
1974 acl local_dst hdr(host) -i localhost
1975 block if invalid_src || local_dst
1976
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001977 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001978
1979
1980capture cookie <name> len <length>
1981 Capture and log a cookie in the request and in the response.
1982 May be used in sections : defaults | frontend | listen | backend
1983 no | yes | yes | no
1984 Arguments :
1985 <name> is the beginning of the name of the cookie to capture. In order
1986 to match the exact name, simply suffix the name with an equal
1987 sign ('='). The full name will appear in the logs, which is
1988 useful with application servers which adjust both the cookie name
1989 and value (eg: ASPSESSIONXXXXX).
1990
1991 <length> is the maximum number of characters to report in the logs, which
1992 include the cookie name, the equal sign and the value, all in the
1993 standard "name=value" form. The string will be truncated on the
1994 right if it exceeds <length>.
1995
1996 Only the first cookie is captured. Both the "cookie" request headers and the
1997 "set-cookie" response headers are monitored. This is particularly useful to
1998 check for application bugs causing session crossing or stealing between
1999 users, because generally the user's cookies can only change on a login page.
2000
2001 When the cookie was not presented by the client, the associated log column
2002 will report "-". When a request does not cause a cookie to be assigned by the
2003 server, a "-" is reported in the response column.
2004
2005 The capture is performed in the frontend only because it is necessary that
2006 the log format does not change for a given frontend depending on the
2007 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002008 "capture cookie" statement in a frontend. The maximum capture length is set
2009 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2010 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002011
2012 Example:
2013 capture cookie ASPSESSION len 32
2014
2015 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002016 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002017
2018
2019capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002020 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002021 May be used in sections : defaults | frontend | listen | backend
2022 no | yes | yes | no
2023 Arguments :
2024 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002025 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002026 appear in the requests, with the first letter of each word in
2027 upper case. The header name will not appear in the logs, only the
2028 value is reported, but the position in the logs is respected.
2029
2030 <length> is the maximum number of characters to extract from the value and
2031 report in the logs. The string will be truncated on the right if
2032 it exceeds <length>.
2033
Willy Tarreau4460d032012-11-21 23:37:37 +01002034 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002035 value will be added to the logs between braces ('{}'). If multiple headers
2036 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002037 in the same order they were declared in the configuration. Non-existent
2038 headers will be logged just as an empty string. Common uses for request
2039 header captures include the "Host" field in virtual hosting environments, the
2040 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002041 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002042 environments to find where the request came from.
2043
2044 Note that when capturing headers such as "User-agent", some spaces may be
2045 logged, making the log analysis more difficult. Thus be careful about what
2046 you log if you know your log parser is not smart enough to rely on the
2047 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002048
Willy Tarreau0900abb2012-11-22 00:21:46 +01002049 There is no limit to the number of captured request headers nor to their
2050 length, though it is wise to keep them low to limit memory usage per session.
2051 In order to keep log format consistent for a same frontend, header captures
2052 can only be declared in a frontend. It is not possible to specify a capture
2053 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002054
2055 Example:
2056 capture request header Host len 15
2057 capture request header X-Forwarded-For len 15
2058 capture request header Referrer len 15
2059
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002060 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002061 about logging.
2062
2063
2064capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002065 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002066 May be used in sections : defaults | frontend | listen | backend
2067 no | yes | yes | no
2068 Arguments :
2069 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002070 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002071 appear in the response, with the first letter of each word in
2072 upper case. The header name will not appear in the logs, only the
2073 value is reported, but the position in the logs is respected.
2074
2075 <length> is the maximum number of characters to extract from the value and
2076 report in the logs. The string will be truncated on the right if
2077 it exceeds <length>.
2078
Willy Tarreau4460d032012-11-21 23:37:37 +01002079 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002080 result will be added to the logs between braces ('{}') after the captured
2081 request headers. If multiple headers are captured, they will be delimited by
2082 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002083 the configuration. Non-existent headers will be logged just as an empty
2084 string. Common uses for response header captures include the "Content-length"
2085 header which indicates how many bytes are expected to be returned, the
2086 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002087
Willy Tarreau0900abb2012-11-22 00:21:46 +01002088 There is no limit to the number of captured response headers nor to their
2089 length, though it is wise to keep them low to limit memory usage per session.
2090 In order to keep log format consistent for a same frontend, header captures
2091 can only be declared in a frontend. It is not possible to specify a capture
2092 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002093
2094 Example:
2095 capture response header Content-length len 9
2096 capture response header Location len 15
2097
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002098 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002099 about logging.
2100
2101
Cyril Bontéf0c60612010-02-06 14:44:47 +01002102clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002103 Set the maximum inactivity time on the client side.
2104 May be used in sections : defaults | frontend | listen | backend
2105 yes | yes | yes | no
2106 Arguments :
2107 <timeout> is the timeout value is specified in milliseconds by default, but
2108 can be in any other unit if the number is suffixed by the unit,
2109 as explained at the top of this document.
2110
2111 The inactivity timeout applies when the client is expected to acknowledge or
2112 send data. In HTTP mode, this timeout is particularly important to consider
2113 during the first phase, when the client sends the request, and during the
2114 response while it is reading data sent by the server. The value is specified
2115 in milliseconds by default, but can be in any other unit if the number is
2116 suffixed by the unit, as specified at the top of this document. In TCP mode
2117 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2118 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002119 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002120 losses by specifying timeouts that are slightly above multiples of 3 seconds
2121 (eg: 4 or 5 seconds).
2122
2123 This parameter is specific to frontends, but can be specified once for all in
2124 "defaults" sections. This is in fact one of the easiest solutions not to
2125 forget about it. An unspecified timeout results in an infinite timeout, which
2126 is not recommended. Such a usage is accepted and works but reports a warning
2127 during startup because it may results in accumulation of expired sessions in
2128 the system if the system's timeouts are not configured either.
2129
2130 This parameter is provided for compatibility but is currently deprecated.
2131 Please use "timeout client" instead.
2132
Willy Tarreau036fae02008-01-06 13:24:40 +01002133 See also : "timeout client", "timeout http-request", "timeout server", and
2134 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002135
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002136compression algo <algorithm> ...
2137compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002138compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002139 Enable HTTP compression.
2140 May be used in sections : defaults | frontend | listen | backend
2141 yes | yes | yes | yes
2142 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002143 algo is followed by the list of supported compression algorithms.
2144 type is followed by the list of MIME types that will be compressed.
2145 offload makes haproxy work as a compression offloader only (see notes).
2146
2147 The currently supported algorithms are :
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002148 identity this is mostly for debugging, and it was useful for developing
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002149 the compression feature. Identity does not apply any change on
2150 data.
2151
2152 gzip applies gzip compression. This setting is only available when
2153 support for zlib was built in.
2154
2155 deflate same as gzip, but with deflate algorithm and zlib format.
2156 Note that this algorithm has ambiguous support on many browsers
2157 and no support at all from recent ones. It is strongly
2158 recommended not to use it for anything else than experimentation.
2159 This setting is only available when support for zlib was built
2160 in.
2161
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002162 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002163 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002164 If backend servers support HTTP compression, these directives
2165 will be no-op: haproxy will see the compressed response and will not
2166 compress again. If backend servers do not support HTTP compression and
2167 there is Accept-Encoding header in request, haproxy will compress the
2168 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002169
2170 The "offload" setting makes haproxy remove the Accept-Encoding header to
2171 prevent backend servers from compressing responses. It is strongly
2172 recommended not to do this because this means that all the compression work
2173 will be done on the single point where haproxy is located. However in some
2174 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002175 with broken HTTP compression implementation which can't be turned off.
2176 In that case haproxy can be used to prevent that gateway from emitting
2177 invalid payloads. In this case, simply removing the header in the
2178 configuration does not work because it applies before the header is parsed,
2179 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreau4cf57af2014-07-12 16:37:02 +02002180 then be used for such scenarios. Note: for now, the "offload" setting is
2181 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002182
William Lallemand05097442012-11-20 12:14:28 +01002183 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002184 * the request does not advertise a supported compression algorithm in the
2185 "Accept-Encoding" header
2186 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002187 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002188 * response header "Transfer-Encoding" contains "chunked" (Temporary
2189 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002190 * response contain neither a "Content-Length" header nor a
2191 "Transfer-Encoding" whose last value is "chunked"
2192 * response contains a "Content-Type" header whose first value starts with
2193 "multipart"
2194 * the response contains the "no-transform" value in the "Cache-control"
2195 header
2196 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2197 and later
2198 * The response contains a "Content-Encoding" header, indicating that the
2199 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002200
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002201 Note: The compression does not rewrite Etag headers, and does not emit the
2202 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002203
William Lallemand82fe75c2012-10-23 10:25:10 +02002204 Examples :
2205 compression algo gzip
2206 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002207
Cyril Bontéf0c60612010-02-06 14:44:47 +01002208contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002209 Set the maximum time to wait for a connection attempt to a server to succeed.
2210 May be used in sections : defaults | frontend | listen | backend
2211 yes | no | yes | yes
2212 Arguments :
2213 <timeout> is the timeout value is specified in milliseconds by default, but
2214 can be in any other unit if the number is suffixed by the unit,
2215 as explained at the top of this document.
2216
2217 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002218 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002219 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01002220 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
2221 connect timeout also presets the queue timeout to the same value if this one
2222 has not been specified. Historically, the contimeout was also used to set the
2223 tarpit timeout in a listen section, which is not possible in a pure frontend.
2224
2225 This parameter is specific to backends, but can be specified once for all in
2226 "defaults" sections. This is in fact one of the easiest solutions not to
2227 forget about it. An unspecified timeout results in an infinite timeout, which
2228 is not recommended. Such a usage is accepted and works but reports a warning
2229 during startup because it may results in accumulation of failed sessions in
2230 the system if the system's timeouts are not configured either.
2231
2232 This parameter is provided for backwards compatibility but is currently
2233 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2234 instead.
2235
2236 See also : "timeout connect", "timeout queue", "timeout tarpit",
2237 "timeout server", "contimeout".
2238
2239
Willy Tarreau55165fe2009-05-10 12:02:55 +02002240cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002241 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2242 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002243 Enable cookie-based persistence in a backend.
2244 May be used in sections : defaults | frontend | listen | backend
2245 yes | no | yes | yes
2246 Arguments :
2247 <name> is the name of the cookie which will be monitored, modified or
2248 inserted in order to bring persistence. This cookie is sent to
2249 the client via a "Set-Cookie" header in the response, and is
2250 brought back by the client in a "Cookie" header in all requests.
2251 Special care should be taken to choose a name which does not
2252 conflict with any likely application cookie. Also, if the same
2253 backends are subject to be used by the same clients (eg:
2254 HTTP/HTTPS), care should be taken to use different cookie names
2255 between all backends if persistence between them is not desired.
2256
2257 rewrite This keyword indicates that the cookie will be provided by the
2258 server and that haproxy will have to modify its value to set the
2259 server's identifier in it. This mode is handy when the management
2260 of complex combinations of "Set-cookie" and "Cache-control"
2261 headers is left to the application. The application can then
2262 decide whether or not it is appropriate to emit a persistence
2263 cookie. Since all responses should be monitored, this mode only
2264 works in HTTP close mode. Unless the application behaviour is
2265 very complex and/or broken, it is advised not to start with this
2266 mode for new deployments. This keyword is incompatible with
2267 "insert" and "prefix".
2268
2269 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02002270 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002271
Willy Tarreaua79094d2010-08-31 22:54:15 +02002272 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002273 server. When used without the "preserve" option, if the server
2274 emits a cookie with the same name, it will be remove before
2275 processing. For this reason, this mode can be used to upgrade
2276 existing configurations running in the "rewrite" mode. The cookie
2277 will only be a session cookie and will not be stored on the
2278 client's disk. By default, unless the "indirect" option is added,
2279 the server will see the cookies emitted by the client. Due to
2280 caching effects, it is generally wise to add the "nocache" or
2281 "postonly" keywords (see below). The "insert" keyword is not
2282 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002283
2284 prefix This keyword indicates that instead of relying on a dedicated
2285 cookie for the persistence, an existing one will be completed.
2286 This may be needed in some specific environments where the client
2287 does not support more than one single cookie and the application
2288 already needs it. In this case, whenever the server sets a cookie
2289 named <name>, it will be prefixed with the server's identifier
2290 and a delimiter. The prefix will be removed from all client
2291 requests so that the server still finds the cookie it emitted.
2292 Since all requests and responses are subject to being modified,
2293 this mode requires the HTTP close mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02002294 not compatible with "rewrite" and "insert". Note: it is highly
2295 recommended not to use "indirect" with "prefix", otherwise server
2296 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002297
Willy Tarreaua79094d2010-08-31 22:54:15 +02002298 indirect When this option is specified, no cookie will be emitted to a
2299 client which already has a valid one for the server which has
2300 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002301 it will be removed, unless the "preserve" option is also set. In
2302 "insert" mode, this will additionally remove cookies from the
2303 requests transmitted to the server, making the persistence
2304 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02002305 Note: it is highly recommended not to use "indirect" with
2306 "prefix", otherwise server cookie updates would not be sent to
2307 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002308
2309 nocache This option is recommended in conjunction with the insert mode
2310 when there is a cache between the client and HAProxy, as it
2311 ensures that a cacheable response will be tagged non-cacheable if
2312 a cookie needs to be inserted. This is important because if all
2313 persistence cookies are added on a cacheable home page for
2314 instance, then all customers will then fetch the page from an
2315 outer cache and will all share the same persistence cookie,
2316 leading to one server receiving much more traffic than others.
2317 See also the "insert" and "postonly" options.
2318
2319 postonly This option ensures that cookie insertion will only be performed
2320 on responses to POST requests. It is an alternative to the
2321 "nocache" option, because POST responses are not cacheable, so
2322 this ensures that the persistence cookie will never get cached.
2323 Since most sites do not need any sort of persistence before the
2324 first POST which generally is a login request, this is a very
2325 efficient method to optimize caching without risking to find a
2326 persistence cookie in the cache.
2327 See also the "insert" and "nocache" options.
2328
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002329 preserve This option may only be used with "insert" and/or "indirect". It
2330 allows the server to emit the persistence cookie itself. In this
2331 case, if a cookie is found in the response, haproxy will leave it
2332 untouched. This is useful in order to end persistence after a
2333 logout request for instance. For this, the server just has to
2334 emit a cookie with an invalid value (eg: empty) or with a date in
2335 the past. By combining this mechanism with the "disable-on-404"
2336 check option, it is possible to perform a completely graceful
2337 shutdown because users will definitely leave the server after
2338 they logout.
2339
Willy Tarreau4992dd22012-05-31 21:02:17 +02002340 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
2341 when a cookie is inserted. This attribute is used so that a
2342 user agent doesn't share the cookie with non-HTTP components.
2343 Please check RFC6265 for more information on this attribute.
2344
2345 secure This option tells haproxy to add a "Secure" cookie attribute when
2346 a cookie is inserted. This attribute is used so that a user agent
2347 never emits this cookie over non-secure channels, which means
2348 that a cookie learned with this flag will be presented only over
2349 SSL/TLS connections. Please check RFC6265 for more information on
2350 this attribute.
2351
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002352 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002353 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01002354 name. If the domain begins with a dot, the browser is allowed to
2355 use it for any host ending with that name. It is also possible to
2356 specify several domain names by invoking this option multiple
2357 times. Some browsers might have small limits on the number of
2358 domains, so be careful when doing that. For the record, sending
2359 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002360
Willy Tarreau996a92c2010-10-13 19:30:47 +02002361 maxidle This option allows inserted cookies to be ignored after some idle
2362 time. It only works with insert-mode cookies. When a cookie is
2363 sent to the client, the date this cookie was emitted is sent too.
2364 Upon further presentations of this cookie, if the date is older
2365 than the delay indicated by the parameter (in seconds), it will
2366 be ignored. Otherwise, it will be refreshed if needed when the
2367 response is sent to the client. This is particularly useful to
2368 prevent users who never close their browsers from remaining for
2369 too long on the same server (eg: after a farm size change). When
2370 this option is set and a cookie has no date, it is always
2371 accepted, but gets refreshed in the response. This maintains the
2372 ability for admins to access their sites. Cookies that have a
2373 date in the future further than 24 hours are ignored. Doing so
2374 lets admins fix timezone issues without risking kicking users off
2375 the site.
2376
2377 maxlife This option allows inserted cookies to be ignored after some life
2378 time, whether they're in use or not. It only works with insert
2379 mode cookies. When a cookie is first sent to the client, the date
2380 this cookie was emitted is sent too. Upon further presentations
2381 of this cookie, if the date is older than the delay indicated by
2382 the parameter (in seconds), it will be ignored. If the cookie in
2383 the request has no date, it is accepted and a date will be set.
2384 Cookies that have a date in the future further than 24 hours are
2385 ignored. Doing so lets admins fix timezone issues without risking
2386 kicking users off the site. Contrary to maxidle, this value is
2387 not refreshed, only the first visit date counts. Both maxidle and
2388 maxlife may be used at the time. This is particularly useful to
2389 prevent users who never close their browsers from remaining for
2390 too long on the same server (eg: after a farm size change). This
2391 is stronger than the maxidle method in that it forces a
2392 redispatch after some absolute delay.
2393
Willy Tarreau0ba27502007-12-24 16:55:16 +01002394 There can be only one persistence cookie per HTTP backend, and it can be
2395 declared in a defaults section. The value of the cookie will be the value
2396 indicated after the "cookie" keyword in a "server" statement. If no cookie
2397 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002398
Willy Tarreau0ba27502007-12-24 16:55:16 +01002399 Examples :
2400 cookie JSESSIONID prefix
2401 cookie SRV insert indirect nocache
2402 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02002403 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01002404
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002405 See also : "appsession", "balance source", "capture cookie", "server"
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002406 and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002407
Willy Tarreau983e01e2010-01-11 18:42:06 +01002408
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002409default-server [param*]
2410 Change default options for a server in a backend
2411 May be used in sections : defaults | frontend | listen | backend
2412 yes | no | yes | yes
2413 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01002414 <param*> is a list of parameters for this server. The "default-server"
2415 keyword accepts an important number of options and has a complete
2416 section dedicated to it. Please refer to section 5 for more
2417 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002418
Willy Tarreau983e01e2010-01-11 18:42:06 +01002419 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002420 default-server inter 1000 weight 13
2421
2422 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01002423
Willy Tarreau983e01e2010-01-11 18:42:06 +01002424
Willy Tarreau0ba27502007-12-24 16:55:16 +01002425default_backend <backend>
2426 Specify the backend to use when no "use_backend" rule has been matched.
2427 May be used in sections : defaults | frontend | listen | backend
2428 yes | yes | yes | no
2429 Arguments :
2430 <backend> is the name of the backend to use.
2431
2432 When doing content-switching between frontend and backends using the
2433 "use_backend" keyword, it is often useful to indicate which backend will be
2434 used when no rule has matched. It generally is the dynamic backend which
2435 will catch all undetermined requests.
2436
Willy Tarreau0ba27502007-12-24 16:55:16 +01002437 Example :
2438
2439 use_backend dynamic if url_dyn
2440 use_backend static if url_css url_img extension_img
2441 default_backend dynamic
2442
Willy Tarreau2769aa02007-12-27 18:26:09 +01002443 See also : "use_backend", "reqsetbe", "reqisetbe"
2444
Willy Tarreau0ba27502007-12-24 16:55:16 +01002445
Baptiste Assmann27f51342013-10-09 06:51:49 +02002446description <string>
2447 Describe a listen, frontend or backend.
2448 May be used in sections : defaults | frontend | listen | backend
2449 no | yes | yes | yes
2450 Arguments : string
2451
2452 Allows to add a sentence to describe the related object in the HAProxy HTML
2453 stats page. The description will be printed on the right of the object name
2454 it describes.
2455 No need to backslash spaces in the <string> arguments.
2456
2457
Willy Tarreau0ba27502007-12-24 16:55:16 +01002458disabled
2459 Disable a proxy, frontend or backend.
2460 May be used in sections : defaults | frontend | listen | backend
2461 yes | yes | yes | yes
2462 Arguments : none
2463
2464 The "disabled" keyword is used to disable an instance, mainly in order to
2465 liberate a listening port or to temporarily disable a service. The instance
2466 will still be created and its configuration will be checked, but it will be
2467 created in the "stopped" state and will appear as such in the statistics. It
2468 will not receive any traffic nor will it send any health-checks or logs. It
2469 is possible to disable many instances at once by adding the "disabled"
2470 keyword in a "defaults" section.
2471
2472 See also : "enabled"
2473
2474
Willy Tarreau5ce94572010-06-07 14:35:41 +02002475dispatch <address>:<port>
2476 Set a default server address
2477 May be used in sections : defaults | frontend | listen | backend
2478 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02002479 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02002480
2481 <address> is the IPv4 address of the default server. Alternatively, a
2482 resolvable hostname is supported, but this name will be resolved
2483 during start-up.
2484
2485 <ports> is a mandatory port specification. All connections will be sent
2486 to this port, and it is not permitted to use port offsets as is
2487 possible with normal servers.
2488
Willy Tarreau787aed52011-04-15 06:45:37 +02002489 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02002490 server can take the connection. In the past it was used to forward non
2491 persistent connections to an auxiliary load balancer. Due to its simple
2492 syntax, it has also been used for simple TCP relays. It is recommended not to
2493 use it for more clarity, and to use the "server" directive instead.
2494
2495 See also : "server"
2496
2497
Willy Tarreau0ba27502007-12-24 16:55:16 +01002498enabled
2499 Enable a proxy, frontend or backend.
2500 May be used in sections : defaults | frontend | listen | backend
2501 yes | yes | yes | yes
2502 Arguments : none
2503
2504 The "enabled" keyword is used to explicitly enable an instance, when the
2505 defaults has been set to "disabled". This is very rarely used.
2506
2507 See also : "disabled"
2508
2509
2510errorfile <code> <file>
2511 Return a file contents instead of errors generated by HAProxy
2512 May be used in sections : defaults | frontend | listen | backend
2513 yes | yes | yes | yes
2514 Arguments :
2515 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002516 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002517
2518 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002519 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01002520 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01002521 error pages, and to use absolute paths, since files are read
2522 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002523
2524 It is important to understand that this keyword is not meant to rewrite
2525 errors returned by the server, but errors detected and returned by HAProxy.
2526 This is why the list of supported errors is limited to a small set.
2527
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002528 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2529
Willy Tarreau0ba27502007-12-24 16:55:16 +01002530 The files are returned verbatim on the TCP socket. This allows any trick such
2531 as redirections to another URL or site, as well as tricks to clean cookies,
2532 force enable or disable caching, etc... The package provides default error
2533 files returning the same contents as default errors.
2534
Willy Tarreau59140a22009-02-22 12:02:30 +01002535 The files should not exceed the configured buffer size (BUFSIZE), which
2536 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
2537 not to put any reference to local contents (eg: images) in order to avoid
2538 loops between the client and HAProxy when all servers are down, causing an
2539 error to be returned instead of an image. For better HTTP compliance, it is
2540 recommended that all header lines end with CR-LF and not LF alone.
2541
Willy Tarreau0ba27502007-12-24 16:55:16 +01002542 The files are read at the same time as the configuration and kept in memory.
2543 For this reason, the errors continue to be returned even when the process is
2544 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01002545 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002546 403 status code and interrogating a blocked URL.
2547
2548 See also : "errorloc", "errorloc302", "errorloc303"
2549
Willy Tarreau59140a22009-02-22 12:02:30 +01002550 Example :
2551 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau2705a612014-05-23 17:38:34 +02002552 errorfile 408 /dev/null # workaround Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01002553 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
2554 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
2555
Willy Tarreau2769aa02007-12-27 18:26:09 +01002556
2557errorloc <code> <url>
2558errorloc302 <code> <url>
2559 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2560 May be used in sections : defaults | frontend | listen | backend
2561 yes | yes | yes | yes
2562 Arguments :
2563 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002564 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002565
2566 <url> it is the exact contents of the "Location" header. It may contain
2567 either a relative URI to an error page hosted on the same site,
2568 or an absolute URI designating an error page on another site.
2569 Special care should be given to relative URIs to avoid redirect
2570 loops if the URI itself may generate the same error (eg: 500).
2571
2572 It is important to understand that this keyword is not meant to rewrite
2573 errors returned by the server, but errors detected and returned by HAProxy.
2574 This is why the list of supported errors is limited to a small set.
2575
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002576 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2577
Willy Tarreau2769aa02007-12-27 18:26:09 +01002578 Note that both keyword return the HTTP 302 status code, which tells the
2579 client to fetch the designated URL using the same HTTP method. This can be
2580 quite problematic in case of non-GET methods such as POST, because the URL
2581 sent to the client might not be allowed for something other than GET. To
2582 workaround this problem, please use "errorloc303" which send the HTTP 303
2583 status code, indicating to the client that the URL must be fetched with a GET
2584 request.
2585
2586 See also : "errorfile", "errorloc303"
2587
2588
2589errorloc303 <code> <url>
2590 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2591 May be used in sections : defaults | frontend | listen | backend
2592 yes | yes | yes | yes
2593 Arguments :
2594 <code> is the HTTP status code. Currently, HAProxy is capable of
2595 generating codes 400, 403, 408, 500, 502, 503, and 504.
2596
2597 <url> it is the exact contents of the "Location" header. It may contain
2598 either a relative URI to an error page hosted on the same site,
2599 or an absolute URI designating an error page on another site.
2600 Special care should be given to relative URIs to avoid redirect
2601 loops if the URI itself may generate the same error (eg: 500).
2602
2603 It is important to understand that this keyword is not meant to rewrite
2604 errors returned by the server, but errors detected and returned by HAProxy.
2605 This is why the list of supported errors is limited to a small set.
2606
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002607 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2608
Willy Tarreau2769aa02007-12-27 18:26:09 +01002609 Note that both keyword return the HTTP 303 status code, which tells the
2610 client to fetch the designated URL using the same HTTP GET method. This
2611 solves the usual problems associated with "errorloc" and the 302 code. It is
2612 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002613 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002614
2615 See also : "errorfile", "errorloc", "errorloc302"
2616
2617
Willy Tarreau4de91492010-01-22 19:10:05 +01002618force-persist { if | unless } <condition>
2619 Declare a condition to force persistence on down servers
2620 May be used in sections: defaults | frontend | listen | backend
2621 no | yes | yes | yes
2622
2623 By default, requests are not dispatched to down servers. It is possible to
2624 force this using "option persist", but it is unconditional and redispatches
2625 to a valid server if "option redispatch" is set. That leaves with very little
2626 possibilities to force some requests to reach a server which is artificially
2627 marked down for maintenance operations.
2628
2629 The "force-persist" statement allows one to declare various ACL-based
2630 conditions which, when met, will cause a request to ignore the down status of
2631 a server and still try to connect to it. That makes it possible to start a
2632 server, still replying an error to the health checks, and run a specially
2633 configured browser to test the service. Among the handy methods, one could
2634 use a specific source IP address, or a specific cookie. The cookie also has
2635 the advantage that it can easily be added/removed on the browser from a test
2636 page. Once the service is validated, it is then possible to open the service
2637 to the world by returning a valid response to health checks.
2638
2639 The forced persistence is enabled when an "if" condition is met, or unless an
2640 "unless" condition is met. The final redispatch is always disabled when this
2641 is used.
2642
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002643 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002644 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01002645
2646
Willy Tarreau2769aa02007-12-27 18:26:09 +01002647fullconn <conns>
2648 Specify at what backend load the servers will reach their maxconn
2649 May be used in sections : defaults | frontend | listen | backend
2650 yes | no | yes | yes
2651 Arguments :
2652 <conns> is the number of connections on the backend which will make the
2653 servers use the maximal number of connections.
2654
Willy Tarreau198a7442008-01-17 12:05:32 +01002655 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01002656 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01002657 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01002658 load. The server will then always accept at least <minconn> connections,
2659 never more than <maxconn>, and the limit will be on the ramp between both
2660 values when the backend has less than <conns> concurrent connections. This
2661 makes it possible to limit the load on the servers during normal loads, but
2662 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002663 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002664
Willy Tarreaufbb78422011-06-05 15:38:35 +02002665 Since it's hard to get this value right, haproxy automatically sets it to
2666 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01002667 backend (based on "use_backend" and "default_backend" rules). That way it's
2668 safe to leave it unset. However, "use_backend" involving dynamic names are
2669 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02002670
Willy Tarreau2769aa02007-12-27 18:26:09 +01002671 Example :
2672 # The servers will accept between 100 and 1000 concurrent connections each
2673 # and the maximum of 1000 will be reached when the backend reaches 10000
2674 # connections.
2675 backend dynamic
2676 fullconn 10000
2677 server srv1 dyn1:80 minconn 100 maxconn 1000
2678 server srv2 dyn2:80 minconn 100 maxconn 1000
2679
2680 See also : "maxconn", "server"
2681
2682
2683grace <time>
2684 Maintain a proxy operational for some time after a soft stop
2685 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01002686 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01002687 Arguments :
2688 <time> is the time (by default in milliseconds) for which the instance
2689 will remain operational with the frontend sockets still listening
2690 when a soft-stop is received via the SIGUSR1 signal.
2691
2692 This may be used to ensure that the services disappear in a certain order.
2693 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002694 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01002695 needed by the equipment to detect the failure.
2696
2697 Note that currently, there is very little benefit in using this parameter,
2698 and it may in fact complicate the soft-reconfiguration process more than
2699 simplify it.
2700
Willy Tarreau0ba27502007-12-24 16:55:16 +01002701
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002702hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002703 Specify a method to use for mapping hashes to servers
2704 May be used in sections : defaults | frontend | listen | backend
2705 yes | no | yes | yes
2706 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04002707 <method> is the method used to select a server from the hash computed by
2708 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002709
Bhaskar98634f02013-10-29 23:30:51 -04002710 map-based the hash table is a static array containing all alive servers.
2711 The hashes will be very smooth, will consider weights, but
2712 will be static in that weight changes while a server is up
2713 will be ignored. This means that there will be no slow start.
2714 Also, since a server is selected by its position in the array,
2715 most mappings are changed when the server count changes. This
2716 means that when a server goes up or down, or when a server is
2717 added to a farm, most connections will be redistributed to
2718 different servers. This can be inconvenient with caches for
2719 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01002720
Bhaskar98634f02013-10-29 23:30:51 -04002721 consistent the hash table is a tree filled with many occurrences of each
2722 server. The hash key is looked up in the tree and the closest
2723 server is chosen. This hash is dynamic, it supports changing
2724 weights while the servers are up, so it is compatible with the
2725 slow start feature. It has the advantage that when a server
2726 goes up or down, only its associations are moved. When a
2727 server is added to the farm, only a few part of the mappings
2728 are redistributed, making it an ideal method for caches.
2729 However, due to its principle, the distribution will never be
2730 very smooth and it may sometimes be necessary to adjust a
2731 server's weight or its ID to get a more balanced distribution.
2732 In order to get the same distribution on multiple load
2733 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002734 same IDs. Note: consistent hash uses sdbm and avalanche if no
2735 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04002736
2737 <function> is the hash function to be used :
2738
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002739 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04002740 reimplementation of ndbm) database library. It was found to do
2741 well in scrambling bits, causing better distribution of the keys
2742 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002743 function with good distribution, unless the total server weight
2744 is a multiple of 64, in which case applying the avalanche
2745 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04002746
2747 djb2 this function was first proposed by Dan Bernstein many years ago
2748 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002749 function provides a better distribution than sdbm. It generally
2750 works well with text-based inputs though it can perform extremely
2751 poorly with numeric-only input or when the total server weight is
2752 a multiple of 33, unless the avalanche modifier is also used.
2753
Willy Tarreaua0f42712013-11-14 14:30:35 +01002754 wt6 this function was designed for haproxy while testing other
2755 functions in the past. It is not as smooth as the other ones, but
2756 is much less sensible to the input data set or to the number of
2757 servers. It can make sense as an alternative to sdbm+avalanche or
2758 djb2+avalanche for consistent hashing or when hashing on numeric
2759 data such as a source IP address or a visitor identifier in a URL
2760 parameter.
2761
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002762 <modifier> indicates an optional method applied after hashing the key :
2763
2764 avalanche This directive indicates that the result from the hash
2765 function above should not be used in its raw form but that
2766 a 4-byte full avalanche hash must be applied first. The
2767 purpose of this step is to mix the resulting bits from the
2768 previous hash in order to avoid any undesired effect when
2769 the input contains some limited values or when the number of
2770 servers is a multiple of one of the hash's components (64
2771 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
2772 result less predictable, but it's also not as smooth as when
2773 using the original function. Some testing might be needed
2774 with some workloads. This hash is one of the many proposed
2775 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002776
Bhaskar98634f02013-10-29 23:30:51 -04002777 The default hash type is "map-based" and is recommended for most usages. The
2778 default function is "sdbm", the selection of a function should be based on
2779 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002780
2781 See also : "balance", "server"
2782
2783
Willy Tarreau0ba27502007-12-24 16:55:16 +01002784http-check disable-on-404
2785 Enable a maintenance mode upon HTTP/404 response to health-checks
2786 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01002787 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01002788 Arguments : none
2789
2790 When this option is set, a server which returns an HTTP code 404 will be
2791 excluded from further load-balancing, but will still receive persistent
2792 connections. This provides a very convenient method for Web administrators
2793 to perform a graceful shutdown of their servers. It is also important to note
2794 that a server which is detected as failed while it was in this mode will not
2795 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
2796 will immediately be reinserted into the farm. The status on the stats page
2797 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01002798 option only works in conjunction with the "httpchk" option. If this option
2799 is used with "http-check expect", then it has precedence over it so that 404
2800 responses will still be considered as soft-stop.
2801
2802 See also : "option httpchk", "http-check expect"
2803
2804
2805http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002806 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01002807 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02002808 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01002809 Arguments :
2810 <match> is a keyword indicating how to look for a specific pattern in the
2811 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002812 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01002813 exclamation mark ("!") to negate the match. Spaces are allowed
2814 between the exclamation mark and the keyword. See below for more
2815 details on the supported keywords.
2816
2817 <pattern> is the pattern to look for. It may be a string or a regular
2818 expression. If the pattern contains spaces, they must be escaped
2819 with the usual backslash ('\').
2820
2821 By default, "option httpchk" considers that response statuses 2xx and 3xx
2822 are valid, and that others are invalid. When "http-check expect" is used,
2823 it defines what is considered valid or invalid. Only one "http-check"
2824 statement is supported in a backend. If a server fails to respond or times
2825 out, the check obviously fails. The available matches are :
2826
2827 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002828 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002829 response's status code is exactly this string. If the
2830 "status" keyword is prefixed with "!", then the response
2831 will be considered invalid if the status code matches.
2832
2833 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002834 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002835 response's status code matches the expression. If the
2836 "rstatus" keyword is prefixed with "!", then the response
2837 will be considered invalid if the status code matches.
2838 This is mostly used to check for multiple codes.
2839
2840 string <string> : test the exact string match in the HTTP response body.
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 body contains this exact string. If the
2843 "string" keyword is prefixed with "!", then the response
2844 will be considered invalid if the body contains this
2845 string. This can be used to look for a mandatory word at
2846 the end of a dynamic page, or to detect a failure when a
2847 specific error appears on the check page (eg: a stack
2848 trace).
2849
2850 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002851 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002852 response's body matches this expression. If the "rstring"
2853 keyword is prefixed with "!", then the response will be
2854 considered invalid if the body matches the expression.
2855 This can be used to look for a mandatory word at the end
2856 of a dynamic page, or to detect a failure when a specific
2857 error appears on the check page (eg: a stack trace).
2858
2859 It is important to note that the responses will be limited to a certain size
2860 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
2861 Thus, too large responses may not contain the mandatory pattern when using
2862 "string" or "rstring". If a large response is absolutely required, it is
2863 possible to change the default max size by setting the global variable.
2864 However, it is worth keeping in mind that parsing very large responses can
2865 waste some CPU cycles, especially when regular expressions are used, and that
2866 it is always better to focus the checks on smaller resources.
2867
Cyril Bontéa448e162015-01-30 00:07:07 +01002868 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
2869 will automatically append a "Connection: close" header, meaning that this
2870 header should not be present in the request provided by "option httpchk".
2871
Willy Tarreaubd741542010-03-16 18:46:54 +01002872 Last, if "http-check expect" is combined with "http-check disable-on-404",
2873 then this last one has precedence when the server responds with 404.
2874
2875 Examples :
2876 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002877 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01002878
2879 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002880 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01002881
2882 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002883 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01002884
2885 # check that we have a correct hexadecimal tag before /html
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002886 http-check expect rstring <!--tag:[0-9a-f]*</html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01002887
Willy Tarreaubd741542010-03-16 18:46:54 +01002888 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01002889
2890
Willy Tarreauef781042010-01-27 11:53:01 +01002891http-check send-state
2892 Enable emission of a state header with HTTP health checks
2893 May be used in sections : defaults | frontend | listen | backend
2894 yes | no | yes | yes
2895 Arguments : none
2896
2897 When this option is set, haproxy will systematically send a special header
2898 "X-Haproxy-Server-State" with a list of parameters indicating to each server
2899 how they are seen by haproxy. This can be used for instance when a server is
2900 manipulated without access to haproxy and the operator needs to know whether
2901 haproxy still sees it up or not, or if the server is the last one in a farm.
2902
2903 The header is composed of fields delimited by semi-colons, the first of which
2904 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
2905 checks on the total number before transition, just as appears in the stats
2906 interface. Next headers are in the form "<variable>=<value>", indicating in
2907 no specific order some values available in the stats interface :
2908 - a variable "name", containing the name of the backend followed by a slash
2909 ("/") then the name of the server. This can be used when a server is
2910 checked in multiple backends.
2911
2912 - a variable "node" containing the name of the haproxy node, as set in the
2913 global "node" variable, otherwise the system's hostname if unspecified.
2914
2915 - a variable "weight" indicating the weight of the server, a slash ("/")
2916 and the total weight of the farm (just counting usable servers). This
2917 helps to know if other servers are available to handle the load when this
2918 one fails.
2919
2920 - a variable "scur" indicating the current number of concurrent connections
2921 on the server, followed by a slash ("/") then the total number of
2922 connections on all servers of the same backend.
2923
2924 - a variable "qcur" indicating the current number of requests in the
2925 server's queue.
2926
2927 Example of a header received by the application server :
2928 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
2929 scur=13/22; qcur=0
2930
2931 See also : "option httpchk", "http-check disable-on-404"
2932
Willy Tarreauccbcc372012-12-27 12:37:57 +01002933http-request { allow | deny | tarpit | auth [realm <realm>] | redirect <rule> |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02002934 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002935 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06002936 replace-header <name> <match-regex> <replace-fmt> |
2937 replace-value <name> <match-regex> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02002938 set-tos <tos> | set-mark <mark> |
2939 add-acl(<file name>) <key fmt> |
2940 del-acl(<file name>) <key fmt> |
2941 del-map(<file name>) <key fmt> |
2942 set-map(<file name>) <key fmt> <value fmt>
2943 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01002944 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002945 Access control for Layer 7 requests
2946
2947 May be used in sections: defaults | frontend | listen | backend
2948 no | yes | yes | yes
2949
Willy Tarreau20b0de52012-12-24 15:45:22 +01002950 The http-request statement defines a set of rules which apply to layer 7
2951 processing. The rules are evaluated in their declaration order when they are
2952 met in a frontend, listen or backend section. Any rule may optionally be
2953 followed by an ACL-based condition, in which case it will only be evaluated
2954 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002955
Willy Tarreau20b0de52012-12-24 15:45:22 +01002956 The first keyword is the rule's action. Currently supported actions include :
2957 - "allow" : this stops the evaluation of the rules and lets the request
2958 pass the check. No further "http-request" rules are evaluated.
2959
2960 - "deny" : this stops the evaluation of the rules and immediately rejects
2961 the request and emits an HTTP 403 error. No further "http-request" rules
2962 are evaluated.
2963
Willy Tarreauccbcc372012-12-27 12:37:57 +01002964 - "tarpit" : this stops the evaluation of the rules and immediately blocks
2965 the request without responding for a delay specified by "timeout tarpit"
2966 or "timeout connect" if the former is not set. After that delay, if the
2967 client is still connected, an HTTP error 500 is returned so that the
2968 client does not suspect it has been tarpitted. Logs will report the flags
2969 "PT". The goal of the tarpit rule is to slow down robots during an attack
2970 when they're limited on the number of concurrent requests. It can be very
2971 efficient against very dumb robots, and will significantly reduce the
2972 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002973 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreauccbcc372012-12-27 12:37:57 +01002974 front firewall to support insane number of concurrent connections.
2975
Willy Tarreau20b0de52012-12-24 15:45:22 +01002976 - "auth" : this stops the evaluation of the rules and immediately responds
2977 with an HTTP 401 or 407 error code to invite the user to present a valid
2978 user name and password. No further "http-request" rules are evaluated. An
2979 optional "realm" parameter is supported, it sets the authentication realm
2980 that is returned with the response (typically the application's name).
2981
Willy Tarreau81499eb2012-12-27 12:19:02 +01002982 - "redirect" : this performs an HTTP redirection based on a redirect rule.
2983 This is exactly the same as the "redirect" statement except that it
2984 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01002985 "http-request" rules and that these rules use the "log-format" strings.
2986 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01002987
Willy Tarreau20b0de52012-12-24 15:45:22 +01002988 - "add-header" appends an HTTP header field whose name is specified in
2989 <name> and whose value is defined by <fmt> which follows the log-format
2990 rules (see Custom Log Format in section 8.2.4). This is particularly
2991 useful to pass connection-specific information to the server (eg: the
2992 client's SSL certificate), or to combine several headers into one. This
2993 rule is not final, so it is possible to add other similar rules. Note
2994 that header addition is performed immediately, so one rule might reuse
2995 the resulting header from a previous rule.
2996
2997 - "set-header" does the same as "add-header" except that the header name
2998 is first removed if it existed. This is useful when passing security
2999 information to the server, where the header must not be manipulated by
Willy Tarreaufcf75d22015-01-21 20:39:27 +01003000 external users. Note that the new value is computed before the removal so
3001 it is possible to concatenate a value to an existing header.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003002
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003003 - "del-header" removes all HTTP header fields whose name is specified in
3004 <name>.
3005
Sasha Pachev218f0642014-06-16 12:05:59 -06003006 - "replace-header" matches the regular expression in all occurrences of
3007 header field <name> according to <match-regex>, and replaces them with
3008 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3009 and work like in <fmt> arguments in "add-header". The match is only
3010 case-sensitive. It is important to understand that this action only
3011 considers whole header lines, regardless of the number of values they
3012 may contain. This usage is suited to headers naturally containing commas
3013 in their value, such as If-Modified-Since and so on.
3014
3015 Example:
3016
3017 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
3018
3019 applied to:
3020
3021 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3022
3023 outputs:
3024
3025 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3026
3027 assuming the backend IP is 192.168.1.20
3028
3029 - "replace-value" works like "replace-header" except that it matches the
3030 regex against every comma-delimited value of the header field <name>
3031 instead of the entire header. This is suited for all headers which are
3032 allowed to carry more than one value. An example could be the Accept
3033 header.
3034
3035 Example:
3036
3037 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
3038
3039 applied to:
3040
3041 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
3042
3043 outputs:
3044
3045 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
3046
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003047 - "set-nice" sets the "nice" factor of the current request being processed.
3048 It only has effect against the other requests being processed at the same
3049 time. The default value is 0, unless altered by the "nice" setting on the
3050 "bind" line. The accepted range is -1024..1024. The higher the value, the
3051 nicest the request will be. Lower values will make the request more
3052 important than other ones. This can be useful to improve the speed of
3053 some requests, or lower the priority of non-important requests. Using
3054 this setting without prior experimentation can cause some major slowdown.
3055
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003056 - "set-log-level" is used to change the log level of the current request
3057 when a certain condition is met. Valid levels are the 8 syslog levels
3058 (see the "log" keyword) plus the special level "silent" which disables
3059 logging for this request. This rule is not final so the last matching
3060 rule wins. This rule can be useful to disable health checks coming from
3061 another equipment.
3062
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003063 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3064 the client to the value passed in <tos> on platforms which support this.
3065 This value represents the whole 8 bits of the IP TOS field, and can be
3066 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3067 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3068 bits are always 0. This can be used to adjust some routing behaviour on
3069 border routers based on some information from the request. See RFC 2474,
3070 2597, 3260 and 4594 for more information.
3071
Willy Tarreau51347ed2013-06-11 19:34:13 +02003072 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3073 client to the value passed in <mark> on platforms which support it. This
3074 value is an unsigned 32 bit value which can be matched by netfilter and
3075 by the routing table. It can be expressed both in decimal or hexadecimal
3076 format (prefixed by "0x"). This can be useful to force certain packets to
3077 take a different route (for example a cheaper network path for bulk
3078 downloads). This works on Linux kernels 2.6.32 and above and requires
3079 admin privileges.
3080
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003081 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3082 from a file (even a dummy empty file). The file name of the ACL to be
3083 updated is passed between parentheses. It takes one argument: <key fmt>,
3084 which follows log-format rules, to collect content of the new entry. It
3085 performs a lookup in the ACL before insertion, to avoid duplicated (or
3086 more) values. This lookup is done by a linear search and can be expensive
3087 with large lists! It is the equivalent of the "add acl" command from the
3088 stats socket, but can be triggered by an HTTP request.
3089
3090 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3091 from a file (even a dummy empty file). The file name of the ACL to be
3092 updated is passed between parentheses. It takes one argument: <key fmt>,
3093 which follows log-format rules, to collect content of the entry to delete.
3094 It is the equivalent of the "del acl" command from the stats socket, but
3095 can be triggered by an HTTP request.
3096
3097 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3098 from a file (even a dummy empty file). The file name of the MAP 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 takes one argument: "file name" It is the equivalent of the "del map"
3102 command from the stats socket, but can be triggered by an HTTP request.
3103
3104 - "set-map" is used to add a new entry into 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 2 arguments: <key fmt>,
3107 which follows log-format rules, used to collect MAP key, and <value fmt>,
3108 which follows log-format rules, used to collect content for the new entry.
3109 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3110 more) values. This lookup is done by a linear search and can be expensive
3111 with large lists! It is the equivalent of the "set map" command from the
3112 stats socket, but can be triggered by an HTTP request.
3113
Willy Tarreau20b0de52012-12-24 15:45:22 +01003114 There is no limit to the number of http-request statements per instance.
3115
3116 It is important to know that http-request rules are processed very early in
3117 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
3118 rules. That way, headers added by "add-header"/"set-header" are visible by
3119 almost all further ACL rules.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003120
3121 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01003122 acl nagios src 192.168.129.3
3123 acl local_net src 192.168.0.0/16
3124 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003125
Cyril Bonté78caf842010-03-10 22:41:43 +01003126 http-request allow if nagios
3127 http-request allow if local_net auth_ok
3128 http-request auth realm Gimme if local_net auth_ok
3129 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003130
Cyril Bonté78caf842010-03-10 22:41:43 +01003131 Example:
3132 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01003133 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003134
Willy Tarreau20b0de52012-12-24 15:45:22 +01003135 Example:
3136 http-request set-header X-Haproxy-Current-Date %T
3137 http-request set-header X-SSL %[ssl_fc]
3138 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id]
3139 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
3140 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
3141 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
3142 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
3143 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
3144 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
3145
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003146 Example:
3147 acl key req.hdr(X-Add-Acl-Key) -m found
3148 acl add path /addacl
3149 acl del path /delacl
3150
3151 acl myhost hdr(Host) -f myhost.lst
3152
3153 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
3154 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
3155
3156 Example:
3157 acl value req.hdr(X-Value) -m found
3158 acl setmap path /setmap
3159 acl delmap path /delmap
3160
3161 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3162
3163 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
3164 http-request del-map(map.lst) %[src] if delmap
3165
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02003166 See also : "stats http-request", section 3.4 about userlists and section 7
3167 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01003168
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003169http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003170 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003171 replace-header <name> <regex-match> <replace-fmt> |
3172 replace-value <name> <regex-match> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003173 set-log-level <level> | set-mark <mark> | set-tos <tos> |
3174 add-acl(<file name>) <key fmt> |
3175 del-acl(<file name>) <key fmt> |
3176 del-map(<file name>) <key fmt> |
3177 set-map(<file name>) <key fmt> <value fmt>
3178 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02003179 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003180 Access control for Layer 7 responses
3181
3182 May be used in sections: defaults | frontend | listen | backend
3183 no | yes | yes | yes
3184
3185 The http-response statement defines a set of rules which apply to layer 7
3186 processing. The rules are evaluated in their declaration order when they are
3187 met in a frontend, listen or backend section. Any rule may optionally be
3188 followed by an ACL-based condition, in which case it will only be evaluated
3189 if the condition is true. Since these rules apply on responses, the backend
3190 rules are applied first, followed by the frontend's rules.
3191
3192 The first keyword is the rule's action. Currently supported actions include :
3193 - "allow" : this stops the evaluation of the rules and lets the response
3194 pass the check. No further "http-response" rules are evaluated for the
3195 current section.
3196
3197 - "deny" : this stops the evaluation of the rules and immediately rejects
3198 the response and emits an HTTP 502 error. No further "http-response"
3199 rules are evaluated.
3200
3201 - "add-header" appends an HTTP header field whose name is specified in
3202 <name> and whose value is defined by <fmt> which follows the log-format
3203 rules (see Custom Log Format in section 8.2.4). This may be used to send
3204 a cookie to a client for example, or to pass some internal information.
3205 This rule is not final, so it is possible to add other similar rules.
3206 Note that header addition is performed immediately, so one rule might
3207 reuse the resulting header from a previous rule.
3208
3209 - "set-header" does the same as "add-header" except that the header name
3210 is first removed if it existed. This is useful when passing security
3211 information to the server, where the header must not be manipulated by
3212 external users.
3213
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003214 - "del-header" removes all HTTP header fields whose name is specified in
3215 <name>.
3216
Sasha Pachev218f0642014-06-16 12:05:59 -06003217 - "replace-header" matches the regular expression in all occurrences of
3218 header field <name> according to <match-regex>, and replaces them with
3219 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3220 and work like in <fmt> arguments in "add-header". The match is only
3221 case-sensitive. It is important to understand that this action only
3222 considers whole header lines, regardless of the number of values they
3223 may contain. This usage is suited to headers naturally containing commas
3224 in their value, such as Set-Cookie, Expires and so on.
3225
3226 Example:
3227
3228 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
3229
3230 applied to:
3231
3232 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
3233
3234 outputs:
3235
3236 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
3237
3238 assuming the backend IP is 192.168.1.20.
3239
3240 - "replace-value" works like "replace-header" except that it matches the
3241 regex against every comma-delimited value of the header field <name>
3242 instead of the entire header. This is suited for all headers which are
3243 allowed to carry more than one value. An example could be the Accept
3244 header.
3245
3246 Example:
3247
3248 http-response replace-value Cache-control ^public$ private
3249
3250 applied to:
3251
3252 Cache-Control: max-age=3600, public
3253
3254 outputs:
3255
3256 Cache-Control: max-age=3600, private
3257
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003258 - "set-nice" sets the "nice" factor of the current request being processed.
3259 It only has effect against the other requests being processed at the same
3260 time. The default value is 0, unless altered by the "nice" setting on the
3261 "bind" line. The accepted range is -1024..1024. The higher the value, the
3262 nicest the request will be. Lower values will make the request more
3263 important than other ones. This can be useful to improve the speed of
3264 some requests, or lower the priority of non-important requests. Using
3265 this setting without prior experimentation can cause some major slowdown.
3266
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003267 - "set-log-level" is used to change the log level of the current request
3268 when a certain condition is met. Valid levels are the 8 syslog levels
3269 (see the "log" keyword) plus the special level "silent" which disables
3270 logging for this request. This rule is not final so the last matching
3271 rule wins. This rule can be useful to disable health checks coming from
3272 another equipment.
3273
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003274 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3275 the client to the value passed in <tos> on platforms which support this.
3276 This value represents the whole 8 bits of the IP TOS field, and can be
3277 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3278 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3279 bits are always 0. This can be used to adjust some routing behaviour on
3280 border routers based on some information from the request. See RFC 2474,
3281 2597, 3260 and 4594 for more information.
3282
Willy Tarreau51347ed2013-06-11 19:34:13 +02003283 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3284 client to the value passed in <mark> on platforms which support it. This
3285 value is an unsigned 32 bit value which can be matched by netfilter and
3286 by the routing table. It can be expressed both in decimal or hexadecimal
3287 format (prefixed by "0x"). This can be useful to force certain packets to
3288 take a different route (for example a cheaper network path for bulk
3289 downloads). This works on Linux kernels 2.6.32 and above and requires
3290 admin privileges.
3291
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003292 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3293 from a file (even a dummy empty file). The file name of the ACL to be
3294 updated is passed between parentheses. It takes one argument: <key fmt>,
3295 which follows log-format rules, to collect content of the new entry. It
3296 performs a lookup in the ACL before insertion, to avoid duplicated (or
3297 more) values. This lookup is done by a linear search and can be expensive
3298 with large lists! It is the equivalent of the "add acl" command from the
3299 stats socket, but can be triggered by an HTTP response.
3300
3301 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3302 from a file (even a dummy empty file). The file name of the ACL to be
3303 updated is passed between parentheses. It takes one argument: <key fmt>,
3304 which follows log-format rules, to collect content of the entry to delete.
3305 It is the equivalent of the "del acl" command from the stats socket, but
3306 can be triggered by an HTTP response.
3307
3308 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3309 from a file (even a dummy empty file). The file name of the MAP 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 takes one argument: "file name" It is the equivalent of the "del map"
3313 command from the stats socket, but can be triggered by an HTTP response.
3314
3315 - "set-map" is used to add a new entry into 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 2 arguments: <key fmt>,
3318 which follows log-format rules, used to collect MAP key, and <value fmt>,
3319 which follows log-format rules, used to collect content for the new entry.
3320 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3321 more) values. This lookup is done by a linear search and can be expensive
3322 with large lists! It is the equivalent of the "set map" command from the
3323 stats socket, but can be triggered by an HTTP response.
3324
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003325 There is no limit to the number of http-response statements per instance.
3326
Godbach09250262013-07-02 01:19:15 +08003327 It is important to know that http-response rules are processed very early in
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003328 the HTTP processing, before "reqdel" or "reqrep" rules. That way, headers
3329 added by "add-header"/"set-header" are visible by almost all further ACL
3330 rules.
3331
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003332 Example:
3333 acl key_acl res.hdr(X-Acl-Key) -m found
3334
3335 acl myhost hdr(Host) -f myhost.lst
3336
3337 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3338 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3339
3340 Example:
3341 acl value res.hdr(X-Value) -m found
3342
3343 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3344
3345 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
3346 http-response del-map(map.lst) %[src] if ! value
3347
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003348 See also : "http-request", section 3.4 about userlists and section 7 about
3349 ACL usage.
3350
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02003351
Mark Lamourinec2247f02012-01-04 13:02:01 -05003352http-send-name-header [<header>]
3353 Add the server name to a request. Use the header string given by <header>
3354
3355 May be used in sections: defaults | frontend | listen | backend
3356 yes | no | yes | yes
3357
3358 Arguments :
3359
3360 <header> The header string to use to send the server name
3361
3362 The "http-send-name-header" statement causes the name of the target
3363 server to be added to the headers of an HTTP request. The name
3364 is added with the header string proved.
3365
3366 See also : "server"
3367
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003368id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02003369 Set a persistent ID to a proxy.
3370 May be used in sections : defaults | frontend | listen | backend
3371 no | yes | yes | yes
3372 Arguments : none
3373
3374 Set a persistent ID for the proxy. This ID must be unique and positive.
3375 An unused ID will automatically be assigned if unset. The first assigned
3376 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003377
3378
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003379ignore-persist { if | unless } <condition>
3380 Declare a condition to ignore persistence
3381 May be used in sections: defaults | frontend | listen | backend
3382 no | yes | yes | yes
3383
3384 By default, when cookie persistence is enabled, every requests containing
3385 the cookie are unconditionally persistent (assuming the target server is up
3386 and running).
3387
3388 The "ignore-persist" statement allows one to declare various ACL-based
3389 conditions which, when met, will cause a request to ignore persistence.
3390 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003391 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003392 persistence for a specific User-Agent (for example, some web crawler bots).
3393
3394 Combined with "appsession", it can also help reduce HAProxy memory usage, as
3395 the appsession table won't grow if persistence is ignored.
3396
3397 The persistence is ignored when an "if" condition is met, or unless an
3398 "unless" condition is met.
3399
3400 See also : "force-persist", "cookie", and section 7 about ACL usage.
3401
3402
Willy Tarreau2769aa02007-12-27 18:26:09 +01003403log global
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003404log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02003405no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01003406 Enable per-instance logging of events and traffic.
3407 May be used in sections : defaults | frontend | listen | backend
3408 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02003409
3410 Prefix :
3411 no should be used when the logger list must be flushed. For example,
3412 if you don't want to inherit from the default logger list. This
3413 prefix does not allow arguments.
3414
Willy Tarreau2769aa02007-12-27 18:26:09 +01003415 Arguments :
3416 global should be used when the instance's logging parameters are the
3417 same as the global ones. This is the most common usage. "global"
3418 replaces <address>, <facility> and <level> with those of the log
3419 entries found in the "global" section. Only one "log global"
3420 statement may be used per instance, and this form takes no other
3421 parameter.
3422
3423 <address> indicates where to send the logs. It takes the same format as
3424 for the "global" section's logs, and can be one of :
3425
3426 - An IPv4 address optionally followed by a colon (':') and a UDP
3427 port. If no port is specified, 514 is used by default (the
3428 standard syslog port).
3429
David du Colombier24bb5f52011-03-17 10:40:23 +01003430 - An IPv6 address followed by a colon (':') and optionally a UDP
3431 port. If no port is specified, 514 is used by default (the
3432 standard syslog port).
3433
Willy Tarreau2769aa02007-12-27 18:26:09 +01003434 - A filesystem path to a UNIX domain socket, keeping in mind
3435 considerations for chroot (be sure the path is accessible
3436 inside the chroot) and uid/gid (be sure the path is
3437 appropriately writeable).
3438
Willy Tarreaudad36a32013-03-11 01:20:04 +01003439 Any part of the address string may reference any number of
3440 environment variables by preceding their name with a dollar
3441 sign ('$') and optionally enclosing them with braces ('{}'),
3442 similarly to what is done in Bourne shell.
3443
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003444 <length> is an optional maximum line length. Log lines larger than this
3445 value will be truncated before being sent. The reason is that
3446 syslog servers act differently on log line length. All servers
3447 support the default value of 1024, but some servers simply drop
3448 larger lines while others do log them. If a server supports long
3449 lines, it may make sense to set this value here in order to avoid
3450 truncating long lines. Similarly, if a server drops long lines,
3451 it is preferable to truncate them before sending them. Accepted
3452 values are 80 to 65535 inclusive. The default value of 1024 is
3453 generally fine for all standard usages. Some specific cases of
3454 long captures or JSON-formated logs may require larger values.
3455
Willy Tarreau2769aa02007-12-27 18:26:09 +01003456 <facility> must be one of the 24 standard syslog facilities :
3457
3458 kern user mail daemon auth syslog lpr news
3459 uucp cron auth2 ftp ntp audit alert cron2
3460 local0 local1 local2 local3 local4 local5 local6 local7
3461
3462 <level> is optional and can be specified to filter outgoing messages. By
3463 default, all messages are sent. If a level is specified, only
3464 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003465 will be sent. An optional minimum level can be specified. If it
3466 is set, logs emitted with a more severe level than this one will
3467 be capped to this level. This is used to avoid sending "emerg"
3468 messages on all terminals on some default syslog configurations.
3469 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003470
3471 emerg alert crit err warning notice info debug
3472
William Lallemand0f99e342011-10-12 17:50:54 +02003473 It is important to keep in mind that it is the frontend which decides what to
3474 log from a connection, and that in case of content switching, the log entries
3475 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003476
3477 However, backend log declaration define how and where servers status changes
3478 will be logged. Level "notice" will be used to indicate a server going up,
3479 "warning" will be used for termination signals and definitive service
3480 termination, and "alert" will be used for when a server goes down.
3481
3482 Note : According to RFC3164, messages are truncated to 1024 bytes before
3483 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003484
3485 Example :
3486 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003487 log 127.0.0.1:514 local0 notice # only send important events
3488 log 127.0.0.1:514 local0 notice notice # same but limit output level
Willy Tarreaudad36a32013-03-11 01:20:04 +01003489 log ${LOCAL_SYSLOG}:514 local0 notice # send to local server
3490
Willy Tarreau2769aa02007-12-27 18:26:09 +01003491
William Lallemand48940402012-01-30 16:47:22 +01003492log-format <string>
Willy Tarreau14104732015-01-07 14:55:17 +01003493 Specifies the log format string to use for traffic logs
3494 May be used in sections: defaults | frontend | listen | backend
3495 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01003496
Willy Tarreau14104732015-01-07 14:55:17 +01003497 This directive specifies the log format string that will be used for all logs
3498 resulting from traffic passing through the frontend using this line. If the
3499 directive is used in a defaults section, all subsequent frontends will use
3500 the same log format. Please see section 8.2.4 which covers the log format
3501 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01003502
Willy Tarreau2769aa02007-12-27 18:26:09 +01003503
Willy Tarreauc35362a2014-04-25 13:58:37 +02003504max-keep-alive-queue <value>
3505 Set the maximum server queue size for maintaining keep-alive connections
3506 May be used in sections: defaults | frontend | listen | backend
3507 yes | no | yes | yes
3508
3509 HTTP keep-alive tries to reuse the same server connection whenever possible,
3510 but sometimes it can be counter-productive, for example if a server has a lot
3511 of connections while other ones are idle. This is especially true for static
3512 servers.
3513
3514 The purpose of this setting is to set a threshold on the number of queued
3515 connections at which haproxy stops trying to reuse the same server and prefers
3516 to find another one. The default value, -1, means there is no limit. A value
3517 of zero means that keep-alive requests will never be queued. For very close
3518 servers which can be reached with a low latency and which are not sensible to
3519 breaking keep-alive, a low value is recommended (eg: local static server can
3520 use a value of 10 or less). For remote servers suffering from a high latency,
3521 higher values might be needed to cover for the latency and/or the cost of
3522 picking a different server.
3523
3524 Note that this has no impact on responses which are maintained to the same
3525 server consecutively to a 401 response. They will still go to the same server
3526 even if they have to be queued.
3527
3528 See also : "option http-server-close", "option prefer-last-server", server
3529 "maxconn" and cookie persistence.
3530
3531
Willy Tarreau2769aa02007-12-27 18:26:09 +01003532maxconn <conns>
3533 Fix the maximum number of concurrent connections on a frontend
3534 May be used in sections : defaults | frontend | listen | backend
3535 yes | yes | yes | no
3536 Arguments :
3537 <conns> is the maximum number of concurrent connections the frontend will
3538 accept to serve. Excess connections will be queued by the system
3539 in the socket's listen queue and will be served once a connection
3540 closes.
3541
3542 If the system supports it, it can be useful on big sites to raise this limit
3543 very high so that haproxy manages connection queues, instead of leaving the
3544 clients with unanswered connection attempts. This value should not exceed the
3545 global maxconn. Also, keep in mind that a connection contains two buffers
3546 of 8kB each, as well as some other data resulting in about 17 kB of RAM being
3547 consumed per established connection. That means that a medium system equipped
3548 with 1GB of RAM can withstand around 40000-50000 concurrent connections if
3549 properly tuned.
3550
3551 Also, when <conns> is set to large values, it is possible that the servers
3552 are not sized to accept such loads, and for this reason it is generally wise
3553 to assign them some reasonable connection limits.
3554
Vincent Bernat6341be52012-06-27 17:18:30 +02003555 By default, this value is set to 2000.
3556
Willy Tarreau2769aa02007-12-27 18:26:09 +01003557 See also : "server", global section's "maxconn", "fullconn"
3558
3559
3560mode { tcp|http|health }
3561 Set the running mode or protocol of the instance
3562 May be used in sections : defaults | frontend | listen | backend
3563 yes | yes | yes | yes
3564 Arguments :
3565 tcp The instance will work in pure TCP mode. A full-duplex connection
3566 will be established between clients and servers, and no layer 7
3567 examination will be performed. This is the default mode. It
3568 should be used for SSL, SSH, SMTP, ...
3569
3570 http The instance will work in HTTP mode. The client request will be
3571 analyzed in depth before connecting to any server. Any request
3572 which is not RFC-compliant will be rejected. Layer 7 filtering,
3573 processing and switching will be possible. This is the mode which
3574 brings HAProxy most of its value.
3575
3576 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02003577 to incoming connections and close the connection. Alternatively,
3578 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
3579 instead. Nothing will be logged in either case. This mode is used
3580 to reply to external components health checks. This mode is
3581 deprecated and should not be used anymore as it is possible to do
3582 the same and even better by combining TCP or HTTP modes with the
3583 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003584
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003585 When doing content switching, it is mandatory that the frontend and the
3586 backend are in the same mode (generally HTTP), otherwise the configuration
3587 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003588
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003589 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003590 defaults http_instances
3591 mode http
3592
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003593 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003594
Willy Tarreau0ba27502007-12-24 16:55:16 +01003595
Cyril Bontéf0c60612010-02-06 14:44:47 +01003596monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01003597 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003598 May be used in sections : defaults | frontend | listen | backend
3599 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01003600 Arguments :
3601 if <cond> the monitor request will fail if the condition is satisfied,
3602 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003603 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01003604 are met, for instance a low number of servers both in a
3605 backend and its backup.
3606
3607 unless <cond> the monitor request will succeed only if the condition is
3608 satisfied, and will fail otherwise. Such a condition may be
3609 based on a test on the presence of a minimum number of active
3610 servers in a list of backends.
3611
3612 This statement adds a condition which can force the response to a monitor
3613 request to report a failure. By default, when an external component queries
3614 the URI dedicated to monitoring, a 200 response is returned. When one of the
3615 conditions above is met, haproxy will return 503 instead of 200. This is
3616 very useful to report a site failure to an external component which may base
3617 routing advertisements between multiple sites on the availability reported by
3618 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003619 criterion. Note that "monitor fail" only works in HTTP mode. Both status
3620 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003621
3622 Example:
3623 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01003624 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01003625 acl site_dead nbsrv(dynamic) lt 2
3626 acl site_dead nbsrv(static) lt 2
3627 monitor-uri /site_alive
3628 monitor fail if site_dead
3629
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003630 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003631
3632
3633monitor-net <source>
3634 Declare a source network which is limited to monitor requests
3635 May be used in sections : defaults | frontend | listen | backend
3636 yes | yes | yes | no
3637 Arguments :
3638 <source> is the source IPv4 address or network which will only be able to
3639 get monitor responses to any request. It can be either an IPv4
3640 address, a host name, or an address followed by a slash ('/')
3641 followed by a mask.
3642
3643 In TCP mode, any connection coming from a source matching <source> will cause
3644 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003645 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01003646 forwarding the connection to a remote server.
3647
3648 In HTTP mode, a connection coming from a source matching <source> will be
3649 accepted, the following response will be sent without waiting for a request,
3650 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
3651 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02003652 running without forwarding the request to a backend server. Note that this
3653 response is sent in raw format, without any transformation. This is important
3654 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003655
Willy Tarreau82569f92012-09-27 23:48:56 +02003656 Monitor requests are processed very early, just after tcp-request connection
3657 ACLs which are the only ones able to block them. These connections are short
3658 lived and never wait for any data from the client. They cannot be logged, and
3659 it is the intended purpose. They are only used to report HAProxy's health to
3660 an upper component, nothing more. Please note that "monitor fail" rules do
3661 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01003662
Willy Tarreau95cd2832010-03-04 23:36:33 +01003663 Last, please note that only one "monitor-net" statement can be specified in
3664 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003665
Willy Tarreau2769aa02007-12-27 18:26:09 +01003666 Example :
3667 # addresses .252 and .253 are just probing us.
3668 frontend www
3669 monitor-net 192.168.0.252/31
3670
3671 See also : "monitor fail", "monitor-uri"
3672
3673
3674monitor-uri <uri>
3675 Intercept a URI used by external components' monitor requests
3676 May be used in sections : defaults | frontend | listen | backend
3677 yes | yes | yes | no
3678 Arguments :
3679 <uri> is the exact URI which we want to intercept to return HAProxy's
3680 health status instead of forwarding the request.
3681
3682 When an HTTP request referencing <uri> will be received on a frontend,
3683 HAProxy will not forward it nor log it, but instead will return either
3684 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
3685 conditions defined with "monitor fail". This is normally enough for any
3686 front-end HTTP probe to detect that the service is UP and running without
3687 forwarding the request to a backend server. Note that the HTTP method, the
3688 version and all headers are ignored, but the request must at least be valid
3689 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
3690
3691 Monitor requests are processed very early. It is not possible to block nor
3692 divert them using ACLs. They cannot be logged either, and it is the intended
3693 purpose. They are only used to report HAProxy's health to an upper component,
3694 nothing more. However, it is possible to add any number of conditions using
3695 "monitor fail" and ACLs so that the result can be adjusted to whatever check
3696 can be imagined (most often the number of available servers in a backend).
3697
3698 Example :
3699 # Use /haproxy_test to report haproxy's status
3700 frontend www
3701 mode http
3702 monitor-uri /haproxy_test
3703
3704 See also : "monitor fail", "monitor-net"
3705
Willy Tarreau0ba27502007-12-24 16:55:16 +01003706
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003707option abortonclose
3708no option abortonclose
3709 Enable or disable early dropping of aborted requests pending in queues.
3710 May be used in sections : defaults | frontend | listen | backend
3711 yes | no | yes | yes
3712 Arguments : none
3713
3714 In presence of very high loads, the servers will take some time to respond.
3715 The per-instance connection queue will inflate, and the response time will
3716 increase respective to the size of the queue times the average per-session
3717 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01003718 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003719 the queue, and slowing down other users, and the servers as well, because the
3720 request will eventually be served, then aborted at the first error
3721 encountered while delivering the response.
3722
3723 As there is no way to distinguish between a full STOP and a simple output
3724 close on the client side, HTTP agents should be conservative and consider
3725 that the client might only have closed its output channel while waiting for
3726 the response. However, this introduces risks of congestion when lots of users
3727 do the same, and is completely useless nowadays because probably no client at
3728 all will close the session while waiting for the response. Some HTTP agents
Willy Tarreaud72758d2010-01-12 10:42:19 +01003729 support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003730 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01003731 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003732 of being the single component to break rare but valid traffic is extremely
3733 low, which adds to the temptation to be able to abort a session early while
3734 still not served and not pollute the servers.
3735
3736 In HAProxy, the user can choose the desired behaviour using the option
3737 "abortonclose". By default (without the option) the behaviour is HTTP
3738 compliant and aborted requests will be served. But when the option is
3739 specified, a session with an incoming channel closed will be aborted while
3740 it is still possible, either pending in the queue for a connection slot, or
3741 during the connection establishment if the server has not yet acknowledged
3742 the connection request. This considerably reduces the queue size and the load
3743 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01003744 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003745
3746 If this option has been enabled in a "defaults" section, it can be disabled
3747 in a specific instance by prepending the "no" keyword before it.
3748
3749 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
3750
3751
Willy Tarreau4076a152009-04-02 15:18:36 +02003752option accept-invalid-http-request
3753no option accept-invalid-http-request
3754 Enable or disable relaxing of HTTP request parsing
3755 May be used in sections : defaults | frontend | listen | backend
3756 yes | yes | yes | no
3757 Arguments : none
3758
Willy Tarreau55645552015-05-01 13:26:00 +02003759 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02003760 means that invalid characters in header names are not permitted and cause an
3761 error to be returned to the client. This is the desired behaviour as such
3762 forbidden characters are essentially used to build attacks exploiting server
3763 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3764 server will emit invalid header names for whatever reason (configuration,
3765 implementation) and the issue will not be immediately fixed. In such a case,
3766 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01003767 even if that does not make sense, by specifying this option. Similarly, the
3768 list of characters allowed to appear in a URI is well defined by RFC3986, and
3769 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
3770 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
3771 not allowed at all. Haproxy always blocks a number of them (0..32, 127). The
Willy Tarreau55645552015-05-01 13:26:00 +02003772 remaining ones are blocked by default unless this option is enabled. This
3773 option also relaxes the test on the HTTP version format, it allows multiple
3774 digits for both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02003775
3776 This option should never be enabled by default as it hides application bugs
3777 and open security breaches. It should only be deployed after a problem has
3778 been confirmed.
3779
3780 When this option is enabled, erroneous header names will still be accepted in
3781 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01003782 analysis using the "show errors" request on the UNIX stats socket. Similarly,
3783 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02003784 also helps confirming that the issue has been solved.
3785
3786 If this option has been enabled in a "defaults" section, it can be disabled
3787 in a specific instance by prepending the "no" keyword before it.
3788
3789 See also : "option accept-invalid-http-response" and "show errors" on the
3790 stats socket.
3791
3792
3793option accept-invalid-http-response
3794no option accept-invalid-http-response
3795 Enable or disable relaxing of HTTP response parsing
3796 May be used in sections : defaults | frontend | listen | backend
3797 yes | no | yes | yes
3798 Arguments : none
3799
Willy Tarreau55645552015-05-01 13:26:00 +02003800 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02003801 means that invalid characters in header names are not permitted and cause an
3802 error to be returned to the client. This is the desired behaviour as such
3803 forbidden characters are essentially used to build attacks exploiting server
3804 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3805 server will emit invalid header names for whatever reason (configuration,
3806 implementation) and the issue will not be immediately fixed. In such a case,
3807 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau55645552015-05-01 13:26:00 +02003808 even if that does not make sense, by specifying this option. This option also
3809 relaxes the test on the HTTP version format, it allows multiple digits for
3810 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02003811
3812 This option should never be enabled by default as it hides application bugs
3813 and open security breaches. It should only be deployed after a problem has
3814 been confirmed.
3815
3816 When this option is enabled, erroneous header names will still be accepted in
3817 responses, but the complete response will be captured in order to permit
3818 later analysis using the "show errors" request on the UNIX stats socket.
3819 Doing this also helps confirming that the issue has been solved.
3820
3821 If this option has been enabled in a "defaults" section, it can be disabled
3822 in a specific instance by prepending the "no" keyword before it.
3823
3824 See also : "option accept-invalid-http-request" and "show errors" on the
3825 stats socket.
3826
3827
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003828option allbackups
3829no option allbackups
3830 Use either all backup servers at a time or only the first one
3831 May be used in sections : defaults | frontend | listen | backend
3832 yes | no | yes | yes
3833 Arguments : none
3834
3835 By default, the first operational backup server gets all traffic when normal
3836 servers are all down. Sometimes, it may be preferred to use multiple backups
3837 at once, because one will not be enough. When "option allbackups" is enabled,
3838 the load balancing will be performed among all backup servers when all normal
3839 ones are unavailable. The same load balancing algorithm will be used and the
3840 servers' weights will be respected. Thus, there will not be any priority
3841 order between the backup servers anymore.
3842
3843 This option is mostly used with static server farms dedicated to return a
3844 "sorry" page when an application is completely offline.
3845
3846 If this option has been enabled in a "defaults" section, it can be disabled
3847 in a specific instance by prepending the "no" keyword before it.
3848
3849
3850option checkcache
3851no option checkcache
Godbach7056a352013-12-11 20:01:07 +08003852 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003853 May be used in sections : defaults | frontend | listen | backend
3854 yes | no | yes | yes
3855 Arguments : none
3856
3857 Some high-level frameworks set application cookies everywhere and do not
3858 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003859 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003860 high risk of session crossing or stealing between users traversing the same
3861 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003862 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003863
3864 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003865 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01003866 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003867 response to check if there's a risk of caching a cookie on a client-side
3868 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01003869 to the client are :
3870 - all those without "Set-Cookie" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003871 - all those with a return code other than 200, 203, 206, 300, 301, 410,
Willy Tarreau198a7442008-01-17 12:05:32 +01003872 provided that the server has not set a "Cache-control: public" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003873 - all those that come from a POST request, provided that the server has not
3874 set a 'Cache-Control: public' header ;
3875 - those with a 'Pragma: no-cache' header
3876 - those with a 'Cache-control: private' header
3877 - those with a 'Cache-control: no-store' header
3878 - those with a 'Cache-control: max-age=0' header
3879 - those with a 'Cache-control: s-maxage=0' header
3880 - those with a 'Cache-control: no-cache' header
3881 - those with a 'Cache-control: no-cache="set-cookie"' header
3882 - those with a 'Cache-control: no-cache="set-cookie,' header
3883 (allowing other fields after set-cookie)
3884
3885 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01003886 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003887 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003888 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003889 that admins are informed that there's something to be fixed.
3890
3891 Due to the high impact on the application, the application should be tested
3892 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003893 good practice to always activate it during tests, even if it is not used in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003894 production, as it will report potentially dangerous application behaviours.
3895
3896 If this option has been enabled in a "defaults" section, it can be disabled
3897 in a specific instance by prepending the "no" keyword before it.
3898
3899
3900option clitcpka
3901no option clitcpka
3902 Enable or disable the sending of TCP keepalive packets on the client side
3903 May be used in sections : defaults | frontend | listen | backend
3904 yes | yes | yes | no
3905 Arguments : none
3906
3907 When there is a firewall or any session-aware component between a client and
3908 a server, and when the protocol involves very long sessions with long idle
3909 periods (eg: remote desktops), there is a risk that one of the intermediate
3910 components decides to expire a session which has remained idle for too long.
3911
3912 Enabling socket-level TCP keep-alives makes the system regularly send packets
3913 to the other end of the connection, leaving it active. The delay between
3914 keep-alive probes is controlled by the system only and depends both on the
3915 operating system and its tuning parameters.
3916
3917 It is important to understand that keep-alive packets are neither emitted nor
3918 received at the application level. It is only the network stacks which sees
3919 them. For this reason, even if one side of the proxy already uses keep-alives
3920 to maintain its connection alive, those keep-alive packets will not be
3921 forwarded to the other side of the proxy.
3922
3923 Please note that this has nothing to do with HTTP keep-alive.
3924
3925 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
3926 client side of a connection, which should help when session expirations are
3927 noticed between HAProxy and a client.
3928
3929 If this option has been enabled in a "defaults" section, it can be disabled
3930 in a specific instance by prepending the "no" keyword before it.
3931
3932 See also : "option srvtcpka", "option tcpka"
3933
3934
Willy Tarreau0ba27502007-12-24 16:55:16 +01003935option contstats
3936 Enable continuous traffic statistics updates
3937 May be used in sections : defaults | frontend | listen | backend
3938 yes | yes | yes | no
3939 Arguments : none
3940
3941 By default, counters used for statistics calculation are incremented
3942 only when a session finishes. It works quite well when serving small
3943 objects, but with big ones (for example large images or archives) or
3944 with A/V streaming, a graph generated from haproxy counters looks like
3945 a hedgehog. With this option enabled counters get incremented continuously,
3946 during a whole session. Recounting touches a hotpath directly so
3947 it is not enabled by default, as it has small performance impact (~0.5%).
3948
3949
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003950option dontlog-normal
3951no option dontlog-normal
3952 Enable or disable logging of normal, successful connections
3953 May be used in sections : defaults | frontend | listen | backend
3954 yes | yes | yes | no
3955 Arguments : none
3956
3957 There are large sites dealing with several thousand connections per second
3958 and for which logging is a major pain. Some of them are even forced to turn
3959 logs off and cannot debug production issues. Setting this option ensures that
3960 normal connections, those which experience no error, no timeout, no retry nor
3961 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
3962 mode, the response status code is checked and return codes 5xx will still be
3963 logged.
3964
3965 It is strongly discouraged to use this option as most of the time, the key to
3966 complex issues is in the normal logs which will not be logged here. If you
3967 need to separate logs, see the "log-separate-errors" option instead.
3968
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003969 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003970 logging.
3971
3972
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003973option dontlognull
3974no option dontlognull
3975 Enable or disable logging of null connections
3976 May be used in sections : defaults | frontend | listen | backend
3977 yes | yes | yes | no
3978 Arguments : none
3979
3980 In certain environments, there are components which will regularly connect to
3981 various systems to ensure that they are still alive. It can be the case from
3982 another load balancer as well as from monitoring systems. By default, even a
3983 simple port probe or scan will produce a log. If those connections pollute
3984 the logs too much, it is possible to enable option "dontlognull" to indicate
3985 that a connection on which no data has been transferred will not be logged,
Willy Tarreau1c3a6122015-05-01 15:37:53 +02003986 which typically corresponds to those probes. Note that errors will still be
3987 returned to the client and accounted for in the stats. If this is not what is
3988 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003989
3990 It is generally recommended not to use this option in uncontrolled
3991 environments (eg: internet), otherwise scans and other malicious activities
3992 would not be logged.
3993
3994 If this option has been enabled in a "defaults" section, it can be disabled
3995 in a specific instance by prepending the "no" keyword before it.
3996
Willy Tarreau1c3a6122015-05-01 15:37:53 +02003997 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
3998 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003999
4000
4001option forceclose
4002no option forceclose
4003 Enable or disable active connection closing after response is transferred.
4004 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01004005 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004006 Arguments : none
4007
4008 Some HTTP servers do not necessarily close the connections when they receive
4009 the "Connection: close" set by "option httpclose", and if the client does not
4010 close either, then the connection remains open till the timeout expires. This
4011 causes high number of simultaneous connections on the servers and shows high
4012 global session times in the logs.
4013
4014 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01004015 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01004016 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004017
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004018 This option may also be combined with "option http-pretend-keepalive", which
4019 will disable sending of the "Connection: close" header, but will still cause
4020 the connection to be closed once the whole response is received.
4021
Cyril Bonté653dcd62014-02-20 00:13:15 +01004022 This option disables and replaces any previous "option httpclose", "option
4023 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004024
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004025 If this option has been enabled in a "defaults" section, it can be disabled
4026 in a specific instance by prepending the "no" keyword before it.
4027
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004028 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004029
4030
Willy Tarreau87cf5142011-08-19 22:57:24 +02004031option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004032 Enable insertion of the X-Forwarded-For header to requests sent to servers
4033 May be used in sections : defaults | frontend | listen | backend
4034 yes | yes | yes | yes
4035 Arguments :
4036 <network> is an optional argument used to disable this option for sources
4037 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02004038 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01004039 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004040
4041 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
4042 their client address. This is sometimes annoying when the client's IP address
4043 is expected in server logs. To solve this problem, the well-known HTTP header
4044 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
4045 This header contains a value representing the client's IP address. Since this
4046 header is always appended at the end of the existing header list, the server
4047 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02004048 the server's manual to find how to enable use of this standard header. Note
4049 that only the last occurrence of the header must be used, since it is really
4050 possible that the client has already brought one.
4051
Willy Tarreaud72758d2010-01-12 10:42:19 +01004052 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02004053 the default "X-Forwarded-For". This can be useful where you might already
Willy Tarreaud72758d2010-01-12 10:42:19 +01004054 have a "X-Forwarded-For" header from a different application (eg: stunnel),
4055 and you need preserve it. Also if your backend server doesn't use the
Ross Westaf72a1d2008-08-03 10:51:45 +02004056 "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers
4057 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01004058
4059 Sometimes, a same HAProxy instance may be shared between a direct client
4060 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4061 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4062 header for a known source address or network by adding the "except" keyword
4063 followed by the network address. In this case, any source IP matching the
4064 network will not cause an addition of this header. Most common uses are with
4065 private networks or 127.0.0.1.
4066
Willy Tarreau87cf5142011-08-19 22:57:24 +02004067 Alternatively, the keyword "if-none" states that the header will only be
4068 added if it is not present. This should only be used in perfectly trusted
4069 environment, as this might cause a security issue if headers reaching haproxy
4070 are under the control of the end-user.
4071
Willy Tarreauc27debf2008-01-06 08:57:02 +01004072 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02004073 least one of them uses it, the header will be added. Note that the backend's
4074 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02004075 both are defined. In the case of the "if-none" argument, if at least one of
4076 the frontend or the backend does not specify it, it wants the addition to be
4077 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004078
Ross Westaf72a1d2008-08-03 10:51:45 +02004079 Examples :
Willy Tarreauc27debf2008-01-06 08:57:02 +01004080 # Public HTTP address also used by stunnel on the same machine
4081 frontend www
4082 mode http
4083 option forwardfor except 127.0.0.1 # stunnel already adds the header
4084
Ross Westaf72a1d2008-08-03 10:51:45 +02004085 # Those servers want the IP Address in X-Client
4086 backend www
4087 mode http
4088 option forwardfor header X-Client
4089
Willy Tarreau87cf5142011-08-19 22:57:24 +02004090 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01004091 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01004092
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004093
Willy Tarreau1c3a6122015-05-01 15:37:53 +02004094option http-ignore-probes
4095no option http-ignore-probes
4096 Enable or disable logging of null connections and request timeouts
4097 May be used in sections : defaults | frontend | listen | backend
4098 yes | yes | yes | no
4099 Arguments : none
4100
4101 Recently some browsers started to implement a "pre-connect" feature
4102 consisting in speculatively connecting to some recently visited web sites
4103 just in case the user would like to visit them. This results in many
4104 connections being established to web sites, which end up in 408 Request
4105 Timeout if the timeout strikes first, or 400 Bad Request when the browser
4106 decides to close them first. These ones pollute the log and feed the error
4107 counters. There was already "option dontlognull" but it's insufficient in
4108 this case. Instead, this option does the following things :
4109 - prevent any 400/408 message from being sent to the client if nothing
4110 was received over a connection before it was closed ;
4111 - prevent any log from being emitted in this situation ;
4112 - prevent any error counter from being incremented
4113
4114 That way the empty connection is silently ignored. Note that it is better
4115 not to use this unless it is clear that it is needed, because it will hide
4116 real problems. The most common reason for not receiving a request and seeing
4117 a 408 is due to an MTU inconsistency between the client and an intermediary
4118 element such as a VPN, which blocks too large packets. These issues are
4119 generally seen with POST requests as well as GET with large cookies. The logs
4120 are often the only way to detect them.
4121
4122 If this option has been enabled in a "defaults" section, it can be disabled
4123 in a specific instance by prepending the "no" keyword before it.
4124
4125 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
4126
4127
Willy Tarreau16bfb022010-01-16 19:48:41 +01004128option http-keep-alive
4129no option http-keep-alive
4130 Enable or disable HTTP keep-alive from client to server
4131 May be used in sections : defaults | frontend | listen | backend
4132 yes | yes | yes | yes
4133 Arguments : none
4134
Willy Tarreau70dffda2014-01-30 03:07:23 +01004135 By default HAProxy operates in keep-alive mode with regards to persistent
4136 connections: for each connection it processes each request and response, and
4137 leaves the connection idle on both sides between the end of a response and the
4138 start of a new request. This mode may be changed by several options such as
4139 "option http-server-close", "option forceclose", "option httpclose" or
4140 "option http-tunnel". This option allows to set back the keep-alive mode,
4141 which can be useful when another mode was used in a defaults section.
4142
4143 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
4144 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01004145 network) and the fastest session reuse on the server side at the expense
4146 of maintaining idle connections to the servers. In general, it is possible
4147 with this option to achieve approximately twice the request rate that the
4148 "http-server-close" option achieves on small objects. There are mainly two
4149 situations where this option may be useful :
4150
4151 - when the server is non-HTTP compliant and authenticates the connection
4152 instead of requests (eg: NTLM authentication)
4153
4154 - when the cost of establishing the connection to the server is significant
4155 compared to the cost of retrieving the associated object from the server.
4156
4157 This last case can happen when the server is a fast static server of cache.
4158 In this case, the server will need to be properly tuned to support high enough
4159 connection counts because connections will last until the client sends another
4160 request.
4161
4162 If the client request has to go to another backend or another server due to
4163 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01004164 immediately be closed and a new one re-opened. Option "prefer-last-server" is
4165 available to try optimize server selection so that if the server currently
4166 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01004167
4168 In general it is preferred to use "option http-server-close" with application
4169 servers, and some static servers might benefit from "option http-keep-alive".
4170
4171 At the moment, logs will not indicate whether requests came from the same
4172 session or not. The accept date reported in the logs corresponds to the end
4173 of the previous request, and the request time corresponds to the time spent
4174 waiting for a new request. The keep-alive request time is still bound to the
4175 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4176 not set.
4177
Cyril Bonté653dcd62014-02-20 00:13:15 +01004178 This option disables and replaces any previous "option httpclose", "option
4179 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01004180 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01004181 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004182
4183 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01004184 "option prefer-last-server", "option http-pretend-keepalive",
4185 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004186
4187
Willy Tarreau96e31212011-05-30 18:10:30 +02004188option http-no-delay
4189no option http-no-delay
4190 Instruct the system to favor low interactive delays over performance in HTTP
4191 May be used in sections : defaults | frontend | listen | backend
4192 yes | yes | yes | yes
4193 Arguments : none
4194
4195 In HTTP, each payload is unidirectional and has no notion of interactivity.
4196 Any agent is expected to queue data somewhat for a reasonably low delay.
4197 There are some very rare server-to-server applications that abuse the HTTP
4198 protocol and expect the payload phase to be highly interactive, with many
4199 interleaved data chunks in both directions within a single request. This is
4200 absolutely not supported by the HTTP specification and will not work across
4201 most proxies or servers. When such applications attempt to do this through
4202 haproxy, it works but they will experience high delays due to the network
4203 optimizations which favor performance by instructing the system to wait for
4204 enough data to be available in order to only send full packets. Typical
4205 delays are around 200 ms per round trip. Note that this only happens with
4206 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
4207 affected.
4208
4209 When "option http-no-delay" is present in either the frontend or the backend
4210 used by a connection, all such optimizations will be disabled in order to
4211 make the exchanges as fast as possible. Of course this offers no guarantee on
4212 the functionality, as it may break at any other place. But if it works via
4213 HAProxy, it will work as fast as possible. This option should never be used
4214 by default, and should never be used at all unless such a buggy application
4215 is discovered. The impact of using this option is an increase of bandwidth
4216 usage and CPU usage, which may significantly lower performance in high
4217 latency environments.
4218
4219
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004220option http-pretend-keepalive
4221no option http-pretend-keepalive
4222 Define whether haproxy will announce keepalive to the server or not
4223 May be used in sections : defaults | frontend | listen | backend
4224 yes | yes | yes | yes
4225 Arguments : none
4226
4227 When running with "option http-server-close" or "option forceclose", haproxy
4228 adds a "Connection: close" header to the request forwarded to the server.
4229 Unfortunately, when some servers see this header, they automatically refrain
4230 from using the chunked encoding for responses of unknown length, while this
4231 is totally unrelated. The immediate effect is that this prevents haproxy from
4232 maintaining the client connection alive. A second effect is that a client or
4233 a cache could receive an incomplete response without being aware of it, and
4234 consider the response complete.
4235
4236 By setting "option http-pretend-keepalive", haproxy will make the server
4237 believe it will keep the connection alive. The server will then not fall back
4238 to the abnormal undesired above. When haproxy gets the whole response, it
4239 will close the connection with the server just as it would do with the
4240 "forceclose" option. That way the client gets a normal response and the
4241 connection is correctly closed on the server side.
4242
4243 It is recommended not to enable this option by default, because most servers
4244 will more efficiently close the connection themselves after the last packet,
4245 and release its buffers slightly earlier. Also, the added packet on the
4246 network could slightly reduce the overall peak performance. However it is
4247 worth noting that when this option is enabled, haproxy will have slightly
4248 less work to do. So if haproxy is the bottleneck on the whole architecture,
4249 enabling this option might save a few CPU cycles.
4250
4251 This option may be set both in a frontend and in a backend. It is enabled if
4252 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004253 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02004254 keepalive to be announced to the server and close to be announced to the
4255 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004256
4257 If this option has been enabled in a "defaults" section, it can be disabled
4258 in a specific instance by prepending the "no" keyword before it.
4259
Willy Tarreau16bfb022010-01-16 19:48:41 +01004260 See also : "option forceclose", "option http-server-close", and
4261 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004262
Willy Tarreauc27debf2008-01-06 08:57:02 +01004263
Willy Tarreaub608feb2010-01-02 22:47:18 +01004264option http-server-close
4265no option http-server-close
4266 Enable or disable HTTP connection closing on the server side
4267 May be used in sections : defaults | frontend | listen | backend
4268 yes | yes | yes | yes
4269 Arguments : none
4270
Willy Tarreau70dffda2014-01-30 03:07:23 +01004271 By default HAProxy operates in keep-alive mode with regards to persistent
4272 connections: for each connection it processes each request and response, and
4273 leaves the connection idle on both sides between the end of a response and
4274 the start of a new request. This mode may be changed by several options such
4275 as "option http-server-close", "option forceclose", "option httpclose" or
4276 "option http-tunnel". Setting "option http-server-close" enables HTTP
4277 connection-close mode on the server side while keeping the ability to support
4278 HTTP keep-alive and pipelining on the client side. This provides the lowest
4279 latency on the client side (slow network) and the fastest session reuse on
4280 the server side to save server resources, similarly to "option forceclose".
4281 It also permits non-keepalive capable servers to be served in keep-alive mode
4282 to the clients if they conform to the requirements of RFC2616. Please note
4283 that some servers do not always conform to those requirements when they see
4284 "Connection: close" in the request. The effect will be that keep-alive will
4285 never be used. A workaround consists in enabling "option
4286 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004287
4288 At the moment, logs will not indicate whether requests came from the same
4289 session or not. The accept date reported in the logs corresponds to the end
4290 of the previous request, and the request time corresponds to the time spent
4291 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01004292 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4293 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004294
4295 This option may be set both in a frontend and in a backend. It is enabled if
4296 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004297 It disables and replaces any previous "option httpclose", "option forceclose",
4298 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01004299 ("Proxies") to see how this option combines with others when frontend and
4300 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004301
4302 If this option has been enabled in a "defaults" section, it can be disabled
4303 in a specific instance by prepending the "no" keyword before it.
4304
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004305 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01004306 "option httpclose", "option http-keep-alive", and
4307 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004308
4309
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004310option http-tunnel
4311no option http-tunnel
4312 Disable or enable HTTP connection processing after first transaction
4313 May be used in sections : defaults | frontend | listen | backend
4314 yes | yes | yes | yes
4315 Arguments : none
4316
Willy Tarreau70dffda2014-01-30 03:07:23 +01004317 By default HAProxy operates in keep-alive mode with regards to persistent
4318 connections: for each connection it processes each request and response, and
4319 leaves the connection idle on both sides between the end of a response and
4320 the start of a new request. This mode may be changed by several options such
4321 as "option http-server-close", "option forceclose", "option httpclose" or
4322 "option http-tunnel".
4323
4324 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004325 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01004326 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
4327 is normally not needed anymore unless in very specific cases such as when
4328 using an in-house protocol that looks like HTTP but is not compatible, or
4329 just to log one request per client in order to reduce log size. Note that
4330 everything which works at the HTTP level, including header parsing/addition,
4331 cookie processing or content switching will only work for the first request
4332 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004333
4334 If this option has been enabled in a "defaults" section, it can be disabled
4335 in a specific instance by prepending the "no" keyword before it.
4336
4337 See also : "option forceclose", "option http-server-close",
4338 "option httpclose", "option http-keep-alive", and
4339 "1.1. The HTTP transaction model".
4340
4341
Willy Tarreau88d349d2010-01-25 12:15:43 +01004342option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01004343no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01004344 Make use of non-standard Proxy-Connection header instead of Connection
4345 May be used in sections : defaults | frontend | listen | backend
4346 yes | yes | yes | no
4347 Arguments : none
4348
4349 While RFC2616 explicitly states that HTTP/1.1 agents must use the
4350 Connection header to indicate their wish of persistent or non-persistent
4351 connections, both browsers and proxies ignore this header for proxied
4352 connections and make use of the undocumented, non-standard Proxy-Connection
4353 header instead. The issue begins when trying to put a load balancer between
4354 browsers and such proxies, because there will be a difference between what
4355 haproxy understands and what the client and the proxy agree on.
4356
4357 By setting this option in a frontend, haproxy can automatically switch to use
4358 that non-standard header if it sees proxied requests. A proxied request is
4359 defined here as one where the URI begins with neither a '/' nor a '*'. The
4360 choice of header only affects requests passing through proxies making use of
4361 one of the "httpclose", "forceclose" and "http-server-close" options. Note
4362 that this option can only be specified in a frontend and will affect the
4363 request along its whole life.
4364
Willy Tarreau844a7e72010-01-31 21:46:18 +01004365 Also, when this option is set, a request which requires authentication will
4366 automatically switch to use proxy authentication headers if it is itself a
4367 proxied request. That makes it possible to check or enforce authentication in
4368 front of an existing proxy.
4369
Willy Tarreau88d349d2010-01-25 12:15:43 +01004370 This option should normally never be used, except in front of a proxy.
4371
4372 See also : "option httpclose", "option forceclose" and "option
4373 http-server-close".
4374
4375
Willy Tarreaud63335a2010-02-26 12:56:52 +01004376option httpchk
4377option httpchk <uri>
4378option httpchk <method> <uri>
4379option httpchk <method> <uri> <version>
4380 Enable HTTP protocol to check on the servers health
4381 May be used in sections : defaults | frontend | listen | backend
4382 yes | no | yes | yes
4383 Arguments :
4384 <method> is the optional HTTP method used with the requests. When not set,
4385 the "OPTIONS" method is used, as it generally requires low server
4386 processing and is easy to filter out from the logs. Any method
4387 may be used, though it is not recommended to invent non-standard
4388 ones.
4389
4390 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
4391 which is accessible by default on almost any server, but may be
4392 changed to any other URI. Query strings are permitted.
4393
4394 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
4395 but some servers might behave incorrectly in HTTP 1.0, so turning
4396 it to HTTP/1.1 may sometimes help. Note that the Host field is
4397 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
4398 after "\r\n" following the version string.
4399
4400 By default, server health checks only consist in trying to establish a TCP
4401 connection. When "option httpchk" is specified, a complete HTTP request is
4402 sent once the TCP connection is established, and responses 2xx and 3xx are
4403 considered valid, while all other ones indicate a server failure, including
4404 the lack of any response.
4405
4406 The port and interval are specified in the server configuration.
4407
4408 This option does not necessarily require an HTTP backend, it also works with
4409 plain TCP backends. This is particularly useful to check simple scripts bound
4410 to some dedicated ports using the inetd daemon.
4411
4412 Examples :
4413 # Relay HTTPS traffic to Apache instance and check service availability
4414 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
4415 backend https_relay
4416 mode tcp
4417 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
4418 server apache1 192.168.1.1:443 check port 80
4419
Simon Hormanafc47ee2013-11-25 10:46:35 +09004420 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
4421 "option pgsql-check", "http-check" and the "check", "port" and
4422 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01004423
4424
Willy Tarreauc27debf2008-01-06 08:57:02 +01004425option httpclose
4426no option httpclose
4427 Enable or disable passive HTTP connection closing
4428 May be used in sections : defaults | frontend | listen | backend
4429 yes | yes | yes | yes
4430 Arguments : none
4431
Willy Tarreau70dffda2014-01-30 03:07:23 +01004432 By default HAProxy operates in keep-alive mode with regards to persistent
4433 connections: for each connection it processes each request and response, and
4434 leaves the connection idle on both sides between the end of a response and
4435 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01004436 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01004437 "option http-tunnel".
4438
4439 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
4440 if a "Connection: close" header is already set in each direction, and will
4441 add one if missing. Each end should react to this by actively closing the TCP
4442 connection after each transfer, thus resulting in a switch to the HTTP close
4443 mode. Any "Connection" header different from "close" will also be removed.
4444 Note that this option is deprecated since what it does is very cheap but not
4445 reliable. Using "option http-server-close" or "option forceclose" is strongly
4446 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004447
4448 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004449 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01004450 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
4451 it is possible to use the "option forceclose" which actively closes the
4452 request connection once the server responds. Option "forceclose" also
4453 releases the server connection earlier because it does not have to wait for
4454 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004455
4456 This option may be set both in a frontend and in a backend. It is enabled if
4457 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004458 It disables and replaces any previous "option http-server-close",
4459 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01004460 check section 4 ("Proxies") to see how this option combines with others when
4461 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004462
4463 If this option has been enabled in a "defaults" section, it can be disabled
4464 in a specific instance by prepending the "no" keyword before it.
4465
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004466 See also : "option forceclose", "option http-server-close" and
4467 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01004468
4469
Emeric Brun3a058f32009-06-30 18:26:00 +02004470option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004471 Enable logging of HTTP request, session state and timers
4472 May be used in sections : defaults | frontend | listen | backend
4473 yes | yes | yes | yes
Emeric Brun3a058f32009-06-30 18:26:00 +02004474 Arguments :
4475 clf if the "clf" argument is added, then the output format will be
4476 the CLF format instead of HAProxy's default HTTP format. You can
4477 use this when you need to feed HAProxy's logs through a specific
4478 log analyser which only support the CLF format and which is not
4479 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004480
4481 By default, the log output format is very poor, as it only contains the
4482 source and destination addresses, and the instance name. By specifying
4483 "option httplog", each log line turns into a much richer format including,
4484 but not limited to, the HTTP request, the connection timers, the session
4485 status, the connections numbers, the captured headers and cookies, the
4486 frontend, backend and server name, and of course the source address and
4487 ports.
4488
4489 This option may be set either in the frontend or the backend.
4490
PiBa-NL211c2e92014-12-11 21:31:54 +01004491 Specifying only "option httplog" will automatically clear the 'clf' mode
4492 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02004493
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004494 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004495
Willy Tarreau55165fe2009-05-10 12:02:55 +02004496
4497option http_proxy
4498no option http_proxy
4499 Enable or disable plain HTTP proxy mode
4500 May be used in sections : defaults | frontend | listen | backend
4501 yes | yes | yes | yes
4502 Arguments : none
4503
4504 It sometimes happens that people need a pure HTTP proxy which understands
4505 basic proxy requests without caching nor any fancy feature. In this case,
4506 it may be worth setting up an HAProxy instance with the "option http_proxy"
4507 set. In this mode, no server is declared, and the connection is forwarded to
4508 the IP address and port found in the URL after the "http://" scheme.
4509
4510 No host address resolution is performed, so this only works when pure IP
4511 addresses are passed. Since this option's usage perimeter is rather limited,
4512 it will probably be used only by experts who know they need exactly it. Last,
4513 if the clients are susceptible of sending keep-alive requests, it will be
Cyril Bonté2409e682010-12-14 22:47:51 +01004514 needed to add "option httpclose" to ensure that all requests will correctly
Willy Tarreau55165fe2009-05-10 12:02:55 +02004515 be analyzed.
4516
4517 If this option has been enabled in a "defaults" section, it can be disabled
4518 in a specific instance by prepending the "no" keyword before it.
4519
4520 Example :
4521 # this backend understands HTTP proxy requests and forwards them directly.
4522 backend direct_forward
4523 option httpclose
4524 option http_proxy
4525
4526 See also : "option httpclose"
4527
Willy Tarreau211ad242009-10-03 21:45:07 +02004528
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004529option independent-streams
4530no option independent-streams
4531 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004532 May be used in sections : defaults | frontend | listen | backend
4533 yes | yes | yes | yes
4534 Arguments : none
4535
4536 By default, when data is sent over a socket, both the write timeout and the
4537 read timeout for that socket are refreshed, because we consider that there is
4538 activity on that socket, and we have no other means of guessing if we should
4539 receive data or not.
4540
4541 While this default behaviour is desirable for almost all applications, there
4542 exists a situation where it is desirable to disable it, and only refresh the
4543 read timeout if there are incoming data. This happens on sessions with large
4544 timeouts and low amounts of exchanged data such as telnet session. If the
4545 server suddenly disappears, the output data accumulates in the system's
4546 socket buffers, both timeouts are correctly refreshed, and there is no way
4547 to know the server does not receive them, so we don't timeout. However, when
4548 the underlying protocol always echoes sent data, it would be enough by itself
4549 to detect the issue using the read timeout. Note that this problem does not
4550 happen with more verbose protocols because data won't accumulate long in the
4551 socket buffers.
4552
4553 When this option is set on the frontend, it will disable read timeout updates
4554 on data sent to the client. There probably is little use of this case. When
4555 the option is set on the backend, it will disable read timeout updates on
4556 data sent to the server. Doing so will typically break large HTTP posts from
4557 slow lines, so use it with caution.
4558
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004559 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004560 with a spelling mistake. This spelling is still supported but
4561 deprecated.
4562
Willy Tarreauce887fd2012-05-12 12:50:00 +02004563 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004564
4565
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02004566option ldap-check
4567 Use LDAPv3 health checks for server testing
4568 May be used in sections : defaults | frontend | listen | backend
4569 yes | no | yes | yes
4570 Arguments : none
4571
4572 It is possible to test that the server correctly talks LDAPv3 instead of just
4573 testing that it accepts the TCP connection. When this option is set, an
4574 LDAPv3 anonymous simple bind message is sent to the server, and the response
4575 is analyzed to find an LDAPv3 bind response message.
4576
4577 The server is considered valid only when the LDAP response contains success
4578 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
4579
4580 Logging of bind requests is server dependent see your documentation how to
4581 configure it.
4582
4583 Example :
4584 option ldap-check
4585
4586 See also : "option httpchk"
4587
4588
Willy Tarreau211ad242009-10-03 21:45:07 +02004589option log-health-checks
4590no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02004591 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02004592 May be used in sections : defaults | frontend | listen | backend
4593 yes | no | yes | yes
4594 Arguments : none
4595
Willy Tarreaubef1b322014-05-13 21:01:39 +02004596 By default, failed health check are logged if server is UP and successful
4597 health checks are logged if server is DOWN, so the amount of additional
4598 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02004599
Willy Tarreaubef1b322014-05-13 21:01:39 +02004600 When this option is enabled, any change of the health check status or to
4601 the server's health will be logged, so that it becomes possible to know
4602 that a server was failing occasional checks before crashing, or exactly when
4603 it failed to respond a valid HTTP status, then when the port started to
4604 reject connections, then when the server stopped responding at all.
4605
4606 Note that status changes not caused by health checks (eg: enable/disable on
4607 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02004608
Willy Tarreaubef1b322014-05-13 21:01:39 +02004609 See also: "option httpchk", "option ldap-check", "option mysql-check",
4610 "option pgsql-check", "option redis-check", "option smtpchk",
4611 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02004612
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004613
4614option log-separate-errors
4615no option log-separate-errors
4616 Change log level for non-completely successful connections
4617 May be used in sections : defaults | frontend | listen | backend
4618 yes | yes | yes | no
4619 Arguments : none
4620
4621 Sometimes looking for errors in logs is not easy. This option makes haproxy
4622 raise the level of logs containing potentially interesting information such
4623 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
4624 level changes from "info" to "err". This makes it possible to log them
4625 separately to a different file with most syslog daemons. Be careful not to
4626 remove them from the original file, otherwise you would lose ordering which
4627 provides very important information.
4628
4629 Using this option, large sites dealing with several thousand connections per
4630 second may log normal traffic to a rotating buffer and only archive smaller
4631 error logs.
4632
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004633 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004634 logging.
4635
Willy Tarreauc27debf2008-01-06 08:57:02 +01004636
4637option logasap
4638no option logasap
4639 Enable or disable early logging of HTTP requests
4640 May be used in sections : defaults | frontend | listen | backend
4641 yes | yes | yes | no
4642 Arguments : none
4643
4644 By default, HTTP requests are logged upon termination so that the total
4645 transfer time and the number of bytes appear in the logs. When large objects
4646 are being transferred, it may take a while before the request appears in the
4647 logs. Using "option logasap", the request gets logged as soon as the server
4648 sends the complete headers. The only missing information in the logs will be
4649 the total number of bytes which will indicate everything except the amount
4650 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004651 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01004652 "Content-Length" response header so that the logs at least indicate how many
4653 bytes are expected to be transferred.
4654
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004655 Examples :
4656 listen http_proxy 0.0.0.0:80
4657 mode http
4658 option httplog
4659 option logasap
4660 log 192.168.2.200 local3
4661
4662 >>> Feb 6 12:14:14 localhost \
4663 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
4664 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
4665 "GET /image.iso HTTP/1.0"
4666
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004667 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01004668 logging.
4669
4670
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004671option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004672 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004673 May be used in sections : defaults | frontend | listen | backend
4674 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004675 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004676 <username> This is the username which will be used when connecting to MySQL
4677 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004678 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004679
4680 If you specify a username, the check consists of sending two MySQL packet,
4681 one Client Authentication packet, and one QUIT packet, to correctly close
4682 MySQL session. We then parse the MySQL Handshake Initialisation packet and/or
4683 Error packet. It is a basic but useful test which does not produce error nor
4684 aborted connect on the server. However, it requires adding an authorization
4685 in the MySQL table, like this :
4686
4687 USE mysql;
4688 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
4689 FLUSH PRIVILEGES;
4690
4691 If you don't specify a username (it is deprecated and not recommended), the
4692 check only consists in parsing the Mysql Handshake Initialisation packet or
4693 Error packet, we don't send anything in this mode. It was reported that it
4694 can generate lockout if check is too frequent and/or if there is not enough
4695 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
4696 value as if a connection is established successfully within fewer than MySQL
4697 "max_connect_errors" attempts after a previous connection was interrupted,
4698 the error count for the host is cleared to zero. If HAProxy's server get
4699 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
4700
4701 Remember that this does not check database presence nor database consistency.
4702 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004703
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02004704 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004705
4706 Most often, an incoming MySQL server needs to see the client's IP address for
4707 various purposes, including IP privilege matching and connection logging.
4708 When possible, it is often wise to masquerade the client's IP address when
4709 connecting to the server using the "usesrc" argument of the "source" keyword,
4710 which requires the cttproxy feature to be compiled in, and the MySQL server
4711 to route the client via the machine hosting haproxy.
4712
4713 See also: "option httpchk"
4714
4715
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004716option nolinger
4717no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004718 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004719 May be used in sections: defaults | frontend | listen | backend
4720 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004721 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004722
4723 When clients or servers abort connections in a dirty way (eg: they are
4724 physically disconnected), the session timeouts triggers and the session is
4725 closed. But it will remain in FIN_WAIT1 state for some time in the system,
4726 using some resources and possibly limiting the ability to establish newer
4727 connections.
4728
4729 When this happens, it is possible to activate "option nolinger" which forces
4730 the system to immediately remove any socket's pending data on close. Thus,
4731 the session is instantly purged from the system's tables. This usually has
4732 side effects such as increased number of TCP resets due to old retransmits
4733 getting immediately rejected. Some firewalls may sometimes complain about
4734 this too.
4735
4736 For this reason, it is not recommended to use this option when not absolutely
4737 needed. You know that you need it when you have thousands of FIN_WAIT1
4738 sessions on your system (TIME_WAIT ones do not count).
4739
4740 This option may be used both on frontends and backends, depending on the side
4741 where it is required. Use it on the frontend for clients, and on the backend
4742 for servers.
4743
4744 If this option has been enabled in a "defaults" section, it can be disabled
4745 in a specific instance by prepending the "no" keyword before it.
4746
4747
Willy Tarreau55165fe2009-05-10 12:02:55 +02004748option originalto [ except <network> ] [ header <name> ]
4749 Enable insertion of the X-Original-To header to requests sent to servers
4750 May be used in sections : defaults | frontend | listen | backend
4751 yes | yes | yes | yes
4752 Arguments :
4753 <network> is an optional argument used to disable this option for sources
4754 matching <network>
4755 <name> an optional argument to specify a different "X-Original-To"
4756 header name.
4757
4758 Since HAProxy can work in transparent mode, every request from a client can
4759 be redirected to the proxy and HAProxy itself can proxy every request to a
4760 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
4761 be lost. This is annoying when you want access rules based on destination ip
4762 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
4763 added by HAProxy to all requests sent to the server. This header contains a
4764 value representing the original destination IP address. Since this must be
4765 configured to always use the last occurrence of this header only. Note that
4766 only the last occurrence of the header must be used, since it is really
4767 possible that the client has already brought one.
4768
4769 The keyword "header" may be used to supply a different header name to replace
4770 the default "X-Original-To". This can be useful where you might already
4771 have a "X-Original-To" header from a different application, and you need
4772 preserve it. Also if your backend server doesn't use the "X-Original-To"
4773 header and requires different one.
4774
4775 Sometimes, a same HAProxy instance may be shared between a direct client
4776 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4777 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4778 header for a known source address or network by adding the "except" keyword
4779 followed by the network address. In this case, any source IP matching the
4780 network will not cause an addition of this header. Most common uses are with
4781 private networks or 127.0.0.1.
4782
4783 This option may be specified either in the frontend or in the backend. If at
4784 least one of them uses it, the header will be added. Note that the backend's
4785 setting of the header subargument takes precedence over the frontend's if
4786 both are defined.
4787
Willy Tarreau55165fe2009-05-10 12:02:55 +02004788 Examples :
4789 # Original Destination address
4790 frontend www
4791 mode http
4792 option originalto except 127.0.0.1
4793
4794 # Those servers want the IP Address in X-Client-Dst
4795 backend www
4796 mode http
4797 option originalto header X-Client-Dst
4798
Willy Tarreau87cf5142011-08-19 22:57:24 +02004799 See also : "option httpclose", "option http-server-close",
4800 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02004801
4802
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004803option persist
4804no option persist
4805 Enable or disable forced persistence on down servers
4806 May be used in sections: defaults | frontend | listen | backend
4807 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004808 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004809
4810 When an HTTP request reaches a backend with a cookie which references a dead
4811 server, by default it is redispatched to another server. It is possible to
4812 force the request to be sent to the dead server first using "option persist"
4813 if absolutely needed. A common use case is when servers are under extreme
4814 load and spend their time flapping. In this case, the users would still be
4815 directed to the server they opened the session on, in the hope they would be
4816 correctly served. It is recommended to use "option redispatch" in conjunction
4817 with this option so that in the event it would not be possible to connect to
4818 the server at all (server definitely dead), the client would finally be
4819 redirected to another valid server.
4820
4821 If this option has been enabled in a "defaults" section, it can be disabled
4822 in a specific instance by prepending the "no" keyword before it.
4823
Willy Tarreau4de91492010-01-22 19:10:05 +01004824 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004825
4826
Willy Tarreau0c122822013-12-15 18:49:01 +01004827option pgsql-check [ user <username> ]
4828 Use PostgreSQL health checks for server testing
4829 May be used in sections : defaults | frontend | listen | backend
4830 yes | no | yes | yes
4831 Arguments :
4832 <username> This is the username which will be used when connecting to
4833 PostgreSQL server.
4834
4835 The check sends a PostgreSQL StartupMessage and waits for either
4836 Authentication request or ErrorResponse message. It is a basic but useful
4837 test which does not produce error nor aborted connect on the server.
4838 This check is identical with the "mysql-check".
4839
4840 See also: "option httpchk"
4841
4842
Willy Tarreau9420b122013-12-15 18:58:25 +01004843option prefer-last-server
4844no option prefer-last-server
4845 Allow multiple load balanced requests to remain on the same server
4846 May be used in sections: defaults | frontend | listen | backend
4847 yes | no | yes | yes
4848 Arguments : none
4849
4850 When the load balancing algorithm in use is not deterministic, and a previous
4851 request was sent to a server to which haproxy still holds a connection, it is
4852 sometimes desirable that subsequent requests on a same session go to the same
4853 server as much as possible. Note that this is different from persistence, as
4854 we only indicate a preference which haproxy tries to apply without any form
4855 of warranty. The real use is for keep-alive connections sent to servers. When
4856 this option is used, haproxy will try to reuse the same connection that is
4857 attached to the server instead of rebalancing to another server, causing a
4858 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01004859 not make much sense to use this in combination with hashing algorithms. Note,
4860 haproxy already automatically tries to stick to a server which sends a 401 or
4861 to a proxy which sends a 407 (authentication required). This is mandatory for
4862 use with the broken NTLM authentication challenge, and significantly helps in
4863 troubleshooting some faulty applications. Option prefer-last-server might be
4864 desirable in these environments as well, to avoid redistributing the traffic
4865 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01004866
4867 If this option has been enabled in a "defaults" section, it can be disabled
4868 in a specific instance by prepending the "no" keyword before it.
4869
4870 See also: "option http-keep-alive"
4871
4872
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004873option redispatch
4874no option redispatch
4875 Enable or disable session redistribution in case of connection failure
4876 May be used in sections: defaults | frontend | listen | backend
4877 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004878 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004879
4880 In HTTP mode, if a server designated by a cookie is down, clients may
4881 definitely stick to it because they cannot flush the cookie, so they will not
4882 be able to access the service anymore.
4883
4884 Specifying "option redispatch" will allow the proxy to break their
4885 persistence and redistribute them to a working server.
4886
4887 It also allows to retry last connection to another server in case of multiple
4888 connection failures. Of course, it requires having "retries" set to a nonzero
4889 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01004890
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004891 This form is the preferred form, which replaces both the "redispatch" and
4892 "redisp" keywords.
4893
4894 If this option has been enabled in a "defaults" section, it can be disabled
4895 in a specific instance by prepending the "no" keyword before it.
4896
Willy Tarreau4de91492010-01-22 19:10:05 +01004897 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004898
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004899
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02004900option redis-check
4901 Use redis health checks for server testing
4902 May be used in sections : defaults | frontend | listen | backend
4903 yes | no | yes | yes
4904 Arguments : none
4905
4906 It is possible to test that the server correctly talks REDIS protocol instead
4907 of just testing that it accepts the TCP connection. When this option is set,
4908 a PING redis command is sent to the server, and the response is analyzed to
4909 find the "+PONG" response message.
4910
4911 Example :
4912 option redis-check
4913
4914 See also : "option httpchk"
4915
4916
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004917option smtpchk
4918option smtpchk <hello> <domain>
4919 Use SMTP health checks for server testing
4920 May be used in sections : defaults | frontend | listen | backend
4921 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01004922 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004923 <hello> is an optional argument. It is the "hello" command to use. It can
4924 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
4925 values will be turned into the default command ("HELO").
4926
4927 <domain> is the domain name to present to the server. It may only be
4928 specified (and is mandatory) if the hello command has been
4929 specified. By default, "localhost" is used.
4930
4931 When "option smtpchk" is set, the health checks will consist in TCP
4932 connections followed by an SMTP command. By default, this command is
4933 "HELO localhost". The server's return code is analyzed and only return codes
4934 starting with a "2" will be considered as valid. All other responses,
4935 including a lack of response will constitute an error and will indicate a
4936 dead server.
4937
4938 This test is meant to be used with SMTP servers or relays. Depending on the
4939 request, it is possible that some servers do not log each connection attempt,
4940 so you may want to experiment to improve the behaviour. Using telnet on port
4941 25 is often easier than adjusting the configuration.
4942
4943 Most often, an incoming SMTP server needs to see the client's IP address for
4944 various purposes, including spam filtering, anti-spoofing and logging. When
4945 possible, it is often wise to masquerade the client's IP address when
4946 connecting to the server using the "usesrc" argument of the "source" keyword,
4947 which requires the cttproxy feature to be compiled in.
4948
4949 Example :
4950 option smtpchk HELO mydomain.org
4951
4952 See also : "option httpchk", "source"
4953
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004954
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02004955option socket-stats
4956no option socket-stats
4957
4958 Enable or disable collecting & providing separate statistics for each socket.
4959 May be used in sections : defaults | frontend | listen | backend
4960 yes | yes | yes | no
4961
4962 Arguments : none
4963
4964
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004965option splice-auto
4966no option splice-auto
4967 Enable or disable automatic kernel acceleration on sockets in both directions
4968 May be used in sections : defaults | frontend | listen | backend
4969 yes | yes | yes | yes
4970 Arguments : none
4971
4972 When this option is enabled either on a frontend or on a backend, haproxy
4973 will automatically evaluate the opportunity to use kernel tcp splicing to
4974 forward data between the client and the server, in either direction. Haproxy
4975 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004976 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004977 are not much aggressive in order to limit excessive use of splicing. This
4978 option requires splicing to be enabled at compile time, and may be globally
4979 disabled with the global option "nosplice". Since splice uses pipes, using it
4980 requires that there are enough spare pipes.
4981
4982 Important note: kernel-based TCP splicing is a Linux-specific feature which
4983 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
4984 transfer data between sockets without copying these data to user-space, thus
4985 providing noticeable performance gains and CPU cycles savings. Since many
4986 early implementations are buggy, corrupt data and/or are inefficient, this
4987 feature is not enabled by default, and it should be used with extreme care.
4988 While it is not possible to detect the correctness of an implementation,
4989 2.6.29 is the first version offering a properly working implementation. In
4990 case of doubt, splicing may be globally disabled using the global "nosplice"
4991 keyword.
4992
4993 Example :
4994 option splice-auto
4995
4996 If this option has been enabled in a "defaults" section, it can be disabled
4997 in a specific instance by prepending the "no" keyword before it.
4998
4999 See also : "option splice-request", "option splice-response", and global
5000 options "nosplice" and "maxpipes"
5001
5002
5003option splice-request
5004no option splice-request
5005 Enable or disable automatic kernel acceleration on sockets for requests
5006 May be used in sections : defaults | frontend | listen | backend
5007 yes | yes | yes | yes
5008 Arguments : none
5009
5010 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005011 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01005012 the client to the server. It might still use the recv/send scheme if there
5013 are no spare pipes left. This option requires splicing to be enabled at
5014 compile time, and may be globally disabled with the global option "nosplice".
5015 Since splice uses pipes, using it requires that there are enough spare pipes.
5016
5017 Important note: see "option splice-auto" for usage limitations.
5018
5019 Example :
5020 option splice-request
5021
5022 If this option has been enabled in a "defaults" section, it can be disabled
5023 in a specific instance by prepending the "no" keyword before it.
5024
5025 See also : "option splice-auto", "option splice-response", and global options
5026 "nosplice" and "maxpipes"
5027
5028
5029option splice-response
5030no option splice-response
5031 Enable or disable automatic kernel acceleration on sockets for responses
5032 May be used in sections : defaults | frontend | listen | backend
5033 yes | yes | yes | yes
5034 Arguments : none
5035
5036 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005037 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01005038 the server to the client. It might still use the recv/send scheme if there
5039 are no spare pipes left. This option requires splicing to be enabled at
5040 compile time, and may be globally disabled with the global option "nosplice".
5041 Since splice uses pipes, using it requires that there are enough spare pipes.
5042
5043 Important note: see "option splice-auto" for usage limitations.
5044
5045 Example :
5046 option splice-response
5047
5048 If this option has been enabled in a "defaults" section, it can be disabled
5049 in a specific instance by prepending the "no" keyword before it.
5050
5051 See also : "option splice-auto", "option splice-request", and global options
5052 "nosplice" and "maxpipes"
5053
5054
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005055option srvtcpka
5056no option srvtcpka
5057 Enable or disable the sending of TCP keepalive packets on the server side
5058 May be used in sections : defaults | frontend | listen | backend
5059 yes | no | yes | yes
5060 Arguments : none
5061
5062 When there is a firewall or any session-aware component between a client and
5063 a server, and when the protocol involves very long sessions with long idle
5064 periods (eg: remote desktops), there is a risk that one of the intermediate
5065 components decides to expire a session which has remained idle for too long.
5066
5067 Enabling socket-level TCP keep-alives makes the system regularly send packets
5068 to the other end of the connection, leaving it active. The delay between
5069 keep-alive probes is controlled by the system only and depends both on the
5070 operating system and its tuning parameters.
5071
5072 It is important to understand that keep-alive packets are neither emitted nor
5073 received at the application level. It is only the network stacks which sees
5074 them. For this reason, even if one side of the proxy already uses keep-alives
5075 to maintain its connection alive, those keep-alive packets will not be
5076 forwarded to the other side of the proxy.
5077
5078 Please note that this has nothing to do with HTTP keep-alive.
5079
5080 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
5081 server side of a connection, which should help when session expirations are
5082 noticed between HAProxy and a server.
5083
5084 If this option has been enabled in a "defaults" section, it can be disabled
5085 in a specific instance by prepending the "no" keyword before it.
5086
5087 See also : "option clitcpka", "option tcpka"
5088
5089
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005090option ssl-hello-chk
5091 Use SSLv3 client hello health checks for server testing
5092 May be used in sections : defaults | frontend | listen | backend
5093 yes | no | yes | yes
5094 Arguments : none
5095
5096 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
5097 possible to test that the server correctly talks SSL instead of just testing
5098 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
5099 SSLv3 client hello messages are sent once the connection is established to
5100 the server, and the response is analyzed to find an SSL server hello message.
5101 The server is considered valid only when the response contains this server
5102 hello message.
5103
5104 All servers tested till there correctly reply to SSLv3 client hello messages,
5105 and most servers tested do not even log the requests containing only hello
5106 messages, which is appreciable.
5107
Willy Tarreau763a95b2012-10-04 23:15:39 +02005108 Note that this check works even when SSL support was not built into haproxy
5109 because it forges the SSL message. When SSL support is available, it is best
5110 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005111
Willy Tarreau763a95b2012-10-04 23:15:39 +02005112 See also: "option httpchk", "check-ssl"
5113
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005114
Willy Tarreaued179852013-12-16 01:07:00 +01005115option tcp-check
5116 Perform health checks using tcp-check send/expect sequences
5117 May be used in sections: defaults | frontend | listen | backend
5118 yes | no | yes | yes
5119
5120 This health check method is intended to be combined with "tcp-check" command
5121 lists in order to support send/expect types of health check sequences.
5122
5123 TCP checks currently support 4 modes of operations :
5124 - no "tcp-check" directive : the health check only consists in a connection
5125 attempt, which remains the default mode.
5126
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005127 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005128 used to send a string along with a connection opening. With some
5129 protocols, it helps sending a "QUIT" message for example that prevents
5130 the server from logging a connection error for each health check. The
5131 check result will still be based on the ability to open the connection
5132 only.
5133
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005134 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01005135 The connection is opened and haproxy waits for the server to present some
5136 contents which must validate some rules. The check result will be based
5137 on the matching between the contents and the rules. This is suited for
5138 POP, IMAP, SMTP, FTP, SSH, TELNET.
5139
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005140 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005141 used to test a hello-type protocol. Haproxy sends a message, the server
5142 responds and its response is analysed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005143 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01005144 suited for protocols which require a binding or a request/response model.
5145 LDAP, MySQL, Redis and SSL are example of such protocols, though they
5146 already all have their dedicated checks with a deeper understanding of
5147 the respective protocols.
5148 In this mode, many questions may be sent and many answers may be
5149 analysed.
5150
5151 Examples :
5152 # perform a POP check (analyse only server's banner)
5153 option tcp-check
5154 tcp-check expect string +OK\ POP3\ ready
5155
5156 # perform an IMAP check (analyse only server's banner)
5157 option tcp-check
5158 tcp-check expect string *\ OK\ IMAP4\ ready
5159
5160 # look for the redis master server after ensuring it speaks well
5161 # redis protocol, then it exits properly.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005162 # (send a command then analyse the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01005163 option tcp-check
5164 tcp-check send PING\r\n
5165 tcp-check expect +PONG
5166 tcp-check send info\ replication\r\n
5167 tcp-check expect string role:master
5168 tcp-check send QUIT\r\n
5169 tcp-check expect string +OK
5170
5171 forge a HTTP request, then analyse the response
5172 (send many headers before analyzing)
5173 option tcp-check
5174 tcp-check send HEAD\ /\ HTTP/1.1\r\n
5175 tcp-check send Host:\ www.mydomain.com\r\n
5176 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
5177 tcp-check send \r\n
5178 tcp-check expect rstring HTTP/1\..\ (2..|3..)
5179
5180
5181 See also : "tcp-check expect", "tcp-check send"
5182
5183
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005184option tcp-smart-accept
5185no option tcp-smart-accept
5186 Enable or disable the saving of one ACK packet during the accept sequence
5187 May be used in sections : defaults | frontend | listen | backend
5188 yes | yes | yes | no
5189 Arguments : none
5190
5191 When an HTTP connection request comes in, the system acknowledges it on
5192 behalf of HAProxy, then the client immediately sends its request, and the
5193 system acknowledges it too while it is notifying HAProxy about the new
5194 connection. HAProxy then reads the request and responds. This means that we
5195 have one TCP ACK sent by the system for nothing, because the request could
5196 very well be acknowledged by HAProxy when it sends its response.
5197
5198 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
5199 sending this useless ACK on platforms which support it (currently at least
5200 Linux). It must not cause any problem, because the system will send it anyway
5201 after 40 ms if the response takes more time than expected to come.
5202
5203 During complex network debugging sessions, it may be desirable to disable
5204 this optimization because delayed ACKs can make troubleshooting more complex
5205 when trying to identify where packets are delayed. It is then possible to
5206 fall back to normal behaviour by specifying "no option tcp-smart-accept".
5207
5208 It is also possible to force it for non-HTTP proxies by simply specifying
5209 "option tcp-smart-accept". For instance, it can make sense with some services
5210 such as SMTP where the server speaks first.
5211
5212 It is recommended to avoid forcing this option in a defaults section. In case
5213 of doubt, consider setting it back to automatic values by prepending the
5214 "default" keyword before it, or disabling it using the "no" keyword.
5215
Willy Tarreaud88edf22009-06-14 15:48:17 +02005216 See also : "option tcp-smart-connect"
5217
5218
5219option tcp-smart-connect
5220no option tcp-smart-connect
5221 Enable or disable the saving of one ACK packet during the connect sequence
5222 May be used in sections : defaults | frontend | listen | backend
5223 yes | no | yes | yes
5224 Arguments : none
5225
5226 On certain systems (at least Linux), HAProxy can ask the kernel not to
5227 immediately send an empty ACK upon a connection request, but to directly
5228 send the buffer request instead. This saves one packet on the network and
5229 thus boosts performance. It can also be useful for some servers, because they
5230 immediately get the request along with the incoming connection.
5231
5232 This feature is enabled when "option tcp-smart-connect" is set in a backend.
5233 It is not enabled by default because it makes network troubleshooting more
5234 complex.
5235
5236 It only makes sense to enable it with protocols where the client speaks first
5237 such as HTTP. In other situations, if there is no data to send in place of
5238 the ACK, a normal ACK is sent.
5239
5240 If this option has been enabled in a "defaults" section, it can be disabled
5241 in a specific instance by prepending the "no" keyword before it.
5242
5243 See also : "option tcp-smart-accept"
5244
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005245
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005246option tcpka
5247 Enable or disable the sending of TCP keepalive packets on both sides
5248 May be used in sections : defaults | frontend | listen | backend
5249 yes | yes | yes | yes
5250 Arguments : none
5251
5252 When there is a firewall or any session-aware component between a client and
5253 a server, and when the protocol involves very long sessions with long idle
5254 periods (eg: remote desktops), there is a risk that one of the intermediate
5255 components decides to expire a session which has remained idle for too long.
5256
5257 Enabling socket-level TCP keep-alives makes the system regularly send packets
5258 to the other end of the connection, leaving it active. The delay between
5259 keep-alive probes is controlled by the system only and depends both on the
5260 operating system and its tuning parameters.
5261
5262 It is important to understand that keep-alive packets are neither emitted nor
5263 received at the application level. It is only the network stacks which sees
5264 them. For this reason, even if one side of the proxy already uses keep-alives
5265 to maintain its connection alive, those keep-alive packets will not be
5266 forwarded to the other side of the proxy.
5267
5268 Please note that this has nothing to do with HTTP keep-alive.
5269
5270 Using option "tcpka" enables the emission of TCP keep-alive probes on both
5271 the client and server sides of a connection. Note that this is meaningful
5272 only in "defaults" or "listen" sections. If this option is used in a
5273 frontend, only the client side will get keep-alives, and if this option is
5274 used in a backend, only the server side will get keep-alives. For this
5275 reason, it is strongly recommended to explicitly use "option clitcpka" and
5276 "option srvtcpka" when the configuration is split between frontends and
5277 backends.
5278
5279 See also : "option clitcpka", "option srvtcpka"
5280
Willy Tarreau844e3c52008-01-11 16:28:18 +01005281
5282option tcplog
5283 Enable advanced logging of TCP connections with session state and timers
5284 May be used in sections : defaults | frontend | listen | backend
5285 yes | yes | yes | yes
5286 Arguments : none
5287
5288 By default, the log output format is very poor, as it only contains the
5289 source and destination addresses, and the instance name. By specifying
5290 "option tcplog", each log line turns into a much richer format including, but
5291 not limited to, the connection timers, the session status, the connections
5292 numbers, the frontend, backend and server name, and of course the source
5293 address and ports. This option is useful for pure TCP proxies in order to
5294 find which of the client or server disconnects or times out. For normal HTTP
5295 proxies, it's better to use "option httplog" which is even more complete.
5296
5297 This option may be set either in the frontend or the backend.
5298
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005299 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005300
5301
Willy Tarreau844e3c52008-01-11 16:28:18 +01005302option transparent
5303no option transparent
5304 Enable client-side transparent proxying
5305 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01005306 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01005307 Arguments : none
5308
5309 This option was introduced in order to provide layer 7 persistence to layer 3
5310 load balancers. The idea is to use the OS's ability to redirect an incoming
5311 connection for a remote address to a local process (here HAProxy), and let
5312 this process know what address was initially requested. When this option is
5313 used, sessions without cookies will be forwarded to the original destination
5314 IP address of the incoming request (which should match that of another
5315 equipment), while requests with cookies will still be forwarded to the
5316 appropriate server.
5317
5318 Note that contrary to a common belief, this option does NOT make HAProxy
5319 present the client's IP to the server when establishing the connection.
5320
Willy Tarreaua1146052011-03-01 09:51:54 +01005321 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005322 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005323
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005324
Emeric Brun647caf12009-06-30 17:57:00 +02005325persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02005326persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02005327 Enable RDP cookie-based persistence
5328 May be used in sections : defaults | frontend | listen | backend
5329 yes | no | yes | yes
5330 Arguments :
5331 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02005332 default cookie name "msts" will be used. There currently is no
5333 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02005334
5335 This statement enables persistence based on an RDP cookie. The RDP cookie
5336 contains all information required to find the server in the list of known
5337 servers. So when this option is set in the backend, the request is analysed
5338 and if an RDP cookie is found, it is decoded. If it matches a known server
5339 which is still UP (or if "option persist" is set), then the connection is
5340 forwarded to this server.
5341
5342 Note that this only makes sense in a TCP backend, but for this to work, the
5343 frontend must have waited long enough to ensure that an RDP cookie is present
5344 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005345 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02005346 a single "listen" section.
5347
Willy Tarreau61e28f22010-05-16 22:31:05 +02005348 Also, it is important to understand that the terminal server will emit this
5349 RDP cookie only if it is configured for "token redirection mode", which means
5350 that the "IP address redirection" option is disabled.
5351
Emeric Brun647caf12009-06-30 17:57:00 +02005352 Example :
5353 listen tse-farm
5354 bind :3389
5355 # wait up to 5s for an RDP cookie in the request
5356 tcp-request inspect-delay 5s
5357 tcp-request content accept if RDP_COOKIE
5358 # apply RDP cookie persistence
5359 persist rdp-cookie
5360 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02005361 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02005362 balance rdp-cookie
5363 server srv1 1.1.1.1:3389
5364 server srv2 1.1.1.2:3389
5365
Simon Hormanab814e02011-06-24 14:50:20 +09005366 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
5367 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02005368
5369
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005370rate-limit sessions <rate>
5371 Set a limit on the number of new sessions accepted per second on a frontend
5372 May be used in sections : defaults | frontend | listen | backend
5373 yes | yes | yes | no
5374 Arguments :
5375 <rate> The <rate> parameter is an integer designating the maximum number
5376 of new sessions per second to accept on the frontend.
5377
5378 When the frontend reaches the specified number of new sessions per second, it
5379 stops accepting new connections until the rate drops below the limit again.
5380 During this time, the pending sessions will be kept in the socket's backlog
5381 (in system buffers) and haproxy will not even be aware that sessions are
5382 pending. When applying very low limit on a highly loaded service, it may make
5383 sense to increase the socket's backlog using the "backlog" keyword.
5384
5385 This feature is particularly efficient at blocking connection-based attacks
5386 or service abuse on fragile servers. Since the session rate is measured every
5387 millisecond, it is extremely accurate. Also, the limit applies immediately,
5388 no delay is needed at all to detect the threshold.
5389
5390 Example : limit the connection rate on SMTP to 10 per second max
5391 listen smtp
5392 mode tcp
5393 bind :25
5394 rate-limit sessions 10
5395 server 127.0.0.1:1025
5396
Willy Tarreaua17c2d92011-07-25 08:16:20 +02005397 Note : when the maximum rate is reached, the frontend's status is not changed
5398 but its sockets appear as "WAITING" in the statistics if the
5399 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005400
5401 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
5402
5403
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005404redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
5405redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
5406redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005407 Return an HTTP redirection if/unless a condition is matched
5408 May be used in sections : defaults | frontend | listen | backend
5409 no | yes | yes | yes
5410
5411 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01005412 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005413
Willy Tarreau0140f252008-11-19 21:07:09 +01005414 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005415 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005416 the HTTP "Location" header. When used in an "http-request" rule,
5417 <loc> value follows the log-format rules and can include some
5418 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005419
5420 <pfx> With "redirect prefix", the "Location" header is built from the
5421 concatenation of <pfx> and the complete URI path, including the
5422 query string, unless the "drop-query" option is specified (see
5423 below). As a special case, if <pfx> equals exactly "/", then
5424 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005425 redirect to the same URL (for instance, to insert a cookie). When
5426 used in an "http-request" rule, <pfx> value follows the log-format
5427 rules and can include some dynamic values (see Custom Log Format
5428 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005429
5430 <sch> With "redirect scheme", then the "Location" header is built by
5431 concatenating <sch> with "://" then the first occurrence of the
5432 "Host" header, and then the URI path, including the query string
5433 unless the "drop-query" option is specified (see below). If no
5434 path is found or if the path is "*", then "/" is used instead. If
5435 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005436 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005437 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005438 HTTPS. When used in an "http-request" rule, <sch> value follows
5439 the log-format rules and can include some dynamic values (see
5440 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01005441
5442 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005443 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
5444 with 302 used by default if no code is specified. 301 means
5445 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmann6c7351b2015-08-03 11:42:50 +02005446 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005447 cache the redirection. 303 is equivalent to 302 except that the
5448 browser will fetch the location with a GET method. 307 is just
5449 like 302 but makes it clear that the same method must be reused.
5450 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01005451
5452 <option> There are several options which can be specified to adjust the
5453 expected behaviour of a redirection :
5454
5455 - "drop-query"
5456 When this keyword is used in a prefix-based redirection, then the
5457 location will be set without any possible query-string, which is useful
5458 for directing users to a non-secure page for instance. It has no effect
5459 with a location-type redirect.
5460
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005461 - "append-slash"
5462 This keyword may be used in conjunction with "drop-query" to redirect
5463 users who use a URL not ending with a '/' to the same one with the '/'.
5464 It can be useful to ensure that search engines will only see one URL.
5465 For this, a return code 301 is preferred.
5466
Willy Tarreau0140f252008-11-19 21:07:09 +01005467 - "set-cookie NAME[=value]"
5468 A "Set-Cookie" header will be added with NAME (and optionally "=value")
5469 to the response. This is sometimes used to indicate that a user has
5470 been seen, for instance to protect against some types of DoS. No other
5471 cookie option is added, so the cookie will be a session cookie. Note
5472 that for a browser, a sole cookie name without an equal sign is
5473 different from a cookie with an equal sign.
5474
5475 - "clear-cookie NAME[=]"
5476 A "Set-Cookie" header will be added with NAME (and optionally "="), but
5477 with the "Max-Age" attribute set to zero. This will tell the browser to
5478 delete this cookie. It is useful for instance on logout pages. It is
5479 important to note that clearing the cookie "NAME" will not remove a
5480 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
5481 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005482
5483 Example: move the login URL only to HTTPS.
5484 acl clear dst_port 80
5485 acl secure dst_port 8080
5486 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01005487 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01005488 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01005489 acl cookie_set hdr_sub(cookie) SEEN=1
5490
5491 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01005492 redirect prefix https://mysite.com if login_page !secure
5493 redirect prefix http://mysite.com drop-query if login_page !uid_given
5494 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01005495 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005496
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005497 Example: send redirects for request for articles without a '/'.
5498 acl missing_slash path_reg ^/article/[^/]*$
5499 redirect code 301 prefix / drop-query append-slash if missing_slash
5500
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005501 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01005502 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005503
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005504 Example: append 'www.' prefix in front of all hosts not having it
5505 http-request redirect code 301 location www.%[hdr(host)]%[req.uri] \
5506 unless { hdr_beg(host) -i www }
5507
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005508 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005509
5510
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005511redisp (deprecated)
5512redispatch (deprecated)
5513 Enable or disable session redistribution in case of connection failure
5514 May be used in sections: defaults | frontend | listen | backend
5515 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005516 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005517
5518 In HTTP mode, if a server designated by a cookie is down, clients may
5519 definitely stick to it because they cannot flush the cookie, so they will not
5520 be able to access the service anymore.
5521
5522 Specifying "redispatch" will allow the proxy to break their persistence and
5523 redistribute them to a working server.
5524
5525 It also allows to retry last connection to another server in case of multiple
5526 connection failures. Of course, it requires having "retries" set to a nonzero
5527 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01005528
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005529 This form is deprecated, do not use it in any new configuration, use the new
5530 "option redispatch" instead.
5531
5532 See also : "option redispatch"
5533
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005534
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005535reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005536 Add a header at the end of the HTTP request
5537 May be used in sections : defaults | frontend | listen | backend
5538 no | yes | yes | yes
5539 Arguments :
5540 <string> is the complete line to be added. Any space or known delimiter
5541 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005542 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005543
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005544 <cond> is an optional matching condition built from ACLs. It makes it
5545 possible to ignore this rule when other conditions are not met.
5546
Willy Tarreau303c0352008-01-17 19:01:39 +01005547 A new line consisting in <string> followed by a line feed will be added after
5548 the last header of an HTTP request.
5549
5550 Header transformations only apply to traffic which passes through HAProxy,
5551 and not to traffic generated by HAProxy, such as health-checks or error
5552 responses.
5553
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005554 Example : add "X-Proto: SSL" to requests coming via port 81
5555 acl is-ssl dst_port 81
5556 reqadd X-Proto:\ SSL if is-ssl
5557
5558 See also: "rspadd", section 6 about HTTP header manipulation, and section 7
5559 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005560
5561
Willy Tarreau5321c422010-01-28 20:35:13 +01005562reqallow <search> [{if | unless} <cond>]
5563reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005564 Definitely allow an HTTP request if a line matches a regular expression
5565 May be used in sections : defaults | frontend | listen | backend
5566 no | yes | yes | yes
5567 Arguments :
5568 <search> is the regular expression applied to HTTP headers and to the
5569 request line. This is an extended regular expression. Parenthesis
5570 grouping is supported and no preliminary backslash is required.
5571 Any space or known delimiter must be escaped using a backslash
5572 ('\'). The pattern applies to a full line at a time. The
5573 "reqallow" keyword strictly matches case while "reqiallow"
5574 ignores case.
5575
Willy Tarreau5321c422010-01-28 20:35:13 +01005576 <cond> is an optional matching condition built from ACLs. It makes it
5577 possible to ignore this rule when other conditions are not met.
5578
Willy Tarreau303c0352008-01-17 19:01:39 +01005579 A request containing any line which matches extended regular expression
5580 <search> will mark the request as allowed, even if any later test would
5581 result in a deny. The test applies both to the request line and to request
5582 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005583 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005584
5585 It is easier, faster and more powerful to use ACLs to write access policies.
5586 Reqdeny, reqallow and reqpass should be avoided in new designs.
5587
5588 Example :
5589 # allow www.* but refuse *.local
5590 reqiallow ^Host:\ www\.
5591 reqideny ^Host:\ .*\.local
5592
Willy Tarreau5321c422010-01-28 20:35:13 +01005593 See also: "reqdeny", "block", section 6 about HTTP header manipulation, and
5594 section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005595
5596
Willy Tarreau5321c422010-01-28 20:35:13 +01005597reqdel <search> [{if | unless} <cond>]
5598reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005599 Delete all headers matching a regular expression in an HTTP request
5600 May be used in sections : defaults | frontend | listen | backend
5601 no | yes | yes | yes
5602 Arguments :
5603 <search> is the regular expression applied to HTTP headers and to the
5604 request line. This is an extended regular expression. Parenthesis
5605 grouping is supported and no preliminary backslash is required.
5606 Any space or known delimiter must be escaped using a backslash
5607 ('\'). The pattern applies to a full line at a time. The "reqdel"
5608 keyword strictly matches case while "reqidel" ignores case.
5609
Willy Tarreau5321c422010-01-28 20:35:13 +01005610 <cond> is an optional matching condition built from ACLs. It makes it
5611 possible to ignore this rule when other conditions are not met.
5612
Willy Tarreau303c0352008-01-17 19:01:39 +01005613 Any header line matching extended regular expression <search> in the request
5614 will be completely deleted. Most common use of this is to remove unwanted
5615 and/or dangerous headers or cookies from a request before passing it to the
5616 next servers.
5617
5618 Header transformations only apply to traffic which passes through HAProxy,
5619 and not to traffic generated by HAProxy, such as health-checks or error
5620 responses. Keep in mind that header names are not case-sensitive.
5621
5622 Example :
5623 # remove X-Forwarded-For header and SERVER cookie
5624 reqidel ^X-Forwarded-For:.*
5625 reqidel ^Cookie:.*SERVER=
5626
Willy Tarreau5321c422010-01-28 20:35:13 +01005627 See also: "reqadd", "reqrep", "rspdel", section 6 about HTTP header
5628 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005629
5630
Willy Tarreau5321c422010-01-28 20:35:13 +01005631reqdeny <search> [{if | unless} <cond>]
5632reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005633 Deny an HTTP request if a line matches a regular expression
5634 May be used in sections : defaults | frontend | listen | backend
5635 no | yes | yes | yes
5636 Arguments :
5637 <search> is the regular expression applied to HTTP headers and to the
5638 request line. This is an extended regular expression. Parenthesis
5639 grouping is supported and no preliminary backslash is required.
5640 Any space or known delimiter must be escaped using a backslash
5641 ('\'). The pattern applies to a full line at a time. The
5642 "reqdeny" keyword strictly matches case while "reqideny" ignores
5643 case.
5644
Willy Tarreau5321c422010-01-28 20:35:13 +01005645 <cond> is an optional matching condition built from ACLs. It makes it
5646 possible to ignore this rule when other conditions are not met.
5647
Willy Tarreau303c0352008-01-17 19:01:39 +01005648 A request containing any line which matches extended regular expression
5649 <search> will mark the request as denied, even if any later test would
5650 result in an allow. The test applies both to the request line and to request
5651 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005652 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005653
Willy Tarreauced27012008-01-17 20:35:34 +01005654 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005655 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01005656 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01005657
Willy Tarreau303c0352008-01-17 19:01:39 +01005658 It is easier, faster and more powerful to use ACLs to write access policies.
5659 Reqdeny, reqallow and reqpass should be avoided in new designs.
5660
5661 Example :
5662 # refuse *.local, then allow www.*
5663 reqideny ^Host:\ .*\.local
5664 reqiallow ^Host:\ www\.
5665
Willy Tarreau5321c422010-01-28 20:35:13 +01005666 See also: "reqallow", "rspdeny", "block", section 6 about HTTP header
5667 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005668
5669
Willy Tarreau5321c422010-01-28 20:35:13 +01005670reqpass <search> [{if | unless} <cond>]
5671reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005672 Ignore any HTTP request line matching a regular expression in next rules
5673 May be used in sections : defaults | frontend | listen | backend
5674 no | yes | yes | yes
5675 Arguments :
5676 <search> is the regular expression applied to HTTP headers and to the
5677 request line. This is an extended regular expression. Parenthesis
5678 grouping is supported and no preliminary backslash is required.
5679 Any space or known delimiter must be escaped using a backslash
5680 ('\'). The pattern applies to a full line at a time. The
5681 "reqpass" keyword strictly matches case while "reqipass" ignores
5682 case.
5683
Willy Tarreau5321c422010-01-28 20:35:13 +01005684 <cond> is an optional matching condition built from ACLs. It makes it
5685 possible to ignore this rule when other conditions are not met.
5686
Willy Tarreau303c0352008-01-17 19:01:39 +01005687 A request containing any line which matches extended regular expression
5688 <search> will skip next rules, without assigning any deny or allow verdict.
5689 The test applies both to the request line and to request headers. Keep in
5690 mind that URLs in request line are case-sensitive while header names are not.
5691
5692 It is easier, faster and more powerful to use ACLs to write access policies.
5693 Reqdeny, reqallow and reqpass should be avoided in new designs.
5694
5695 Example :
5696 # refuse *.local, then allow www.*, but ignore "www.private.local"
5697 reqipass ^Host:\ www.private\.local
5698 reqideny ^Host:\ .*\.local
5699 reqiallow ^Host:\ www\.
5700
Willy Tarreau5321c422010-01-28 20:35:13 +01005701 See also: "reqallow", "reqdeny", "block", section 6 about HTTP header
5702 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005703
5704
Willy Tarreau5321c422010-01-28 20:35:13 +01005705reqrep <search> <string> [{if | unless} <cond>]
5706reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005707 Replace a regular expression with a string in an HTTP request line
5708 May be used in sections : defaults | frontend | listen | backend
5709 no | yes | yes | yes
5710 Arguments :
5711 <search> is the regular expression applied to HTTP headers and to the
5712 request line. This is an extended regular expression. Parenthesis
5713 grouping is supported and no preliminary backslash is required.
5714 Any space or known delimiter must be escaped using a backslash
5715 ('\'). The pattern applies to a full line at a time. The "reqrep"
5716 keyword strictly matches case while "reqirep" ignores case.
5717
5718 <string> is the complete line to be added. Any space or known delimiter
5719 must be escaped using a backslash ('\'). References to matched
5720 pattern groups are possible using the common \N form, with N
5721 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005722 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005723
Willy Tarreau5321c422010-01-28 20:35:13 +01005724 <cond> is an optional matching condition built from ACLs. It makes it
5725 possible to ignore this rule when other conditions are not met.
5726
Willy Tarreau303c0352008-01-17 19:01:39 +01005727 Any line matching extended regular expression <search> in the request (both
5728 the request line and header lines) will be completely replaced with <string>.
5729 Most common use of this is to rewrite URLs or domain names in "Host" headers.
5730
5731 Header transformations only apply to traffic which passes through HAProxy,
5732 and not to traffic generated by HAProxy, such as health-checks or error
5733 responses. Note that for increased readability, it is suggested to add enough
5734 spaces between the request and the response. Keep in mind that URLs in
5735 request line are case-sensitive while header names are not.
5736
5737 Example :
5738 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04005739 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01005740 # replace "www.mydomain.com" with "www" in the host name.
5741 reqirep ^Host:\ www.mydomain.com Host:\ www
5742
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04005743 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", section 6 about
5744 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005745
5746
Willy Tarreau5321c422010-01-28 20:35:13 +01005747reqtarpit <search> [{if | unless} <cond>]
5748reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005749 Tarpit an HTTP request containing a line matching a regular expression
5750 May be used in sections : defaults | frontend | listen | backend
5751 no | yes | yes | yes
5752 Arguments :
5753 <search> is the regular expression applied to HTTP headers and to the
5754 request line. This is an extended regular expression. Parenthesis
5755 grouping is supported and no preliminary backslash is required.
5756 Any space or known delimiter must be escaped using a backslash
5757 ('\'). The pattern applies to a full line at a time. The
5758 "reqtarpit" keyword strictly matches case while "reqitarpit"
5759 ignores case.
5760
Willy Tarreau5321c422010-01-28 20:35:13 +01005761 <cond> is an optional matching condition built from ACLs. It makes it
5762 possible to ignore this rule when other conditions are not met.
5763
Willy Tarreau303c0352008-01-17 19:01:39 +01005764 A request containing any line which matches extended regular expression
5765 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01005766 be kept open for a pre-defined time, then will return an HTTP error 500 so
5767 that the attacker does not suspect it has been tarpitted. The status 500 will
5768 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01005769 delay is defined by "timeout tarpit", or "timeout connect" if the former is
5770 not set.
5771
5772 The goal of the tarpit is to slow down robots attacking servers with
5773 identifiable requests. Many robots limit their outgoing number of connections
5774 and stay connected waiting for a reply which can take several minutes to
5775 come. Depending on the environment and attack, it may be particularly
5776 efficient at reducing the load on the network and firewalls.
5777
Willy Tarreau5321c422010-01-28 20:35:13 +01005778 Examples :
Willy Tarreau303c0352008-01-17 19:01:39 +01005779 # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but
5780 # block all others.
5781 reqipass ^User-Agent:\.*(Mozilla|MSIE)
5782 reqitarpit ^User-Agent:
5783
Willy Tarreau5321c422010-01-28 20:35:13 +01005784 # block bad guys
5785 acl badguys src 10.1.0.3 172.16.13.20/28
5786 reqitarpit . if badguys
5787
5788 See also: "reqallow", "reqdeny", "reqpass", section 6 about HTTP header
5789 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005790
5791
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02005792retries <value>
5793 Set the number of retries to perform on a server after a connection failure
5794 May be used in sections: defaults | frontend | listen | backend
5795 yes | no | yes | yes
5796 Arguments :
5797 <value> is the number of times a connection attempt should be retried on
5798 a server when a connection either is refused or times out. The
5799 default value is 3.
5800
5801 It is important to understand that this value applies to the number of
5802 connection attempts, not full requests. When a connection has effectively
5803 been established to a server, there will be no more retry.
5804
5805 In order to avoid immediate reconnections to a server which is restarting,
5806 a turn-around timer of 1 second is applied before a retry occurs.
5807
5808 When "option redispatch" is set, the last retry may be performed on another
5809 server even if a cookie references a different server.
5810
5811 See also : "option redispatch"
5812
5813
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005814rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005815 Add a header at the end of the HTTP response
5816 May be used in sections : defaults | frontend | listen | backend
5817 no | yes | yes | yes
5818 Arguments :
5819 <string> is the complete line to be added. Any space or known delimiter
5820 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005821 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005822
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005823 <cond> is an optional matching condition built from ACLs. It makes it
5824 possible to ignore this rule when other conditions are not met.
5825
Willy Tarreau303c0352008-01-17 19:01:39 +01005826 A new line consisting in <string> followed by a line feed will be added after
5827 the last header of an HTTP response.
5828
5829 Header transformations only apply to traffic which passes through HAProxy,
5830 and not to traffic generated by HAProxy, such as health-checks or error
5831 responses.
5832
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005833 See also: "reqadd", section 6 about HTTP header manipulation, and section 7
5834 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005835
5836
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005837rspdel <search> [{if | unless} <cond>]
5838rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005839 Delete all headers matching a regular expression in an HTTP response
5840 May be used in sections : defaults | frontend | listen | backend
5841 no | yes | yes | yes
5842 Arguments :
5843 <search> is the regular expression applied to HTTP headers and to the
5844 response line. This is an extended regular expression, so
5845 parenthesis grouping is supported and no preliminary backslash
5846 is required. Any space or known delimiter must be escaped using
5847 a backslash ('\'). The pattern applies to a full line at a time.
5848 The "rspdel" keyword strictly matches case while "rspidel"
5849 ignores case.
5850
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005851 <cond> is an optional matching condition built from ACLs. It makes it
5852 possible to ignore this rule when other conditions are not met.
5853
Willy Tarreau303c0352008-01-17 19:01:39 +01005854 Any header line matching extended regular expression <search> in the response
5855 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005856 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01005857 client.
5858
5859 Header transformations only apply to traffic which passes through HAProxy,
5860 and not to traffic generated by HAProxy, such as health-checks or error
5861 responses. Keep in mind that header names are not case-sensitive.
5862
5863 Example :
5864 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02005865 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01005866
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005867 See also: "rspadd", "rsprep", "reqdel", section 6 about HTTP header
5868 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005869
5870
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005871rspdeny <search> [{if | unless} <cond>]
5872rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005873 Block an HTTP response if a line matches a regular expression
5874 May be used in sections : defaults | frontend | listen | backend
5875 no | yes | yes | yes
5876 Arguments :
5877 <search> is the regular expression applied to HTTP headers and to the
5878 response line. This is an extended regular expression, so
5879 parenthesis grouping is supported and no preliminary backslash
5880 is required. Any space or known delimiter must be escaped using
5881 a backslash ('\'). The pattern applies to a full line at a time.
5882 The "rspdeny" keyword strictly matches case while "rspideny"
5883 ignores case.
5884
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005885 <cond> is an optional matching condition built from ACLs. It makes it
5886 possible to ignore this rule when other conditions are not met.
5887
Willy Tarreau303c0352008-01-17 19:01:39 +01005888 A response containing any line which matches extended regular expression
5889 <search> will mark the request as denied. The test applies both to the
5890 response line and to response headers. Keep in mind that header names are not
5891 case-sensitive.
5892
5893 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01005894 block the response before it reaches the client. If a response is denied, it
5895 will be replaced with an HTTP 502 error so that the client never retrieves
5896 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01005897
5898 It is easier, faster and more powerful to use ACLs to write access policies.
5899 Rspdeny should be avoided in new designs.
5900
5901 Example :
5902 # Ensure that no content type matching ms-word will leak
5903 rspideny ^Content-type:\.*/ms-word
5904
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005905 See also: "reqdeny", "acl", "block", section 6 about HTTP header manipulation
5906 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005907
5908
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005909rsprep <search> <string> [{if | unless} <cond>]
5910rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005911 Replace a regular expression with a string in an HTTP response line
5912 May be used in sections : defaults | frontend | listen | backend
5913 no | yes | yes | yes
5914 Arguments :
5915 <search> is the regular expression applied to HTTP headers and to the
5916 response line. This is an extended regular expression, so
5917 parenthesis grouping is supported and no preliminary backslash
5918 is required. Any space or known delimiter must be escaped using
5919 a backslash ('\'). The pattern applies to a full line at a time.
5920 The "rsprep" keyword strictly matches case while "rspirep"
5921 ignores case.
5922
5923 <string> is the complete line to be added. Any space or known delimiter
5924 must be escaped using a backslash ('\'). References to matched
5925 pattern groups are possible using the common \N form, with N
5926 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005927 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005928
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005929 <cond> is an optional matching condition built from ACLs. It makes it
5930 possible to ignore this rule when other conditions are not met.
5931
Willy Tarreau303c0352008-01-17 19:01:39 +01005932 Any line matching extended regular expression <search> in the response (both
5933 the response line and header lines) will be completely replaced with
5934 <string>. Most common use of this is to rewrite Location headers.
5935
5936 Header transformations only apply to traffic which passes through HAProxy,
5937 and not to traffic generated by HAProxy, such as health-checks or error
5938 responses. Note that for increased readability, it is suggested to add enough
5939 spaces between the request and the response. Keep in mind that header names
5940 are not case-sensitive.
5941
5942 Example :
5943 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
5944 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
5945
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005946 See also: "rspadd", "rspdel", "reqrep", section 6 about HTTP header
5947 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005948
5949
David du Colombier486df472011-03-17 10:40:26 +01005950server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005951 Declare a server in a backend
5952 May be used in sections : defaults | frontend | listen | backend
5953 no | no | yes | yes
5954 Arguments :
5955 <name> is the internal name assigned to this server. This name will
Cyril Bonté941a0c62012-10-15 19:44:24 +02005956 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05005957 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005958
David du Colombier486df472011-03-17 10:40:26 +01005959 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
5960 resolvable hostname is supported, but this name will be resolved
5961 during start-up. Address "0.0.0.0" or "*" has a special meaning.
5962 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02005963 address as the one from the client connection. This is useful in
5964 transparent proxy architectures where the client's connection is
5965 intercepted and haproxy must forward to the original destination
5966 address. This is more or less what the "transparent" keyword does
5967 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01005968 to report statistics. Optionally, an address family prefix may be
5969 used before the address to force the family regardless of the
5970 address format, which can be useful to specify a path to a unix
5971 socket with no slash ('/'). Currently supported prefixes are :
5972 - 'ipv4@' -> address is always IPv4
5973 - 'ipv6@' -> address is always IPv6
5974 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02005975 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01005976 Any part of the address string may reference any number of
5977 environment variables by preceding their name with a dollar
5978 sign ('$') and optionally enclosing them with braces ('{}'),
5979 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005980
Willy Tarreaub6205fd2012-09-24 12:27:33 +02005981 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005982 be sent to this port. If unset, the same port the client
5983 connected to will be used. The port may also be prefixed by a "+"
5984 or a "-". In this case, the server's port will be determined by
5985 adding this value to the client's port.
5986
5987 <param*> is a list of parameters for this server. The "server" keywords
5988 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005989 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005990
5991 Examples :
5992 server first 10.1.1.1:1080 cookie first check inter 1000
5993 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01005994 server transp ipv4@
Willy Tarreaudad36a32013-03-11 01:20:04 +01005995 server backup ${SRV_BACKUP}:1080 backup
5996 server www1_dc1 ${LAN_DC1}.101:80
5997 server www1_dc2 ${LAN_DC2}.101:80
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005998
Mark Lamourinec2247f02012-01-04 13:02:01 -05005999 See also: "default-server", "http-send-name-header" and section 5 about
6000 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006001
6002
6003source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02006004source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006005source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006006 Set the source address for outgoing connections
6007 May be used in sections : defaults | frontend | listen | backend
6008 yes | no | yes | yes
6009 Arguments :
6010 <addr> is the IPv4 address HAProxy will bind to before connecting to a
6011 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01006012
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006013 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01006014 the most appropriate address to reach its destination. Optionally
6015 an address family prefix may be used before the address to force
6016 the family regardless of the address format, which can be useful
6017 to specify a path to a unix socket with no slash ('/'). Currently
6018 supported prefixes are :
6019 - 'ipv4@' -> address is always IPv4
6020 - 'ipv6@' -> address is always IPv6
6021 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02006022 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01006023 Any part of the address string may reference any number of
6024 environment variables by preceding their name with a dollar
6025 sign ('$') and optionally enclosing them with braces ('{}'),
6026 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006027
6028 <port> is an optional port. It is normally not needed but may be useful
6029 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02006030 the system will select a free port. Note that port ranges are not
6031 supported in the backend. If you want to force port ranges, you
6032 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006033
6034 <addr2> is the IP address to present to the server when connections are
6035 forwarded in full transparent proxy mode. This is currently only
6036 supported on some patched Linux kernels. When this address is
6037 specified, clients connecting to the server will be presented
6038 with this address, while health checks will still use the address
6039 <addr>.
6040
6041 <port2> is the optional port to present to the server when connections
6042 are forwarded in full transparent proxy mode (see <addr2> above).
6043 The default value of zero means the system will select a free
6044 port.
6045
Willy Tarreaubce70882009-09-07 11:51:47 +02006046 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
6047 This is the name of a comma-separated header list which can
6048 contain multiple IP addresses. By default, the last occurrence is
6049 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01006050 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02006051 by previous proxy, typically Stunnel. In order to use another
6052 occurrence from the last one, please see the <occ> parameter
6053 below. When the header (or occurrence) is not found, no binding
6054 is performed so that the proxy's default IP address is used. Also
6055 keep in mind that the header name is case insensitive, as for any
6056 HTTP header.
6057
6058 <occ> is the occurrence number of a value to be used in a multi-value
6059 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006060 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02006061 address. Positive values indicate a position from the first
6062 occurrence, 1 being the first one. Negative values indicate
6063 positions relative to the last one, -1 being the last one. This
6064 is helpful for situations where an X-Forwarded-For header is set
6065 at the entry point of an infrastructure and must be used several
6066 proxy layers away. When this value is not specified, -1 is
6067 assumed. Passing a zero here disables the feature.
6068
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006069 <name> is an optional interface name to which to bind to for outgoing
6070 traffic. On systems supporting this features (currently, only
6071 Linux), this allows one to bind all traffic to the server to
6072 this interface even if it is not the one the system would select
6073 based on routing tables. This should be used with extreme care.
6074 Note that using this option requires root privileges.
6075
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006076 The "source" keyword is useful in complex environments where a specific
6077 address only is allowed to connect to the servers. It may be needed when a
6078 private address must be used through a public gateway for instance, and it is
6079 known that the system cannot determine the adequate source address by itself.
6080
6081 An extension which is available on certain patched Linux kernels may be used
6082 through the "usesrc" optional keyword. It makes it possible to connect to the
6083 servers with an IP address which does not belong to the system itself. This
6084 is called "full transparent proxy mode". For this to work, the destination
6085 servers have to route their traffic back to this address through the machine
6086 running HAProxy, and IP forwarding must generally be enabled on this machine.
6087
6088 In this "full transparent proxy" mode, it is possible to force a specific IP
6089 address to be presented to the servers. This is not much used in fact. A more
6090 common use is to tell HAProxy to present the client's IP address. For this,
6091 there are two methods :
6092
6093 - present the client's IP and port addresses. This is the most transparent
6094 mode, but it can cause problems when IP connection tracking is enabled on
6095 the machine, because a same connection may be seen twice with different
6096 states. However, this solution presents the huge advantage of not
6097 limiting the system to the 64k outgoing address+port couples, because all
6098 of the client ranges may be used.
6099
6100 - present only the client's IP address and select a spare port. This
6101 solution is still quite elegant but slightly less transparent (downstream
6102 firewalls logs will not match upstream's). It also presents the downside
6103 of limiting the number of concurrent connections to the usual 64k ports.
6104 However, since the upstream and downstream ports are different, local IP
6105 connection tracking on the machine will not be upset by the reuse of the
6106 same session.
6107
6108 Note that depending on the transparent proxy technology used, it may be
6109 required to force the source address. In fact, cttproxy version 2 requires an
6110 IP address in <addr> above, and does not support setting of "0.0.0.0" as the
6111 IP address because it creates NAT entries which much match the exact outgoing
6112 address. Tproxy version 4 and some other kernel patches which work in pure
6113 forwarding mode generally will not have this limitation.
6114
6115 This option sets the default source for all servers in the backend. It may
6116 also be specified in a "defaults" section. Finer source address specification
6117 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006118 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006119
Baptiste Assmannea31f222015-07-17 21:59:42 +02006120 In order to work, "usesrc" requires root privileges.
6121
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006122 Examples :
6123 backend private
6124 # Connect to the servers using our 192.168.1.200 source address
6125 source 192.168.1.200
6126
6127 backend transparent_ssl1
6128 # Connect to the SSL farm from the client's source address
6129 source 192.168.1.200 usesrc clientip
6130
6131 backend transparent_ssl2
6132 # Connect to the SSL farm from the client's source address and port
6133 # not recommended if IP conntrack is present on the local machine.
6134 source 192.168.1.200 usesrc client
6135
6136 backend transparent_ssl3
6137 # Connect to the SSL farm from the client's source address. It
6138 # is more conntrack-friendly.
6139 source 192.168.1.200 usesrc clientip
6140
6141 backend transparent_smtp
6142 # Connect to the SMTP farm from the client's source address/port
6143 # with Tproxy version 4.
6144 source 0.0.0.0 usesrc clientip
6145
Willy Tarreaubce70882009-09-07 11:51:47 +02006146 backend transparent_http
6147 # Connect to the servers using the client's IP as seen by previous
6148 # proxy.
6149 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
6150
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006151 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006152 the Linux kernel on www.balabit.com, the "bind" keyword.
6153
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006154
Willy Tarreau844e3c52008-01-11 16:28:18 +01006155srvtimeout <timeout> (deprecated)
6156 Set the maximum inactivity time on the server side.
6157 May be used in sections : defaults | frontend | listen | backend
6158 yes | no | yes | yes
6159 Arguments :
6160 <timeout> is the timeout value specified in milliseconds by default, but
6161 can be in any other unit if the number is suffixed by the unit,
6162 as explained at the top of this document.
6163
6164 The inactivity timeout applies when the server is expected to acknowledge or
6165 send data. In HTTP mode, this timeout is particularly important to consider
6166 during the first phase of the server's response, when it has to send the
6167 headers, as it directly represents the server's processing time for the
6168 request. To find out what value to put there, it's often good to start with
6169 what would be considered as unacceptable response times, then check the logs
6170 to observe the response time distribution, and adjust the value accordingly.
6171
6172 The value is specified in milliseconds by default, but can be in any other
6173 unit if the number is suffixed by the unit, as specified at the top of this
6174 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
6175 recommended that the client timeout remains equal to the server timeout in
6176 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006177 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01006178 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreaud72758d2010-01-12 10:42:19 +01006179 seconds (eg: 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01006180
6181 This parameter is specific to backends, but can be specified once for all in
6182 "defaults" sections. This is in fact one of the easiest solutions not to
6183 forget about it. An unspecified timeout results in an infinite timeout, which
6184 is not recommended. Such a usage is accepted and works but reports a warning
6185 during startup because it may results in accumulation of expired sessions in
6186 the system if the system's timeouts are not configured either.
6187
6188 This parameter is provided for compatibility but is currently deprecated.
6189 Please use "timeout server" instead.
6190
Willy Tarreauce887fd2012-05-12 12:50:00 +02006191 See also : "timeout server", "timeout tunnel", "timeout client" and
6192 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01006193
6194
Cyril Bonté66c327d2010-10-12 00:14:37 +02006195stats admin { if | unless } <cond>
6196 Enable statistics admin level if/unless a condition is matched
6197 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006198 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02006199
6200 This statement enables the statistics admin level if/unless a condition is
6201 matched.
6202
6203 The admin level allows to enable/disable servers from the web interface. By
6204 default, statistics page is read-only for security reasons.
6205
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006206 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6207 unless you know what you do : memory is not shared between the
6208 processes, which can result in random behaviours.
6209
Cyril Bonté23b39d92011-02-10 22:54:44 +01006210 Currently, the POST request is limited to the buffer size minus the reserved
6211 buffer space, which means that if the list of servers is too long, the
6212 request won't be processed. It is recommended to alter few servers at a
6213 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006214
6215 Example :
6216 # statistics admin level only for localhost
6217 backend stats_localhost
6218 stats enable
6219 stats admin if LOCALHOST
6220
6221 Example :
6222 # statistics admin level always enabled because of the authentication
6223 backend stats_auth
6224 stats enable
6225 stats auth admin:AdMiN123
6226 stats admin if TRUE
6227
6228 Example :
6229 # statistics admin level depends on the authenticated user
6230 userlist stats-auth
6231 group admin users admin
6232 user admin insecure-password AdMiN123
6233 group readonly users haproxy
6234 user haproxy insecure-password haproxy
6235
6236 backend stats_auth
6237 stats enable
6238 acl AUTH http_auth(stats-auth)
6239 acl AUTH_ADMIN http_auth_group(stats-auth) admin
6240 stats http-request auth unless AUTH
6241 stats admin if AUTH_ADMIN
6242
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006243 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
6244 "bind-process", section 3.4 about userlists and section 7 about
6245 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006246
6247
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006248stats auth <user>:<passwd>
6249 Enable statistics with authentication and grant access to an account
6250 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006251 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006252 Arguments :
6253 <user> is a user name to grant access to
6254
6255 <passwd> is the cleartext password associated to this user
6256
6257 This statement enables statistics with default settings, and restricts access
6258 to declared users only. It may be repeated as many times as necessary to
6259 allow as many users as desired. When a user tries to access the statistics
6260 without a valid account, a "401 Forbidden" response will be returned so that
6261 the browser asks the user to provide a valid user and password. The real
6262 which will be returned to the browser is configurable using "stats realm".
6263
6264 Since the authentication method is HTTP Basic Authentication, the passwords
6265 circulate in cleartext on the network. Thus, it was decided that the
6266 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006267 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006268
6269 It is also possible to reduce the scope of the proxies which appear in the
6270 report using "stats scope".
6271
6272 Though this statement alone is enough to enable statistics reporting, it is
6273 recommended to set all other settings in order to avoid relying on default
6274 unobvious parameters.
6275
6276 Example :
6277 # public access (limited to this backend only)
6278 backend public_www
6279 server srv1 192.168.0.1:80
6280 stats enable
6281 stats hide-version
6282 stats scope .
6283 stats uri /admin?stats
6284 stats realm Haproxy\ Statistics
6285 stats auth admin1:AdMiN123
6286 stats auth admin2:AdMiN321
6287
6288 # internal monitoring access (unlimited)
6289 backend private_monitoring
6290 stats enable
6291 stats uri /admin?stats
6292 stats refresh 5s
6293
6294 See also : "stats enable", "stats realm", "stats scope", "stats uri"
6295
6296
6297stats enable
6298 Enable statistics reporting with default settings
6299 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006300 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006301 Arguments : none
6302
6303 This statement enables statistics reporting with default settings defined
6304 at build time. Unless stated otherwise, these settings are used :
6305 - stats uri : /haproxy?stats
6306 - stats realm : "HAProxy Statistics"
6307 - stats auth : no authentication
6308 - stats scope : no restriction
6309
6310 Though this statement alone is enough to enable statistics reporting, it is
6311 recommended to set all other settings in order to avoid relying on default
6312 unobvious parameters.
6313
6314 Example :
6315 # public access (limited to this backend only)
6316 backend public_www
6317 server srv1 192.168.0.1:80
6318 stats enable
6319 stats hide-version
6320 stats scope .
6321 stats uri /admin?stats
6322 stats realm Haproxy\ Statistics
6323 stats auth admin1:AdMiN123
6324 stats auth admin2:AdMiN321
6325
6326 # internal monitoring access (unlimited)
6327 backend private_monitoring
6328 stats enable
6329 stats uri /admin?stats
6330 stats refresh 5s
6331
6332 See also : "stats auth", "stats realm", "stats uri"
6333
6334
Willy Tarreaud63335a2010-02-26 12:56:52 +01006335stats hide-version
6336 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006337 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006338 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006339 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006340
Willy Tarreaud63335a2010-02-26 12:56:52 +01006341 By default, the stats page reports some useful status information along with
6342 the statistics. Among them is HAProxy's version. However, it is generally
6343 considered dangerous to report precise version to anyone, as it can help them
6344 target known weaknesses with specific attacks. The "stats hide-version"
6345 statement removes the version from the statistics report. This is recommended
6346 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006347
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006348 Though this statement alone is enough to enable statistics reporting, it is
6349 recommended to set all other settings in order to avoid relying on default
6350 unobvious parameters.
6351
Willy Tarreaud63335a2010-02-26 12:56:52 +01006352 Example :
6353 # public access (limited to this backend only)
6354 backend public_www
6355 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006356 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006357 stats hide-version
6358 stats scope .
6359 stats uri /admin?stats
6360 stats realm Haproxy\ Statistics
6361 stats auth admin1:AdMiN123
6362 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006363
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006364 # internal monitoring access (unlimited)
6365 backend private_monitoring
6366 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006367 stats uri /admin?stats
6368 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01006369
Willy Tarreaud63335a2010-02-26 12:56:52 +01006370 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006371
Willy Tarreau983e01e2010-01-11 18:42:06 +01006372
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02006373stats http-request { allow | deny | auth [realm <realm>] }
6374 [ { if | unless } <condition> ]
6375 Access control for statistics
6376
6377 May be used in sections: defaults | frontend | listen | backend
6378 no | no | yes | yes
6379
6380 As "http-request", these set of options allow to fine control access to
6381 statistics. Each option may be followed by if/unless and acl.
6382 First option with matched condition (or option without condition) is final.
6383 For "deny" a 403 error will be returned, for "allow" normal processing is
6384 performed, for "auth" a 401/407 error code is returned so the client
6385 should be asked to enter a username and password.
6386
6387 There is no fixed limit to the number of http-request statements per
6388 instance.
6389
6390 See also : "http-request", section 3.4 about userlists and section 7
6391 about ACL usage.
6392
6393
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006394stats realm <realm>
6395 Enable statistics and set authentication realm
6396 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006397 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006398 Arguments :
6399 <realm> is the name of the HTTP Basic Authentication realm reported to
6400 the browser. The browser uses it to display it in the pop-up
6401 inviting the user to enter a valid username and password.
6402
6403 The realm is read as a single word, so any spaces in it should be escaped
6404 using a backslash ('\').
6405
6406 This statement is useful only in conjunction with "stats auth" since it is
6407 only related to authentication.
6408
6409 Though this statement alone is enough to enable statistics reporting, it is
6410 recommended to set all other settings in order to avoid relying on default
6411 unobvious parameters.
6412
6413 Example :
6414 # public access (limited to this backend only)
6415 backend public_www
6416 server srv1 192.168.0.1:80
6417 stats enable
6418 stats hide-version
6419 stats scope .
6420 stats uri /admin?stats
6421 stats realm Haproxy\ Statistics
6422 stats auth admin1:AdMiN123
6423 stats auth admin2:AdMiN321
6424
6425 # internal monitoring access (unlimited)
6426 backend private_monitoring
6427 stats enable
6428 stats uri /admin?stats
6429 stats refresh 5s
6430
6431 See also : "stats auth", "stats enable", "stats uri"
6432
6433
6434stats refresh <delay>
6435 Enable statistics with automatic refresh
6436 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006437 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006438 Arguments :
6439 <delay> is the suggested refresh delay, specified in seconds, which will
6440 be returned to the browser consulting the report page. While the
6441 browser is free to apply any delay, it will generally respect it
6442 and refresh the page this every seconds. The refresh interval may
6443 be specified in any other non-default time unit, by suffixing the
6444 unit after the value, as explained at the top of this document.
6445
6446 This statement is useful on monitoring displays with a permanent page
6447 reporting the load balancer's activity. When set, the HTML report page will
6448 include a link "refresh"/"stop refresh" so that the user can select whether
6449 he wants automatic refresh of the page or not.
6450
6451 Though this statement alone is enough to enable statistics reporting, it is
6452 recommended to set all other settings in order to avoid relying on default
6453 unobvious parameters.
6454
6455 Example :
6456 # public access (limited to this backend only)
6457 backend public_www
6458 server srv1 192.168.0.1:80
6459 stats enable
6460 stats hide-version
6461 stats scope .
6462 stats uri /admin?stats
6463 stats realm Haproxy\ Statistics
6464 stats auth admin1:AdMiN123
6465 stats auth admin2:AdMiN321
6466
6467 # internal monitoring access (unlimited)
6468 backend private_monitoring
6469 stats enable
6470 stats uri /admin?stats
6471 stats refresh 5s
6472
6473 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6474
6475
6476stats scope { <name> | "." }
6477 Enable statistics and limit access scope
6478 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006479 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006480 Arguments :
6481 <name> is the name of a listen, frontend or backend section to be
6482 reported. The special name "." (a single dot) designates the
6483 section in which the statement appears.
6484
6485 When this statement is specified, only the sections enumerated with this
6486 statement will appear in the report. All other ones will be hidden. This
6487 statement may appear as many times as needed if multiple sections need to be
6488 reported. Please note that the name checking is performed as simple string
6489 comparisons, and that it is never checked that a give section name really
6490 exists.
6491
6492 Though this statement alone is enough to enable statistics reporting, it is
6493 recommended to set all other settings in order to avoid relying on default
6494 unobvious parameters.
6495
6496 Example :
6497 # public access (limited to this backend only)
6498 backend public_www
6499 server srv1 192.168.0.1:80
6500 stats enable
6501 stats hide-version
6502 stats scope .
6503 stats uri /admin?stats
6504 stats realm Haproxy\ Statistics
6505 stats auth admin1:AdMiN123
6506 stats auth admin2:AdMiN321
6507
6508 # internal monitoring access (unlimited)
6509 backend private_monitoring
6510 stats enable
6511 stats uri /admin?stats
6512 stats refresh 5s
6513
6514 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6515
Willy Tarreaud63335a2010-02-26 12:56:52 +01006516
Willy Tarreauc9705a12010-07-27 20:05:50 +02006517stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01006518 Enable reporting of a description on the statistics page.
6519 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006520 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006521
Willy Tarreauc9705a12010-07-27 20:05:50 +02006522 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01006523 description from global section is automatically used instead.
6524
6525 This statement is useful for users that offer shared services to their
6526 customers, where node or description should be different for each customer.
6527
6528 Though this statement alone is enough to enable statistics reporting, it is
6529 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006530 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006531
6532 Example :
6533 # internal monitoring access (unlimited)
6534 backend private_monitoring
6535 stats enable
6536 stats show-desc Master node for Europe, Asia, Africa
6537 stats uri /admin?stats
6538 stats refresh 5s
6539
6540 See also: "show-node", "stats enable", "stats uri" and "description" in
6541 global section.
6542
6543
6544stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02006545 Enable reporting additional information on the statistics page
6546 May be used in sections : defaults | frontend | listen | backend
6547 yes | yes | yes | yes
6548 Arguments : none
6549
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006550 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01006551 - cap: capabilities (proxy)
6552 - mode: one of tcp, http or health (proxy)
6553 - id: SNMP ID (proxy, socket, server)
6554 - IP (socket, server)
6555 - cookie (backend, server)
6556
6557 Though this statement alone is enough to enable statistics reporting, it is
6558 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006559 unobvious parameters. Default behaviour is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006560
6561 See also: "stats enable", "stats uri".
6562
6563
6564stats show-node [ <name> ]
6565 Enable reporting of a host name on the statistics page.
6566 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006567 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006568 Arguments:
6569 <name> is an optional name to be reported. If unspecified, the
6570 node name from global section is automatically used instead.
6571
6572 This statement is useful for users that offer shared services to their
6573 customers, where node or description might be different on a stats page
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006574 provided for each customer. Default behaviour is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006575
6576 Though this statement alone is enough to enable statistics reporting, it is
6577 recommended to set all other settings in order to avoid relying on default
6578 unobvious parameters.
6579
6580 Example:
6581 # internal monitoring access (unlimited)
6582 backend private_monitoring
6583 stats enable
6584 stats show-node Europe-1
6585 stats uri /admin?stats
6586 stats refresh 5s
6587
6588 See also: "show-desc", "stats enable", "stats uri", and "node" in global
6589 section.
6590
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006591
6592stats uri <prefix>
6593 Enable statistics and define the URI prefix to access them
6594 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006595 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006596 Arguments :
6597 <prefix> is the prefix of any URI which will be redirected to stats. This
6598 prefix may contain a question mark ('?') to indicate part of a
6599 query string.
6600
6601 The statistics URI is intercepted on the relayed traffic, so it appears as a
6602 page within the normal application. It is strongly advised to ensure that the
6603 selected URI will never appear in the application, otherwise it will never be
6604 possible to reach it in the application.
6605
6606 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006607 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006608 It is generally a good idea to include a question mark in the URI so that
6609 intermediate proxies refrain from caching the results. Also, since any string
6610 beginning with the prefix will be accepted as a stats request, the question
6611 mark helps ensuring that no valid URI will begin with the same words.
6612
6613 It is sometimes very convenient to use "/" as the URI prefix, and put that
6614 statement in a "listen" instance of its own. That makes it easy to dedicate
6615 an address or a port to statistics only.
6616
6617 Though this statement alone is enough to enable statistics reporting, it is
6618 recommended to set all other settings in order to avoid relying on default
6619 unobvious parameters.
6620
6621 Example :
6622 # public access (limited to this backend only)
6623 backend public_www
6624 server srv1 192.168.0.1:80
6625 stats enable
6626 stats hide-version
6627 stats scope .
6628 stats uri /admin?stats
6629 stats realm Haproxy\ Statistics
6630 stats auth admin1:AdMiN123
6631 stats auth admin2:AdMiN321
6632
6633 # internal monitoring access (unlimited)
6634 backend private_monitoring
6635 stats enable
6636 stats uri /admin?stats
6637 stats refresh 5s
6638
6639 See also : "stats auth", "stats enable", "stats realm"
6640
6641
Willy Tarreaud63335a2010-02-26 12:56:52 +01006642stick match <pattern> [table <table>] [{if | unless} <cond>]
6643 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006644 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01006645 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006646
6647 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006648 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006649 describes what elements of the incoming request or connection
6650 will be analysed in the hope to find a matching entry in a
6651 stickiness table. This rule is mandatory.
6652
6653 <table> is an optional stickiness table name. If unspecified, the same
6654 backend's table is used. A stickiness table is declared using
6655 the "stick-table" statement.
6656
6657 <cond> is an optional matching condition. It makes it possible to match
6658 on a certain criterion only when other conditions are met (or
6659 not met). For instance, it could be used to match on a source IP
6660 address except when a request passes through a known proxy, in
6661 which case we'd match on a header containing that IP address.
6662
6663 Some protocols or applications require complex stickiness rules and cannot
6664 always simply rely on cookies nor hashing. The "stick match" statement
6665 describes a rule to extract the stickiness criterion from an incoming request
6666 or connection. See section 7 for a complete list of possible patterns and
6667 transformation rules.
6668
6669 The table has to be declared using the "stick-table" statement. It must be of
6670 a type compatible with the pattern. By default it is the one which is present
6671 in the same backend. It is possible to share a table with other backends by
6672 referencing it using the "table" keyword. If another table is referenced,
6673 the server's ID inside the backends are used. By default, all server IDs
6674 start at 1 in each backend, so the server ordering is enough. But in case of
6675 doubt, it is highly recommended to force server IDs using their "id" setting.
6676
6677 It is possible to restrict the conditions where a "stick match" statement
6678 will apply, using "if" or "unless" followed by a condition. See section 7 for
6679 ACL based conditions.
6680
6681 There is no limit on the number of "stick match" statements. The first that
6682 applies and matches will cause the request to be directed to the same server
6683 as was used for the request which created the entry. That way, multiple
6684 matches can be used as fallbacks.
6685
6686 The stick rules are checked after the persistence cookies, so they will not
6687 affect stickiness if a cookie has already been used to select a server. That
6688 way, it becomes very easy to insert cookies and match on IP addresses in
6689 order to maintain stickiness between HTTP and HTTPS.
6690
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006691 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6692 unless you know what you do : memory is not shared between the
6693 processes, which can result in random behaviours.
6694
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006695 Example :
6696 # forward SMTP users to the same server they just used for POP in the
6697 # last 30 minutes
6698 backend pop
6699 mode tcp
6700 balance roundrobin
6701 stick store-request src
6702 stick-table type ip size 200k expire 30m
6703 server s1 192.168.1.1:110
6704 server s2 192.168.1.1:110
6705
6706 backend smtp
6707 mode tcp
6708 balance roundrobin
6709 stick match src table pop
6710 server s1 192.168.1.1:25
6711 server s2 192.168.1.1:25
6712
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006713 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006714 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006715
6716
6717stick on <pattern> [table <table>] [{if | unless} <condition>]
6718 Define a request pattern to associate a user to a server
6719 May be used in sections : defaults | frontend | listen | backend
6720 no | no | yes | yes
6721
6722 Note : This form is exactly equivalent to "stick match" followed by
6723 "stick store-request", all with the same arguments. Please refer
6724 to both keywords for details. It is only provided as a convenience
6725 for writing more maintainable configurations.
6726
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006727 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6728 unless you know what you do : memory is not shared between the
6729 processes, which can result in random behaviours.
6730
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006731 Examples :
6732 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01006733 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006734
6735 # ...is strictly equivalent to this one :
6736 stick match src table pop if !localhost
6737 stick store-request src table pop if !localhost
6738
6739
6740 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
6741 # well as HTTP without cookie. Share the same table between both accesses.
6742 backend http
6743 mode http
6744 balance roundrobin
6745 stick on src table https
6746 cookie SRV insert indirect nocache
6747 server s1 192.168.1.1:80 cookie s1
6748 server s2 192.168.1.1:80 cookie s2
6749
6750 backend https
6751 mode tcp
6752 balance roundrobin
6753 stick-table type ip size 200k expire 30m
6754 stick on src
6755 server s1 192.168.1.1:443
6756 server s2 192.168.1.1:443
6757
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006758 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006759
6760
6761stick store-request <pattern> [table <table>] [{if | unless} <condition>]
6762 Define a request pattern used to create an entry in a stickiness table
6763 May be used in sections : defaults | frontend | listen | backend
6764 no | no | yes | yes
6765
6766 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006767 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006768 describes what elements of the incoming request or connection
6769 will be analysed, extracted and stored in the table once a
6770 server is selected.
6771
6772 <table> is an optional stickiness table name. If unspecified, the same
6773 backend's table is used. A stickiness table is declared using
6774 the "stick-table" statement.
6775
6776 <cond> is an optional storage condition. It makes it possible to store
6777 certain criteria only when some conditions are met (or not met).
6778 For instance, it could be used to store the source IP address
6779 except when the request passes through a known proxy, in which
6780 case we'd store a converted form of a header containing that IP
6781 address.
6782
6783 Some protocols or applications require complex stickiness rules and cannot
6784 always simply rely on cookies nor hashing. The "stick store-request" statement
6785 describes a rule to decide what to extract from the request and when to do
6786 it, in order to store it into a stickiness table for further requests to
6787 match it using the "stick match" statement. Obviously the extracted part must
6788 make sense and have a chance to be matched in a further request. Storing a
6789 client's IP address for instance often makes sense. Storing an ID found in a
6790 URL parameter also makes sense. Storing a source port will almost never make
6791 any sense because it will be randomly matched. See section 7 for a complete
6792 list of possible patterns and transformation rules.
6793
6794 The table has to be declared using the "stick-table" statement. It must be of
6795 a type compatible with the pattern. By default it is the one which is present
6796 in the same backend. It is possible to share a table with other backends by
6797 referencing it using the "table" keyword. If another table is referenced,
6798 the server's ID inside the backends are used. By default, all server IDs
6799 start at 1 in each backend, so the server ordering is enough. But in case of
6800 doubt, it is highly recommended to force server IDs using their "id" setting.
6801
6802 It is possible to restrict the conditions where a "stick store-request"
6803 statement will apply, using "if" or "unless" followed by a condition. This
6804 condition will be evaluated while parsing the request, so any criteria can be
6805 used. See section 7 for ACL based conditions.
6806
6807 There is no limit on the number of "stick store-request" statements, but
6808 there is a limit of 8 simultaneous stores per request or response. This
6809 makes it possible to store up to 8 criteria, all extracted from either the
6810 request or the response, regardless of the number of rules. Only the 8 first
6811 ones which match will be kept. Using this, it is possible to feed multiple
6812 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01006813 another protocol or access method. Using multiple store-request rules with
6814 the same table is possible and may be used to find the best criterion to rely
6815 on, by arranging the rules by decreasing preference order. Only the first
6816 extracted criterion for a given table will be stored. All subsequent store-
6817 request rules referencing the same table will be skipped and their ACLs will
6818 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006819
6820 The "store-request" rules are evaluated once the server connection has been
6821 established, so that the table will contain the real server that processed
6822 the request.
6823
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006824 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6825 unless you know what you do : memory is not shared between the
6826 processes, which can result in random behaviours.
6827
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006828 Example :
6829 # forward SMTP users to the same server they just used for POP in the
6830 # last 30 minutes
6831 backend pop
6832 mode tcp
6833 balance roundrobin
6834 stick store-request src
6835 stick-table type ip size 200k expire 30m
6836 server s1 192.168.1.1:110
6837 server s2 192.168.1.1:110
6838
6839 backend smtp
6840 mode tcp
6841 balance roundrobin
6842 stick match src table pop
6843 server s1 192.168.1.1:25
6844 server s2 192.168.1.1:25
6845
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006846 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006847 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006848
6849
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006850stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02006851 size <size> [expire <expire>] [nopurge] [peers <peersect>]
6852 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08006853 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006854 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02006855 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006856
6857 Arguments :
6858 ip a table declared with "type ip" will only store IPv4 addresses.
6859 This form is very compact (about 50 bytes per entry) and allows
6860 very fast entry lookup and stores with almost no overhead. This
6861 is mainly used to store client source IP addresses.
6862
David du Colombier9a6d3c92011-03-17 10:40:24 +01006863 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
6864 This form is very compact (about 60 bytes per entry) and allows
6865 very fast entry lookup and stores with almost no overhead. This
6866 is mainly used to store client source IP addresses.
6867
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006868 integer a table declared with "type integer" will store 32bit integers
6869 which can represent a client identifier found in a request for
6870 instance.
6871
6872 string a table declared with "type string" will store substrings of up
6873 to <len> characters. If the string provided by the pattern
6874 extractor is larger than <len>, it will be truncated before
6875 being stored. During matching, at most <len> characters will be
6876 compared between the string in the table and the extracted
6877 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006878 to 32 characters.
6879
6880 binary a table declared with "type binary" will store binary blocks
6881 of <len> bytes. If the block provided by the pattern
6882 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02006883 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006884 is shorter than <len>, it will be padded by 0. When not
6885 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006886
6887 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006888 "string" type table (See type "string" above). Or the number
6889 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006890 changing this parameter as memory usage will proportionally
6891 increase.
6892
6893 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01006894 value directly impacts memory usage. Count approximately
6895 50 bytes per entry, plus the size of a string if any. The size
6896 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006897
6898 [nopurge] indicates that we refuse to purge older entries when the table
6899 is full. When not specified and the table is full when haproxy
6900 wants to store an entry in it, it will flush a few of the oldest
6901 entries in order to release some space for the new ones. This is
6902 most often the desired behaviour. In some specific cases, it
6903 be desirable to refuse new entries instead of purging the older
6904 ones. That may be the case when the amount of data to store is
6905 far above the hardware limits and we prefer not to offer access
6906 to new clients than to reject the ones already connected. When
6907 using this parameter, be sure to properly set the "expire"
6908 parameter (see below).
6909
Emeric Brunf099e792010-09-27 12:05:28 +02006910 <peersect> is the name of the peers section to use for replication. Entries
6911 which associate keys to server IDs are kept synchronized with
6912 the remote peers declared in this section. All entries are also
6913 automatically learned from the local peer (old process) during a
6914 soft restart.
6915
Willy Tarreauc85ad792015-05-01 19:21:02 +02006916 NOTE : each peers section may be referenced only by tables
6917 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006918
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006919 <expire> defines the maximum duration of an entry in the table since it
6920 was last created, refreshed or matched. The expiration delay is
6921 defined using the standard time format, similarly as the various
6922 timeouts. The maximum duration is slightly above 24 days. See
6923 section 2.2 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02006924 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006925 be removed once full. Be sure not to use the "nopurge" parameter
6926 if not expiration delay is specified.
6927
Willy Tarreau08d5f982010-06-06 13:34:54 +02006928 <data_type> is used to store additional information in the stick-table. This
6929 may be used by ACLs in order to control various criteria related
6930 to the activity of the client matching the stick-table. For each
6931 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02006932 that the additional data can fit. Several data types may be
6933 stored with an entry. Multiple data types may be specified after
6934 the "store" keyword, as a comma-separated list. Alternatively,
6935 it is possible to repeat the "store" keyword followed by one or
6936 several data types. Except for the "server_id" type which is
6937 automatically detected and enabled, all data types must be
6938 explicitly declared to be stored. If an ACL references a data
6939 type which is not stored, the ACL will simply not match. Some
6940 data types require an argument which must be passed just after
6941 the type between parenthesis. See below for the supported data
6942 types and their arguments.
6943
6944 The data types that can be stored with an entry are the following :
6945 - server_id : this is an integer which holds the numeric ID of the server a
6946 request was assigned to. It is used by the "stick match", "stick store",
6947 and "stick on" rules. It is automatically enabled when referenced.
6948
6949 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
6950 integer which may be used for anything. Most of the time it will be used
6951 to put a special tag on some entries, for instance to note that a
6952 specific behaviour was detected and must be known for future matches.
6953
Willy Tarreauba2ffd12013-05-29 15:54:14 +02006954 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
6955 over a period. It is a positive 32-bit integer integer which may be used
6956 for anything. Just like <gpc0>, it counts events, but instead of keeping
6957 a cumulative count, it maintains the rate at which the counter is
6958 incremented. Most of the time it will be used to measure the frequency of
6959 occurrence of certain events (eg: requests to a specific URL).
6960
Willy Tarreauc9705a12010-07-27 20:05:50 +02006961 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
6962 the absolute number of connections received from clients which matched
6963 this entry. It does not mean the connections were accepted, just that
6964 they were received.
6965
6966 - conn_cur : Current Connections. It is a positive 32-bit integer which
6967 stores the concurrent connection counts for the entry. It is incremented
6968 once an incoming connection matches the entry, and decremented once the
6969 connection leaves. That way it is possible to know at any time the exact
6970 number of concurrent connections for an entry.
6971
6972 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6973 integer parameter <period> which indicates in milliseconds the length
6974 of the period over which the average is measured. It reports the average
6975 incoming connection rate over that period, in connections per period. The
6976 result is an integer which can be matched using ACLs.
6977
6978 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
6979 the absolute number of sessions received from clients which matched this
6980 entry. A session is a connection that was accepted by the layer 4 rules.
6981
6982 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6983 integer parameter <period> which indicates in milliseconds the length
6984 of the period over which the average is measured. It reports the average
6985 incoming session rate over that period, in sessions per period. The
6986 result is an integer which can be matched using ACLs.
6987
6988 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
6989 counts the absolute number of HTTP requests received from clients which
6990 matched this entry. It does not matter whether they are valid requests or
6991 not. Note that this is different from sessions when keep-alive is used on
6992 the client side.
6993
6994 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6995 integer parameter <period> which indicates in milliseconds the length
6996 of the period over which the average is measured. It reports the average
6997 HTTP request rate over that period, in requests per period. The result is
6998 an integer which can be matched using ACLs. It does not matter whether
6999 they are valid requests or not. Note that this is different from sessions
7000 when keep-alive is used on the client side.
7001
7002 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
7003 counts the absolute number of HTTP requests errors induced by clients
7004 which matched this entry. Errors are counted on invalid and truncated
7005 requests, as well as on denied or tarpitted requests, and on failed
7006 authentications. If the server responds with 4xx, then the request is
7007 also counted as an error since it's an error triggered by the client
7008 (eg: vulnerability scan).
7009
7010 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7011 integer parameter <period> which indicates in milliseconds the length
7012 of the period over which the average is measured. It reports the average
7013 HTTP request error rate over that period, in requests per period (see
7014 http_err_cnt above for what is accounted as an error). The result is an
7015 integer which can be matched using ACLs.
7016
7017 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
7018 integer which counts the cumulated amount of bytes received from clients
7019 which matched this entry. Headers are included in the count. This may be
7020 used to limit abuse of upload features on photo or video servers.
7021
7022 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7023 integer parameter <period> which indicates in milliseconds the length
7024 of the period over which the average is measured. It reports the average
7025 incoming bytes rate over that period, in bytes per period. It may be used
7026 to detect users which upload too much and too fast. Warning: with large
7027 uploads, it is possible that the amount of uploaded data will be counted
7028 once upon termination, thus causing spikes in the average transfer speed
7029 instead of having a smooth one. This may partially be smoothed with
7030 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
7031 recommended for better fairness.
7032
7033 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
7034 integer which counts the cumulated amount of bytes sent to clients which
7035 matched this entry. Headers are included in the count. This may be used
7036 to limit abuse of bots sucking the whole site.
7037
7038 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
7039 an integer parameter <period> which indicates in milliseconds the length
7040 of the period over which the average is measured. It reports the average
7041 outgoing bytes rate over that period, in bytes per period. It may be used
7042 to detect users which download too much and too fast. Warning: with large
7043 transfers, it is possible that the amount of transferred data will be
7044 counted once upon termination, thus causing spikes in the average
7045 transfer speed instead of having a smooth one. This may partially be
7046 smoothed with "option contstats" though this is not perfect yet. Use of
7047 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02007048
Willy Tarreauc00cdc22010-06-06 16:48:26 +02007049 There is only one stick-table per proxy. At the moment of writing this doc,
7050 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007051 to be required, simply create a dummy backend with a stick-table in it and
7052 reference it.
7053
7054 It is important to understand that stickiness based on learning information
7055 has some limitations, including the fact that all learned associations are
7056 lost upon restart. In general it can be good as a complement but not always
7057 as an exclusive stickiness.
7058
Willy Tarreauc9705a12010-07-27 20:05:50 +02007059 Last, memory requirements may be important when storing many data types.
7060 Indeed, storing all indicators above at once in each entry requires 116 bytes
7061 per entry, or 116 MB for a 1-million entries table. This is definitely not
7062 something that can be ignored.
7063
7064 Example:
7065 # Keep track of counters of up to 1 million IP addresses over 5 minutes
7066 # and store a general purpose counter and the average connection rate
7067 # computed over a sliding window of 30 seconds.
7068 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
7069
7070 See also : "stick match", "stick on", "stick store-request", section 2.2
David du Colombiera13d1b92011-03-17 10:40:22 +01007071 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007072
7073
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007074stick store-response <pattern> [table <table>] [{if | unless} <condition>]
7075 Define a request pattern used to create an entry in a stickiness table
7076 May be used in sections : defaults | frontend | listen | backend
7077 no | no | yes | yes
7078
7079 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007080 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007081 describes what elements of the response or connection will
7082 be analysed, extracted and stored in the table once a
7083 server is selected.
7084
7085 <table> is an optional stickiness table name. If unspecified, the same
7086 backend's table is used. A stickiness table is declared using
7087 the "stick-table" statement.
7088
7089 <cond> is an optional storage condition. It makes it possible to store
7090 certain criteria only when some conditions are met (or not met).
7091 For instance, it could be used to store the SSL session ID only
7092 when the response is a SSL server hello.
7093
7094 Some protocols or applications require complex stickiness rules and cannot
7095 always simply rely on cookies nor hashing. The "stick store-response"
7096 statement describes a rule to decide what to extract from the response and
7097 when to do it, in order to store it into a stickiness table for further
7098 requests to match it using the "stick match" statement. Obviously the
7099 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007100 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007101 See section 7 for a complete list of possible patterns and transformation
7102 rules.
7103
7104 The table has to be declared using the "stick-table" statement. It must be of
7105 a type compatible with the pattern. By default it is the one which is present
7106 in the same backend. It is possible to share a table with other backends by
7107 referencing it using the "table" keyword. If another table is referenced,
7108 the server's ID inside the backends are used. By default, all server IDs
7109 start at 1 in each backend, so the server ordering is enough. But in case of
7110 doubt, it is highly recommended to force server IDs using their "id" setting.
7111
7112 It is possible to restrict the conditions where a "stick store-response"
7113 statement will apply, using "if" or "unless" followed by a condition. This
7114 condition will be evaluated while parsing the response, so any criteria can
7115 be used. See section 7 for ACL based conditions.
7116
7117 There is no limit on the number of "stick store-response" statements, but
7118 there is a limit of 8 simultaneous stores per request or response. This
7119 makes it possible to store up to 8 criteria, all extracted from either the
7120 request or the response, regardless of the number of rules. Only the 8 first
7121 ones which match will be kept. Using this, it is possible to feed multiple
7122 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01007123 another protocol or access method. Using multiple store-response rules with
7124 the same table is possible and may be used to find the best criterion to rely
7125 on, by arranging the rules by decreasing preference order. Only the first
7126 extracted criterion for a given table will be stored. All subsequent store-
7127 response rules referencing the same table will be skipped and their ACLs will
7128 not be evaluated. However, even if a store-request rule references a table, a
7129 store-response rule may also use the same table. This means that each table
7130 may learn exactly one element from the request and one element from the
7131 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007132
7133 The table will contain the real server that processed the request.
7134
7135 Example :
7136 # Learn SSL session ID from both request and response and create affinity.
7137 backend https
7138 mode tcp
7139 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02007140 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007141 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007142
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007143 acl clienthello req_ssl_hello_type 1
7144 acl serverhello rep_ssl_hello_type 2
7145
7146 # use tcp content accepts to detects ssl client and server hello.
7147 tcp-request inspect-delay 5s
7148 tcp-request content accept if clienthello
7149
7150 # no timeout on response inspect delay by default.
7151 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007152
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007153 # SSL session ID (SSLID) may be present on a client or server hello.
7154 # Its length is coded on 1 byte at offset 43 and its value starts
7155 # at offset 44.
7156
7157 # Match and learn on request if client hello.
7158 stick on payload_lv(43,1) if clienthello
7159
7160 # Learn on response if server hello.
7161 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02007162
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007163 server s1 192.168.1.1:443
7164 server s2 192.168.1.1:443
7165
7166 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
7167 extraction.
7168
7169
Willy Tarreau938c7fe2014-04-25 14:21:39 +02007170tcp-check connect [params*]
7171 Opens a new connection
7172 May be used in sections: defaults | frontend | listen | backend
7173 no | no | yes | yes
7174
7175 When an application lies on more than a single TCP port or when HAProxy
7176 load-balance many services in a single backend, it makes sense to probe all
7177 the services individually before considering a server as operational.
7178
7179 When there are no TCP port configured on the server line neither server port
7180 directive, then the 'tcp-check connect port <port>' must be the first step
7181 of the sequence.
7182
7183 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
7184 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
7185 do.
7186
7187 Parameters :
7188 They are optional and can be used to describe how HAProxy should open and
7189 use the TCP connection.
7190
7191 port if not set, check port or server port is used.
7192 It tells HAProxy where to open the connection to.
7193 <port> must be a valid TCP port source integer, from 1 to 65535.
7194
7195 send-proxy send a PROXY protocol string
7196
7197 ssl opens a ciphered connection
7198
7199 Examples:
7200 # check HTTP and HTTPs services on a server.
7201 # first open port 80 thanks to server line port directive, then
7202 # tcp-check opens port 443, ciphered and run a request on it:
7203 option tcp-check
7204 tcp-check connect
7205 tcp-check send GET\ /\ HTTP/1.0\r\n
7206 tcp-check send Host:\ haproxy.1wt.eu\r\n
7207 tcp-check send \r\n
7208 tcp-check expect rstring (2..|3..)
7209 tcp-check connect port 443 ssl
7210 tcp-check send GET\ /\ HTTP/1.0\r\n
7211 tcp-check send Host:\ haproxy.1wt.eu\r\n
7212 tcp-check send \r\n
7213 tcp-check expect rstring (2..|3..)
7214 server www 10.0.0.1 check port 80
7215
7216 # check both POP and IMAP from a single server:
7217 option tcp-check
7218 tcp-check connect port 110
7219 tcp-check expect string +OK\ POP3\ ready
7220 tcp-check connect port 143
7221 tcp-check expect string *\ OK\ IMAP4\ ready
7222 server mail 10.0.0.1 check
7223
7224 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
7225
7226
7227tcp-check expect [!] <match> <pattern>
7228 Specify data to be collected and analysed during a generic health check
7229 May be used in sections: defaults | frontend | listen | backend
7230 no | no | yes | yes
7231
7232 Arguments :
7233 <match> is a keyword indicating how to look for a specific pattern in the
7234 response. The keyword may be one of "string", "rstring" or
7235 binary.
7236 The keyword may be preceded by an exclamation mark ("!") to negate
7237 the match. Spaces are allowed between the exclamation mark and the
7238 keyword. See below for more details on the supported keywords.
7239
7240 <pattern> is the pattern to look for. It may be a string or a regular
7241 expression. If the pattern contains spaces, they must be escaped
7242 with the usual backslash ('\').
7243 If the match is set to binary, then the pattern must be passed as
7244 a serie of hexadecimal digits in an even number. Each sequence of
7245 two digits will represent a byte. The hexadecimal digits may be
7246 used upper or lower case.
7247
7248
7249 The available matches are intentionally similar to their http-check cousins :
7250
7251 string <string> : test the exact string matches in the response buffer.
7252 A health check response will be considered valid if the
7253 response's buffer contains this exact string. If the
7254 "string" keyword is prefixed with "!", then the response
7255 will be considered invalid if the body contains this
7256 string. This can be used to look for a mandatory pattern
7257 in a protocol response, or to detect a failure when a
7258 specific error appears in a protocol banner.
7259
7260 rstring <regex> : test a regular expression on the response buffer.
7261 A health check response will be considered valid if the
7262 response's buffer matches this expression. If the
7263 "rstring" keyword is prefixed with "!", then the response
7264 will be considered invalid if the body matches the
7265 expression.
7266
7267 binary <hexstring> : test the exact string in its hexadecimal form matches
7268 in the response buffer. A health check response will
7269 be considered valid if the response's buffer contains
7270 this exact hexadecimal string.
7271 Purpose is to match data on binary protocols.
7272
7273 It is important to note that the responses will be limited to a certain size
7274 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
7275 Thus, too large responses may not contain the mandatory pattern when using
7276 "string", "rstring" or binary. If a large response is absolutely required, it
7277 is possible to change the default max size by setting the global variable.
7278 However, it is worth keeping in mind that parsing very large responses can
7279 waste some CPU cycles, especially when regular expressions are used, and that
7280 it is always better to focus the checks on smaller resources. Also, in its
7281 current state, the check will not find any string nor regex past a null
7282 character in the response. Similarly it is not possible to request matching
7283 the null character.
7284
7285 Examples :
7286 # perform a POP check
7287 option tcp-check
7288 tcp-check expect string +OK\ POP3\ ready
7289
7290 # perform an IMAP check
7291 option tcp-check
7292 tcp-check expect string *\ OK\ IMAP4\ ready
7293
7294 # look for the redis master server
7295 option tcp-check
7296 tcp-check send PING\r\n
7297 tcp-check expect +PONG
7298 tcp-check send info\ replication\r\n
7299 tcp-check expect string role:master
7300 tcp-check send QUIT\r\n
7301 tcp-check expect string +OK
7302
7303
7304 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
7305 "tcp-check send-binary", "http-check expect", tune.chksize
7306
7307
7308tcp-check send <data>
7309 Specify a string to be sent as a question during a generic health check
7310 May be used in sections: defaults | frontend | listen | backend
7311 no | no | yes | yes
7312
7313 <data> : the data to be sent as a question during a generic health check
7314 session. For now, <data> must be a string.
7315
7316 Examples :
7317 # look for the redis master server
7318 option tcp-check
7319 tcp-check send info\ replication\r\n
7320 tcp-check expect string role:master
7321
7322 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7323 "tcp-check send-binary", tune.chksize
7324
7325
7326tcp-check send-binary <hexastring>
7327 Specify an hexa digits string to be sent as a binary question during a raw
7328 tcp health check
7329 May be used in sections: defaults | frontend | listen | backend
7330 no | no | yes | yes
7331
7332 <data> : the data to be sent as a question during a generic health check
7333 session. For now, <data> must be a string.
7334 <hexastring> : test the exact string in its hexadecimal form matches in the
7335 response buffer. A health check response will be considered
7336 valid if the response's buffer contains this exact
7337 hexadecimal string.
7338 Purpose is to send binary data to ask on binary protocols.
7339
7340 Examples :
7341 # redis check in binary
7342 option tcp-check
7343 tcp-check send-binary 50494e470d0a # PING\r\n
7344 tcp-check expect binary 2b504F4e47 # +PONG
7345
7346
7347 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7348 "tcp-check send", tune.chksize
7349
7350
Willy Tarreaue9656522010-08-17 15:40:09 +02007351tcp-request connection <action> [{if | unless} <condition>]
7352 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02007353 May be used in sections : defaults | frontend | listen | backend
7354 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02007355 Arguments :
7356 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007357 actions include : "accept", "reject", "track-sc0", "track-sc1",
7358 "track-sc2", and "expect-proxy". See below for more details.
Willy Tarreau1a687942010-05-23 22:40:30 +02007359
Willy Tarreaue9656522010-08-17 15:40:09 +02007360 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007361
7362 Immediately after acceptance of a new incoming connection, it is possible to
7363 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02007364 or dropped or have its counters tracked. Those conditions cannot make use of
7365 any data contents because the connection has not been read from yet, and the
7366 buffers are not yet allocated. This is used to selectively and very quickly
7367 accept or drop connections from various sources with a very low overhead. If
7368 some contents need to be inspected in order to take the decision, the
7369 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007370
Willy Tarreaue9656522010-08-17 15:40:09 +02007371 The "tcp-request connection" rules are evaluated in their exact declaration
7372 order. If no rule matches or if there is no rule, the default action is to
7373 accept the incoming connection. There is no specific limit to the number of
7374 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007375
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007376 Five types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02007377 - accept :
7378 accepts the connection if the condition is true (when used with "if")
7379 or false (when used with "unless"). The first such rule executed ends
7380 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007381
Willy Tarreaue9656522010-08-17 15:40:09 +02007382 - reject :
7383 rejects the connection if the condition is true (when used with "if")
7384 or false (when used with "unless"). The first such rule executed ends
7385 the rules evaluation. Rejected connections do not even become a
7386 session, which is why they are accounted separately for in the stats,
7387 as "denied connections". They are not considered for the session
7388 rate-limit and are not logged either. The reason is that these rules
7389 should only be used to filter extremely high connection rates such as
7390 the ones encountered during a massive DDoS attack. Under these extreme
7391 conditions, the simple action of logging each event would make the
7392 system collapse and would considerably lower the filtering capacity. If
7393 logging is absolutely desired, then "tcp-request content" rules should
7394 be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007395
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007396 - expect-proxy layer4 :
7397 configures the client-facing connection to receive a PROXY protocol
7398 header before any byte is read from the socket. This is equivalent to
7399 having the "accept-proxy" keyword on the "bind" line, except that using
7400 the TCP rule allows the PROXY protocol to be accepted only for certain
7401 IP address ranges using an ACL. This is convenient when multiple layers
7402 of load balancers are passed through by traffic coming from public
7403 hosts.
7404
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007405 - capture <sample> len <length> :
7406 This only applies to "tcp-request content" rules. It captures sample
7407 expression <sample> from the request buffer, and converts it to a
7408 string of at most <len> characters. The resulting string is stored into
7409 the next request "capture" slot, so it will possibly appear next to
7410 some captured HTTP headers. It will then automatically appear in the
7411 logs, and it will be possible to extract it using sample fetch rules to
7412 feed it into headers or anything. The length should be limited given
7413 that this size will be allocated for each capture during the whole
7414 session life. Since it applies to Please check section 7.3 (Fetching
7415 samples) and "capture request header" for more information.
7416
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007417 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02007418 enables tracking of sticky counters from current connection. These
7419 rules do not stop evaluation and do not change default action. Two sets
7420 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007421 first "track-sc0" rule executed enables tracking of the counters of the
7422 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02007423 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007424 set. The first "track-sc2" rule executed enables tracking of the
7425 counters of the specified table as the third set. It is a recommended
7426 practice to use the first set of counters for the per-frontend counters
7427 and the second set for the per-backend ones. But this is just a
7428 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007429
Willy Tarreaue9656522010-08-17 15:40:09 +02007430 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007431 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02007432 in section 7.3. It describes what elements of the incoming
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007433 request or connection will be analysed, extracted, combined,
7434 and used to select which table entry to update the counters.
7435 Note that "tcp-request connection" cannot use content-based
7436 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007437
Willy Tarreaue9656522010-08-17 15:40:09 +02007438 <table> is an optional table to be used instead of the default one,
7439 which is the stick-table declared in the current proxy. All
7440 the counters for the matches and updates for the key will
7441 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007442
Willy Tarreaue9656522010-08-17 15:40:09 +02007443 Once a "track-sc*" rule is executed, the key is looked up in the table
7444 and if it is not found, an entry is allocated for it. Then a pointer to
7445 that entry is kept during all the session's life, and this entry's
7446 counters are updated as often as possible, every time the session's
7447 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007448 Counters are only updated for events that happen after the tracking has
7449 been started. For example, connection counters will not be updated when
7450 tracking layer 7 information, since the connection event happens before
7451 layer7 information is extracted.
7452
Willy Tarreaue9656522010-08-17 15:40:09 +02007453 If the entry tracks concurrent connection counters, one connection is
7454 counted for as long as the entry is tracked, and the entry will not
7455 expire during that time. Tracking counters also provides a performance
7456 advantage over just checking the keys, because only one table lookup is
7457 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007458
Willy Tarreaue9656522010-08-17 15:40:09 +02007459 Note that the "if/unless" condition is optional. If no condition is set on
7460 the action, it is simply performed unconditionally. That can be useful for
7461 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007462
Willy Tarreaue9656522010-08-17 15:40:09 +02007463 Example: accept all connections from white-listed hosts, reject too fast
7464 connection without counting them, and track accepted connections.
7465 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007466
Willy Tarreaue9656522010-08-17 15:40:09 +02007467 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007468 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007469 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007470
Willy Tarreaue9656522010-08-17 15:40:09 +02007471 Example: accept all connections from white-listed hosts, count all other
7472 connections and reject too fast ones. This results in abusive ones
7473 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007474
Willy Tarreaue9656522010-08-17 15:40:09 +02007475 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007476 tcp-request connection track-sc0 src
7477 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007478
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007479 Example: enable the PROXY protocol for traffic coming from all known proxies.
7480
7481 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
7482
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007483 See section 7 about ACL usage.
7484
Willy Tarreaue9656522010-08-17 15:40:09 +02007485 See also : "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007486
7487
Willy Tarreaue9656522010-08-17 15:40:09 +02007488tcp-request content <action> [{if | unless} <condition>]
7489 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02007490 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007491 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02007492 Arguments :
7493 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007494 actions include : "accept", "reject", "track-sc0", "track-sc1",
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007495 "track-sc2" and "capture". See "tcp-request connection" above
7496 for their signification.
Willy Tarreau62644772008-07-16 18:36:06 +02007497
Willy Tarreaue9656522010-08-17 15:40:09 +02007498 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02007499
Willy Tarreaue9656522010-08-17 15:40:09 +02007500 A request's contents can be analysed at an early stage of request processing
7501 called "TCP content inspection". During this stage, ACL-based rules are
7502 evaluated every time the request contents are updated, until either an
7503 "accept" or a "reject" rule matches, or the TCP request inspection delay
7504 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02007505
Willy Tarreaue9656522010-08-17 15:40:09 +02007506 The first difference between these rules and "tcp-request connection" rules
7507 is that "tcp-request content" rules can make use of contents to take a
7508 decision. Most often, these decisions will consider a protocol recognition or
7509 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01007510 both frontends and backends. In case of HTTP keep-alive with the client, all
7511 tcp-request content rules are evaluated again, so haproxy keeps a record of
7512 what sticky counters were assigned by a "tcp-request connection" versus a
7513 "tcp-request content" rule, and flushes all the content-related ones after
7514 processing an HTTP request, so that they may be evaluated again by the rules
7515 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007516 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01007517 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02007518
Willy Tarreaue9656522010-08-17 15:40:09 +02007519 Content-based rules are evaluated in their exact declaration order. If no
7520 rule matches or if there is no rule, the default action is to accept the
7521 contents. There is no specific limit to the number of rules which may be
7522 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02007523
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007524 Four types of actions are supported :
7525 - accept : the request is accepted
7526 - reject : the request is rejected and the connection is closed
7527 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007528 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Willy Tarreau62644772008-07-16 18:36:06 +02007529
Willy Tarreaue9656522010-08-17 15:40:09 +02007530 They have the same meaning as their counter-parts in "tcp-request connection"
7531 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02007532
Willy Tarreauf3338342014-01-28 21:40:28 +01007533 While there is nothing mandatory about it, it is recommended to use the
7534 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
7535 content" rules in the frontend, and track-sc2 for "tcp-request content"
7536 rules in the backend, because that makes the configuration more readable
7537 and easier to troubleshoot, but this is just a guideline and all counters
7538 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02007539
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007540 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02007541 the action, it is simply performed unconditionally. That can be useful for
7542 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02007543
Willy Tarreaue9656522010-08-17 15:40:09 +02007544 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02007545 rules, since HTTP-specific ACL matches are able to preliminarily parse the
7546 contents of a buffer before extracting the required data. If the buffered
7547 contents do not parse as a valid HTTP message, then the ACL does not match.
7548 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01007549 processing, so there is no risk of parsing something differently. In an HTTP
7550 backend connected to from an HTTP frontend, it is guaranteed that HTTP
7551 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02007552
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007553 Tracking layer7 information is also possible provided that the information
Willy Tarreau531485c2014-09-16 15:48:15 +02007554 are present when the rule is processed. The rule processing engine is able to
7555 wait until the inspect delay expires when the data to be tracked is not yet
7556 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007557
Willy Tarreau62644772008-07-16 18:36:06 +02007558 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02007559 # Accept HTTP requests containing a Host header saying "example.com"
7560 # and reject everything else.
7561 acl is_host_com hdr(Host) -i example.com
7562 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02007563 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02007564 tcp-request content reject
7565
7566 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02007567 # reject SMTP connection if client speaks first
7568 tcp-request inspect-delay 30s
7569 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007570 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02007571
7572 # Forward HTTPS connection only if client speaks
7573 tcp-request inspect-delay 30s
7574 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007575 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02007576 tcp-request content reject
7577
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007578 Example:
7579 # Track the last IP from X-Forwarded-For
7580 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007581 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007582
7583 Example:
7584 # track request counts per "base" (concatenation of Host+URL)
7585 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007586 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007587
Willy Tarreaue9656522010-08-17 15:40:09 +02007588 Example: track per-frontend and per-backend counters, block abusers at the
7589 frontend when the backend detects abuse.
7590
7591 frontend http
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007592 # Use General Purpose Couter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02007593 # protecting all our sites
7594 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007595 tcp-request connection track-sc0 src
7596 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02007597 ...
7598 use_backend http_dynamic if { path_end .php }
7599
7600 backend http_dynamic
7601 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007602 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02007603 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007604 acl click_too_fast sc1_http_req_rate gt 10
7605 acl mark_as_abuser sc0_inc_gpc0 gt 0
7606 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02007607 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02007608
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007609 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02007610
Willy Tarreaue9656522010-08-17 15:40:09 +02007611 See also : "tcp-request connection", "tcp-request inspect-delay"
Willy Tarreau62644772008-07-16 18:36:06 +02007612
7613
7614tcp-request inspect-delay <timeout>
7615 Set the maximum allowed time to wait for data during content inspection
7616 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007617 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02007618 Arguments :
7619 <timeout> is the timeout value specified in milliseconds by default, but
7620 can be in any other unit if the number is suffixed by the unit,
7621 as explained at the top of this document.
7622
7623 People using haproxy primarily as a TCP relay are often worried about the
7624 risk of passing any type of protocol to a server without any analysis. In
7625 order to be able to analyze the request contents, we must first withhold
7626 the data then analyze them. This statement simply enables withholding of
7627 data for at most the specified amount of time.
7628
Willy Tarreaufb356202010-08-03 14:02:05 +02007629 TCP content inspection applies very early when a connection reaches a
7630 frontend, then very early when the connection is forwarded to a backend. This
7631 means that a connection may experience a first delay in the frontend and a
7632 second delay in the backend if both have tcp-request rules.
7633
Willy Tarreau62644772008-07-16 18:36:06 +02007634 Note that when performing content inspection, haproxy will evaluate the whole
7635 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007636 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02007637 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01007638 contents are definitive. If no delay is set, haproxy will not wait at all
7639 and will immediately apply a verdict based on the available information.
7640 Obviously this is unlikely to be very useful and might even be racy, so such
7641 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02007642
7643 As soon as a rule matches, the request is released and continues as usual. If
7644 the timeout is reached and no rule matches, the default policy will be to let
7645 it pass through unaffected.
7646
7647 For most protocols, it is enough to set it to a few seconds, as most clients
7648 send the full request immediately upon connection. Add 3 or more seconds to
7649 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01007650 to use large values, for instance to ensure that the client never talks
Willy Tarreau62644772008-07-16 18:36:06 +02007651 before the server (eg: SMTP), or to wait for a client to talk before passing
7652 data to the server (eg: SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02007653 least the inspection delay, otherwise it will expire first. If the client
7654 closes the connection or if the buffer is full, the delay immediately expires
7655 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02007656
Willy Tarreau55165fe2009-05-10 12:02:55 +02007657 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02007658 "timeout client".
7659
7660
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007661tcp-response content <action> [{if | unless} <condition>]
7662 Perform an action on a session response depending on a layer 4-7 condition
7663 May be used in sections : defaults | frontend | listen | backend
7664 no | no | yes | yes
7665 Arguments :
7666 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007667 actions include : "accept", "close", "reject".
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007668
7669 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
7670
7671 Response contents can be analysed at an early stage of response processing
7672 called "TCP content inspection". During this stage, ACL-based rules are
7673 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007674 "accept", "close" or a "reject" rule matches, or a TCP response inspection
7675 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007676
7677 Most often, these decisions will consider a protocol recognition or validity.
7678
7679 Content-based rules are evaluated in their exact declaration order. If no
7680 rule matches or if there is no rule, the default action is to accept the
7681 contents. There is no specific limit to the number of rules which may be
7682 inserted.
7683
7684 Two types of actions are supported :
7685 - accept :
7686 accepts the response if the condition is true (when used with "if")
7687 or false (when used with "unless"). The first such rule executed ends
7688 the rules evaluation.
7689
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007690 - close :
7691 immediately closes the connection with the server if the condition is
7692 true (when used with "if"), or false (when used with "unless"). The
7693 first such rule executed ends the rules evaluation. The main purpose of
7694 this action is to force a connection to be finished between a client
7695 and a server after an exchange when the application protocol expects
7696 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007697 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007698 protocols.
7699
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007700 - reject :
7701 rejects the response if the condition is true (when used with "if")
7702 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007703 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007704
7705 Note that the "if/unless" condition is optional. If no condition is set on
7706 the action, it is simply performed unconditionally. That can be useful for
7707 for changing the default action to a reject.
7708
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007709 It is perfectly possible to match layer 7 contents with "tcp-response
7710 content" rules, but then it is important to ensure that a full response has
7711 been buffered, otherwise no contents will match. In order to achieve this,
7712 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007713 period.
7714
7715 See section 7 about ACL usage.
7716
7717 See also : "tcp-request content", "tcp-response inspect-delay"
7718
7719
7720tcp-response inspect-delay <timeout>
7721 Set the maximum allowed time to wait for a response during content inspection
7722 May be used in sections : defaults | frontend | listen | backend
7723 no | no | yes | yes
7724 Arguments :
7725 <timeout> is the timeout value specified in milliseconds by default, but
7726 can be in any other unit if the number is suffixed by the unit,
7727 as explained at the top of this document.
7728
7729 See also : "tcp-response content", "tcp-request inspect-delay".
7730
7731
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007732timeout check <timeout>
7733 Set additional check timeout, but only after a connection has been already
7734 established.
7735
7736 May be used in sections: defaults | frontend | listen | backend
7737 yes | no | yes | yes
7738 Arguments:
7739 <timeout> is the timeout value specified in milliseconds by default, but
7740 can be in any other unit if the number is suffixed by the unit,
7741 as explained at the top of this document.
7742
7743 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
7744 for check and "timeout check" as an additional read timeout. The "min" is
7745 used so that people running with *very* long "timeout connect" (eg. those
7746 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01007747 (Please also note that there is no valid reason to have such long connect
7748 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
7749 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007750
7751 If "timeout check" is not set haproxy uses "inter" for complete check
7752 timeout (connect + read) exactly like all <1.3.15 version.
7753
7754 In most cases check request is much simpler and faster to handle than normal
7755 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01007756 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007757
7758 This parameter is specific to backends, but can be specified once for all in
7759 "defaults" sections. This is in fact one of the easiest solutions not to
7760 forget about it.
7761
Willy Tarreau41a340d2008-01-22 12:25:31 +01007762 See also: "timeout connect", "timeout queue", "timeout server",
7763 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007764
7765
Willy Tarreau0ba27502007-12-24 16:55:16 +01007766timeout client <timeout>
7767timeout clitimeout <timeout> (deprecated)
7768 Set the maximum inactivity time on the client side.
7769 May be used in sections : defaults | frontend | listen | backend
7770 yes | yes | yes | no
7771 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007772 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007773 can be in any other unit if the number is suffixed by the unit,
7774 as explained at the top of this document.
7775
7776 The inactivity timeout applies when the client is expected to acknowledge or
7777 send data. In HTTP mode, this timeout is particularly important to consider
7778 during the first phase, when the client sends the request, and during the
7779 response while it is reading data sent by the server. The value is specified
7780 in milliseconds by default, but can be in any other unit if the number is
7781 suffixed by the unit, as specified at the top of this document. In TCP mode
7782 (and to a lesser extent, in HTTP mode), it is highly recommended that the
7783 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007784 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01007785 losses by specifying timeouts that are slightly above multiples of 3 seconds
Willy Tarreauce887fd2012-05-12 12:50:00 +02007786 (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
7787 sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02007788 which overrides "timeout client" and "timeout server" for tunnels, as well as
7789 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007790
7791 This parameter is specific to frontends, but can be specified once for all in
7792 "defaults" sections. This is in fact one of the easiest solutions not to
7793 forget about it. An unspecified timeout results in an infinite timeout, which
7794 is not recommended. Such a usage is accepted and works but reports a warning
7795 during startup because it may results in accumulation of expired sessions in
7796 the system if the system's timeouts are not configured either.
7797
7798 This parameter replaces the old, deprecated "clitimeout". It is recommended
7799 to use it to write new configurations. The form "timeout clitimeout" is
7800 provided only by backwards compatibility but its use is strongly discouraged.
7801
Willy Tarreauce887fd2012-05-12 12:50:00 +02007802 See also : "clitimeout", "timeout server", "timeout tunnel".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007803
7804
Willy Tarreau05cdd962014-05-10 14:30:07 +02007805timeout client-fin <timeout>
7806 Set the inactivity timeout on the client side for half-closed connections.
7807 May be used in sections : defaults | frontend | listen | backend
7808 yes | yes | yes | no
7809 Arguments :
7810 <timeout> is the timeout value specified in milliseconds by default, but
7811 can be in any other unit if the number is suffixed by the unit,
7812 as explained at the top of this document.
7813
7814 The inactivity timeout applies when the client is expected to acknowledge or
7815 send data while one direction is already shut down. This timeout is different
7816 from "timeout client" in that it only applies to connections which are closed
7817 in one direction. This is particularly useful to avoid keeping connections in
7818 FIN_WAIT state for too long when clients do not disconnect cleanly. This
7819 problem is particularly common long connections such as RDP or WebSocket.
7820 Note that this timeout can override "timeout tunnel" when a connection shuts
7821 down in one direction.
7822
7823 This parameter is specific to frontends, but can be specified once for all in
7824 "defaults" sections. By default it is not set, so half-closed connections
7825 will use the other timeouts (timeout.client or timeout.tunnel).
7826
7827 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
7828
7829
Willy Tarreau0ba27502007-12-24 16:55:16 +01007830timeout connect <timeout>
7831timeout contimeout <timeout> (deprecated)
7832 Set the maximum time to wait for a connection attempt to a server to succeed.
7833 May be used in sections : defaults | frontend | listen | backend
7834 yes | no | yes | yes
7835 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007836 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007837 can be in any other unit if the number is suffixed by the unit,
7838 as explained at the top of this document.
7839
7840 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007841 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01007842 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01007843 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007844 connect timeout also presets both queue and tarpit timeouts to the same value
7845 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007846
7847 This parameter is specific to backends, but can be specified once for all in
7848 "defaults" sections. This is in fact one of the easiest solutions not to
7849 forget about it. An unspecified timeout results in an infinite timeout, which
7850 is not recommended. Such a usage is accepted and works but reports a warning
7851 during startup because it may results in accumulation of failed sessions in
7852 the system if the system's timeouts are not configured either.
7853
7854 This parameter replaces the old, deprecated "contimeout". It is recommended
7855 to use it to write new configurations. The form "timeout contimeout" is
7856 provided only by backwards compatibility but its use is strongly discouraged.
7857
Willy Tarreau41a340d2008-01-22 12:25:31 +01007858 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
7859 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007860
7861
Willy Tarreaub16a5742010-01-10 14:46:16 +01007862timeout http-keep-alive <timeout>
7863 Set the maximum allowed time to wait for a new HTTP request to appear
7864 May be used in sections : defaults | frontend | listen | backend
7865 yes | yes | yes | yes
7866 Arguments :
7867 <timeout> is the timeout value specified in milliseconds by default, but
7868 can be in any other unit if the number is suffixed by the unit,
7869 as explained at the top of this document.
7870
7871 By default, the time to wait for a new request in case of keep-alive is set
7872 by "timeout http-request". However this is not always convenient because some
7873 people want very short keep-alive timeouts in order to release connections
7874 faster, and others prefer to have larger ones but still have short timeouts
7875 once the request has started to present itself.
7876
7877 The "http-keep-alive" timeout covers these needs. It will define how long to
7878 wait for a new HTTP request to start coming after a response was sent. Once
7879 the first byte of request has been seen, the "http-request" timeout is used
7880 to wait for the complete request to come. Note that empty lines prior to a
7881 new request do not refresh the timeout and are not counted as a new request.
7882
7883 There is also another difference between the two timeouts : when a connection
7884 expires during timeout http-keep-alive, no error is returned, the connection
7885 just closes. If the connection expires in "http-request" while waiting for a
7886 connection to complete, a HTTP 408 error is returned.
7887
7888 In general it is optimal to set this value to a few tens to hundreds of
7889 milliseconds, to allow users to fetch all objects of a page at once but
7890 without waiting for further clicks. Also, if set to a very small value (eg:
7891 1 millisecond) it will probably only accept pipelined requests but not the
7892 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +02007893 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +01007894
7895 If this parameter is not set, the "http-request" timeout applies, and if both
7896 are not set, "timeout client" still applies at the lower level. It should be
7897 set in the frontend to take effect, unless the frontend is in TCP mode, in
7898 which case the HTTP backend's timeout will be used.
7899
7900 See also : "timeout http-request", "timeout client".
7901
7902
Willy Tarreau036fae02008-01-06 13:24:40 +01007903timeout http-request <timeout>
7904 Set the maximum allowed time to wait for a complete HTTP request
7905 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007906 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +01007907 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007908 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +01007909 can be in any other unit if the number is suffixed by the unit,
7910 as explained at the top of this document.
7911
7912 In order to offer DoS protection, it may be required to lower the maximum
7913 accepted time to receive a complete HTTP request without affecting the client
7914 timeout. This helps protecting against established connections on which
7915 nothing is sent. The client timeout cannot offer a good protection against
7916 this abuse because it is an inactivity timeout, which means that if the
7917 attacker sends one character every now and then, the timeout will not
7918 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +02007919 types, the request will be aborted if it does not complete in time. When the
7920 timeout expires, an HTTP 408 response is sent to the client to inform it
7921 about the problem, and the connection is closed. The logs will report
7922 termination codes "cR". Some recent browsers are having problems with this
7923 standard, well-documented behaviour, so it might be needed to hide the 408
Willy Tarreau1c3a6122015-05-01 15:37:53 +02007924 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
7925 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +01007926
7927 Note that this timeout only applies to the header part of the request, and
7928 not to any data. As soon as the empty line is received, this timeout is not
Willy Tarreaub16a5742010-01-10 14:46:16 +01007929 used anymore. It is used again on keep-alive connections to wait for a second
7930 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +01007931
7932 Generally it is enough to set it to a few seconds, as most clients send the
7933 full request immediately upon connection. Add 3 or more seconds to cover TCP
7934 retransmits but that's all. Setting it to very low values (eg: 50 ms) will
7935 generally work on local networks as long as there are no packet losses. This
7936 will prevent people from sending bare HTTP requests using telnet.
7937
7938 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007939 chunk of the incoming request. It should be set in the frontend to take
7940 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
7941 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +01007942
Willy Tarreau1c3a6122015-05-01 15:37:53 +02007943 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
7944 "timeout client".
Willy Tarreau036fae02008-01-06 13:24:40 +01007945
Willy Tarreau844e3c52008-01-11 16:28:18 +01007946
7947timeout queue <timeout>
7948 Set the maximum time to wait in the queue for a connection slot to be free
7949 May be used in sections : defaults | frontend | listen | backend
7950 yes | no | yes | yes
7951 Arguments :
7952 <timeout> is the timeout value specified in milliseconds by default, but
7953 can be in any other unit if the number is suffixed by the unit,
7954 as explained at the top of this document.
7955
7956 When a server's maxconn is reached, connections are left pending in a queue
7957 which may be server-specific or global to the backend. In order not to wait
7958 indefinitely, a timeout is applied to requests pending in the queue. If the
7959 timeout is reached, it is considered that the request will almost never be
7960 served, so it is dropped and a 503 error is returned to the client.
7961
7962 The "timeout queue" statement allows to fix the maximum time for a request to
7963 be left pending in a queue. If unspecified, the same value as the backend's
7964 connection timeout ("timeout connect") is used, for backwards compatibility
7965 with older versions with no "timeout queue" parameter.
7966
7967 See also : "timeout connect", "contimeout".
7968
7969
7970timeout server <timeout>
7971timeout srvtimeout <timeout> (deprecated)
7972 Set the maximum inactivity time on the server side.
7973 May be used in sections : defaults | frontend | listen | backend
7974 yes | no | yes | yes
7975 Arguments :
7976 <timeout> is the timeout value specified in milliseconds by default, but
7977 can be in any other unit if the number is suffixed by the unit,
7978 as explained at the top of this document.
7979
7980 The inactivity timeout applies when the server is expected to acknowledge or
7981 send data. In HTTP mode, this timeout is particularly important to consider
7982 during the first phase of the server's response, when it has to send the
7983 headers, as it directly represents the server's processing time for the
7984 request. To find out what value to put there, it's often good to start with
7985 what would be considered as unacceptable response times, then check the logs
7986 to observe the response time distribution, and adjust the value accordingly.
7987
7988 The value is specified in milliseconds by default, but can be in any other
7989 unit if the number is suffixed by the unit, as specified at the top of this
7990 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
7991 recommended that the client timeout remains equal to the server timeout in
7992 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007993 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01007994 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreauce887fd2012-05-12 12:50:00 +02007995 seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed
7996 with short-lived sessions (eg: WebSocket and HTTP), it's worth considering
7997 "timeout tunnel", which overrides "timeout client" and "timeout server" for
7998 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007999
8000 This parameter is specific to backends, but can be specified once for all in
8001 "defaults" sections. This is in fact one of the easiest solutions not to
8002 forget about it. An unspecified timeout results in an infinite timeout, which
8003 is not recommended. Such a usage is accepted and works but reports a warning
8004 during startup because it may results in accumulation of expired sessions in
8005 the system if the system's timeouts are not configured either.
8006
8007 This parameter replaces the old, deprecated "srvtimeout". It is recommended
8008 to use it to write new configurations. The form "timeout srvtimeout" is
8009 provided only by backwards compatibility but its use is strongly discouraged.
8010
Willy Tarreauce887fd2012-05-12 12:50:00 +02008011 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008012
Willy Tarreau05cdd962014-05-10 14:30:07 +02008013
8014timeout server-fin <timeout>
8015 Set the inactivity timeout on the server side for half-closed connections.
8016 May be used in sections : defaults | frontend | listen | backend
8017 yes | no | yes | yes
8018 Arguments :
8019 <timeout> is the timeout value specified in milliseconds by default, but
8020 can be in any other unit if the number is suffixed by the unit,
8021 as explained at the top of this document.
8022
8023 The inactivity timeout applies when the server is expected to acknowledge or
8024 send data while one direction is already shut down. This timeout is different
8025 from "timeout server" in that it only applies to connections which are closed
8026 in one direction. This is particularly useful to avoid keeping connections in
8027 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
8028 This problem is particularly common long connections such as RDP or WebSocket.
8029 Note that this timeout can override "timeout tunnel" when a connection shuts
8030 down in one direction. This setting was provided for completeness, but in most
8031 situations, it should not be needed.
8032
8033 This parameter is specific to backends, but can be specified once for all in
8034 "defaults" sections. By default it is not set, so half-closed connections
8035 will use the other timeouts (timeout.server or timeout.tunnel).
8036
8037 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
8038
Willy Tarreau844e3c52008-01-11 16:28:18 +01008039
8040timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +01008041 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +01008042 May be used in sections : defaults | frontend | listen | backend
8043 yes | yes | yes | yes
8044 Arguments :
8045 <timeout> is the tarpit duration specified in milliseconds by default, but
8046 can be in any other unit if the number is suffixed by the unit,
8047 as explained at the top of this document.
8048
8049 When a connection is tarpitted using "reqtarpit", it is maintained open with
8050 no activity for a certain amount of time, then closed. "timeout tarpit"
8051 defines how long it will be maintained open.
8052
8053 The value is specified in milliseconds by default, but can be in any other
8054 unit if the number is suffixed by the unit, as specified at the top of this
8055 document. If unspecified, the same value as the backend's connection timeout
8056 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +01008057 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008058
8059 See also : "timeout connect", "contimeout".
8060
8061
Willy Tarreauce887fd2012-05-12 12:50:00 +02008062timeout tunnel <timeout>
8063 Set the maximum inactivity time on the client and server side for tunnels.
8064 May be used in sections : defaults | frontend | listen | backend
8065 yes | no | yes | yes
8066 Arguments :
8067 <timeout> is the timeout value specified in milliseconds by default, but
8068 can be in any other unit if the number is suffixed by the unit,
8069 as explained at the top of this document.
8070
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008071 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +02008072 between a client and a server, and the connection remains inactive in both
8073 directions. This timeout supersedes both the client and server timeouts once
8074 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
8075 analyser remains attached to either connection (eg: tcp content rules are
8076 accepted). In HTTP, this timeout is used when a connection is upgraded (eg:
8077 when switching to the WebSocket protocol, or forwarding a CONNECT request
8078 to a proxy), or after the first response when no keepalive/close option is
8079 specified.
8080
Willy Tarreau05cdd962014-05-10 14:30:07 +02008081 Since this timeout is usually used in conjunction with long-lived connections,
8082 it usually is a good idea to also set "timeout client-fin" to handle the
8083 situation where a client suddenly disappears from the net and does not
8084 acknowledge a close, or sends a shutdown and does not acknowledge pending
8085 data anymore. This can happen in lossy networks where firewalls are present,
8086 and is detected by the presence of large amounts of sessions in a FIN_WAIT
8087 state.
8088
Willy Tarreauce887fd2012-05-12 12:50:00 +02008089 The value is specified in milliseconds by default, but can be in any other
8090 unit if the number is suffixed by the unit, as specified at the top of this
8091 document. Whatever the expected normal idle time, it is a good practice to
8092 cover at least one or several TCP packet losses by specifying timeouts that
8093 are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum).
8094
8095 This parameter is specific to backends, but can be specified once for all in
8096 "defaults" sections. This is in fact one of the easiest solutions not to
8097 forget about it.
8098
8099 Example :
8100 defaults http
8101 option http-server-close
8102 timeout connect 5s
8103 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +02008104 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +02008105 timeout server 30s
8106 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
8107
Willy Tarreau05cdd962014-05-10 14:30:07 +02008108 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +02008109
8110
Willy Tarreau844e3c52008-01-11 16:28:18 +01008111transparent (deprecated)
8112 Enable client-side transparent proxying
8113 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008114 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008115 Arguments : none
8116
8117 This keyword was introduced in order to provide layer 7 persistence to layer
8118 3 load balancers. The idea is to use the OS's ability to redirect an incoming
8119 connection for a remote address to a local process (here HAProxy), and let
8120 this process know what address was initially requested. When this option is
8121 used, sessions without cookies will be forwarded to the original destination
8122 IP address of the incoming request (which should match that of another
8123 equipment), while requests with cookies will still be forwarded to the
8124 appropriate server.
8125
8126 The "transparent" keyword is deprecated, use "option transparent" instead.
8127
8128 Note that contrary to a common belief, this option does NOT make HAProxy
8129 present the client's IP to the server when establishing the connection.
8130
Willy Tarreau844e3c52008-01-11 16:28:18 +01008131 See also: "option transparent"
8132
William Lallemanda73203e2012-03-12 12:48:57 +01008133unique-id-format <string>
8134 Generate a unique ID for each request.
8135 May be used in sections : defaults | frontend | listen | backend
8136 yes | yes | yes | no
8137 Arguments :
8138 <string> is a log-format string.
8139
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008140 This keyword creates a ID for each request using the custom log format. A
8141 unique ID is useful to trace a request passing through many components of
8142 a complex infrastructure. The newly created ID may also be logged using the
8143 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +01008144
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008145 The format should be composed from elements that are guaranteed to be
8146 unique when combined together. For instance, if multiple haproxy instances
8147 are involved, it might be important to include the node name. It is often
8148 needed to log the incoming connection's source and destination addresses
8149 and ports. Note that since multiple requests may be performed over the same
8150 connection, including a request counter may help differentiate them.
8151 Similarly, a timestamp may protect against a rollover of the counter.
8152 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +01008153
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008154 It is recommended to use hexadecimal notation for many fields since it
8155 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +01008156
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008157 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008158
Julien Vehentf21be322014-03-07 08:27:34 -05008159 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008160
8161 will generate:
8162
8163 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8164
8165 See also: "unique-id-header"
8166
8167unique-id-header <name>
8168 Add a unique ID header in the HTTP request.
8169 May be used in sections : defaults | frontend | listen | backend
8170 yes | yes | yes | no
8171 Arguments :
8172 <name> is the name of the header.
8173
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008174 Add a unique-id header in the HTTP request sent to the server, using the
8175 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +01008176
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008177 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008178
Julien Vehentf21be322014-03-07 08:27:34 -05008179 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008180 unique-id-header X-Unique-ID
8181
8182 will generate:
8183
8184 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8185
8186 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +01008187
Willy Tarreauf51658d2014-04-23 01:21:56 +02008188use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008189 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008190 May be used in sections : defaults | frontend | listen | backend
8191 no | yes | yes | no
8192 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008193 <backend> is the name of a valid backend or "listen" section, or a
8194 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008195
Willy Tarreauf51658d2014-04-23 01:21:56 +02008196 <condition> is a condition composed of ACLs, as described in section 7. If
8197 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008198
8199 When doing content-switching, connections arrive on a frontend and are then
8200 dispatched to various backends depending on a number of conditions. The
8201 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008202 "use_backend" keyword. While it is normally used with HTTP processing, it can
8203 also be used in pure TCP, either without content using stateless ACLs (eg:
8204 source address validation) or combined with a "tcp-request" rule to wait for
8205 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008206
8207 There may be as many "use_backend" rules as desired. All of these rules are
8208 evaluated in their declaration order, and the first one which matches will
8209 assign the backend.
8210
8211 In the first form, the backend will be used if the condition is met. In the
8212 second form, the backend will be used if the condition is not met. If no
8213 condition is valid, the backend defined with "default_backend" will be used.
8214 If no default backend is defined, either the servers in the same section are
8215 used (in case of a "listen" section) or, in case of a frontend, no server is
8216 used and a 503 service unavailable response is returned.
8217
Willy Tarreau51aecc72009-07-12 09:47:04 +02008218 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008219 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +02008220 and backend processing will immediately follow, or the backend will wait for
8221 a complete HTTP request to get in. This feature is useful when a frontend
8222 must decode several protocols on a unique port, one of them being HTTP.
8223
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008224 When <backend> is a simple name, it is resolved at configuration time, and an
8225 error is reported if the specified backend does not exist. If <backend> is
8226 a log-format string instead, no check may be done at configuration time, so
8227 the backend name is resolved dynamically at run time. If the resulting
8228 backend name does not correspond to any valid backend, no other rule is
8229 evaluated, and the default_backend directive is applied instead. Note that
8230 when using dynamic backend names, it is highly recommended to use a prefix
8231 that no other backend uses in order to ensure that an unauthorized backend
8232 cannot be forced from the request.
8233
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008234 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008235 used to detect the association between frontends and backends to compute the
8236 backend's "fullconn" setting. This cannot be done for dynamic names.
8237
8238 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
8239 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008240
Willy Tarreau036fae02008-01-06 13:24:40 +01008241
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008242use-server <server> if <condition>
8243use-server <server> unless <condition>
8244 Only use a specific server if/unless an ACL-based condition is matched.
8245 May be used in sections : defaults | frontend | listen | backend
8246 no | no | yes | yes
8247 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008248 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008249
8250 <condition> is a condition composed of ACLs, as described in section 7.
8251
8252 By default, connections which arrive to a backend are load-balanced across
8253 the available servers according to the configured algorithm, unless a
8254 persistence mechanism such as a cookie is used and found in the request.
8255
8256 Sometimes it is desirable to forward a particular request to a specific
8257 server without having to declare a dedicated backend for this server. This
8258 can be achieved using the "use-server" rules. These rules are evaluated after
8259 the "redirect" rules and before evaluating cookies, and they have precedence
8260 on them. There may be as many "use-server" rules as desired. All of these
8261 rules are evaluated in their declaration order, and the first one which
8262 matches will assign the server.
8263
8264 If a rule designates a server which is down, and "option persist" is not used
8265 and no force-persist rule was validated, it is ignored and evaluation goes on
8266 with the next rules until one matches.
8267
8268 In the first form, the server will be used if the condition is met. In the
8269 second form, the server will be used if the condition is not met. If no
8270 condition is valid, the processing continues and the server will be assigned
8271 according to other persistence mechanisms.
8272
8273 Note that even if a rule is matched, cookie processing is still performed but
8274 does not assign the server. This allows prefixed cookies to have their prefix
8275 stripped.
8276
8277 The "use-server" statement works both in HTTP and TCP mode. This makes it
8278 suitable for use with content-based inspection. For instance, a server could
8279 be selected in a farm according to the TLS SNI field. And if these servers
8280 have their weight set to zero, they will not be used for other traffic.
8281
8282 Example :
8283 # intercept incoming TLS requests based on the SNI field
8284 use-server www if { req_ssl_sni -i www.example.com }
8285 server www 192.168.0.1:443 weight 0
8286 use-server mail if { req_ssl_sni -i mail.example.com }
8287 server mail 192.168.0.1:587 weight 0
8288 use-server imap if { req_ssl_sni -i imap.example.com }
8289 server mail 192.168.0.1:993 weight 0
8290 # all the rest is forwarded to this server
8291 server default 192.168.0.2:443 check
8292
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008293 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008294
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008295
82965. Bind and Server options
8297--------------------------
8298
8299The "bind", "server" and "default-server" keywords support a number of settings
8300depending on some build options and on the system HAProxy was built on. These
8301settings generally each consist in one word sometimes followed by a value,
8302written on the same line as the "bind" or "server" line. All these options are
8303described in this section.
8304
8305
83065.1. Bind options
8307-----------------
8308
8309The "bind" keyword supports a certain number of settings which are all passed
8310as arguments on the same line. The order in which those arguments appear makes
8311no importance, provided that they appear after the bind address. All of these
8312parameters are optional. Some of them consist in a single words (booleans),
8313while other ones expect a value after them. In this case, the value must be
8314provided immediately after the setting name.
8315
8316The currently supported settings are the following ones.
8317
8318accept-proxy
8319 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +02008320 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
8321 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008322 3/4 addresses of the incoming connection to be used everywhere an address is
8323 used, with the only exception of "tcp-request connection" rules which will
8324 only see the real connection address. Logs will reflect the addresses
8325 indicated in the protocol, unless it is violated, in which case the real
8326 address will still be used. This keyword combined with support from external
8327 components can be used as an efficient and reliable alternative to the
8328 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +02008329 usable. See also "tcp-request connection expect-proxy" for a finer-grained
8330 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008331
Willy Tarreauab861d32013-04-02 02:30:41 +02008332alpn <protocols>
8333 This enables the TLS ALPN extension and advertises the specified protocol
8334 list as supported on top of ALPN. The protocol list consists in a comma-
8335 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
8336 quotes). This requires that the SSL library is build with support for TLS
8337 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
8338 initial NPN extension.
8339
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008340backlog <backlog>
8341 Sets the socket's backlog to this value. If unspecified, the frontend's
8342 backlog is used instead, which generally defaults to the maxconn value.
8343
Emeric Brun7fb34422012-09-28 15:26:15 +02008344ecdhe <named curve>
8345 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +01008346 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
8347 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +02008348
Emeric Brunfd33a262012-10-11 16:28:27 +02008349ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008350 This setting is only available when support for OpenSSL was built in. It
8351 designates a PEM file from which to load CA certificates used to verify
8352 client's certificate.
8353
Emeric Brunb6dc9342012-09-28 17:55:37 +02008354ca-ignore-err [all|<errorID>,...]
8355 This setting is only available when support for OpenSSL was built in.
8356 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
8357 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
8358 error is ignored.
8359
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008360ciphers <ciphers>
8361 This setting is only available when support for OpenSSL was built in. It sets
8362 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008363 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008364 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
8365 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
8366
Emeric Brunfd33a262012-10-11 16:28:27 +02008367crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008368 This setting is only available when support for OpenSSL was built in. It
8369 designates a PEM file from which to load certificate revocation list used
8370 to verify client's certificate.
8371
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008372crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +00008373 This setting is only available when support for OpenSSL was built in. It
8374 designates a PEM file containing both the required certificates and any
8375 associated private keys. This file can be built by concatenating multiple
8376 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
8377 requires an intermediate certificate, this can also be concatenated into this
8378 file.
8379
8380 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
8381 are loaded.
8382
8383 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté62043c92015-01-25 00:16:08 +01008384 that directory will be loaded in alphabetic order unless their name ends with
8385 '.issuer' or '.ocsp' (reserved extensions). This directive may be specified
8386 multiple times in order to load certificates from multiple files or
8387 directories. The certificates will be presented to clients who provide a valid
8388 TLS Server Name Indication field matching one of their CN or alt subjects.
8389 Wildcards are supported, where a wildcard character '*' is used instead of the
8390 first hostname component (eg: *.example.org matches www.example.org but not
Alex Davies0fbf0162013-03-02 16:04:50 +00008391 www.sub.example.org).
8392
8393 If no SNI is provided by the client or if the SSL library does not support
8394 TLS extensions, or if the client provides an SNI hostname which does not
8395 match any certificate, then the first loaded certificate will be presented.
8396 This means that when loading certificates from a directory, it is highly
Cyril Bonté62043c92015-01-25 00:16:08 +01008397 recommended to load the default one first as a file or to ensure that it will
8398 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +00008399
Emeric Brune032bfa2012-09-28 13:01:45 +02008400 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008401
Alex Davies0fbf0162013-03-02 16:04:50 +00008402 Some CAs (such as Godaddy) offer a drop down list of server types that do not
8403 include HAProxy when obtaining a certificate. If this happens be sure to
Godbach8bf60a12014-04-21 21:42:41 +08008404 choose a webserver that the CA believes requires an intermediate CA (for
Alex Davies0fbf0162013-03-02 16:04:50 +00008405 Godaddy, selection Apache Tomcat will get the correct bundle, but many
8406 others, e.g. nginx, result in a wrong bundle that will not work for some
8407 clients).
8408
Emeric Brun4147b2e2014-06-16 18:36:30 +02008409 For each PEM file, haproxy checks for the presence of file at the same path
8410 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
8411 Status Request extension (also known as "OCSP stapling") is automatically
8412 enabled. The content of this file is optional. If not empty, it must contain
8413 a valid OCSP Response in DER format. In order to be valid an OCSP Response
8414 must comply with the following rules: it has to indicate a good status,
8415 it has to be a single response for the certificate of the PEM file, and it
8416 has to be valid at the moment of addition. If these rules are not respected
8417 the OCSP Response is ignored and a warning is emitted. In order to identify
8418 which certificate an OCSP Response applies to, the issuer's certificate is
8419 necessary. If the issuer's certificate is not found in the PEM file, it will
8420 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
8421 if it exists otherwise it will fail with an error.
8422
Emeric Brunb6dc9342012-09-28 17:55:37 +02008423crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +00008424 This setting is only available when support for OpenSSL was built in. Sets a
8425 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008426 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +00008427 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +02008428
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008429crt-list <file>
8430 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008431 designates a list of PEM file with an optional list of SNI filter per
8432 certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008433
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008434 <crtfile> [[!]<snifilter> ...]
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008435
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008436 Wildcards are supported in the SNI filter. Negative filter are also supported,
8437 only useful in combination with a wildcard filter to exclude a particular SNI.
8438 The certificates will be presented to clients who provide a valid TLS Server
8439 Name Indication field matching one of the SNI filters. If no SNI filter is
8440 specified, the CN and alt subjects are used. This directive may be specified
8441 multiple times. See the "crt" option for more information. The default
8442 certificate is still needed to meet OpenSSL expectations. If it is not used,
8443 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008444
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008445defer-accept
8446 Is an optional keyword which is supported only on certain Linux kernels. It
8447 states that a connection will only be accepted once some data arrive on it,
8448 or at worst after the first retransmit. This should be used only on protocols
8449 for which the client talks first (eg: HTTP). It can slightly improve
8450 performance by ensuring that most of the request is already available when
8451 the connection is accepted. On the other hand, it will not be able to detect
8452 connections which don't talk. It is important to note that this option is
8453 broken in all kernels up to 2.6.31, as the connection is never accepted until
8454 the client talks. This can cause issues with front firewalls which would see
8455 an established connection while the proxy will only see it in SYN_RECV. This
8456 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
8457
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008458force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008459 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008460 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun42a3e202014-10-30 15:56:50 +01008461 for high connection rates. This option is also available on global statement
8462 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008463
8464force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008465 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008466 this listener. This option is also available on global statement
8467 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008468
8469force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008470 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008471 this listener. This option is also available on global statement
8472 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008473
8474force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008475 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008476 this listener. This option is also available on global statement
8477 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008478
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008479gid <gid>
8480 Sets the group of the UNIX sockets to the designated system gid. It can also
8481 be set by default in the global section's "unix-bind" statement. Note that
8482 some platforms simply ignore this. This setting is equivalent to the "group"
8483 setting except that the group ID is used instead of its name. This setting is
8484 ignored by non UNIX sockets.
8485
8486group <group>
8487 Sets the group of the UNIX sockets to the designated system group. It can
8488 also be set by default in the global section's "unix-bind" statement. Note
8489 that some platforms simply ignore this. This setting is equivalent to the
8490 "gid" setting except that the group name is used instead of its gid. This
8491 setting is ignored by non UNIX sockets.
8492
8493id <id>
8494 Fixes the socket ID. By default, socket IDs are automatically assigned, but
8495 sometimes it is more convenient to fix them to ease monitoring. This value
8496 must be strictly positive and unique within the listener/frontend. This
8497 option can only be used when defining only a single socket.
8498
8499interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +01008500 Restricts the socket to a specific interface. When specified, only packets
8501 received from that particular interface are processed by the socket. This is
8502 currently only supported on Linux. The interface must be a primary system
8503 interface, not an aliased interface. It is also possible to bind multiple
8504 frontends to the same address if they are bound to different interfaces. Note
8505 that binding to a network interface requires root privileges. This parameter
8506 is only compatible with TCPv4/TCPv6 sockets.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008507
Willy Tarreauabb175f2012-09-24 12:43:26 +02008508level <level>
8509 This setting is used with the stats sockets only to restrict the nature of
8510 the commands that can be issued on the socket. It is ignored by other
8511 sockets. <level> can be one of :
8512 - "user" is the least privileged level ; only non-sensitive stats can be
8513 read, and no change is allowed. It would make sense on systems where it
8514 is not easy to restrict access to the socket.
8515 - "operator" is the default level and fits most common uses. All data can
8516 be read, and only non-sensitive changes are permitted (eg: clear max
8517 counters).
8518 - "admin" should be used with care, as everything is permitted (eg: clear
8519 all counters).
8520
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008521maxconn <maxconn>
8522 Limits the sockets to this number of concurrent connections. Extraneous
8523 connections will remain in the system's backlog until a connection is
8524 released. If unspecified, the limit will be the same as the frontend's
8525 maxconn. Note that in case of port ranges or multiple addresses, the same
8526 value will be applied to each socket. This setting enables different
8527 limitations on expensive sockets, for instance SSL entries which may easily
8528 eat all memory.
8529
8530mode <mode>
8531 Sets the octal mode used to define access permissions on the UNIX socket. It
8532 can also be set by default in the global section's "unix-bind" statement.
8533 Note that some platforms simply ignore this. This setting is ignored by non
8534 UNIX sockets.
8535
8536mss <maxseg>
8537 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
8538 connections. This can be used to force a lower MSS for certain specific
8539 ports, for instance for connections passing through a VPN. Note that this
8540 relies on a kernel feature which is theoretically supported under Linux but
8541 was buggy in all versions prior to 2.6.28. It may or may not work on other
8542 operating systems. It may also not change the advertised value but change the
8543 effective size of outgoing segments. The commonly advertised value for TCPv4
8544 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
8545 positive, it will be used as the advertised MSS. If it is negative, it will
8546 indicate by how much to reduce the incoming connection's advertised MSS for
8547 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
8548
8549name <name>
8550 Sets an optional name for these sockets, which will be reported on the stats
8551 page.
8552
8553nice <nice>
8554 Sets the 'niceness' of connections initiated from the socket. Value must be
8555 in the range -1024..1024 inclusive, and defaults to zero. Positive values
8556 means that such connections are more friendly to others and easily offer
8557 their place in the scheduler. On the opposite, negative values mean that
8558 connections want to run with a higher priority than others. The difference
8559 only happens under high loads when the system is close to saturation.
8560 Negative values are appropriate for low-latency or administration services,
8561 and high values are generally recommended for CPU intensive tasks such as SSL
8562 processing or bulk transfers which are less sensible to latency. For example,
8563 it may make sense to use a positive value for an SMTP socket and a negative
8564 one for an RDP socket.
8565
Emeric Brun9b3009b2012-10-05 11:55:06 +02008566no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008567 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008568 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008569 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun42a3e202014-10-30 15:56:50 +01008570 be enabled using any configuration option. This option is also available on
8571 global statement "ssl-default-bind-options". See also "force-tls*",
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008572 and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008573
Emeric Brun90ad8722012-10-02 14:00:59 +02008574no-tls-tickets
8575 This setting is only available when support for OpenSSL was built in. It
8576 disables the stateless session resumption (RFC 5077 TLS Ticket
8577 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01008578 session resumption is more expensive in CPU usage. This option is also
8579 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +02008580
Emeric Brun9b3009b2012-10-05 11:55:06 +02008581no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008582 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008583 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008584 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008585 cannot be enabled using any configuration option. This option is also
8586 available on global statement "ssl-default-bind-options". See also
8587 "force-tlsv*", and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008588
Emeric Brun9b3009b2012-10-05 11:55:06 +02008589no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +02008590 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008591 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008592 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008593 cannot be enabled using any configuration option. This option is also
8594 available on global statement "ssl-default-bind-options". See also
8595 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008596
Emeric Brun9b3009b2012-10-05 11:55:06 +02008597no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +02008598 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008599 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008600 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008601 cannot be enabled using any configuration option. This option is also
8602 available on global statement "ssl-default-bind-options". See also
8603 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008604
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008605npn <protocols>
8606 This enables the NPN TLS extension and advertises the specified protocol list
8607 as supported on top of NPN. The protocol list consists in a comma-delimited
8608 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
8609 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +02008610 enabled (check with haproxy -vv). Note that the NPN extension has been
8611 replaced with the ALPN extension (see the "alpn" keyword).
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008612
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008613process [ all | odd | even | <number 1-64>[-<number 1-64>] ]
8614 This restricts the list of processes on which this listener is allowed to
8615 run. It does not enforce any process but eliminates those which do not match.
8616 If the frontend uses a "bind-process" setting, the intersection between the
8617 two is applied. If in the end the listener is not allowed to run on any
8618 remaining process, a warning is emitted, and the listener will either run on
8619 the first process of the listener if a single process was specified, or on
8620 all of its processes if multiple processes were specified. For the unlikely
Willy Tarreauae302532014-05-07 19:22:24 +02008621 case where several ranges are needed, this directive may be repeated. The
8622 main purpose of this directive is to be used with the stats sockets and have
8623 one different socket per process. The second purpose is to have multiple bind
8624 lines sharing the same IP:port but not the same process in a listener, so
8625 that the system can distribute the incoming connections into multiple queues
8626 and allow a smoother inter-process load balancing. Currently Linux 3.9 and
8627 above is known for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008628
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008629ssl
8630 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008631 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008632 certificate is necessary (see "crt" above). All contents in the buffers will
8633 appear in clear text, so that ACLs and HTTP processing will only have access
8634 to deciphered contents.
8635
Emmanuel Hocdet65623372013-01-24 17:17:15 +01008636strict-sni
8637 This setting is only available when support for OpenSSL was built in. The
8638 SSL/TLS negotiation is allow only if the client provided an SNI which match
8639 a certificate. The default certificate is not used.
8640 See the "crt" option for more information.
8641
Willy Tarreau1c862c52012-10-05 16:21:00 +02008642tfo
Lukas Tribus0defb902013-02-13 23:35:39 +01008643 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +02008644 enables TCP Fast Open on the listening socket, which means that clients which
8645 support this feature will be able to send a request and receive a response
8646 during the 3-way handshake starting from second connection, thus saving one
8647 round-trip after the first connection. This only makes sense with protocols
8648 that use high connection rates and where each round trip matters. This can
8649 possibly cause issues with many firewalls which do not accept data on SYN
8650 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +02008651 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
8652 need to build HAProxy with USE_TFO=1 if your libc doesn't define
8653 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +02008654
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008655transparent
8656 Is an optional keyword which is supported only on certain Linux kernels. It
8657 indicates that the addresses will be bound even if they do not belong to the
8658 local machine, and that packets targeting any of these addresses will be
8659 intercepted just as if the addresses were locally configured. This normally
8660 requires that IP forwarding is enabled. Caution! do not use this with the
8661 default address '*', as it would redirect any traffic for the specified port.
8662 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
8663 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
8664 kernel version. Some distribution kernels include backports of the feature,
8665 so check for support with your vendor.
8666
Willy Tarreau77e3af92012-11-24 15:07:23 +01008667v4v6
8668 Is an optional keyword which is supported only on most recent systems
8669 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
8670 and IPv6 when it uses the default address. Doing so is sometimes necessary
8671 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008672 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +01008673
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008674v6only
8675 Is an optional keyword which is supported only on most recent systems
8676 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
8677 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +01008678 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
8679 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008680
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008681uid <uid>
8682 Sets the owner of the UNIX sockets to the designated system uid. It can also
8683 be set by default in the global section's "unix-bind" statement. Note that
8684 some platforms simply ignore this. This setting is equivalent to the "user"
8685 setting except that the user numeric ID is used instead of its name. This
8686 setting is ignored by non UNIX sockets.
8687
8688user <user>
8689 Sets the owner of the UNIX sockets to the designated system user. It can also
8690 be set by default in the global section's "unix-bind" statement. Note that
8691 some platforms simply ignore this. This setting is equivalent to the "uid"
8692 setting except that the user name is used instead of its uid. This setting is
8693 ignored by non UNIX sockets.
8694
Emeric Brun1a073b42012-09-28 17:07:34 +02008695verify [none|optional|required]
8696 This setting is only available when support for OpenSSL was built in. If set
8697 to 'none', client certificate is not requested. This is the default. In other
8698 cases, a client certificate is requested. If the client does not provide a
8699 certificate after the request and if 'verify' is set to 'required', then the
8700 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +02008701 certificate provided by the client is always verified using CAs from
8702 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
8703 is aborted, regardless of the 'verify' option, unless the error code exactly
8704 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008705
Willy Tarreaub6205fd2012-09-24 12:27:33 +020087065.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +01008707------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02008708
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008709The "server" and "default-server" keywords support a certain number of settings
8710which are all passed as arguments on the server line. The order in which those
8711arguments appear does not count, and they are all optional. Some of those
8712settings are single words (booleans) while others expect one or several values
8713after them. In this case, the values must immediately follow the setting name.
8714Except default-server, all those settings must be specified after the server's
8715address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +02008716
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008717 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008718 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +02008719
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008720The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008721
Willy Tarreauceb4ac92012-04-28 00:41:46 +02008722addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008723 Using the "addr" parameter, it becomes possible to use a different IP address
8724 to send health-checks. On some servers, it may be desirable to dedicate an IP
8725 address to specific component able to perform complex tests which are more
8726 suitable to health-checks than the application. This parameter is ignored if
8727 the "check" parameter is not set. See also the "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008728
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008729 Supported in default-server: No
8730
Simon Hormand60d6912013-11-25 10:46:36 +09008731agent-check
8732 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +01008733 health check. An agent health check is performed by making a TCP connection
8734 to the port set by the "agent-port" parameter and reading an ASCII string.
8735 The string is made of a series of words delimited by spaces, tabs or commas
8736 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +09008737
Willy Tarreau81f5d942013-12-09 20:51:51 +01008738 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +09008739 Values in this format will set the weight proportional to the initial
Willy Tarreau47067ca2014-10-07 15:27:33 +02008740 weight of a server as configured when haproxy starts. Note that a zero
8741 weight is reported on the stats page as "DRAIN" since it has the same
8742 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +09008743
Willy Tarreau81f5d942013-12-09 20:51:51 +01008744 - The word "ready". This will turn the server's administrative state to the
8745 READY mode, thus cancelling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +09008746
Willy Tarreau81f5d942013-12-09 20:51:51 +01008747 - The word "drain". This will turn the server's administrative state to the
8748 DRAIN mode, thus it will not accept any new connections other than those
8749 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +09008750
Willy Tarreau81f5d942013-12-09 20:51:51 +01008751 - The word "maint". This will turn the server's administrative state to the
8752 MAINT mode, thus it will not accept any new connections at all, and health
8753 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +09008754
Willy Tarreau81f5d942013-12-09 20:51:51 +01008755 - The words "down", "failed", or "stopped", optionally followed by a
8756 description string after a sharp ('#'). All of these mark the server's
8757 operating state as DOWN, but since the word itself is reported on the stats
8758 page, the difference allows an administrator to know if the situation was
8759 expected or not : the service may intentionally be stopped, may appear up
8760 but fail some validity tests, or may be seen as down (eg: missing process,
8761 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +09008762
Willy Tarreau81f5d942013-12-09 20:51:51 +01008763 - The word "up" sets back the server's operating state as UP if health checks
8764 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +09008765
Willy Tarreau81f5d942013-12-09 20:51:51 +01008766 Parameters which are not advertised by the agent are not changed. For
8767 example, an agent might be designed to monitor CPU usage and only report a
8768 relative weight and never interact with the operating status. Similarly, an
8769 agent could be designed as an end-user interface with 3 radio buttons
8770 allowing an administrator to change only the administrative state. However,
8771 it is important to consider that only the agent may revert its own actions,
8772 so if a server is set to DRAIN mode or to DOWN state using the agent, the
8773 agent must implement the other equivalent actions to bring the service into
8774 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +09008775
Simon Horman2f1f9552013-11-25 10:46:37 +09008776 Failure to connect to the agent is not considered an error as connectivity
8777 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +01008778 parameter. Warning though, it is not a good idea to stop an agent after it
8779 reports "down", since only an agent reporting "up" will be able to turn the
8780 server up again. Note that the CLI on the Unix stats socket is also able to
8781 force an agent's result in order to workaround a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +09008782
Willy Tarreau81f5d942013-12-09 20:51:51 +01008783 Requires the "agent-port" parameter to be set. See also the "agent-inter"
8784 parameter.
Simon Hormand60d6912013-11-25 10:46:36 +09008785
8786 Supported in default-server: No
8787
8788agent-inter <delay>
8789 The "agent-inter" parameter sets the interval between two agent checks
8790 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8791
8792 Just as with every other time-based parameter, it may be entered in any
8793 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
8794 parameter also serves as a timeout for agent checks "timeout check" is
8795 not set. In order to reduce "resonance" effects when multiple servers are
8796 hosted on the same hardware, the agent and health checks of all servers
8797 are started with a small time offset between them. It is also possible to
8798 add some random noise in the agent and health checks interval using the
8799 global "spread-checks" keyword. This makes sense for instance when a lot
8800 of backends use the same servers.
8801
8802 See also the "agent-check" and "agent-port" parameters.
8803
8804 Supported in default-server: Yes
8805
8806agent-port <port>
8807 The "agent-port" parameter sets the TCP port used for agent checks.
8808
8809 See also the "agent-check" and "agent-inter" parameters.
8810
8811 Supported in default-server: Yes
8812
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008813backup
8814 When "backup" is present on a server line, the server is only used in load
8815 balancing when all other non-backup servers are unavailable. Requests coming
8816 with a persistence cookie referencing the server will always be served
8817 though. By default, only the first operational backup server is used, unless
8818 the "allbackups" option is set in the backend. See also the "allbackups"
8819 option.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008820
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008821 Supported in default-server: No
8822
Emeric Brunef42d922012-10-11 16:11:36 +02008823ca-file <cafile>
8824 This setting is only available when support for OpenSSL was built in. It
8825 designates a PEM file from which to load CA certificates used to verify
8826 server's certificate.
8827
8828 Supported in default-server: No
8829
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008830check
8831 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +01008832 always considered available. If "check" is set, the server is available when
8833 accepting periodic TCP connections, to ensure that it is really able to serve
8834 requests. The default address and port to send the tests to are those of the
8835 server, and the default source is the same as the one defined in the
8836 backend. It is possible to change the address using the "addr" parameter, the
8837 port using the "port" parameter, the source address using the "source"
8838 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +09008839 parameters. The request method is define in the backend using the "httpchk",
8840 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
8841 refer to those options and parameters for more information.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008842
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008843 Supported in default-server: No
8844
Willy Tarreau6c16adc2012-10-05 00:04:16 +02008845check-send-proxy
8846 This option forces emission of a PROXY protocol line with outgoing health
8847 checks, regardless of whether the server uses send-proxy or not for the
8848 normal traffic. By default, the PROXY protocol is enabled for health checks
8849 if it is already enabled for normal traffic and if no "port" nor "addr"
8850 directive is present. However, if such a directive is present, the
8851 "check-send-proxy" option needs to be used to force the use of the
8852 protocol. See also the "send-proxy" option for more information.
8853
8854 Supported in default-server: No
8855
Willy Tarreau763a95b2012-10-04 23:15:39 +02008856check-ssl
8857 This option forces encryption of all health checks over SSL, regardless of
8858 whether the server uses SSL or not for the normal traffic. This is generally
8859 used when an explicit "port" or "addr" directive is specified and SSL health
8860 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008861 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +02008862 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
8863 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
8864 All SSL settings are common to health checks and traffic (eg: ciphers).
8865 See the "ssl" option for more information.
8866
8867 Supported in default-server: No
8868
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008869ciphers <ciphers>
8870 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008871 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008872 string is defined in "man 1 ciphers". When SSL is used to communicate with
8873 servers on the local network, it is common to see a weaker set of algorithms
8874 than what is used over the internet. Doing so reduces CPU usage on both the
8875 server and haproxy while still keeping it compatible with deployed software.
8876 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
8877 is needed and just connectivity, using DES can be appropriate.
8878
Willy Tarreau763a95b2012-10-04 23:15:39 +02008879 Supported in default-server: No
8880
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008881cookie <value>
8882 The "cookie" parameter sets the cookie value assigned to the server to
8883 <value>. This value will be checked in incoming requests, and the first
8884 operational server possessing the same value will be selected. In return, in
8885 cookie insertion or rewrite modes, this value will be assigned to the cookie
8886 sent to the client. There is nothing wrong in having several servers sharing
8887 the same cookie value, and it is in fact somewhat common between normal and
8888 backup servers. See also the "cookie" keyword in backend section.
8889
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008890 Supported in default-server: No
8891
Emeric Brunef42d922012-10-11 16:11:36 +02008892crl-file <crlfile>
8893 This setting is only available when support for OpenSSL was built in. It
8894 designates a PEM file from which to load certificate revocation list used
8895 to verify server's certificate.
8896
8897 Supported in default-server: No
8898
Emeric Bruna7aa3092012-10-26 12:58:00 +02008899crt <cert>
8900 This setting is only available when support for OpenSSL was built in.
8901 It designates a PEM file from which to load both a certificate and the
8902 associated private key. This file can be built by concatenating both PEM
8903 files into one. This certificate will be sent if the server send a client
8904 certificate request.
8905
8906 Supported in default-server: No
8907
Willy Tarreau96839092010-03-29 10:02:24 +02008908disabled
8909 The "disabled" keyword starts the server in the "disabled" state. That means
8910 that it is marked down in maintenance mode, and no connection other than the
8911 ones allowed by persist mode will reach it. It is very well suited to setup
8912 new servers, because normal traffic will never reach them, while it is still
8913 possible to test the service by making use of the force-persist mechanism.
8914
8915 Supported in default-server: No
8916
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008917error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +01008918 If health observing is enabled, the "error-limit" parameter specifies the
8919 number of consecutive errors that triggers event selected by the "on-error"
8920 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008921
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008922 Supported in default-server: Yes
8923
8924 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008925
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008926fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008927 The "fall" parameter states that a server will be considered as dead after
8928 <count> consecutive unsuccessful health checks. This value defaults to 3 if
8929 unspecified. See also the "check", "inter" and "rise" parameters.
8930
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008931 Supported in default-server: Yes
8932
Emeric Brun8694b9a2012-10-05 14:39:07 +02008933force-sslv3
8934 This option enforces use of SSLv3 only when SSL is used to communicate with
8935 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun42a3e202014-10-30 15:56:50 +01008936 high connection rates. This option is also available on global statement
8937 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008938
8939 Supported in default-server: No
8940
8941force-tlsv10
8942 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008943 the server. This option is also available on global statement
8944 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008945
8946 Supported in default-server: No
8947
8948force-tlsv11
8949 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008950 the server. This option is also available on global statement
8951 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008952
8953 Supported in default-server: No
8954
8955force-tlsv12
8956 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008957 the server. This option is also available on global statement
8958 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008959
8960 Supported in default-server: No
8961
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008962id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02008963 Set a persistent ID for the server. This ID must be positive and unique for
8964 the proxy. An unused ID will automatically be assigned if unset. The first
8965 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008966
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008967 Supported in default-server: No
8968
8969inter <delay>
8970fastinter <delay>
8971downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008972 The "inter" parameter sets the interval between two consecutive health checks
8973 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8974 It is also possible to use "fastinter" and "downinter" to optimize delays
8975 between checks depending on the server state :
8976
8977 Server state | Interval used
8978 ---------------------------------+-----------------------------------------
8979 UP 100% (non-transitional) | "inter"
8980 ---------------------------------+-----------------------------------------
8981 Transitionally UP (going down), |
8982 Transitionally DOWN (going up), | "fastinter" if set, "inter" otherwise.
8983 or yet unchecked. |
8984 ---------------------------------+-----------------------------------------
8985 DOWN 100% (non-transitional) | "downinter" if set, "inter" otherwise.
8986 ---------------------------------+-----------------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +01008987
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008988 Just as with every other time-based parameter, they can be entered in any
8989 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
8990 serves as a timeout for health checks sent to servers if "timeout check" is
8991 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +09008992 hosted on the same hardware, the agent and health checks of all servers
8993 are started with a small time offset between them. It is also possible to
8994 add some random noise in the agent and health checks interval using the
8995 global "spread-checks" keyword. This makes sense for instance when a lot
8996 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008997
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008998 Supported in default-server: Yes
8999
9000maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009001 The "maxconn" parameter specifies the maximal number of concurrent
9002 connections that will be sent to this server. If the number of incoming
9003 concurrent requests goes higher than this value, they will be queued, waiting
9004 for a connection to be released. This parameter is very important as it can
9005 save fragile servers from going down under extreme loads. If a "minconn"
9006 parameter is specified, the limit becomes dynamic. The default value is "0"
9007 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
9008 the backend's "fullconn" keyword.
9009
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009010 Supported in default-server: Yes
9011
9012maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009013 The "maxqueue" parameter specifies the maximal number of connections which
9014 will wait in the queue for this server. If this limit is reached, next
9015 requests will be redispatched to other servers instead of indefinitely
9016 waiting to be served. This will break persistence but may allow people to
9017 quickly re-log in when the server they try to connect to is dying. The
9018 default value is "0" which means the queue is unlimited. See also the
9019 "maxconn" and "minconn" parameters.
9020
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009021 Supported in default-server: Yes
9022
9023minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009024 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
9025 limit following the backend's load. The server will always accept at least
9026 <minconn> connections, never more than <maxconn>, and the limit will be on
9027 the ramp between both values when the backend has less than <fullconn>
9028 concurrent connections. This makes it possible to limit the load on the
9029 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009030 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009031 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009032
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009033 Supported in default-server: Yes
9034
Emeric Brun9b3009b2012-10-05 11:55:06 +02009035no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009036 This option disables support for SSLv3 when SSL is used to communicate with
9037 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emeric Brun8694b9a2012-10-05 14:39:07 +02009038 using any configuration option. See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009039
Willy Tarreau763a95b2012-10-04 23:15:39 +02009040 Supported in default-server: No
9041
Emeric Brunf9c5c472012-10-11 15:28:34 +02009042no-tls-tickets
9043 This setting is only available when support for OpenSSL was built in. It
9044 disables the stateless session resumption (RFC 5077 TLS Ticket
9045 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01009046 session resumption is more expensive in CPU usage for servers. This option
9047 is also available on global statement "ssl-default-server-options".
Emeric Brunf9c5c472012-10-11 15:28:34 +02009048
9049 Supported in default-server: No
9050
Emeric Brun9b3009b2012-10-05 11:55:06 +02009051no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +02009052 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009053 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9054 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009055 often makes sense to disable it when communicating with local servers. This
9056 option is also available on global statement "ssl-default-server-options".
9057 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009058
Willy Tarreau763a95b2012-10-04 23:15:39 +02009059 Supported in default-server: No
9060
Emeric Brun9b3009b2012-10-05 11:55:06 +02009061no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +02009062 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009063 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9064 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009065 often makes sense to disable it when communicating with local servers. This
9066 option is also available on global statement "ssl-default-server-options".
9067 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009068
Willy Tarreau763a95b2012-10-04 23:15:39 +02009069 Supported in default-server: No
9070
Emeric Brun9b3009b2012-10-05 11:55:06 +02009071no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +02009072 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009073 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9074 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009075 often makes sense to disable it when communicating with local servers. This
9076 option is also available on global statement "ssl-default-server-options".
9077 See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009078
Willy Tarreau763a95b2012-10-04 23:15:39 +02009079 Supported in default-server: No
9080
Simon Hormanfa461682011-06-25 09:39:49 +09009081non-stick
9082 Never add connections allocated to this sever to a stick-table.
9083 This may be used in conjunction with backup to ensure that
9084 stick-table persistence is disabled for backup servers.
9085
Willy Tarreau763a95b2012-10-04 23:15:39 +02009086 Supported in default-server: No
9087
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009088observe <mode>
9089 This option enables health adjusting based on observing communication with
9090 the server. By default this functionality is disabled and enabling it also
9091 requires to enable health checks. There are two supported modes: "layer4" and
9092 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
9093 significant. In layer7, which is only allowed for http proxies, responses
9094 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +01009095 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009096
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009097 Supported in default-server: No
9098
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009099 See also the "check", "on-error" and "error-limit".
9100
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009101on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009102 Select what should happen when enough consecutive errors are detected.
9103 Currently, four modes are available:
9104 - fastinter: force fastinter
9105 - fail-check: simulate a failed check, also forces fastinter (default)
9106 - sudden-death: simulate a pre-fatal failed health check, one more failed
9107 check will mark a server down, forces fastinter
9108 - mark-down: mark the server immediately down and force fastinter
9109
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009110 Supported in default-server: Yes
9111
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009112 See also the "check", "observe" and "error-limit".
9113
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009114on-marked-down <action>
9115 Modify what occurs when a server is marked down.
9116 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009117 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
9118 all connections to the server are immediately terminated when the server
9119 goes down. It might be used if the health check detects more complex cases
9120 than a simple connection status, and long timeouts would cause the service
9121 to remain unresponsive for too long a time. For instance, a health check
9122 might detect that a database is stuck and that there's no chance to reuse
9123 existing connections anymore. Connections killed this way are logged with
9124 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009125
9126 Actions are disabled by default
9127
9128 Supported in default-server: Yes
9129
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009130on-marked-up <action>
9131 Modify what occurs when a server is marked up.
9132 Currently one action is available:
9133 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
9134 done only if the server is not in backup state and if it is not disabled
9135 (it must have an effective weight > 0). This can be used sometimes to force
9136 an active server to take all the traffic back after recovery when dealing
9137 with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble
9138 than it tries to solve (eg: incomplete transactions), so use this feature
9139 with extreme care. Sessions killed because a server comes up are logged
9140 with an 'U' termination code (for "Up").
9141
9142 Actions are disabled by default
9143
9144 Supported in default-server: Yes
9145
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009146port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009147 Using the "port" parameter, it becomes possible to use a different port to
9148 send health-checks. On some servers, it may be desirable to dedicate a port
9149 to a specific component able to perform complex tests which are more suitable
9150 to health-checks than the application. It is common to run a simple script in
9151 inetd for instance. This parameter is ignored if the "check" parameter is not
9152 set. See also the "addr" parameter.
9153
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009154 Supported in default-server: Yes
9155
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009156redir <prefix>
9157 The "redir" parameter enables the redirection mode for all GET and HEAD
9158 requests addressing this server. This means that instead of having HAProxy
9159 forward the request to the server, it will send an "HTTP 302" response with
9160 the "Location" header composed of this prefix immediately followed by the
9161 requested URI beginning at the leading '/' of the path component. That means
9162 that no trailing slash should be used after <prefix>. All invalid requests
9163 will be rejected, and all non-GET or HEAD requests will be normally served by
9164 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009165 mangling nor cookie insertion is possible in the response. However, cookies in
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009166 requests are still analysed, making this solution completely usable to direct
9167 users to a remote location in case of local disaster. Main use consists in
9168 increasing bandwidth for static servers by having the clients directly
9169 connect to them. Note: never use a relative location here, it would cause a
9170 loop between the client and HAProxy!
9171
9172 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
9173
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009174 Supported in default-server: No
9175
9176rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009177 The "rise" parameter states that a server will be considered as operational
9178 after <count> consecutive successful health checks. This value defaults to 2
9179 if unspecified. See also the "check", "inter" and "fall" parameters.
9180
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009181 Supported in default-server: Yes
9182
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009183send-proxy
9184 The "send-proxy" parameter enforces use of the PROXY protocol over any
9185 connection established to this server. The PROXY protocol informs the other
9186 end about the layer 3/4 addresses of the incoming connection, so that it can
9187 know the client's address or the public address it accessed to, whatever the
9188 upper layer protocol. For connections accepted by an "accept-proxy" listener,
9189 the advertised address will be used. Only TCPv4 and TCPv6 address families
9190 are supported. Other families such as Unix sockets, will report an UNKNOWN
9191 family. Servers using this option can fully be chained to another instance of
9192 haproxy listening with an "accept-proxy" setting. This setting must not be
Willy Tarreau6c16adc2012-10-05 00:04:16 +02009193 used if the server isn't aware of the protocol. When health checks are sent
9194 to the server, the PROXY protocol is automatically used when this option is
9195 set, unless there is an explicit "port" or "addr" directive, in which case an
9196 explicit "check-send-proxy" directive would also be needed to use the PROXY
9197 protocol. See also the "accept-proxy" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009198
9199 Supported in default-server: No
9200
David Safb76832014-05-08 23:42:08 -04009201send-proxy-v2
9202 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
9203 over any connection established to this server. The PROXY protocol informs
9204 the other end about the layer 3/4 addresses of the incoming connection, so
9205 that it can know the client's address or the public address it accessed to,
9206 whatever the upper layer protocol. This setting must not be used if the
9207 server isn't aware of this version of the protocol. See also the "send-proxy"
9208 option of the "bind" keyword.
9209
9210 Supported in default-server: No
9211
9212send-proxy-v2-ssl
9213 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9214 2 over any connection established to this server. The PROXY protocol informs
9215 the other end about the layer 3/4 addresses of the incoming connection, so
9216 that it can know the client's address or the public address it accessed to,
9217 whatever the upper layer protocol. In addition, the SSL information extension
9218 of the PROXY protocol is added to the PROXY protocol header. This setting
9219 must not be used if the server isn't aware of this version of the protocol.
9220 See also the "send-proxy-v2" option of the "bind" keyword.
9221
9222 Supported in default-server: No
9223
9224send-proxy-v2-ssl-cn
9225 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9226 2 over any connection established to this server. The PROXY protocol informs
9227 the other end about the layer 3/4 addresses of the incoming connection, so
9228 that it can know the client's address or the public address it accessed to,
9229 whatever the upper layer protocol. In addition, the SSL information extension
9230 of the PROXY protocol, along along with the Common Name from the subject of
9231 the client certificate (if any), is added to the PROXY protocol header. This
9232 setting must not be used if the server isn't aware of this version of the
9233 protocol. See also the "send-proxy-v2" option of the "bind" keyword.
9234
9235 Supported in default-server: No
9236
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009237slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009238 The "slowstart" parameter for a server accepts a value in milliseconds which
9239 indicates after how long a server which has just come back up will run at
9240 full speed. Just as with every other time-based parameter, it can be entered
9241 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
9242 linearly from 0 to 100% during this time. The limitation applies to two
9243 parameters :
9244
9245 - maxconn: the number of connections accepted by the server will grow from 1
9246 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
9247
9248 - weight: when the backend uses a dynamic weighted algorithm, the weight
9249 grows linearly from 1 to 100%. In this case, the weight is updated at every
9250 health-check. For this reason, it is important that the "inter" parameter
9251 is smaller than the "slowstart", in order to maximize the number of steps.
9252
9253 The slowstart never applies when haproxy starts, otherwise it would cause
9254 trouble to running servers. It only applies when a server has been previously
9255 seen as failed.
9256
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009257 Supported in default-server: Yes
9258
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009259source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009260source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009261source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009262 The "source" parameter sets the source address which will be used when
9263 connecting to the server. It follows the exact same parameters and principle
9264 as the backend "source" keyword, except that it only applies to the server
9265 referencing it. Please consult the "source" keyword for details.
9266
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009267 Additionally, the "source" statement on a server line allows one to specify a
9268 source port range by indicating the lower and higher bounds delimited by a
9269 dash ('-'). Some operating systems might require a valid IP address when a
9270 source port range is specified. It is permitted to have the same IP/range for
9271 several servers. Doing so makes it possible to bypass the maximum of 64k
9272 total concurrent connections. The limit will then reach 64k connections per
9273 server.
9274
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009275 Supported in default-server: No
9276
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009277ssl
Willy Tarreau44f65392013-06-25 07:56:20 +02009278 This option enables SSL ciphering on outgoing connections to the server. It
9279 is critical to verify server certificates using "verify" when using SSL to
9280 connect to servers, otherwise the communication is prone to trivial man in
9281 the-middle attacks rendering SSL useless. When this option is used, health
9282 checks are automatically sent in SSL too unless there is a "port" or an
9283 "addr" directive indicating the check should be sent to a different location.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009284 See the "check-ssl" option to force SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +02009285
9286 Supported in default-server: No
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009287
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009288track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +02009289 This option enables ability to set the current state of the server by tracking
9290 another one. It is possible to track a server which itself tracks another
9291 server, provided that at the end of the chain, a server has health checks
9292 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009293 used, it has to be enabled on both proxies.
9294
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009295 Supported in default-server: No
9296
Emeric Brunef42d922012-10-11 16:11:36 +02009297verify [none|required]
9298 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +01009299 to 'none', server certificate is not verified. In the other case, The
9300 certificate provided by the server is verified using CAs from 'ca-file'
9301 and optional CRLs from 'crl-file'. If 'ssl_server_verify' is not specified
9302 in global section, this is the default. On verify failure the handshake
Willy Tarreau44f65392013-06-25 07:56:20 +02009303 is aborted. It is critically important to verify server certificates when
9304 using SSL to connect to servers, otherwise the communication is prone to
9305 trivial man-in-the-middle attacks rendering SSL totally useless.
Emeric Brunef42d922012-10-11 16:11:36 +02009306
9307 Supported in default-server: No
9308
Evan Broderbe554312013-06-27 00:05:25 -07009309verifyhost <hostname>
9310 This setting is only available when support for OpenSSL was built in, and
9311 only takes effect if 'verify required' is also specified. When set, the
9312 hostnames in the subject and subjectAlternateNames of the certificate
9313 provided by the server are checked. If none of the hostnames in the
9314 certificate match the specified hostname, the handshake is aborted. The
9315 hostnames in the server-provided certificate may include wildcards.
9316
9317 Supported in default-server: No
9318
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009319weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009320 The "weight" parameter is used to adjust the server's weight relative to
9321 other servers. All servers will receive a load proportional to their weight
9322 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +02009323 load. The default weight is 1, and the maximal value is 256. A value of 0
9324 means the server will not participate in load-balancing but will still accept
9325 persistent connections. If this parameter is used to distribute the load
9326 according to server's capacity, it is recommended to start with values which
9327 can both grow and shrink, for instance between 10 and 100 to leave enough
9328 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009329
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009330 Supported in default-server: Yes
9331
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009332
93336. HTTP header manipulation
9334---------------------------
9335
9336In HTTP mode, it is possible to rewrite, add or delete some of the request and
9337response headers based on regular expressions. It is also possible to block a
9338request or a response if a particular header matches a regular expression,
9339which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +01009340against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009341
Willy Tarreau70dffda2014-01-30 03:07:23 +01009342If HAProxy encounters an "Informational Response" (status code 1xx), it is able
9343to process all rsp* rules which can allow, deny, rewrite or delete a header,
9344but it will refuse to add a header to any such messages as this is not
9345HTTP-compliant. The reason for still processing headers in such responses is to
9346stop and/or fix any possible information leak which may happen, for instance
9347because another downstream equipment would unconditionally add a header, or if
9348a server name appears there. When such messages are seen, normal processing
9349still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +02009350
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009351This section covers common usage of the following keywords, described in detail
9352in section 4.2 :
9353
9354 - reqadd <string>
9355 - reqallow <search>
9356 - reqiallow <search>
9357 - reqdel <search>
9358 - reqidel <search>
9359 - reqdeny <search>
9360 - reqideny <search>
9361 - reqpass <search>
9362 - reqipass <search>
9363 - reqrep <search> <replace>
9364 - reqirep <search> <replace>
9365 - reqtarpit <search>
9366 - reqitarpit <search>
9367 - rspadd <string>
9368 - rspdel <search>
9369 - rspidel <search>
9370 - rspdeny <search>
9371 - rspideny <search>
9372 - rsprep <search> <replace>
9373 - rspirep <search> <replace>
9374
9375With all these keywords, the same conventions are used. The <search> parameter
9376is a POSIX extended regular expression (regex) which supports grouping through
9377parenthesis (without the backslash). Spaces and other delimiters must be
9378prefixed with a backslash ('\') to avoid confusion with a field delimiter.
9379Other characters may be prefixed with a backslash to change their meaning :
9380
9381 \t for a tab
9382 \r for a carriage return (CR)
9383 \n for a new line (LF)
9384 \ to mark a space and differentiate it from a delimiter
9385 \# to mark a sharp and differentiate it from a comment
9386 \\ to use a backslash in a regex
9387 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
9388 \xXX to write the ASCII hex code XX as in the C language
9389
9390The <replace> parameter contains the string to be used to replace the largest
9391portion of text matching the regex. It can make use of the special characters
9392above, and can reference a substring which is delimited by parenthesis in the
9393regex, by writing a backslash ('\') immediately followed by one digit from 0 to
93949 indicating the group position (0 designating the entire line). This practice
9395is very common to users of the "sed" program.
9396
9397The <string> parameter represents the string which will systematically be added
9398after the last header line. It can also use special character sequences above.
9399
9400Notes related to these keywords :
9401---------------------------------
9402 - these keywords are not always convenient to allow/deny based on header
9403 contents. It is strongly recommended to use ACLs with the "block" keyword
9404 instead, resulting in far more flexible and manageable rules.
9405
9406 - lines are always considered as a whole. It is not possible to reference
9407 a header name only or a value only. This is important because of the way
9408 headers are written (notably the number of spaces after the colon).
9409
9410 - the first line is always considered as a header, which makes it possible to
9411 rewrite or filter HTTP requests URIs or response codes, but in turn makes
9412 it harder to distinguish between headers and request line. The regex prefix
9413 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
9414 ^[^ \t:]*: matches any header name followed by a colon.
9415
9416 - for performances reasons, the number of characters added to a request or to
9417 a response is limited at build time to values between 1 and 4 kB. This
9418 should normally be far more than enough for most usages. If it is too short
9419 on occasional usages, it is possible to gain some space by removing some
9420 useless headers before adding new ones.
9421
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009422 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009423 without the 'i' letter except that they ignore case when matching patterns.
9424
9425 - when a request passes through a frontend then a backend, all req* rules
9426 from the frontend will be evaluated, then all req* rules from the backend
9427 will be evaluated. The reverse path is applied to responses.
9428
9429 - req* statements are applied after "block" statements, so that "block" is
9430 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +01009431 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009432
9433
Willy Tarreau74ca5042013-06-11 23:12:07 +020094347. Using ACLs and fetching samples
9435----------------------------------
9436
9437Haproxy is capable of extracting data from request or response streams, from
9438client or server information, from tables, environmental information etc...
9439The action of extracting such data is called fetching a sample. Once retrieved,
9440these samples may be used for various purposes such as a key to a stick-table,
9441but most common usages consist in matching them against predefined constant
9442data called patterns.
9443
9444
94457.1. ACL basics
9446---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009447
9448The use of Access Control Lists (ACL) provides a flexible solution to perform
9449content switching and generally to take decisions based on content extracted
9450from the request, the response or any environmental status. The principle is
9451simple :
9452
Willy Tarreau74ca5042013-06-11 23:12:07 +02009453 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009454 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +02009455 - apply one or multiple pattern matching methods on this sample
9456 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009457
Willy Tarreau74ca5042013-06-11 23:12:07 +02009458The actions generally consist in blocking a request, selecting a backend, or
9459adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009460
9461In order to define a test, the "acl" keyword is used. The syntax is :
9462
Willy Tarreau74ca5042013-06-11 23:12:07 +02009463 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009464
9465This creates a new ACL <aclname> or completes an existing one with new tests.
9466Those tests apply to the portion of request/response specified in <criterion>
9467and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009468an operator which may be specified before the set of values. Optionally some
9469conversion operators may be applied to the sample, and they will be specified
9470as a comma-delimited list of keywords just after the first keyword. The values
9471are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009472
9473ACL names must be formed from upper and lower case letters, digits, '-' (dash),
9474'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
9475which means that "my_acl" and "My_Acl" are two different ACLs.
9476
9477There is no enforced limit to the number of ACLs. The unused ones do not affect
9478performance, they just consume a small amount of memory.
9479
Willy Tarreau74ca5042013-06-11 23:12:07 +02009480The criterion generally is the name of a sample fetch method, or one of its ACL
9481specific declinations. The default test method is implied by the output type of
9482this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009483methods of a same sample fetch method. The sample fetch methods are the only
9484ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009485
9486Sample fetch methods return data which can be of the following types :
9487 - boolean
9488 - integer (signed or unsigned)
9489 - IPv4 or IPv6 address
9490 - string
9491 - data block
9492
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009493Converters transform any of these data into any of these. For example, some
9494converters might convert a string to a lower-case string while other ones
9495would turn a string to an IPv4 address, or apply a netmask to an IP address.
9496The resulting sample is of the type of the last converter applied to the list,
9497which defaults to the type of the sample fetch method.
9498
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009499Each sample or converter returns data of a specific type, specified with its
9500keyword in this documentation. When an ACL is declared using a standard sample
9501fetch method, certain types automatically involved a default matching method
9502which are summarized in the table below :
9503
9504 +---------------------+-----------------+
9505 | Sample or converter | Default |
9506 | output type | matching method |
9507 +---------------------+-----------------+
9508 | boolean | bool |
9509 +---------------------+-----------------+
9510 | integer | int |
9511 +---------------------+-----------------+
9512 | ip | ip |
9513 +---------------------+-----------------+
9514 | string | str |
9515 +---------------------+-----------------+
9516 | binary | none, use "-m" |
9517 +---------------------+-----------------+
9518
9519Note that in order to match a binary samples, it is mandatory to specify a
9520matching method, see below.
9521
Willy Tarreau74ca5042013-06-11 23:12:07 +02009522The ACL engine can match these types against patterns of the following types :
9523 - boolean
9524 - integer or integer range
9525 - IP address / network
9526 - string (exact, substring, suffix, prefix, subdir, domain)
9527 - regular expression
9528 - hex block
9529
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009530The following ACL flags are currently supported :
9531
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009532 -i : ignore case during matching of all subsequent patterns.
9533 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009534 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009535 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009536 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009537 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +02009538 -- : force end of flags. Useful when a string looks like one of the flags.
9539
Willy Tarreau74ca5042013-06-11 23:12:07 +02009540The "-f" flag is followed by the name of a file from which all lines will be
9541read as individual values. It is even possible to pass multiple "-f" arguments
9542if the patterns are to be loaded from multiple files. Empty lines as well as
9543lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
9544will be stripped. If it is absolutely necessary to insert a valid pattern
9545beginning with a sharp, just prefix it with a space so that it is not taken for
9546a comment. Depending on the data type and match method, haproxy may load the
9547lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
9548exact string matching. In this case, duplicates will automatically be removed.
9549
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009550The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
9551parsed as two column file. The first column contains the patterns used by the
9552ACL, and the second column contain the samples. The sample can be used later by
9553a map. This can be useful in some rare cases where an ACL would just be used to
9554check for the existence of a pattern in a map before a mapping is applied.
9555
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009556The "-u" flag forces the unique id of the ACL. This unique id is used with the
9557socket interface to identify ACL and dynamically change its values. Note that a
9558file is always identified by its name even if an id is set.
9559
Willy Tarreau74ca5042013-06-11 23:12:07 +02009560Also, note that the "-i" flag applies to subsequent entries and not to entries
9561loaded from files preceding it. For instance :
9562
9563 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
9564
9565In this example, each line of "exact-ua.lst" will be exactly matched against
9566the "user-agent" header of the request. Then each line of "generic-ua" will be
9567case-insensitively matched. Then the word "test" will be insensitively matched
9568as well.
9569
9570The "-m" flag is used to select a specific pattern matching method on the input
9571sample. All ACL-specific criteria imply a pattern matching method and generally
9572do not need this flag. However, this flag is useful with generic sample fetch
9573methods to describe how they're going to be matched against the patterns. This
9574is required for sample fetches which return data type for which there is no
9575obvious matching method (eg: string or binary). When "-m" is specified and
9576followed by a pattern matching method name, this method is used instead of the
9577default one for the criterion. This makes it possible to match contents in ways
9578that were not initially planned, or with sample fetch methods which return a
9579string. The matching method also affects the way the patterns are parsed.
9580
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009581The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
9582By default, if the parser cannot parse ip address it considers that the parsed
9583string is maybe a domain name and try dns resolution. The flag "-n" disable this
9584resolution. It is useful for detecting malformed ip lists. Note that if the DNS
9585server is not reachable, the haproxy configuration parsing may last many minutes
9586waiting fir the timeout. During this time no error messages are displayed. The
9587flag "-n" disable this behavior. Note also that during the runtime, this
9588function is disabled for the dynamic acl modifications.
9589
Willy Tarreau74ca5042013-06-11 23:12:07 +02009590There are some restrictions however. Not all methods can be used with all
9591sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
9592be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +02009593
9594 - "found" : only check if the requested sample could be found in the stream,
9595 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +02009596 to pass any pattern to avoid confusion. This matching method is
9597 particularly useful to detect presence of certain contents such
9598 as headers, cookies, etc... even if they are empty and without
9599 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009600
9601 - "bool" : check the value as a boolean. It can only be applied to fetches
9602 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009603 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009604
9605 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +02009606 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009607
9608 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +02009609 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009610
9611 - "bin" : match the contents against an hexadecimal string representing a
9612 binary sequence. This may be used with binary or string samples.
9613
9614 - "len" : match the sample's length as an integer. This may be used with
9615 binary or string samples.
9616
Willy Tarreau74ca5042013-06-11 23:12:07 +02009617 - "str" : exact match : match the contents against a string. This may be
9618 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009619
Willy Tarreau74ca5042013-06-11 23:12:07 +02009620 - "sub" : substring match : check that the contents contain at least one of
9621 the provided string patterns. This may be used with binary or
9622 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009623
Willy Tarreau74ca5042013-06-11 23:12:07 +02009624 - "reg" : regex match : match the contents against a list of regular
9625 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009626
Willy Tarreau74ca5042013-06-11 23:12:07 +02009627 - "beg" : prefix match : check that the contents begin like the provided
9628 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009629
Willy Tarreau74ca5042013-06-11 23:12:07 +02009630 - "end" : suffix match : check that the contents end like the provided
9631 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009632
Willy Tarreau74ca5042013-06-11 23:12:07 +02009633 - "dir" : subdir match : check that a slash-delimited portion of the
9634 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009635 This may be used with binary or string samples.
9636
Willy Tarreau74ca5042013-06-11 23:12:07 +02009637 - "dom" : domain match : check that a dot-delimited portion of the contents
9638 exactly match one of the provided string patterns. This may be
9639 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009640
9641For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
9642request, it is possible to do :
9643
9644 acl jsess_present cook(JSESSIONID) -m found
9645
9646In order to apply a regular expression on the 500 first bytes of data in the
9647buffer, one would use the following acl :
9648
9649 acl script_tag payload(0,500) -m reg -i <script>
9650
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009651On systems where the regex library is much slower when using "-i", it is
9652possible to convert the sample to lowercase before matching, like this :
9653
9654 acl script_tag payload(0,500),lower -m reg <script>
9655
Willy Tarreau74ca5042013-06-11 23:12:07 +02009656All ACL-specific criteria imply a default matching method. Most often, these
9657criteria are composed by concatenating the name of the original sample fetch
9658method and the matching method. For example, "hdr_beg" applies the "beg" match
9659to samples retrieved using the "hdr" fetch method. Since all ACL-specific
9660criteria rely on a sample fetch method, it is always possible instead to use
9661the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009662
Willy Tarreau74ca5042013-06-11 23:12:07 +02009663If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009664the matching method is simply applied to the underlying sample fetch method.
9665For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009666
Willy Tarreau74ca5042013-06-11 23:12:07 +02009667 acl short_form hdr_beg(host) www.
9668 acl alternate1 hdr_beg(host) -m beg www.
9669 acl alternate2 hdr_dom(host) -m beg www.
9670 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009671
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009672
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009673The table below summarizes the compatibility matrix between sample or converter
9674types and the pattern types to fetch against. It indicates for each compatible
9675combination the name of the matching method to be used, surrounded with angle
9676brackets ">" and "<" when the method is the default one and will work by
9677default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009678
Willy Tarreau74ca5042013-06-11 23:12:07 +02009679 +-------------------------------------------------+
9680 | Input sample type |
9681 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009682 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009683 +----------------------+---------+---------+---------+---------+---------+
9684 | none (presence only) | found | found | found | found | found |
9685 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009686 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009687 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009688 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009689 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009690 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009691 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009692 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009693 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009694 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009695 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009696 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009697 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009698 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009699 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009700 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009701 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009702 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009703 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009704 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009705 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009706 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009707 +----------------------+---------+---------+---------+---------+---------+
9708 | hex block | | | | bin | bin |
9709 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +02009710
9711
Willy Tarreau74ca5042013-06-11 23:12:07 +020097127.1.1. Matching booleans
9713------------------------
9714
9715In order to match a boolean, no value is needed and all values are ignored.
9716Boolean matching is used by default for all fetch methods of type "boolean".
9717When boolean matching is used, the fetched value is returned as-is, which means
9718that a boolean "true" will always match and a boolean "false" will never match.
9719
9720Boolean matching may also be enforced using "-m bool" on fetch methods which
9721return an integer value. Then, integer value 0 is converted to the boolean
9722"false" and all other values are converted to "true".
9723
Willy Tarreau6a06a402007-07-15 20:15:28 +02009724
Willy Tarreau74ca5042013-06-11 23:12:07 +020097257.1.2. Matching integers
9726------------------------
9727
9728Integer matching applies by default to integer fetch methods. It can also be
9729enforced on boolean fetches using "-m int". In this case, "false" is converted
9730to the integer 0, and "true" is converted to the integer 1.
9731
9732Integer matching also supports integer ranges and operators. Note that integer
9733matching only applies to positive values. A range is a value expressed with a
9734lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009735
9736For instance, "1024:65535" is a valid range to represent a range of
9737unprivileged ports, and "1024:" would also work. "0:1023" is a valid
9738representation of privileged ports, and ":1023" would also work.
9739
Willy Tarreau62644772008-07-16 18:36:06 +02009740As a special case, some ACL functions support decimal numbers which are in fact
9741two integers separated by a dot. This is used with some version checks for
9742instance. All integer properties apply to those decimal numbers, including
9743ranges and operators.
9744
Willy Tarreau6a06a402007-07-15 20:15:28 +02009745For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +01009746operators with ranges does not make much sense and is strongly discouraged.
9747Similarly, it does not make much sense to perform order comparisons with a set
9748of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009749
Willy Tarreau0ba27502007-12-24 16:55:16 +01009750Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009751
9752 eq : true if the tested value equals at least one value
9753 ge : true if the tested value is greater than or equal to at least one value
9754 gt : true if the tested value is greater than at least one value
9755 le : true if the tested value is less than or equal to at least one value
9756 lt : true if the tested value is less than at least one value
9757
Willy Tarreau0ba27502007-12-24 16:55:16 +01009758For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009759
9760 acl negative-length hdr_val(content-length) lt 0
9761
Willy Tarreau62644772008-07-16 18:36:06 +02009762This one matches SSL versions between 3.0 and 3.1 (inclusive) :
9763
9764 acl sslv3 req_ssl_ver 3:3.1
9765
Willy Tarreau6a06a402007-07-15 20:15:28 +02009766
Willy Tarreau74ca5042013-06-11 23:12:07 +020097677.1.3. Matching strings
9768-----------------------
9769
9770String matching applies to string or binary fetch methods, and exists in 6
9771different forms :
9772
9773 - exact match (-m str) : the extracted string must exactly match the
9774 patterns ;
9775
9776 - substring match (-m sub) : the patterns are looked up inside the
9777 extracted string, and the ACL matches if any of them is found inside ;
9778
9779 - prefix match (-m beg) : the patterns are compared with the beginning of
9780 the extracted string, and the ACL matches if any of them matches.
9781
9782 - suffix match (-m end) : the patterns are compared with the end of the
9783 extracted string, and the ACL matches if any of them matches.
9784
9785 - subdir match (-m sub) : the patterns are looked up inside the extracted
9786 string, delimited with slashes ("/"), and the ACL matches if any of them
9787 matches.
9788
9789 - domain match (-m dom) : the patterns are looked up inside the extracted
9790 string, delimited with dots ("."), and the ACL matches if any of them
9791 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009792
9793String matching applies to verbatim strings as they are passed, with the
9794exception of the backslash ("\") which makes it possible to escape some
9795characters such as the space. If the "-i" flag is passed before the first
9796string, then the matching will be performed ignoring the case. In order
9797to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +01009798before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009799
9800
Willy Tarreau74ca5042013-06-11 23:12:07 +020098017.1.4. Matching regular expressions (regexes)
9802---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009803
9804Just like with string matching, regex matching applies to verbatim strings as
9805they are passed, with the exception of the backslash ("\") which makes it
9806possible to escape some characters such as the space. If the "-i" flag is
9807passed before the first regex, then the matching will be performed ignoring
9808the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +01009809the "--" flag before the first string. Same principle applies of course to
9810match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009811
9812
Willy Tarreau74ca5042013-06-11 23:12:07 +020098137.1.5. Matching arbitrary data blocks
9814-------------------------------------
9815
9816It is possible to match some extracted samples against a binary block which may
9817not safely be represented as a string. For this, the patterns must be passed as
9818a series of hexadecimal digits in an even number, when the match method is set
9819to binary. Each sequence of two digits will represent a byte. The hexadecimal
9820digits may be used upper or lower case.
9821
9822Example :
9823 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
9824 acl hello payload(0,6) -m bin 48656c6c6f0a
9825
9826
98277.1.6. Matching IPv4 and IPv6 addresses
9828---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009829
9830IPv4 addresses values can be specified either as plain addresses or with a
9831netmask appended, in which case the IPv4 address matches whenever it is
9832within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009833host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +01009834difficult to read and debug configurations. If hostnames are used, you should
9835at least ensure that they are present in /etc/hosts so that the configuration
9836does not depend on any random DNS match at the moment the configuration is
9837parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009838
Willy Tarreauceb4ac92012-04-28 00:41:46 +02009839IPv6 may be entered in their usual form, with or without a netmask appended.
9840Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
9841trouble with randomly resolved IP addresses, host names are never allowed in
9842IPv6 patterns.
9843
9844HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
9845following situations :
9846 - tested address is IPv4, pattern address is IPv4, the match applies
9847 in IPv4 using the supplied mask if any.
9848 - tested address is IPv6, pattern address is IPv6, the match applies
9849 in IPv6 using the supplied mask if any.
9850 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
9851 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
9852 ::IPV4 or ::ffff:IPV4, otherwise it fails.
9853 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
9854 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
9855 applied in IPv6 using the supplied IPv6 mask.
9856
Willy Tarreau74ca5042013-06-11 23:12:07 +02009857
98587.2. Using ACLs to form conditions
9859----------------------------------
9860
9861Some actions are only performed upon a valid condition. A condition is a
9862combination of ACLs with operators. 3 operators are supported :
9863
9864 - AND (implicit)
9865 - OR (explicit with the "or" keyword or the "||" operator)
9866 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +02009867
Willy Tarreau74ca5042013-06-11 23:12:07 +02009868A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +02009869
Willy Tarreau74ca5042013-06-11 23:12:07 +02009870 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +02009871
Willy Tarreau74ca5042013-06-11 23:12:07 +02009872Such conditions are generally used after an "if" or "unless" statement,
9873indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +02009874
Willy Tarreau74ca5042013-06-11 23:12:07 +02009875For instance, to block HTTP requests to the "*" URL with methods other than
9876"OPTIONS", as well as POST requests without content-length, and GET or HEAD
9877requests with a content-length greater than 0, and finally every request which
9878is not either GET/HEAD/POST/OPTIONS !
9879
9880 acl missing_cl hdr_cnt(Content-length) eq 0
9881 block if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
9882 block if METH_GET HTTP_CONTENT
9883 block unless METH_GET or METH_POST or METH_OPTIONS
9884
9885To select a different backend for requests to static contents on the "www" site
9886and to every request on the "img", "video", "download" and "ftp" hosts :
9887
9888 acl url_static path_beg /static /images /img /css
9889 acl url_static path_end .gif .png .jpg .css .js
9890 acl host_www hdr_beg(host) -i www
9891 acl host_static hdr_beg(host) -i img. video. download. ftp.
9892
9893 # now use backend "static" for all static-only hosts, and for static urls
9894 # of host "www". Use backend "www" for the rest.
9895 use_backend static if host_static or host_www url_static
9896 use_backend www if host_www
9897
9898It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
9899expressions that are built on the fly without needing to be declared. They must
9900be enclosed between braces, with a space before and after each brace (because
9901the braces must be seen as independent words). Example :
9902
9903 The following rule :
9904
9905 acl missing_cl hdr_cnt(Content-length) eq 0
9906 block if METH_POST missing_cl
9907
9908 Can also be written that way :
9909
9910 block if METH_POST { hdr_cnt(Content-length) eq 0 }
9911
9912It is generally not recommended to use this construct because it's a lot easier
9913to leave errors in the configuration when written that way. However, for very
9914simple rules matching only one source IP address for instance, it can make more
9915sense to use them than to declare ACLs with random names. Another example of
9916good use is the following :
9917
9918 With named ACLs :
9919
9920 acl site_dead nbsrv(dynamic) lt 2
9921 acl site_dead nbsrv(static) lt 2
9922 monitor fail if site_dead
9923
9924 With anonymous ACLs :
9925
9926 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
9927
9928See section 4.2 for detailed help on the "block" and "use_backend" keywords.
9929
9930
99317.3. Fetching samples
9932---------------------
9933
9934Historically, sample fetch methods were only used to retrieve data to match
9935against patterns using ACLs. With the arrival of stick-tables, a new class of
9936sample fetch methods was created, most often sharing the same syntax as their
9937ACL counterpart. These sample fetch methods are also known as "fetches". As
9938of now, ACLs and fetches have converged. All ACL fetch methods have been made
9939available as fetch methods, and ACLs may use any sample fetch method as well.
9940
9941This section details all available sample fetch methods and their output type.
9942Some sample fetch methods have deprecated aliases that are used to maintain
9943compatibility with existing configurations. They are then explicitly marked as
9944deprecated and should not be used in new setups.
9945
9946The ACL derivatives are also indicated when available, with their respective
9947matching methods. These ones all have a well defined default pattern matching
9948method, so it is never necessary (though allowed) to pass the "-m" option to
9949indicate how the sample will be matched using ACLs.
9950
9951As indicated in the sample type versus matching compatibility matrix above,
9952when using a generic sample fetch method in an ACL, the "-m" option is
9953mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
9954the same keyword exists as an ACL keyword and as a standard fetch method, the
9955ACL engine will automatically pick the ACL-only one by default.
9956
9957Some of these keywords support one or multiple mandatory arguments, and one or
9958multiple optional arguments. These arguments are strongly typed and are checked
9959when the configuration is parsed so that there is no risk of running with an
9960incorrect argument (eg: an unresolved backend name). Fetch function arguments
9961are passed between parenthesis and are delimited by commas. When an argument
9962is optional, it will be indicated below between square brackets ('[ ]'). When
9963all arguments are optional, the parenthesis may be omitted.
9964
9965Thus, the syntax of a standard sample fetch method is one of the following :
9966 - name
9967 - name(arg1)
9968 - name(arg1,arg2)
9969
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009970
99717.3.1. Converters
9972-----------------
9973
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009974Sample fetch methods may be combined with transformations to be applied on top
9975of the fetched sample (also called "converters"). These combinations form what
9976is called "sample expressions" and the result is a "sample". Initially this
9977was only supported by "stick on" and "stick store-request" directives but this
9978has now be extended to all places where samples may be used (acls, log-format,
9979unique-id-format, add-header, ...).
9980
9981These transformations are enumerated as a series of specific keywords after the
9982sample fetch method. These keywords may equally be appended immediately after
9983the fetch keyword's argument, delimited by a comma. These keywords can also
9984support some arguments (eg: a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009985
Willy Tarreau74ca5042013-06-11 23:12:07 +02009986The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +01009987
Emeric Brun53d1a982014-04-30 18:21:37 +02009988base64
9989 Converts a binary input sample to a base64 string. It is used to log or
9990 transfer binary content in a way that can be reliably transferred (eg:
9991 an SSL ID can be copied in a header).
9992
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009993hex
9994 Converts a binary input sample to an hex string containing two hex digits per
9995 input byte. It is used to log or transfer hex dumps of some binary input data
9996 in a way that can be reliably transferred (eg: an SSL ID can be copied in a
9997 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +01009998
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009999http_date([<offset>])
10000 Converts an integer supposed to contain a date since epoch to a string
10001 representing this date in a format suitable for use in HTTP header fields. If
10002 an offset value is specified, then it is a number of seconds that is added to
10003 the date before the conversion is operated. This is particularly useful to
10004 emit Date header fields, Expires values in responses when combined with a
10005 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010006
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010007ipmask(<mask>)
10008 Apply a mask to an IPv4 address, and use the result for lookups and storage.
10009 This can be used to make all hosts within a certain mask to share the same
10010 table entries and as such use the same server. The mask can be passed in
10011 dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24).
10012
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010013language(<value>[,<default>])
10014 Returns the value with the highest q-factor from a list as extracted from the
10015 "accept-language" header using "req.fhdr". Values with no q-factor have a
10016 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
10017 belong to the list of semi-colon delimited <values> will be considered. The
10018 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
10019 given list and a default value is provided, it is returned. Note that language
10020 names may have a variant after a dash ('-'). If this variant is present in the
10021 list, it will be matched, but if it is not, only the base language is checked.
10022 The match is case-sensitive, and the output string is always one of those
10023 provided in arguments. The ordering of arguments is meaningless, only the
10024 ordering of the values in the request counts, as the first value among
10025 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020010026
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010027 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010028
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010029 # this configuration switches to the backend matching a
10030 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010031
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010032 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
10033 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
10034 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
10035 use_backend spanish if es
10036 use_backend french if fr
10037 use_backend english if en
10038 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020010039
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010040lower
10041 Convert a string sample to lower case. This can only be placed after a string
10042 sample fetch function or after a transformation keyword returning a string
10043 type. The result is of type string.
10044
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010045map(<map_file>[,<default_value>])
10046map_<match_type>(<map_file>[,<default_value>])
10047map_<match_type>_<output_type>(<map_file>[,<default_value>])
10048 Search the input value from <map_file> using the <match_type> matching method,
10049 and return the associated value converted to the type <output_type>. If the
10050 input value cannot be found in the <map_file>, the converter returns the
10051 <default_value>. If the <default_value> is not set, the converter fails and
10052 acts as if no input value could be fetched. If the <match_type> is not set, it
10053 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
10054 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
10055 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010056
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010057 It is important to avoid overlapping between the keys : IP addresses and
10058 strings are stored in trees, so the first of the finest match will be used.
10059 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010060
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010061 The following array contains the list of all map functions avalaible sorted by
10062 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010063
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010064 input type | match method | output type str | output type int | output type ip
10065 -----------+--------------+-----------------+-----------------+---------------
10066 str | str | map_str | map_str_int | map_str_ip
10067 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020010068 str | beg | map_beg | map_beg_int | map_end_ip
10069 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010070 str | sub | map_sub | map_sub_int | map_sub_ip
10071 -----------+--------------+-----------------+-----------------+---------------
10072 str | dir | map_dir | map_dir_int | map_dir_ip
10073 -----------+--------------+-----------------+-----------------+---------------
10074 str | dom | map_dom | map_dom_int | map_dom_ip
10075 -----------+--------------+-----------------+-----------------+---------------
10076 str | end | map_end | map_end_int | map_end_ip
10077 -----------+--------------+-----------------+-----------------+---------------
10078 str | reg | map_reg | map_reg_int | map_reg_ip
10079 -----------+--------------+-----------------+-----------------+---------------
10080 int | int | map_int | map_int_int | map_int_ip
10081 -----------+--------------+-----------------+-----------------+---------------
10082 ip | ip | map_ip | map_ip_int | map_ip_ip
10083 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010084
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010085 The file contains one key + value per line. Lines which start with '#' are
10086 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
10087 is then the first "word" (series of non-space/tabs characters), and the value
10088 is what follows this series of space/tab till the end of the line excluding
10089 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010090
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010091 Example :
10092
10093 # this is a comment and is ignored
10094 2.22.246.0/23 United Kingdom \n
10095 <-><-----------><--><------------><---->
10096 | | | | `- trailing spaces ignored
10097 | | | `---------- value
10098 | | `-------------------- middle spaces ignored
10099 | `---------------------------- key
10100 `------------------------------------ leading spaces ignored
10101
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010102upper
10103 Convert a string sample to upper case. This can only be placed after a string
10104 sample fetch function or after a transformation keyword returning a string
10105 type. The result is of type string.
10106
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010107
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200101087.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020010109--------------------------------------------
10110
10111A first set of sample fetch methods applies to internal information which does
10112not even relate to any client information. These ones are sometimes used with
10113"monitor-fail" directives to report an internal status to external watchers.
10114The sample fetch methods described in this section are usable anywhere.
10115
10116always_false : boolean
10117 Always returns the boolean "false" value. It may be used with ACLs as a
10118 temporary replacement for another one when adjusting configurations.
10119
10120always_true : boolean
10121 Always returns the boolean "true" value. It may be used with ACLs as a
10122 temporary replacement for another one when adjusting configurations.
10123
10124avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010125 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020010126 divided by the number of active servers. The current backend is used if no
10127 backend is specified. This is very similar to "queue" except that the size of
10128 the farm is considered, in order to give a more accurate measurement of the
10129 time it may take for a new connection to be processed. The main usage is with
10130 ACL to return a sorry page to new users when it becomes certain they will get
10131 a degraded service, or to pass to the backend servers in a header so that
10132 they decide to work in degraded mode or to disable some functions to speed up
10133 the processing a bit. Note that in the event there would not be any active
10134 server anymore, twice the number of queued connections would be considered as
10135 the measured value. This is a fair estimate, as we expect one server to get
10136 back soon anyway, but we still prefer to send new traffic to another backend
10137 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
10138 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010010139
Willy Tarreau74ca5042013-06-11 23:12:07 +020010140be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020010141 Applies to the number of currently established connections on the backend,
10142 possibly including the connection being evaluated. If no backend name is
10143 specified, the current one is used. But it is also possible to check another
10144 backend. It can be used to use a specific farm when the nominal one is full.
10145 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020010146
Willy Tarreau74ca5042013-06-11 23:12:07 +020010147be_sess_rate([<backend>]) : integer
10148 Returns an integer value corresponding to the sessions creation rate on the
10149 backend, in number of new sessions per second. This is used with ACLs to
10150 switch to an alternate backend when an expensive or fragile one reaches too
10151 high a session rate, or to limit abuse of service (eg. prevent sucking of an
10152 online dictionary). It can also be useful to add this element to logs using a
10153 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010154
10155 Example :
10156 # Redirect to an error page if the dictionary is requested too often
10157 backend dynamic
10158 mode http
10159 acl being_scanned be_sess_rate gt 100
10160 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010010161
Willy Tarreau74ca5042013-06-11 23:12:07 +020010162connslots([<backend>]) : integer
10163 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010164 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020010165 connections on all servers and the maximum queue size. This is probably only
10166 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050010167
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010168 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020010169 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010170 usage; see "use_backend" keyword) can be redirected to a different backend.
10171
Willy Tarreau55165fe2009-05-10 12:02:55 +020010172 'connslots' = number of available server connection slots, + number of
10173 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010174
Willy Tarreaua36af912009-10-10 12:02:45 +020010175 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020010176 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020010177 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020010178 you want to be able to differentiate between different backends, and their
10179 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020010180 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020010181 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010182
Willy Tarreau55165fe2009-05-10 12:02:55 +020010183 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
10184 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020010185 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020010186 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010187
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010188date([<offset>]) : integer
10189 Returns the current date as the epoch (number of seconds since 01/01/1970).
10190 If an offset value is specified, then it is a number of seconds that is added
10191 to the current date before returning the value. This is particularly useful
10192 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020010193 It is useful combined with the http_date converter.
10194
10195 Example :
10196
10197 # set an expires header to now+1 hour in every response
10198 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010199
Willy Tarreau595ec542013-06-12 21:34:28 +020010200env(<name>) : string
10201 Returns a string containing the value of environment variable <name>. As a
10202 reminder, environment variables are per-process and are sampled when the
10203 process starts. This can be useful to pass some information to a next hop
10204 server, or with ACLs to take specific action when the process is started a
10205 certain way.
10206
10207 Examples :
10208 # Pass the Via header to next hop with the local hostname in it
10209 http-request add-header Via 1.1\ %[env(HOSTNAME)]
10210
10211 # reject cookie-less requests when the STOP environment variable is set
10212 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
10213
Willy Tarreau74ca5042013-06-11 23:12:07 +020010214fe_conn([<frontend>]) : integer
10215 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010010216 possibly including the connection being evaluated. If no frontend name is
10217 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010218 frontend. It can be used to return a sorry page before hard-blocking, or to
10219 use a specific backend to drain new requests when the farm is considered
10220 full. This is mostly used with ACLs but can also be used to pass some
10221 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
10222 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020010223
Willy Tarreau74ca5042013-06-11 23:12:07 +020010224fe_sess_rate([<frontend>]) : integer
10225 Returns an integer value corresponding to the sessions creation rate on the
10226 frontend, in number of new sessions per second. This is used with ACLs to
10227 limit the incoming session rate to an acceptable range in order to prevent
10228 abuse of service at the earliest moment, for example when combined with other
10229 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
10230 down below the limit. It can also be useful to add this element to logs using
10231 a log-format directive. See also the "rate-limit sessions" directive for use
10232 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010233
10234 Example :
10235 # This frontend limits incoming mails to 10/s with a max of 100
10236 # concurrent connections. We accept any connection below 10/s, and
10237 # force excess clients to wait for 100 ms. Since clients are limited to
10238 # 100 max, there cannot be more than 10 incoming mails per second.
10239 frontend mail
10240 bind :25
10241 mode tcp
10242 maxconn 100
10243 acl too_fast fe_sess_rate ge 10
10244 tcp-request inspect-delay 100ms
10245 tcp-request content accept if ! too_fast
10246 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010010247
Willy Tarreaufad7f072014-11-24 16:02:05 +010010248nbproc : integer
10249 Returns an integer value corresponding to the number of processes that were
10250 started (it equals the global "nbproc" setting). This is useful for logging
10251 and debugging purposes.
10252
Willy Tarreau74ca5042013-06-11 23:12:07 +020010253nbsrv([<backend>]) : integer
10254 Returns an integer value corresponding to the number of usable servers of
10255 either the current backend or the named backend. This is mostly used with
10256 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010010257 switch to an alternate backend when the number of servers is too low to
10258 to handle some load. It is useful to report a failure when combined with
10259 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010260
Willy Tarreaufad7f072014-11-24 16:02:05 +010010261proc : integer
10262 Returns an integer value corresponding to the position of the process calling
10263 the function, between 1 and global.nbproc. This is useful for logging and
10264 debugging purposes.
10265
Willy Tarreau74ca5042013-06-11 23:12:07 +020010266queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010267 Returns the total number of queued connections of the designated backend,
10268 including all the connections in server queues. If no backend name is
10269 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010270 one. This is useful with ACLs or to pass statistics to backend servers. This
10271 can be used to take actions when queuing goes above a known level, generally
10272 indicating a surge of traffic or a massive slowdown on the servers. One
10273 possible action could be to reject new users but still accept old ones. See
10274 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
10275
Willy Tarreau84310e22014-02-14 11:59:04 +010010276rand([<range>]) : integer
10277 Returns a random integer value within a range of <range> possible values,
10278 starting at zero. If the range is not specified, it defaults to 2^32, which
10279 gives numbers between 0 and 4294967295. It can be useful to pass some values
10280 needed to take some routing decisions for example, or just for debugging
10281 purposes. This random must not be used for security purposes.
10282
Willy Tarreau74ca5042013-06-11 23:12:07 +020010283srv_conn([<backend>/]<server>) : integer
10284 Returns an integer value corresponding to the number of currently established
10285 connections on the designated server, possibly including the connection being
10286 evaluated. If <backend> is omitted, then the server is looked up in the
10287 current backend. It can be used to use a specific farm when one server is
10288 full, or to inform the server about our view of the number of active
10289 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
10290 methods.
10291
10292srv_is_up([<backend>/]<server>) : boolean
10293 Returns true when the designated server is UP, and false when it is either
10294 DOWN or in maintenance mode. If <backend> is omitted, then the server is
10295 looked up in the current backend. It is mainly used to take action based on
10296 an external status reported via a health check (eg: a geographical site's
10297 availability). Another possible use which is more of a hack consists in
10298 using dummy servers as boolean variables that can be enabled or disabled from
10299 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
10300
10301srv_sess_rate([<backend>/]<server>) : integer
10302 Returns an integer corresponding to the sessions creation rate on the
10303 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010304 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020010305 used with ACLs but can make sense with logs too. This is used to switch to an
10306 alternate backend when an expensive or fragile one reaches too high a session
10307 rate, or to limit abuse of service (eg. prevent latent requests from
10308 overloading servers).
10309
10310 Example :
10311 # Redirect to a separate back
10312 acl srv1_full srv_sess_rate(be1/srv1) gt 50
10313 acl srv2_full srv_sess_rate(be1/srv2) gt 50
10314 use_backend be2 if srv1_full or srv2_full
10315
Willy Tarreaufad7f072014-11-24 16:02:05 +010010316stopping : boolean
10317 Returns TRUE if the process calling the function is currently stopping. This
10318 can be useful for logging, or for relaxing certain checks or helping close
10319 certain connections upon graceful shutdown.
10320
Willy Tarreau74ca5042013-06-11 23:12:07 +020010321table_avl([<table>]) : integer
10322 Returns the total number of available entries in the current proxy's
10323 stick-table or in the designated stick-table. See also table_cnt.
10324
10325table_cnt([<table>]) : integer
10326 Returns the total number of entries currently in use in the current proxy's
10327 stick-table or in the designated stick-table. See also src_conn_cnt and
10328 table_avl for other entry counting methods.
10329
10330
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200103317.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020010332----------------------------------
10333
10334The layer 4 usually describes just the transport layer which in haproxy is
10335closest to the connection, where no content is yet made available. The fetch
10336methods described here are usable as low as the "tcp-request connection" rule
10337sets unless they require some future information. Those generally include
10338TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010339the incoming connection. For retrieving a value from a sticky counters, the
10340counter number can be explicitly set as 0, 1, or 2 using the pre-defined
10341"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010342argument when using the "sc_" prefix. An optional table may be specified with
10343the "sc*" form, in which case the currently tracked key will be looked up into
10344this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010345
10346be_id : integer
10347 Returns an integer containing the current backend's id. It can be used in
10348 frontends with responses to check which backend processed the request.
10349
10350dst : ip
10351 This is the destination IPv4 address of the connection on the client side,
10352 which is the address the client connected to. It can be useful when running
10353 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
10354 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
10355 RFC 4291.
10356
10357dst_conn : integer
10358 Returns an integer value corresponding to the number of currently established
10359 connections on the same socket including the one being evaluated. It is
10360 normally used with ACLs but can as well be used to pass the information to
10361 servers in an HTTP header or in logs. It can be used to either return a sorry
10362 page before hard-blocking, or to use a specific backend to drain new requests
10363 when the socket is considered saturated. This offers the ability to assign
10364 different limits to different listening ports or addresses. See also the
10365 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010366
Willy Tarreau74ca5042013-06-11 23:12:07 +020010367dst_port : integer
10368 Returns an integer value corresponding to the destination TCP port of the
10369 connection on the client side, which is the port the client connected to.
10370 This might be used when running in transparent mode, when assigning dynamic
10371 ports to some clients for a whole application session, to stick all users to
10372 a same server, or to pass the destination port information to a server using
10373 an HTTP header.
10374
10375fe_id : integer
10376 Returns an integer containing the current frontend's id. It can be used in
10377 backends to check from which backend it was called, or to stick all users
10378 coming via a same frontend to the same server.
10379
Cyril Bonté62ba8702014-04-22 23:52:25 +020010380sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010381sc0_bytes_in_rate([<table>]) : integer
10382sc1_bytes_in_rate([<table>]) : integer
10383sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010384 Returns the average client-to-server bytes rate from the currently tracked
10385 counters, measured in amount of bytes over the period configured in the
10386 table. See also src_bytes_in_rate.
10387
Cyril Bonté62ba8702014-04-22 23:52:25 +020010388sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010389sc0_bytes_out_rate([<table>]) : integer
10390sc1_bytes_out_rate([<table>]) : integer
10391sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010392 Returns the average server-to-client bytes rate from the currently tracked
10393 counters, measured in amount of bytes over the period configured in the
10394 table. See also src_bytes_out_rate.
10395
Cyril Bonté62ba8702014-04-22 23:52:25 +020010396sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010397sc0_clr_gpc0([<table>]) : integer
10398sc1_clr_gpc0([<table>]) : integer
10399sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020010400 Clears the first General Purpose Counter associated to the currently tracked
10401 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010010402 stored value is zero, so first invocation will always return zero. This is
10403 typically used as a second ACL in an expression in order to mark a connection
10404 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010405
10406 # block if 5 consecutive requests continue to come faster than 10 sess
10407 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010408 acl abuse sc0_http_req_rate gt 10
10409 acl kill sc0_inc_gpc0 gt 5
10410 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010411 tcp-request connection accept if !abuse save
10412 tcp-request connection reject if abuse kill
10413
Cyril Bonté62ba8702014-04-22 23:52:25 +020010414sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010415sc0_conn_cnt([<table>]) : integer
10416sc1_conn_cnt([<table>]) : integer
10417sc2_conn_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010418 Returns the cumulated number of incoming connections from currently tracked
10419 counters. See also src_conn_cnt.
10420
Cyril Bonté62ba8702014-04-22 23:52:25 +020010421sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010422sc0_conn_cur([<table>]) : integer
10423sc1_conn_cur([<table>]) : integer
10424sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010425 Returns the current amount of concurrent connections tracking the same
10426 tracked counters. This number is automatically incremented when tracking
10427 begins and decremented when tracking stops. See also src_conn_cur.
10428
Cyril Bonté62ba8702014-04-22 23:52:25 +020010429sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010430sc0_conn_rate([<table>]) : integer
10431sc1_conn_rate([<table>]) : integer
10432sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010433 Returns the average connection rate from the currently tracked counters,
10434 measured in amount of connections over the period configured in the table.
10435 See also src_conn_rate.
10436
Cyril Bonté62ba8702014-04-22 23:52:25 +020010437sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010438sc0_get_gpc0([<table>]) : integer
10439sc1_get_gpc0([<table>]) : integer
10440sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010441 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010442 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010443
Cyril Bonté62ba8702014-04-22 23:52:25 +020010444sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010445sc0_gpc0_rate([<table>]) : integer
10446sc1_gpc0_rate([<table>]) : integer
10447sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010448 Returns the average increment rate of the first General Purpose Counter
10449 associated to the currently tracked counters. It reports the frequency
10450 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010451 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10452 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10453 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020010454
Cyril Bonté62ba8702014-04-22 23:52:25 +020010455sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010456sc0_http_err_cnt([<table>]) : integer
10457sc1_http_err_cnt([<table>]) : integer
10458sc2_http_err_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010459 Returns the cumulated number of HTTP errors from the currently tracked
10460 counters. This includes the both request errors and 4xx error responses.
10461 See also src_http_err_cnt.
10462
Cyril Bonté62ba8702014-04-22 23:52:25 +020010463sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010464sc0_http_err_rate([<table>]) : integer
10465sc1_http_err_rate([<table>]) : integer
10466sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010467 Returns the average rate of HTTP errors from the currently tracked counters,
10468 measured in amount of errors over the period configured in the table. This
10469 includes the both request errors and 4xx error responses. See also
10470 src_http_err_rate.
10471
Cyril Bonté62ba8702014-04-22 23:52:25 +020010472sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010473sc0_http_req_cnt([<table>]) : integer
10474sc1_http_req_cnt([<table>]) : integer
10475sc2_http_req_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010476 Returns the cumulated number of HTTP requests from the currently tracked
10477 counters. This includes every started request, valid or not. See also
10478 src_http_req_cnt.
10479
Cyril Bonté62ba8702014-04-22 23:52:25 +020010480sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010481sc0_http_req_rate([<table>]) : integer
10482sc1_http_req_rate([<table>]) : integer
10483sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010484 Returns the average rate of HTTP requests from the currently tracked
10485 counters, measured in amount of requests over the period configured in
10486 the table. This includes every started request, valid or not. See also
10487 src_http_req_rate.
10488
Cyril Bonté62ba8702014-04-22 23:52:25 +020010489sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010490sc0_inc_gpc0([<table>]) : integer
10491sc1_inc_gpc0([<table>]) : integer
10492sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010493 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010010494 tracked counters, and returns its new value. Before the first invocation,
10495 the stored value is zero, so first invocation will increase it to 1 and will
10496 return 1. This is typically used as a second ACL in an expression in order
10497 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020010498
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010499 acl abuse sc0_http_req_rate gt 10
10500 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010501 tcp-request connection reject if abuse kill
10502
Cyril Bonté62ba8702014-04-22 23:52:25 +020010503sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010504sc0_kbytes_in([<table>]) : integer
10505sc1_kbytes_in([<table>]) : integer
10506sc2_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010507 Returns the total amount of client-to-server data from the currently tracked
10508 counters, measured in kilobytes. The test is currently performed on 32-bit
10509 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020010510
Cyril Bonté62ba8702014-04-22 23:52:25 +020010511sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010512sc0_kbytes_out([<table>]) : integer
10513sc1_kbytes_out([<table>]) : integer
10514sc2_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010515 Returns the total amount of server-to-client data from the currently tracked
10516 counters, measured in kilobytes. The test is currently performed on 32-bit
10517 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020010518
Cyril Bonté62ba8702014-04-22 23:52:25 +020010519sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010520sc0_sess_cnt([<table>]) : integer
10521sc1_sess_cnt([<table>]) : integer
10522sc2_sess_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010523 Returns the cumulated number of incoming connections that were transformed
10524 into sessions, which means that they were accepted by a "tcp-request
10525 connection" rule, from the currently tracked counters. A backend may count
10526 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010527 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020010528 with the client. See also src_sess_cnt.
10529
Cyril Bonté62ba8702014-04-22 23:52:25 +020010530sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010531sc0_sess_rate([<table>]) : integer
10532sc1_sess_rate([<table>]) : integer
10533sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010534 Returns the average session rate from the currently tracked counters,
10535 measured in amount of sessions over the period configured in the table. A
10536 session is a connection that got past the early "tcp-request connection"
10537 rules. A backend may count more sessions than connections because each
10538 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010539 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020010540
Cyril Bonté62ba8702014-04-22 23:52:25 +020010541sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020010542sc0_tracked([<table>]) : boolean
10543sc1_tracked([<table>]) : boolean
10544sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020010545 Returns true if the designated session counter is currently being tracked by
10546 the current session. This can be useful when deciding whether or not we want
10547 to set some values in a header passed to the server.
10548
Cyril Bonté62ba8702014-04-22 23:52:25 +020010549sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010550sc0_trackers([<table>]) : integer
10551sc1_trackers([<table>]) : integer
10552sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010010553 Returns the current amount of concurrent connections tracking the same
10554 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010555 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010010556 that it does not rely on any stored information but on the table's reference
10557 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020010558 may sometimes be more suited for layer7 tracking. It can be used to tell a
10559 server how many concurrent connections there are from a given address for
10560 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010010561
Willy Tarreau74ca5042013-06-11 23:12:07 +020010562so_id : integer
10563 Returns an integer containing the current listening socket's id. It is useful
10564 in frontends involving many "bind" lines, or to stick all users coming via a
10565 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010566
Willy Tarreau74ca5042013-06-11 23:12:07 +020010567src : ip
10568 This is the source IPv4 address of the client of the session. It is of type
10569 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
10570 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
10571 TCP-level source address which is used, and not the address of a client
10572 behind a proxy. However if the "accept-proxy" bind directive is used, it can
10573 be the address of a client behind another PROXY-protocol compatible component
10574 for all rule sets except "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010575
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010576 Example:
10577 # add an HTTP header in requests with the originating address' country
10578 http-request set-header X-Country %[src,map_ip(geoip.lst)]
10579
Willy Tarreau74ca5042013-06-11 23:12:07 +020010580src_bytes_in_rate([<table>]) : integer
10581 Returns the average bytes rate from the incoming connection's source address
10582 in the current proxy's stick-table or in the designated stick-table, measured
10583 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010584 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010585
Willy Tarreau74ca5042013-06-11 23:12:07 +020010586src_bytes_out_rate([<table>]) : integer
10587 Returns the average bytes rate to the incoming connection's source address in
10588 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010589 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010590 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010591
Willy Tarreau74ca5042013-06-11 23:12:07 +020010592src_clr_gpc0([<table>]) : integer
10593 Clears the first General Purpose Counter associated to the incoming
10594 connection's source address in the current proxy's stick-table or in the
10595 designated stick-table, and returns its previous value. If the address is not
10596 found, an entry is created and 0 is returned. This is typically used as a
10597 second ACL in an expression in order to mark a connection when a first ACL
10598 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010599
10600 # block if 5 consecutive requests continue to come faster than 10 sess
10601 # per second, and reset the counter as soon as the traffic slows down.
10602 acl abuse src_http_req_rate gt 10
10603 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010010604 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010605 tcp-request connection accept if !abuse save
10606 tcp-request connection reject if abuse kill
10607
Willy Tarreau74ca5042013-06-11 23:12:07 +020010608src_conn_cnt([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010609 Returns the cumulated number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020010610 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010611 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010612 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010613
Willy Tarreau74ca5042013-06-11 23:12:07 +020010614src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010615 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010616 current incoming connection's source address in the current proxy's
10617 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010618 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010619
Willy Tarreau74ca5042013-06-11 23:12:07 +020010620src_conn_rate([<table>]) : integer
10621 Returns the average connection rate from the incoming connection's source
10622 address in the current proxy's stick-table or in the designated stick-table,
10623 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010624 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010625
Willy Tarreau74ca5042013-06-11 23:12:07 +020010626src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010627 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010628 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010629 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010630 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010631
Willy Tarreau74ca5042013-06-11 23:12:07 +020010632src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010633 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020010634 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010635 stick-table or in the designated stick-table. It reports the frequency
10636 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010637 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10638 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10639 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010640
Willy Tarreau74ca5042013-06-11 23:12:07 +020010641src_http_err_cnt([<table>]) : integer
10642 Returns the cumulated number of HTTP errors from the incoming connection's
10643 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020010644 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010645 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020010646 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010647
Willy Tarreau74ca5042013-06-11 23:12:07 +020010648src_http_err_rate([<table>]) : integer
10649 Returns the average rate of HTTP errors from the incoming connection's source
10650 address in the current proxy's stick-table or in the designated stick-table,
10651 measured in amount of errors over the period configured in the table. This
10652 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010653 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010654
Willy Tarreau74ca5042013-06-11 23:12:07 +020010655src_http_req_cnt([<table>]) : integer
10656 Returns the cumulated number of HTTP requests from the incoming connection's
10657 source address in the current proxy's stick-table or in the designated stick-
10658 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010659 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010660
Willy Tarreau74ca5042013-06-11 23:12:07 +020010661src_http_req_rate([<table>]) : integer
10662 Returns the average rate of HTTP requests from the incoming connection's
10663 source address in the current proxy's stick-table or in the designated stick-
10664 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020010665 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010666 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010667
Willy Tarreau74ca5042013-06-11 23:12:07 +020010668src_inc_gpc0([<table>]) : integer
10669 Increments the first General Purpose Counter associated to the incoming
10670 connection's source address in the current proxy's stick-table or in the
10671 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010672 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010673 This is typically used as a second ACL in an expression in order to mark a
10674 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020010675
10676 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010010677 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010678 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020010679
Willy Tarreau74ca5042013-06-11 23:12:07 +020010680src_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010681 Returns the total amount of data received from the incoming connection's
10682 source address in the current proxy's stick-table or in the designated
10683 stick-table, measured in kilobytes. If the address is not found, zero is
10684 returned. The test is currently performed on 32-bit integers, which limits
10685 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010686
Willy Tarreau74ca5042013-06-11 23:12:07 +020010687src_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010688 Returns the total amount of data sent to the incoming connection's source
10689 address in the current proxy's stick-table or in the designated stick-table,
10690 measured in kilobytes. If the address is not found, zero is returned. The
10691 test is currently performed on 32-bit integers, which limits values to 4
10692 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010693
Willy Tarreau74ca5042013-06-11 23:12:07 +020010694src_port : integer
10695 Returns an integer value corresponding to the TCP source port of the
10696 connection on the client side, which is the port the client connected from.
10697 Usage of this function is very limited as modern protocols do not care much
10698 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010699
Willy Tarreau74ca5042013-06-11 23:12:07 +020010700src_sess_cnt([<table>]) : integer
10701 Returns the cumulated number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020010702 connection's source IPv4 address in the current proxy's stick-table or in the
10703 designated stick-table, that were transformed into sessions, which means that
10704 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010705 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010706
Willy Tarreau74ca5042013-06-11 23:12:07 +020010707src_sess_rate([<table>]) : integer
10708 Returns the average session rate from the incoming connection's source
10709 address in the current proxy's stick-table or in the designated stick-table,
10710 measured in amount of sessions over the period configured in the table. A
10711 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010712 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010713
Willy Tarreau74ca5042013-06-11 23:12:07 +020010714src_updt_conn_cnt([<table>]) : integer
10715 Creates or updates the entry associated to the incoming connection's source
10716 address in the current proxy's stick-table or in the designated stick-table.
10717 This table must be configured to store the "conn_cnt" data type, otherwise
10718 the match will be ignored. The current count is incremented by one, and the
10719 expiration timer refreshed. The updated count is returned, so this match
10720 can't return zero. This was used to reject service abusers based on their
10721 source address. Note: it is recommended to use the more complete "track-sc*"
10722 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010723
10724 Example :
10725 # This frontend limits incoming SSH connections to 3 per 10 second for
10726 # each source address, and rejects excess connections until a 10 second
10727 # silence is observed. At most 20 addresses are tracked.
10728 listen ssh
10729 bind :22
10730 mode tcp
10731 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020010732 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020010733 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010734 server local 127.0.0.1:22
10735
Willy Tarreau74ca5042013-06-11 23:12:07 +020010736srv_id : integer
10737 Returns an integer containing the server's id when processing the response.
10738 While it's almost only used with ACLs, it may be used for logging or
10739 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020010740
Hervé COMMOWICK35ed8012010-12-15 14:04:51 +010010741
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200107427.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020010743----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020010744
Willy Tarreau74ca5042013-06-11 23:12:07 +020010745The layer 5 usually describes just the session layer which in haproxy is
10746closest to the session once all the connection handshakes are finished, but
10747when no content is yet made available. The fetch methods described here are
10748usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010749future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020010750
Emeric Brun645ae792014-04-30 14:21:06 +020010751ssl_bc : boolean
10752 Returns true when the back connection was made via an SSL/TLS transport
10753 layer and is locally deciphered. This means the outgoing connection was made
10754 other a server with the "ssl" option.
10755
10756ssl_bc_alg_keysize : integer
10757 Returns the symmetric cipher key size supported in bits when the outgoing
10758 connection was made over an SSL/TLS transport layer.
10759
10760ssl_bc_cipher : string
10761 Returns the name of the used cipher when the outgoing connection was made
10762 over an SSL/TLS transport layer.
10763
10764ssl_bc_protocol : string
10765 Returns the name of the used protocol when the outgoing connection was made
10766 over an SSL/TLS transport layer.
10767
Emeric Brunb73a9b02014-04-30 18:49:19 +020010768ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020010769 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010770 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10771 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020010772
10773ssl_bc_session_id : binary
10774 Returns the SSL ID of the back connection when the outgoing connection was
10775 made over an SSL/TLS transport layer. It is useful to log if we want to know
10776 if session was reused or not.
10777
10778ssl_bc_use_keysize : integer
10779 Returns the symmetric cipher key size used in bits when the outgoing
10780 connection was made over an SSL/TLS transport layer.
10781
Willy Tarreau74ca5042013-06-11 23:12:07 +020010782ssl_c_ca_err : integer
10783 When the incoming connection was made over an SSL/TLS transport layer,
10784 returns the ID of the first error detected during verification of the client
10785 certificate at depth > 0, or 0 if no error was encountered during this
10786 verification process. Please refer to your SSL library's documentation to
10787 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020010788
Willy Tarreau74ca5042013-06-11 23:12:07 +020010789ssl_c_ca_err_depth : integer
10790 When the incoming connection was made over an SSL/TLS transport layer,
10791 returns the depth in the CA chain of the first error detected during the
10792 verification of the client certificate. If no error is encountered, 0 is
10793 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010794
Emeric Brunb3cc4252014-10-29 19:03:26 +010010795ssl_c_der : binary
10796 Returns the DER formatted certificate presented by the client when the
10797 incoming connection was made over an SSL/TLS transport layer. When used for
10798 an ACL, the value(s) to match against can be passed in hexadecimal form.
10799
Willy Tarreau74ca5042013-06-11 23:12:07 +020010800ssl_c_err : integer
10801 When the incoming connection was made over an SSL/TLS transport layer,
10802 returns the ID of the first error detected during verification at depth 0, or
10803 0 if no error was encountered during this verification process. Please refer
10804 to your SSL library's documentation to find the exhaustive list of error
10805 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020010806
Willy Tarreau74ca5042013-06-11 23:12:07 +020010807ssl_c_i_dn([<entry>[,<occ>]]) : string
10808 When the incoming connection was made over an SSL/TLS transport layer,
10809 returns the full distinguished name of the issuer of the certificate
10810 presented by the client when no <entry> is specified, or the value of the
10811 first given entry found from the beginning of the DN. If a positive/negative
10812 occurrence number is specified as the optional second argument, it returns
10813 the value of the nth given entry value from the beginning/end of the DN.
10814 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
10815 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020010816
Willy Tarreau74ca5042013-06-11 23:12:07 +020010817ssl_c_key_alg : string
10818 Returns the name of the algorithm used to generate the key of the certificate
10819 presented by the client when the incoming connection was made over an SSL/TLS
10820 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020010821
Willy Tarreau74ca5042013-06-11 23:12:07 +020010822ssl_c_notafter : string
10823 Returns the end date presented by the client as a formatted string
10824 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10825 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020010826
Willy Tarreau74ca5042013-06-11 23:12:07 +020010827ssl_c_notbefore : string
10828 Returns the start date presented by the client as a formatted string
10829 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10830 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010831
Willy Tarreau74ca5042013-06-11 23:12:07 +020010832ssl_c_s_dn([<entry>[,<occ>]]) : string
10833 When the incoming connection was made over an SSL/TLS transport layer,
10834 returns the full distinguished name of the subject of the certificate
10835 presented by the client when no <entry> is specified, or the value of the
10836 first given entry found from the beginning of the DN. If a positive/negative
10837 occurrence number is specified as the optional second argument, it returns
10838 the value of the nth given entry value from the beginning/end of the DN.
10839 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
10840 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010841
Willy Tarreau74ca5042013-06-11 23:12:07 +020010842ssl_c_serial : binary
10843 Returns the serial of the certificate presented by the client when the
10844 incoming connection was made over an SSL/TLS transport layer. When used for
10845 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010846
Willy Tarreau74ca5042013-06-11 23:12:07 +020010847ssl_c_sha1 : binary
10848 Returns the SHA-1 fingerprint of the certificate presented by the client when
10849 the incoming connection was made over an SSL/TLS transport layer. This can be
10850 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau9fe4cb62014-07-02 19:01:22 +020010851 Note that the output is binary, so if you want to pass that signature to the
10852 server, you need to encode it in hex or base64, such as in the example below:
10853
10854 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020010855
Willy Tarreau74ca5042013-06-11 23:12:07 +020010856ssl_c_sig_alg : string
10857 Returns the name of the algorithm used to sign the certificate presented by
10858 the client when the incoming connection was made over an SSL/TLS transport
10859 layer.
Emeric Brun87855892012-10-17 17:39:35 +020010860
Willy Tarreau74ca5042013-06-11 23:12:07 +020010861ssl_c_used : boolean
10862 Returns true if current SSL session uses a client certificate even if current
10863 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020010864
Willy Tarreau74ca5042013-06-11 23:12:07 +020010865ssl_c_verify : integer
10866 Returns the verify result error ID when the incoming connection was made over
10867 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
10868 refer to your SSL library's documentation for an exhaustive list of error
10869 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020010870
Willy Tarreau74ca5042013-06-11 23:12:07 +020010871ssl_c_version : integer
10872 Returns the version of the certificate presented by the client when the
10873 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020010874
Emeric Brunb3cc4252014-10-29 19:03:26 +010010875ssl_f_der : binary
10876 Returns the DER formatted certificate presented by the frontend when the
10877 incoming connection was made over an SSL/TLS transport layer. When used for
10878 an ACL, the value(s) to match against can be passed in hexadecimal form.
10879
Willy Tarreau74ca5042013-06-11 23:12:07 +020010880ssl_f_i_dn([<entry>[,<occ>]]) : string
10881 When the incoming connection was made over an SSL/TLS transport layer,
10882 returns the full distinguished name of the issuer of the certificate
10883 presented by the frontend when no <entry> is specified, or the value of the
10884 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020010885 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020010886 the value of the nth given entry value from the beginning/end of the DN.
10887 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
10888 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020010889
Willy Tarreau74ca5042013-06-11 23:12:07 +020010890ssl_f_key_alg : string
10891 Returns the name of the algorithm used to generate the key of the certificate
10892 presented by the frontend when the incoming connection was made over an
10893 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010894
Willy Tarreau74ca5042013-06-11 23:12:07 +020010895ssl_f_notafter : string
10896 Returns the end date presented by the frontend as a formatted string
10897 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10898 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010899
Willy Tarreau74ca5042013-06-11 23:12:07 +020010900ssl_f_notbefore : string
10901 Returns the start date presented by the frontend as a formatted string
10902 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10903 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020010904
Willy Tarreau74ca5042013-06-11 23:12:07 +020010905ssl_f_s_dn([<entry>[,<occ>]]) : string
10906 When the incoming connection was made over an SSL/TLS transport layer,
10907 returns the full distinguished name of the subject of the certificate
10908 presented by the frontend when no <entry> is specified, or the value of the
10909 first given entry found from the beginning of the DN. If a positive/negative
10910 occurrence number is specified as the optional second argument, it returns
10911 the value of the nth given entry value from the beginning/end of the DN.
10912 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
10913 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020010914
Willy Tarreau74ca5042013-06-11 23:12:07 +020010915ssl_f_serial : binary
10916 Returns the serial of the certificate presented by the frontend when the
10917 incoming connection was made over an SSL/TLS transport layer. When used for
10918 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020010919
Emeric Brun55f4fa82014-04-30 17:11:25 +020010920ssl_f_sha1 : binary
10921 Returns the SHA-1 fingerprint of the certificate presented by the frontend
10922 when the incoming connection was made over an SSL/TLS transport layer. This
10923 can be used to know which certificate was chosen using SNI.
10924
Willy Tarreau74ca5042013-06-11 23:12:07 +020010925ssl_f_sig_alg : string
10926 Returns the name of the algorithm used to sign the certificate presented by
10927 the frontend when the incoming connection was made over an SSL/TLS transport
10928 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010929
Willy Tarreau74ca5042013-06-11 23:12:07 +020010930ssl_f_version : integer
10931 Returns the version of the certificate presented by the frontend when the
10932 incoming connection was made over an SSL/TLS transport layer.
10933
10934ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010935 Returns true when the front connection was made via an SSL/TLS transport
10936 layer and is locally deciphered. This means it has matched a socket declared
10937 with a "bind" line having the "ssl" option.
10938
Willy Tarreau74ca5042013-06-11 23:12:07 +020010939 Example :
10940 # This passes "X-Proto: https" to servers when client connects over SSL
10941 listen http-https
10942 bind :80
10943 bind :443 ssl crt /etc/haproxy.pem
10944 http-request add-header X-Proto https if { ssl_fc }
10945
10946ssl_fc_alg_keysize : integer
10947 Returns the symmetric cipher key size supported in bits when the incoming
10948 connection was made over an SSL/TLS transport layer.
10949
10950ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010951 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020010952 incoming connection made via a TLS transport layer and locally deciphered by
10953 haproxy. The result is a string containing the protocol name advertised by
10954 the client. The SSL library must have been built with support for TLS
10955 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
10956 not advertised unless the "alpn" keyword on the "bind" line specifies a
10957 protocol list. Also, nothing forces the client to pick a protocol from this
10958 list, any other one may be requested. The TLS ALPN extension is meant to
10959 replace the TLS NPN extension. See also "ssl_fc_npn".
10960
Willy Tarreau74ca5042013-06-11 23:12:07 +020010961ssl_fc_cipher : string
10962 Returns the name of the used cipher when the incoming connection was made
10963 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020010964
Willy Tarreau74ca5042013-06-11 23:12:07 +020010965ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010966 Returns true if a client certificate is present in an incoming connection over
10967 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010010968 Note: on SSL session resumption with Session ID or TLS ticket, client
10969 certificate is not present in the current connection but may be retrieved
10970 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
10971 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010972
Willy Tarreau74ca5042013-06-11 23:12:07 +020010973ssl_fc_has_sni : boolean
10974 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020010975 in an incoming connection was made over an SSL/TLS transport layer. Returns
10976 true when the incoming connection presents a TLS SNI field. This requires
10977 that the SSL library is build with support for TLS extensions enabled (check
10978 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020010979
Willy Tarreau74ca5042013-06-11 23:12:07 +020010980ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010981 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020010982 made via a TLS transport layer and locally deciphered by haproxy. The result
10983 is a string containing the protocol name advertised by the client. The SSL
10984 library must have been built with support for TLS extensions enabled (check
10985 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
10986 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
10987 forces the client to pick a protocol from this list, any other one may be
10988 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020010989
Willy Tarreau74ca5042013-06-11 23:12:07 +020010990ssl_fc_protocol : string
10991 Returns the name of the used protocol when the incoming connection was made
10992 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020010993
Emeric Brunb73a9b02014-04-30 18:49:19 +020010994ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040010995 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010996 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10997 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040010998
Willy Tarreau74ca5042013-06-11 23:12:07 +020010999ssl_fc_session_id : binary
11000 Returns the SSL ID of the front connection when the incoming connection was
11001 made over an SSL/TLS transport layer. It is useful to stick a given client to
11002 a server. It is important to note that some browsers refresh their session ID
11003 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020011004
Willy Tarreau74ca5042013-06-11 23:12:07 +020011005ssl_fc_sni : string
11006 This extracts the Server Name Indication TLS extension (SNI) field from an
11007 incoming connection made via an SSL/TLS transport layer and locally
11008 deciphered by haproxy. The result (when present) typically is a string
11009 matching the HTTPS host name (253 chars or less). The SSL library must have
11010 been built with support for TLS extensions enabled (check haproxy -vv).
11011
11012 This fetch is different from "req_ssl_sni" above in that it applies to the
11013 connection being deciphered by haproxy and not to SSL contents being blindly
11014 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020011015 requires that the SSL library is build with support for TLS extensions
11016 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020011017
Willy Tarreau74ca5042013-06-11 23:12:07 +020011018 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020011019 ssl_fc_sni_end : suffix match
11020 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020011021
Willy Tarreau74ca5042013-06-11 23:12:07 +020011022ssl_fc_use_keysize : integer
11023 Returns the symmetric cipher key size used in bits when the incoming
11024 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011025
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011026
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200110277.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011028------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011029
Willy Tarreau74ca5042013-06-11 23:12:07 +020011030Fetching samples from buffer contents is a bit different from the previous
11031sample fetches above because the sampled data are ephemeral. These data can
11032only be used when they're available and will be lost when they're forwarded.
11033For this reason, samples fetched from buffer contents during a request cannot
11034be used in a response for example. Even while the data are being fetched, they
11035can change. Sometimes it is necessary to set some delays or combine multiple
11036sample fetch methods to ensure that the expected data are complete and usable,
11037for example through TCP request content inspection. Please see the "tcp-request
11038content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020011039
Willy Tarreau74ca5042013-06-11 23:12:07 +020011040payload(<offset>,<length>) : binary (deprecated)
11041 This is an alias for "req.payload" when used in the context of a request (eg:
11042 "stick on", "stick match"), and for "res.payload" when used in the context of
11043 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011044
Willy Tarreau74ca5042013-06-11 23:12:07 +020011045payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
11046 This is an alias for "req.payload_lv" when used in the context of a request
11047 (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the
11048 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011049
Willy Tarreau74ca5042013-06-11 23:12:07 +020011050req.len : integer
11051req_len : integer (deprecated)
11052 Returns an integer value corresponding to the number of bytes present in the
11053 request buffer. This is mostly used in ACL. It is important to understand
11054 that this test does not return false as long as the buffer is changing. This
11055 means that a check with equality to zero will almost always immediately match
11056 at the beginning of the session, while a test for more data will wait for
11057 that data to come in and return false only when haproxy is certain that no
11058 more data will come in. This test was designed to be used with TCP request
11059 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011060
Willy Tarreau74ca5042013-06-11 23:12:07 +020011061req.payload(<offset>,<length>) : binary
11062 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011063 in the request buffer. As a special case, if the <length> argument is zero,
11064 the the whole buffer from <offset> to the end is extracted. This can be used
11065 with ACLs in order to check for the presence of some content in a buffer at
11066 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011067
Willy Tarreau74ca5042013-06-11 23:12:07 +020011068 ACL alternatives :
11069 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011070
Willy Tarreau74ca5042013-06-11 23:12:07 +020011071req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11072 This extracts a binary block whose size is specified at <offset1> for <length>
11073 bytes, and which starts at <offset2> if specified or just after the length in
11074 the request buffer. The <offset2> parameter also supports relative offsets if
11075 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011076
Willy Tarreau74ca5042013-06-11 23:12:07 +020011077 ACL alternatives :
11078 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011079
Willy Tarreau74ca5042013-06-11 23:12:07 +020011080 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011081
Willy Tarreau74ca5042013-06-11 23:12:07 +020011082req.proto_http : boolean
11083req_proto_http : boolean (deprecated)
11084 Returns true when data in the request buffer look like HTTP and correctly
11085 parses as such. It is the same parser as the common HTTP request parser which
11086 is used so there should be no surprises. The test does not match until the
11087 request is complete, failed or timed out. This test may be used to report the
11088 protocol in TCP logs, but the biggest use is to block TCP request analysis
11089 until a complete HTTP request is present in the buffer, for example to track
11090 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011091
Willy Tarreau74ca5042013-06-11 23:12:07 +020011092 Example:
11093 # track request counts per "base" (concatenation of Host+URL)
11094 tcp-request inspect-delay 10s
11095 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011096 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011097
Willy Tarreau74ca5042013-06-11 23:12:07 +020011098req.rdp_cookie([<name>]) : string
11099rdp_cookie([<name>]) : string (deprecated)
11100 When the request buffer looks like the RDP protocol, extracts the RDP cookie
11101 <name>, or any cookie if unspecified. The parser only checks for the first
11102 cookie, as illustrated in the RDP protocol specification. The cookie name is
11103 case insensitive. Generally the "MSTS" cookie name will be used, as it can
11104 contain the user name of the client connecting to the server if properly
11105 configured on the client. The "MSTSHASH" cookie is often used as well for
11106 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011107
Willy Tarreau74ca5042013-06-11 23:12:07 +020011108 This differs from "balance rdp-cookie" in that any balancing algorithm may be
11109 used and thus the distribution of clients to backend servers is not linked to
11110 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
11111 such as "balance roundrobin" or "balance leastconn" will lead to a more even
11112 distribution of clients to backend servers than the hash used by "balance
11113 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011114
Willy Tarreau74ca5042013-06-11 23:12:07 +020011115 ACL derivatives :
11116 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011117
Willy Tarreau74ca5042013-06-11 23:12:07 +020011118 Example :
11119 listen tse-farm
11120 bind 0.0.0.0:3389
11121 # wait up to 5s for an RDP cookie in the request
11122 tcp-request inspect-delay 5s
11123 tcp-request content accept if RDP_COOKIE
11124 # apply RDP cookie persistence
11125 persist rdp-cookie
11126 # Persist based on the mstshash cookie
11127 # This is only useful makes sense if
11128 # balance rdp-cookie is not used
11129 stick-table type string size 204800
11130 stick on req.rdp_cookie(mstshash)
11131 server srv1 1.1.1.1:3389
11132 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011133
Willy Tarreau74ca5042013-06-11 23:12:07 +020011134 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
11135 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011136
Willy Tarreau74ca5042013-06-11 23:12:07 +020011137req.rdp_cookie_cnt([name]) : integer
11138rdp_cookie_cnt([name]) : integer (deprecated)
11139 Tries to parse the request buffer as RDP protocol, then returns an integer
11140 corresponding to the number of RDP cookies found. If an optional cookie name
11141 is passed, only cookies matching this name are considered. This is mostly
11142 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011143
Willy Tarreau74ca5042013-06-11 23:12:07 +020011144 ACL derivatives :
11145 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011146
Willy Tarreau74ca5042013-06-11 23:12:07 +020011147req.ssl_hello_type : integer
11148req_ssl_hello_type : integer (deprecated)
11149 Returns an integer value containing the type of the SSL hello message found
11150 in the request buffer if the buffer contains data that parse as a complete
11151 SSL (v3 or superior) client hello message. Note that this only applies to raw
11152 contents found in the request buffer and not to contents deciphered via an
11153 SSL data layer, so this will not work with "bind" lines having the "ssl"
11154 option. This is mostly used in ACL to detect presence of an SSL hello message
11155 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011156
Willy Tarreau74ca5042013-06-11 23:12:07 +020011157req.ssl_sni : string
11158req_ssl_sni : string (deprecated)
11159 Returns a string containing the value of the Server Name TLS extension sent
11160 by a client in a TLS stream passing through the request buffer if the buffer
11161 contains data that parse as a complete SSL (v3 or superior) client hello
11162 message. Note that this only applies to raw contents found in the request
11163 buffer and not to contents deciphered via an SSL data layer, so this will not
11164 work with "bind" lines having the "ssl" option. SNI normally contains the
11165 name of the host the client tries to connect to (for recent browsers). SNI is
11166 useful for allowing or denying access to certain hosts when SSL/TLS is used
11167 by the client. This test was designed to be used with TCP request content
11168 inspection. If content switching is needed, it is recommended to first wait
11169 for a complete client hello (type 1), like in the example below. See also
11170 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011171
Willy Tarreau74ca5042013-06-11 23:12:07 +020011172 ACL derivatives :
11173 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011174
Willy Tarreau74ca5042013-06-11 23:12:07 +020011175 Examples :
11176 # Wait for a client hello for at most 5 seconds
11177 tcp-request inspect-delay 5s
11178 tcp-request content accept if { req_ssl_hello_type 1 }
11179 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
11180 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011181
Willy Tarreau74ca5042013-06-11 23:12:07 +020011182res.ssl_hello_type : integer
11183rep_ssl_hello_type : integer (deprecated)
11184 Returns an integer value containing the type of the SSL hello message found
11185 in the response buffer if the buffer contains data that parses as a complete
11186 SSL (v3 or superior) hello message. Note that this only applies to raw
11187 contents found in the response buffer and not to contents deciphered via an
11188 SSL data layer, so this will not work with "server" lines having the "ssl"
11189 option. This is mostly used in ACL to detect presence of an SSL hello message
11190 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau51539362012-05-08 12:46:28 +020011191
Willy Tarreau74ca5042013-06-11 23:12:07 +020011192req.ssl_ver : integer
11193req_ssl_ver : integer (deprecated)
11194 Returns an integer value containing the version of the SSL/TLS protocol of a
11195 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
11196 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
11197 composed of the major version multiplied by 65536, added to the minor
11198 version. Note that this only applies to raw contents found in the request
11199 buffer and not to contents deciphered via an SSL data layer, so this will not
11200 work with "bind" lines having the "ssl" option. The ACL version of the test
11201 matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This
11202 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011203
Willy Tarreau74ca5042013-06-11 23:12:07 +020011204 ACL derivatives :
11205 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011206
Willy Tarreau47e8eba2013-09-11 23:28:46 +020011207res.len : integer
11208 Returns an integer value corresponding to the number of bytes present in the
11209 response buffer. This is mostly used in ACL. It is important to understand
11210 that this test does not return false as long as the buffer is changing. This
11211 means that a check with equality to zero will almost always immediately match
11212 at the beginning of the session, while a test for more data will wait for
11213 that data to come in and return false only when haproxy is certain that no
11214 more data will come in. This test was designed to be used with TCP response
11215 content inspection.
11216
Willy Tarreau74ca5042013-06-11 23:12:07 +020011217res.payload(<offset>,<length>) : binary
11218 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011219 in the response buffer. As a special case, if the <length> argument is zero,
11220 the the whole buffer from <offset> to the end is extracted. This can be used
11221 with ACLs in order to check for the presence of some content in a buffer at
11222 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011223
Willy Tarreau74ca5042013-06-11 23:12:07 +020011224res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11225 This extracts a binary block whose size is specified at <offset1> for <length>
11226 bytes, and which starts at <offset2> if specified or just after the length in
11227 the response buffer. The <offset2> parameter also supports relative offsets
11228 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011229
Willy Tarreau74ca5042013-06-11 23:12:07 +020011230 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011231
Willy Tarreau74ca5042013-06-11 23:12:07 +020011232wait_end : boolean
11233 This fetch either returns true when the inspection period is over, or does
11234 not fetch. It is only used in ACLs, in conjunction with content analysis to
11235 avoid returning a wrong verdict early. It may also be used to delay some
11236 actions, such as a delayed reject for some special addresses. Since it either
11237 stops the rules evaluation or immediately returns true, it is recommended to
11238 use this acl as the last one in a rule. Please note that the default ACL
11239 "WAIT_END" is always usable without prior declaration. This test was designed
11240 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011241
Willy Tarreau74ca5042013-06-11 23:12:07 +020011242 Examples :
11243 # delay every incoming request by 2 seconds
11244 tcp-request inspect-delay 2s
11245 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010011246
Willy Tarreau74ca5042013-06-11 23:12:07 +020011247 # don't immediately tell bad guys they are rejected
11248 tcp-request inspect-delay 10s
11249 acl goodguys src 10.0.0.0/24
11250 acl badguys src 10.0.1.0/24
11251 tcp-request content accept if goodguys
11252 tcp-request content reject if badguys WAIT_END
11253 tcp-request content reject
11254
11255
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200112567.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011257--------------------------------------
11258
11259It is possible to fetch samples from HTTP contents, requests and responses.
11260This application layer is also called layer 7. It is only possible to fetch the
11261data in this section when a full HTTP request or response has been parsed from
11262its respective request or response buffer. This is always the case with all
11263HTTP specific rules and for sections running with "mode http". When using TCP
11264content inspection, it may be necessary to support an inspection delay in order
11265to let the request or response come in first. These fetches may require a bit
11266more CPU resources than the layer 4 ones, but not much since the request and
11267response are indexed.
11268
11269base : string
11270 This returns the concatenation of the first Host header and the path part of
11271 the request, which starts at the first slash and ends before the question
11272 mark. It can be useful in virtual hosted environments to detect URL abuses as
11273 well as to improve shared caches efficiency. Using this with a limited size
11274 stick table also allows one to collect statistics about most commonly
11275 requested objects by host/path. With ACLs it can allow simple content
11276 switching rules involving the host and the path at the same time, such as
11277 "www.example.com/favicon.ico". See also "path" and "uri".
11278
11279 ACL derivatives :
11280 base : exact string match
11281 base_beg : prefix match
11282 base_dir : subdir match
11283 base_dom : domain match
11284 base_end : suffix match
11285 base_len : length match
11286 base_reg : regex match
11287 base_sub : substring match
11288
11289base32 : integer
11290 This returns a 32-bit hash of the value returned by the "base" fetch method
11291 above. This is useful to track per-URL activity on high traffic sites without
11292 having to store all URLs. Instead a shorter hash is stored, saving a lot of
11293 memory. The output type is an unsigned integer.
11294
11295base32+src : binary
11296 This returns the concatenation of the base32 fetch above and the src fetch
11297 below. The resulting type is of type binary, with a size of 8 or 20 bytes
11298 depending on the source address family. This can be used to track per-IP,
11299 per-URL counters.
11300
William Lallemand65ad6e12014-01-31 15:08:02 +010011301capture.req.hdr(<idx>) : string
11302 This extracts the content of the header captured by the "capture request
11303 header", idx is the position of the capture keyword in the configuration.
11304 The first entry is an index of 0. See also: "capture request header".
11305
11306capture.req.method : string
11307 This extracts the METHOD of an HTTP request. It can be used in both request
11308 and response. Unlike "method", it can be used in both request and response
11309 because it's allocated.
11310
11311capture.req.uri : string
11312 This extracts the request's URI, which starts at the first slash and ends
11313 before the first space in the request (without the host part). Unlike "path"
11314 and "url", it can be used in both request and response because it's
11315 allocated.
11316
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011317capture.req.ver : string
11318 This extracts the request's HTTP version and returns either "HTTP/1.0" or
11319 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
11320 logs because it relies on a persistent flag.
11321
William Lallemand65ad6e12014-01-31 15:08:02 +010011322capture.res.hdr(<idx>) : string
11323 This extracts the content of the header captured by the "capture response
11324 header", idx is the position of the capture keyword in the configuration.
11325 The first entry is an index of 0.
11326 See also: "capture response header"
11327
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011328capture.res.ver : string
11329 This extracts the response's HTTP version and returns either "HTTP/1.0" or
11330 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
11331 persistent flag.
11332
Willy Tarreau74ca5042013-06-11 23:12:07 +020011333req.cook([<name>]) : string
11334cook([<name>]) : string (deprecated)
11335 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11336 header line from the request, and returns its value as string. If no name is
11337 specified, the first cookie value is returned. When used with ACLs, all
11338 matching cookies are evaluated. Spaces around the name and the value are
11339 ignored as requested by the Cookie header specification (RFC6265). The cookie
11340 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
11341 well return an empty value if it is present. Use the "found" match to detect
11342 presence. Use the res.cook() variant for response cookies sent by the server.
11343
11344 ACL derivatives :
11345 cook([<name>]) : exact string match
11346 cook_beg([<name>]) : prefix match
11347 cook_dir([<name>]) : subdir match
11348 cook_dom([<name>]) : domain match
11349 cook_end([<name>]) : suffix match
11350 cook_len([<name>]) : length match
11351 cook_reg([<name>]) : regex match
11352 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011353
Willy Tarreau74ca5042013-06-11 23:12:07 +020011354req.cook_cnt([<name>]) : integer
11355cook_cnt([<name>]) : integer (deprecated)
11356 Returns an integer value representing the number of occurrences of the cookie
11357 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011358
Willy Tarreau74ca5042013-06-11 23:12:07 +020011359req.cook_val([<name>]) : integer
11360cook_val([<name>]) : integer (deprecated)
11361 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11362 header line from the request, and converts its value to an integer which is
11363 returned. If no name is specified, the first cookie value is returned. When
11364 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020011365
Willy Tarreau74ca5042013-06-11 23:12:07 +020011366cookie([<name>]) : string (deprecated)
11367 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11368 header line from the request, or a "Set-Cookie" header from the response, and
11369 returns its value as a string. A typical use is to get multiple clients
11370 sharing a same profile use the same server. This can be similar to what
11371 "appsession" does with the "request-learn" statement, but with support for
11372 multi-peer synchronization and state keeping across restarts. If no name is
11373 specified, the first cookie value is returned. This fetch should not be used
11374 anymore and should be replaced by req.cook() or res.cook() instead as it
11375 ambiguously uses the direction based on the context where it is used.
11376 See also : "appsession".
Willy Tarreaud63335a2010-02-26 12:56:52 +010011377
Willy Tarreau74ca5042013-06-11 23:12:07 +020011378hdr([<name>[,<occ>]]) : string
11379 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
11380 used on responses. Please refer to these respective fetches for more details.
11381 In case of doubt about the fetch direction, please use the explicit ones.
11382 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011383 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011384
Willy Tarreau74ca5042013-06-11 23:12:07 +020011385req.fhdr(<name>[,<occ>]) : string
11386 This extracts the last occurrence of header <name> in an HTTP request. When
11387 used from an ACL, all occurrences are iterated over until a match is found.
11388 Optionally, a specific occurrence might be specified as a position number.
11389 Positive values indicate a position from the first occurrence, with 1 being
11390 the first one. Negative values indicate positions relative to the last one,
11391 with -1 being the last one. It differs from req.hdr() in that any commas
11392 present in the value are returned and are not used as delimiters. This is
11393 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011394
Willy Tarreau74ca5042013-06-11 23:12:07 +020011395req.fhdr_cnt([<name>]) : integer
11396 Returns an integer value representing the number of occurrences of request
11397 header field name <name>, or the total number of header fields if <name> is
11398 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
11399 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011400
Willy Tarreau74ca5042013-06-11 23:12:07 +020011401req.hdr([<name>[,<occ>]]) : string
11402 This extracts the last occurrence of header <name> in an HTTP request. When
11403 used from an ACL, all occurrences are iterated over until a match is found.
11404 Optionally, a specific occurrence might be specified as a position number.
11405 Positive values indicate a position from the first occurrence, with 1 being
11406 the first one. Negative values indicate positions relative to the last one,
11407 with -1 being the last one. A typical use is with the X-Forwarded-For header
11408 once converted to IP, associated with an IP stick-table. The function
11409 considers any comma as a delimiter for distinct values. If full-line headers
11410 are desired instead, use req.fhdr(). Please carefully check RFC2616 to know
11411 how certain headers are supposed to be parsed. Also, some of them are case
11412 insensitive (eg: Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010011413
Willy Tarreau74ca5042013-06-11 23:12:07 +020011414 ACL derivatives :
11415 hdr([<name>[,<occ>]]) : exact string match
11416 hdr_beg([<name>[,<occ>]]) : prefix match
11417 hdr_dir([<name>[,<occ>]]) : subdir match
11418 hdr_dom([<name>[,<occ>]]) : domain match
11419 hdr_end([<name>[,<occ>]]) : suffix match
11420 hdr_len([<name>[,<occ>]]) : length match
11421 hdr_reg([<name>[,<occ>]]) : regex match
11422 hdr_sub([<name>[,<occ>]]) : substring match
11423
11424req.hdr_cnt([<name>]) : integer
11425hdr_cnt([<header>]) : integer (deprecated)
11426 Returns an integer value representing the number of occurrences of request
11427 header field name <name>, or the total number of header field values if
11428 <name> is not specified. It is important to remember that one header line may
11429 count as several headers if it has several values. The function considers any
11430 comma as a delimiter for distinct values. If full-line headers are desired
11431 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
11432 detect presence, absence or abuse of a specific header, as well as to block
11433 request smuggling attacks by rejecting requests which contain more than one
11434 of certain headers. See "req.hdr" for more information on header matching.
11435
11436req.hdr_ip([<name>[,<occ>]]) : ip
11437hdr_ip([<name>[,<occ>]]) : ip (deprecated)
11438 This extracts the last occurrence of header <name> in an HTTP request,
11439 converts it to an IPv4 or IPv6 address and returns this address. When used
11440 with ACLs, all occurrences are checked, and if <name> is omitted, every value
11441 of every header is checked. Optionally, a specific occurrence might be
11442 specified as a position number. Positive values indicate a position from the
11443 first occurrence, with 1 being the first one. Negative values indicate
11444 positions relative to the last one, with -1 being the last one. A typical use
11445 is with the X-Forwarded-For and X-Client-IP headers.
11446
11447req.hdr_val([<name>[,<occ>]]) : integer
11448hdr_val([<name>[,<occ>]]) : integer (deprecated)
11449 This extracts the last occurrence of header <name> in an HTTP request, and
11450 converts it to an integer value. When used with ACLs, all occurrences are
11451 checked, and if <name> is omitted, every value of every header is checked.
11452 Optionally, a specific occurrence might be specified as a position number.
11453 Positive values indicate a position from the first occurrence, with 1 being
11454 the first one. Negative values indicate positions relative to the last one,
11455 with -1 being the last one. A typical use is with the X-Forwarded-For header.
11456
11457http_auth(<userlist>) : boolean
11458 Returns a boolean indicating whether the authentication data received from
11459 the client match a username & password stored in the specified userlist. This
11460 fetch function is not really useful outside of ACLs. Currently only http
11461 basic auth is supported.
11462
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011463http_auth_group(<userlist>) : string
11464 Returns a string corresponding to the user name found in the authentication
11465 data received from the client if both the user name and password are valid
11466 according to the specified userlist. The main purpose is to use it in ACLs
11467 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011468 This fetch function is not really useful outside of ACLs. Currently only http
11469 basic auth is supported.
11470
11471 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011472 http_auth_group(<userlist>) : group ...
11473 Returns true when the user extracted from the request and whose password is
11474 valid according to the specified userlist belongs to at least one of the
11475 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011476
11477http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020011478 Returns true when the request being processed is the first one of the
11479 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020011480 from some requests when a request is not the first one, or to help grouping
11481 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020011482
Willy Tarreau74ca5042013-06-11 23:12:07 +020011483method : integer + string
11484 Returns an integer value corresponding to the method in the HTTP request. For
11485 example, "GET" equals 1 (check sources to establish the matching). Value 9
11486 means "other method" and may be converted to a string extracted from the
11487 stream. This should not be used directly as a sample, this is only meant to
11488 be used from ACLs, which transparently convert methods from patterns to these
11489 integer + string values. Some predefined ACL already check for most common
11490 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011491
Willy Tarreau74ca5042013-06-11 23:12:07 +020011492 ACL derivatives :
11493 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011494
Willy Tarreau74ca5042013-06-11 23:12:07 +020011495 Example :
11496 # only accept GET and HEAD requests
11497 acl valid_method method GET HEAD
11498 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020011499
Willy Tarreau74ca5042013-06-11 23:12:07 +020011500path : string
11501 This extracts the request's URL path, which starts at the first slash and
11502 ends before the question mark (without the host part). A typical use is with
11503 prefetch-capable caches, and with portals which need to aggregate multiple
11504 information from databases and keep them in caches. Note that with outgoing
11505 caches, it would be wiser to use "url" instead. With ACLs, it's typically
11506 used to match exact file names (eg: "/login.php"), or directory parts using
11507 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011508
Willy Tarreau74ca5042013-06-11 23:12:07 +020011509 ACL derivatives :
11510 path : exact string match
11511 path_beg : prefix match
11512 path_dir : subdir match
11513 path_dom : domain match
11514 path_end : suffix match
11515 path_len : length match
11516 path_reg : regex match
11517 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011518
Willy Tarreau74ca5042013-06-11 23:12:07 +020011519req.ver : string
11520req_ver : string (deprecated)
11521 Returns the version string from the HTTP request, for example "1.1". This can
11522 be useful for logs, but is mostly there for ACL. Some predefined ACL already
11523 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011524
Willy Tarreau74ca5042013-06-11 23:12:07 +020011525 ACL derivatives :
11526 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020011527
Willy Tarreau74ca5042013-06-11 23:12:07 +020011528res.comp : boolean
11529 Returns the boolean "true" value if the response has been compressed by
11530 HAProxy, otherwise returns boolean "false". This may be used to add
11531 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011532
Willy Tarreau74ca5042013-06-11 23:12:07 +020011533res.comp_algo : string
11534 Returns a string containing the name of the algorithm used if the response
11535 was compressed by HAProxy, for example : "deflate". This may be used to add
11536 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011537
Willy Tarreau74ca5042013-06-11 23:12:07 +020011538res.cook([<name>]) : string
11539scook([<name>]) : string (deprecated)
11540 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11541 header line from the response, and returns its value as string. If no name is
11542 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011543
Willy Tarreau74ca5042013-06-11 23:12:07 +020011544 ACL derivatives :
11545 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011546
Willy Tarreau74ca5042013-06-11 23:12:07 +020011547res.cook_cnt([<name>]) : integer
11548scook_cnt([<name>]) : integer (deprecated)
11549 Returns an integer value representing the number of occurrences of the cookie
11550 <name> in the response, or all cookies if <name> is not specified. This is
11551 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011552
Willy Tarreau74ca5042013-06-11 23:12:07 +020011553res.cook_val([<name>]) : integer
11554scook_val([<name>]) : integer (deprecated)
11555 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11556 header line from the response, and converts its value to an integer which is
11557 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011558
Willy Tarreau74ca5042013-06-11 23:12:07 +020011559res.fhdr([<name>[,<occ>]]) : string
11560 This extracts the last occurrence of header <name> in an HTTP response, or of
11561 the last header if no <name> is specified. Optionally, a specific occurrence
11562 might be specified as a position number. Positive values indicate a position
11563 from the first occurrence, with 1 being the first one. Negative values
11564 indicate positions relative to the last one, with -1 being the last one. It
11565 differs from res.hdr() in that any commas present in the value are returned
11566 and are not used as delimiters. If this is not desired, the res.hdr() fetch
11567 should be used instead. This is sometimes useful with headers such as Date or
11568 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011569
Willy Tarreau74ca5042013-06-11 23:12:07 +020011570res.fhdr_cnt([<name>]) : integer
11571 Returns an integer value representing the number of occurrences of response
11572 header field name <name>, or the total number of header fields if <name> is
11573 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
11574 the number of full line headers and does not stop on commas. If this is not
11575 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011576
Willy Tarreau74ca5042013-06-11 23:12:07 +020011577res.hdr([<name>[,<occ>]]) : string
11578shdr([<name>[,<occ>]]) : string (deprecated)
11579 This extracts the last occurrence of header <name> in an HTTP response, or of
11580 the last header if no <name> is specified. Optionally, a specific occurrence
11581 might be specified as a position number. Positive values indicate a position
11582 from the first occurrence, with 1 being the first one. Negative values
11583 indicate positions relative to the last one, with -1 being the last one. This
11584 can be useful to learn some data into a stick-table. The function considers
11585 any comma as a delimiter for distinct values. If this is not desired, the
11586 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011587
Willy Tarreau74ca5042013-06-11 23:12:07 +020011588 ACL derivatives :
11589 shdr([<name>[,<occ>]]) : exact string match
11590 shdr_beg([<name>[,<occ>]]) : prefix match
11591 shdr_dir([<name>[,<occ>]]) : subdir match
11592 shdr_dom([<name>[,<occ>]]) : domain match
11593 shdr_end([<name>[,<occ>]]) : suffix match
11594 shdr_len([<name>[,<occ>]]) : length match
11595 shdr_reg([<name>[,<occ>]]) : regex match
11596 shdr_sub([<name>[,<occ>]]) : substring match
11597
11598res.hdr_cnt([<name>]) : integer
11599shdr_cnt([<name>]) : integer (deprecated)
11600 Returns an integer value representing the number of occurrences of response
11601 header field name <name>, or the total number of header fields if <name> is
11602 not specified. The function considers any comma as a delimiter for distinct
11603 values. If this is not desired, the res.fhdr_cnt() fetch should be used
11604 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011605
Willy Tarreau74ca5042013-06-11 23:12:07 +020011606res.hdr_ip([<name>[,<occ>]]) : ip
11607shdr_ip([<name>[,<occ>]]) : ip (deprecated)
11608 This extracts the last occurrence of header <name> in an HTTP response,
11609 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
11610 specific occurrence might be specified as a position number. Positive values
11611 indicate a position from the first occurrence, with 1 being the first one.
11612 Negative values indicate positions relative to the last one, with -1 being
11613 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011614
Willy Tarreau74ca5042013-06-11 23:12:07 +020011615res.hdr_val([<name>[,<occ>]]) : integer
11616shdr_val([<name>[,<occ>]]) : integer (deprecated)
11617 This extracts the last occurrence of header <name> in an HTTP response, and
11618 converts it to an integer value. Optionally, a specific occurrence might be
11619 specified as a position number. Positive values indicate a position from the
11620 first occurrence, with 1 being the first one. Negative values indicate
11621 positions relative to the last one, with -1 being the last one. This can be
11622 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011623
Willy Tarreau74ca5042013-06-11 23:12:07 +020011624res.ver : string
11625resp_ver : string (deprecated)
11626 Returns the version string from the HTTP response, for example "1.1". This
11627 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020011628
Willy Tarreau74ca5042013-06-11 23:12:07 +020011629 ACL derivatives :
11630 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011631
Willy Tarreau74ca5042013-06-11 23:12:07 +020011632set-cookie([<name>]) : string (deprecated)
11633 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11634 header line from the response and uses the corresponding value to match. This
11635 can be comparable to what "appsession" does with default options, but with
11636 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011637
Willy Tarreau74ca5042013-06-11 23:12:07 +020011638 This fetch function is deprecated and has been superseded by the "res.cook"
11639 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011640
Willy Tarreau74ca5042013-06-11 23:12:07 +020011641 See also : "appsession"
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011642
Willy Tarreau74ca5042013-06-11 23:12:07 +020011643status : integer
11644 Returns an integer containing the HTTP status code in the HTTP response, for
11645 example, 302. It is mostly used within ACLs and integer ranges, for example,
11646 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011647
Willy Tarreau74ca5042013-06-11 23:12:07 +020011648url : string
11649 This extracts the request's URL as presented in the request. A typical use is
11650 with prefetch-capable caches, and with portals which need to aggregate
11651 multiple information from databases and keep them in caches. With ACLs, using
11652 "path" is preferred over using "url", because clients may send a full URL as
11653 is normally done with proxies. The only real use is to match "*" which does
11654 not match in "path", and for which there is already a predefined ACL. See
11655 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011656
Willy Tarreau74ca5042013-06-11 23:12:07 +020011657 ACL derivatives :
11658 url : exact string match
11659 url_beg : prefix match
11660 url_dir : subdir match
11661 url_dom : domain match
11662 url_end : suffix match
11663 url_len : length match
11664 url_reg : regex match
11665 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011666
Willy Tarreau74ca5042013-06-11 23:12:07 +020011667url_ip : ip
11668 This extracts the IP address from the request's URL when the host part is
11669 presented as an IP address. Its use is very limited. For instance, a
11670 monitoring system might use this field as an alternative for the source IP in
11671 order to test what path a given source address would follow, or to force an
11672 entry in a table for a given source address. With ACLs it can be used to
11673 restrict access to certain systems through a proxy, for example when combined
11674 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011675
Willy Tarreau74ca5042013-06-11 23:12:07 +020011676url_port : integer
11677 This extracts the port part from the request's URL. Note that if the port is
11678 not specified in the request, port 80 is assumed. With ACLs it can be used to
11679 restrict access to certain systems through a proxy, for example when combined
11680 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011681
Willy Tarreau74ca5042013-06-11 23:12:07 +020011682urlp(<name>[,<delim>]) : string
11683url_param(<name>[,<delim>]) : string
11684 This extracts the first occurrence of the parameter <name> in the query
11685 string, which begins after either '?' or <delim>, and which ends before '&',
11686 ';' or <delim>. The parameter name is case-sensitive. The result is a string
11687 corresponding to the value of the parameter <name> as presented in the
11688 request (no URL decoding is performed). This can be used for session
11689 stickiness based on a client ID, to extract an application cookie passed as a
11690 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
11691 this fetch do not iterate over multiple parameters and stop at the first one
11692 as well.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011693
Willy Tarreau74ca5042013-06-11 23:12:07 +020011694 ACL derivatives :
11695 urlp(<name>[,<delim>]) : exact string match
11696 urlp_beg(<name>[,<delim>]) : prefix match
11697 urlp_dir(<name>[,<delim>]) : subdir match
11698 urlp_dom(<name>[,<delim>]) : domain match
11699 urlp_end(<name>[,<delim>]) : suffix match
11700 urlp_len(<name>[,<delim>]) : length match
11701 urlp_reg(<name>[,<delim>]) : regex match
11702 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011703
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011704
Willy Tarreau74ca5042013-06-11 23:12:07 +020011705 Example :
11706 # match http://example.com/foo?PHPSESSIONID=some_id
11707 stick on urlp(PHPSESSIONID)
11708 # match http://example.com/foo;JSESSIONID=some_id
11709 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011710
Willy Tarreau74ca5042013-06-11 23:12:07 +020011711urlp_val(<name>[,<delim>]) : integer
11712 See "urlp" above. This one extracts the URL parameter <name> in the request
11713 and converts it to an integer value. This can be used for session stickiness
11714 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020011715
Willy Tarreau198a7442008-01-17 12:05:32 +010011716
Willy Tarreau74ca5042013-06-11 23:12:07 +0200117177.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011718---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011719
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011720Some predefined ACLs are hard-coded so that they do not have to be declared in
11721every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020011722order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010011723
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011724ACL name Equivalent to Usage
11725---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011726FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020011727HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011728HTTP_1.0 req_ver 1.0 match HTTP version 1.0
11729HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010011730HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
11731HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
11732HTTP_URL_SLASH url_beg / match URL beginning with "/"
11733HTTP_URL_STAR url * match URL equal to "*"
11734LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011735METH_CONNECT method CONNECT match HTTP CONNECT method
11736METH_GET method GET HEAD match HTTP GET or HEAD method
11737METH_HEAD method HEAD match HTTP HEAD method
11738METH_OPTIONS method OPTIONS match HTTP OPTIONS method
11739METH_POST method POST match HTTP POST method
11740METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020011741RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011742REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010011743TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011744WAIT_END wait_end wait for end of content analysis
11745---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011746
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011747
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117488. Logging
11749----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010011750
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011751One of HAProxy's strong points certainly lies is its precise logs. It probably
11752provides the finest level of information available for such a product, which is
11753very important for troubleshooting complex environments. Standard information
11754provided in logs include client ports, TCP/HTTP state timers, precise session
11755state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011756to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011757headers.
11758
11759In order to improve administrators reactivity, it offers a great transparency
11760about encountered problems, both internal and external, and it is possible to
11761send logs to different sources at the same time with different level filters :
11762
11763 - global process-level logs (system errors, start/stop, etc..)
11764 - per-instance system and internal errors (lack of resource, bugs, ...)
11765 - per-instance external troubles (servers up/down, max connections)
11766 - per-instance activity (client connections), either at the establishment or
11767 at the termination.
11768
11769The ability to distribute different levels of logs to different log servers
11770allow several production teams to interact and to fix their problems as soon
11771as possible. For example, the system team might monitor system-wide errors,
11772while the application team might be monitoring the up/down for their servers in
11773real time, and the security team might analyze the activity logs with one hour
11774delay.
11775
11776
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117778.1. Log levels
11778---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011779
Simon Hormandf791f52011-05-29 15:01:10 +090011780TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011781source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090011782HTTP request, HTTP return code, number of bytes transmitted, conditions
11783in which the session ended, and even exchanged cookies values. For example
11784track a particular user's problems. All messages may be sent to up to two
11785syslog servers. Check the "log" keyword in section 4.2 for more information
11786about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011787
11788
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117898.2. Log formats
11790----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011791
William Lallemand48940402012-01-30 16:47:22 +010011792HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090011793and will be detailed in the following sections. A few of them may vary
11794slightly with the configuration, due to indicators specific to certain
11795options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011796
11797 - the default format, which is very basic and very rarely used. It only
11798 provides very basic information about the incoming connection at the moment
11799 it is accepted : source IP:port, destination IP:port, and frontend-name.
11800 This mode will eventually disappear so it will not be described to great
11801 extents.
11802
11803 - the TCP format, which is more advanced. This format is enabled when "option
11804 tcplog" is set on the frontend. HAProxy will then usually wait for the
11805 connection to terminate before logging. This format provides much richer
11806 information, such as timers, connection counts, queue size, etc... This
11807 format is recommended for pure TCP proxies.
11808
11809 - the HTTP format, which is the most advanced for HTTP proxying. This format
11810 is enabled when "option httplog" is set on the frontend. It provides the
11811 same information as the TCP format with some HTTP-specific fields such as
11812 the request, the status code, and captures of headers and cookies. This
11813 format is recommended for HTTP proxies.
11814
Emeric Brun3a058f32009-06-30 18:26:00 +020011815 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
11816 fields arranged in the same order as the CLF format. In this mode, all
11817 timers, captures, flags, etc... appear one per field after the end of the
11818 common fields, in the same order they appear in the standard HTTP format.
11819
William Lallemand48940402012-01-30 16:47:22 +010011820 - the custom log format, allows you to make your own log line.
11821
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011822Next sections will go deeper into details for each of these formats. Format
11823specification will be performed on a "field" basis. Unless stated otherwise, a
11824field is a portion of text delimited by any number of spaces. Since syslog
11825servers are susceptible of inserting fields at the beginning of a line, it is
11826always assumed that the first field is the one containing the process name and
11827identifier.
11828
11829Note : Since log lines may be quite long, the log examples in sections below
11830 might be broken into multiple lines. The example log lines will be
11831 prefixed with 3 closing angle brackets ('>>>') and each time a log is
11832 broken into multiple lines, each non-final line will end with a
11833 backslash ('\') and the next line will start indented by two characters.
11834
11835
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118368.2.1. Default log format
11837-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011838
11839This format is used when no specific option is set. The log is emitted as soon
11840as the connection is accepted. One should note that this currently is the only
11841format which logs the request's destination IP and ports.
11842
11843 Example :
11844 listen www
11845 mode http
11846 log global
11847 server srv1 127.0.0.1:8000
11848
11849 >>> Feb 6 12:12:09 localhost \
11850 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
11851 (www/HTTP)
11852
11853 Field Format Extract from the example above
11854 1 process_name '[' pid ']:' haproxy[14385]:
11855 2 'Connect from' Connect from
11856 3 source_ip ':' source_port 10.0.1.2:33312
11857 4 'to' to
11858 5 destination_ip ':' destination_port 10.0.3.31:8012
11859 6 '(' frontend_name '/' mode ')' (www/HTTP)
11860
11861Detailed fields description :
11862 - "source_ip" is the IP address of the client which initiated the connection.
11863 - "source_port" is the TCP port of the client which initiated the connection.
11864 - "destination_ip" is the IP address the client connected to.
11865 - "destination_port" is the TCP port the client connected to.
11866 - "frontend_name" is the name of the frontend (or listener) which received
11867 and processed the connection.
11868 - "mode is the mode the frontend is operating (TCP or HTTP).
11869
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011870In case of a UNIX socket, the source and destination addresses are marked as
11871"unix:" and the ports reflect the internal ID of the socket which accepted the
11872connection (the same ID as reported in the stats).
11873
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011874It is advised not to use this deprecated format for newer installations as it
11875will eventually disappear.
11876
11877
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118788.2.2. TCP log format
11879---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011880
11881The TCP format is used when "option tcplog" is specified in the frontend, and
11882is the recommended format for pure TCP proxies. It provides a lot of precious
11883information for troubleshooting. Since this format includes timers and byte
11884counts, the log is normally emitted at the end of the session. It can be
11885emitted earlier if "option logasap" is specified, which makes sense in most
11886environments with long sessions such as remote terminals. Sessions which match
11887the "monitor" rules are never logged. It is also possible not to emit logs for
11888sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020011889specifying "option dontlognull" in the frontend. Successful connections will
11890not be logged if "option dontlog-normal" is specified in the frontend. A few
11891fields may slightly vary depending on some configuration options, those are
11892marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011893
11894 Example :
11895 frontend fnt
11896 mode tcp
11897 option tcplog
11898 log global
11899 default_backend bck
11900
11901 backend bck
11902 server srv1 127.0.0.1:8000
11903
11904 >>> Feb 6 12:12:56 localhost \
11905 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
11906 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
11907
11908 Field Format Extract from the example above
11909 1 process_name '[' pid ']:' haproxy[14387]:
11910 2 client_ip ':' client_port 10.0.1.2:33313
11911 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
11912 4 frontend_name fnt
11913 5 backend_name '/' server_name bck/srv1
11914 6 Tw '/' Tc '/' Tt* 0/0/5007
11915 7 bytes_read* 212
11916 8 termination_state --
11917 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
11918 10 srv_queue '/' backend_queue 0/0
11919
11920Detailed fields description :
11921 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011922 connection to haproxy. If the connection was accepted on a UNIX socket
11923 instead, the IP address would be replaced with the word "unix". Note that
11924 when the connection is accepted on a socket configured with "accept-proxy"
11925 and the PROXY protocol is correctly used, then the logs will reflect the
11926 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011927
11928 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011929 If the connection was accepted on a UNIX socket instead, the port would be
11930 replaced with the ID of the accepting socket, which is also reported in the
11931 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011932
11933 - "accept_date" is the exact date when the connection was received by haproxy
11934 (which might be very slightly different from the date observed on the
11935 network if there was some queuing in the system's backlog). This is usually
11936 the same date which may appear in any upstream firewall's log.
11937
11938 - "frontend_name" is the name of the frontend (or listener) which received
11939 and processed the connection.
11940
11941 - "backend_name" is the name of the backend (or listener) which was selected
11942 to manage the connection to the server. This will be the same as the
11943 frontend if no switching rule has been applied, which is common for TCP
11944 applications.
11945
11946 - "server_name" is the name of the last server to which the connection was
11947 sent, which might differ from the first one if there were connection errors
11948 and a redispatch occurred. Note that this server belongs to the backend
11949 which processed the request. If the connection was aborted before reaching
11950 a server, "<NOSRV>" is indicated instead of a server name.
11951
11952 - "Tw" is the total time in milliseconds spent waiting in the various queues.
11953 It can be "-1" if the connection was aborted before reaching the queue.
11954 See "Timers" below for more details.
11955
11956 - "Tc" is the total time in milliseconds spent waiting for the connection to
11957 establish to the final server, including retries. It can be "-1" if the
11958 connection was aborted before a connection could be established. See
11959 "Timers" below for more details.
11960
11961 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011962 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011963 "option logasap" was specified, then the time counting stops at the moment
11964 the log is emitted. In this case, a '+' sign is prepended before the value,
11965 indicating that the final one will be larger. See "Timers" below for more
11966 details.
11967
11968 - "bytes_read" is the total number of bytes transmitted from the server to
11969 the client when the log is emitted. If "option logasap" is specified, the
11970 this value will be prefixed with a '+' sign indicating that the final one
11971 may be larger. Please note that this value is a 64-bit counter, so log
11972 analysis tools must be able to handle it without overflowing.
11973
11974 - "termination_state" is the condition the session was in when the session
11975 ended. This indicates the session state, which side caused the end of
11976 session to happen, and for what reason (timeout, error, ...). The normal
11977 flags should be "--", indicating the session was closed by either end with
11978 no data remaining in buffers. See below "Session state at disconnection"
11979 for more details.
11980
11981 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011982 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011983 limits have been reached. For instance, if actconn is close to 512 when
11984 multiple connection errors occur, chances are high that the system limits
11985 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011986 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011987
11988 - "feconn" is the total number of concurrent connections on the frontend when
11989 the session was logged. It is useful to estimate the amount of resource
11990 required to sustain high loads, and to detect when the frontend's "maxconn"
11991 has been reached. Most often when this value increases by huge jumps, it is
11992 because there is congestion on the backend servers, but sometimes it can be
11993 caused by a denial of service attack.
11994
11995 - "beconn" is the total number of concurrent connections handled by the
11996 backend when the session was logged. It includes the total number of
11997 concurrent connections active on servers as well as the number of
11998 connections pending in queues. It is useful to estimate the amount of
11999 additional servers needed to support high loads for a given application.
12000 Most often when this value increases by huge jumps, it is because there is
12001 congestion on the backend servers, but sometimes it can be caused by a
12002 denial of service attack.
12003
12004 - "srv_conn" is the total number of concurrent connections still active on
12005 the server when the session was logged. It can never exceed the server's
12006 configured "maxconn" parameter. If this value is very often close or equal
12007 to the server's "maxconn", it means that traffic regulation is involved a
12008 lot, meaning that either the server's maxconn value is too low, or that
12009 there aren't enough servers to process the load with an optimal response
12010 time. When only one of the server's "srv_conn" is high, it usually means
12011 that this server has some trouble causing the connections to take longer to
12012 be processed than on other servers.
12013
12014 - "retries" is the number of connection retries experienced by this session
12015 when trying to connect to the server. It must normally be zero, unless a
12016 server is being stopped at the same moment the connection was attempted.
12017 Frequent retries generally indicate either a network problem between
12018 haproxy and the server, or a misconfigured system backlog on the server
12019 preventing new connections from being queued. This field may optionally be
12020 prefixed with a '+' sign, indicating that the session has experienced a
12021 redispatch after the maximal retry count has been reached on the initial
12022 server. In this case, the server name appearing in the log is the one the
12023 connection was redispatched to, and not the first one, though both may
12024 sometimes be the same in case of hashing for instance. So as a general rule
12025 of thumb, when a '+' is present in front of the retry count, this count
12026 should not be attributed to the logged server.
12027
12028 - "srv_queue" is the total number of requests which were processed before
12029 this one in the server queue. It is zero when the request has not gone
12030 through the server queue. It makes it possible to estimate the approximate
12031 server's response time by dividing the time spent in queue by the number of
12032 requests in the queue. It is worth noting that if a session experiences a
12033 redispatch and passes through two server queues, their positions will be
12034 cumulated. A request should not pass through both the server queue and the
12035 backend queue unless a redispatch occurs.
12036
12037 - "backend_queue" is the total number of requests which were processed before
12038 this one in the backend's global queue. It is zero when the request has not
12039 gone through the global queue. It makes it possible to estimate the average
12040 queue length, which easily translates into a number of missing servers when
12041 divided by a server's "maxconn" parameter. It is worth noting that if a
12042 session experiences a redispatch, it may pass twice in the backend's queue,
12043 and then both positions will be cumulated. A request should not pass
12044 through both the server queue and the backend queue unless a redispatch
12045 occurs.
12046
12047
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200120488.2.3. HTTP log format
12049----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012050
12051The HTTP format is the most complete and the best suited for HTTP proxies. It
12052is enabled by when "option httplog" is specified in the frontend. It provides
12053the same level of information as the TCP format with additional features which
12054are specific to the HTTP protocol. Just like the TCP format, the log is usually
12055emitted at the end of the session, unless "option logasap" is specified, which
12056generally only makes sense for download sites. A session which matches the
12057"monitor" rules will never logged. It is also possible not to log sessions for
12058which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012059frontend. Successful connections will not be logged if "option dontlog-normal"
12060is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012061
12062Most fields are shared with the TCP log, some being different. A few fields may
12063slightly vary depending on some configuration options. Those ones are marked
12064with a star ('*') after the field name below.
12065
12066 Example :
12067 frontend http-in
12068 mode http
12069 option httplog
12070 log global
12071 default_backend bck
12072
12073 backend static
12074 server srv1 127.0.0.1:8000
12075
12076 >>> Feb 6 12:14:14 localhost \
12077 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
12078 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 +010012079 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012080
12081 Field Format Extract from the example above
12082 1 process_name '[' pid ']:' haproxy[14389]:
12083 2 client_ip ':' client_port 10.0.1.2:33317
12084 3 '[' accept_date ']' [06/Feb/2009:12:14:14.655]
12085 4 frontend_name http-in
12086 5 backend_name '/' server_name static/srv1
12087 6 Tq '/' Tw '/' Tc '/' Tr '/' Tt* 10/0/30/69/109
12088 7 status_code 200
12089 8 bytes_read* 2750
12090 9 captured_request_cookie -
12091 10 captured_response_cookie -
12092 11 termination_state ----
12093 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
12094 13 srv_queue '/' backend_queue 0/0
12095 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
12096 15 '{' captured_response_headers* '}' {}
12097 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010012098
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012099
12100Detailed fields description :
12101 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012102 connection to haproxy. If the connection was accepted on a UNIX socket
12103 instead, the IP address would be replaced with the word "unix". Note that
12104 when the connection is accepted on a socket configured with "accept-proxy"
12105 and the PROXY protocol is correctly used, then the logs will reflect the
12106 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012107
12108 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012109 If the connection was accepted on a UNIX socket instead, the port would be
12110 replaced with the ID of the accepting socket, which is also reported in the
12111 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012112
12113 - "accept_date" is the exact date when the TCP connection was received by
12114 haproxy (which might be very slightly different from the date observed on
12115 the network if there was some queuing in the system's backlog). This is
12116 usually the same date which may appear in any upstream firewall's log. This
12117 does not depend on the fact that the client has sent the request or not.
12118
12119 - "frontend_name" is the name of the frontend (or listener) which received
12120 and processed the connection.
12121
12122 - "backend_name" is the name of the backend (or listener) which was selected
12123 to manage the connection to the server. This will be the same as the
12124 frontend if no switching rule has been applied.
12125
12126 - "server_name" is the name of the last server to which the connection was
12127 sent, which might differ from the first one if there were connection errors
12128 and a redispatch occurred. Note that this server belongs to the backend
12129 which processed the request. If the request was aborted before reaching a
12130 server, "<NOSRV>" is indicated instead of a server name. If the request was
12131 intercepted by the stats subsystem, "<STATS>" is indicated instead.
12132
12133 - "Tq" is the total time in milliseconds spent waiting for the client to send
12134 a full HTTP request, not counting data. It can be "-1" if the connection
12135 was aborted before a complete request could be received. It should always
12136 be very small because a request generally fits in one single packet. Large
12137 times here generally indicate network trouble between the client and
12138 haproxy. See "Timers" below for more details.
12139
12140 - "Tw" is the total time in milliseconds spent waiting in the various queues.
12141 It can be "-1" if the connection was aborted before reaching the queue.
12142 See "Timers" below for more details.
12143
12144 - "Tc" is the total time in milliseconds spent waiting for the connection to
12145 establish to the final server, including retries. It can be "-1" if the
12146 request was aborted before a connection could be established. See "Timers"
12147 below for more details.
12148
12149 - "Tr" is the total time in milliseconds spent waiting for the server to send
12150 a full HTTP response, not counting data. It can be "-1" if the request was
12151 aborted before a complete response could be received. It generally matches
12152 the server's processing time for the request, though it may be altered by
12153 the amount of data sent by the client to the server. Large times here on
12154 "GET" requests generally indicate an overloaded server. See "Timers" below
12155 for more details.
12156
12157 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012158 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012159 "option logasap" was specified, then the time counting stops at the moment
12160 the log is emitted. In this case, a '+' sign is prepended before the value,
12161 indicating that the final one will be larger. See "Timers" below for more
12162 details.
12163
12164 - "status_code" is the HTTP status code returned to the client. This status
12165 is generally set by the server, but it might also be set by haproxy when
12166 the server cannot be reached or when its response is blocked by haproxy.
12167
12168 - "bytes_read" is the total number of bytes transmitted to the client when
12169 the log is emitted. This does include HTTP headers. If "option logasap" is
12170 specified, the this value will be prefixed with a '+' sign indicating that
12171 the final one may be larger. Please note that this value is a 64-bit
12172 counter, so log analysis tools must be able to handle it without
12173 overflowing.
12174
12175 - "captured_request_cookie" is an optional "name=value" entry indicating that
12176 the client had this cookie in the request. The cookie name and its maximum
12177 length are defined by the "capture cookie" statement in the frontend
12178 configuration. The field is a single dash ('-') when the option is not
12179 set. Only one cookie may be captured, it is generally used to track session
12180 ID exchanges between a client and a server to detect session crossing
12181 between clients due to application bugs. For more details, please consult
12182 the section "Capturing HTTP headers and cookies" below.
12183
12184 - "captured_response_cookie" is an optional "name=value" entry indicating
12185 that the server has returned a cookie with its response. The cookie name
12186 and its maximum length are defined by the "capture cookie" statement in the
12187 frontend configuration. The field is a single dash ('-') when the option is
12188 not set. Only one cookie may be captured, it is generally used to track
12189 session ID exchanges between a client and a server to detect session
12190 crossing between clients due to application bugs. For more details, please
12191 consult the section "Capturing HTTP headers and cookies" below.
12192
12193 - "termination_state" is the condition the session was in when the session
12194 ended. This indicates the session state, which side caused the end of
12195 session to happen, for what reason (timeout, error, ...), just like in TCP
12196 logs, and information about persistence operations on cookies in the last
12197 two characters. The normal flags should begin with "--", indicating the
12198 session was closed by either end with no data remaining in buffers. See
12199 below "Session state at disconnection" for more details.
12200
12201 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012202 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012203 limits have been reached. For instance, if actconn is close to 512 or 1024
12204 when multiple connection errors occur, chances are high that the system
12205 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012206 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012207 system.
12208
12209 - "feconn" is the total number of concurrent connections on the frontend when
12210 the session was logged. It is useful to estimate the amount of resource
12211 required to sustain high loads, and to detect when the frontend's "maxconn"
12212 has been reached. Most often when this value increases by huge jumps, it is
12213 because there is congestion on the backend servers, but sometimes it can be
12214 caused by a denial of service attack.
12215
12216 - "beconn" is the total number of concurrent connections handled by the
12217 backend when the session was logged. It includes the total number of
12218 concurrent connections active on servers as well as the number of
12219 connections pending in queues. It is useful to estimate the amount of
12220 additional servers needed to support high loads for a given application.
12221 Most often when this value increases by huge jumps, it is because there is
12222 congestion on the backend servers, but sometimes it can be caused by a
12223 denial of service attack.
12224
12225 - "srv_conn" is the total number of concurrent connections still active on
12226 the server when the session was logged. It can never exceed the server's
12227 configured "maxconn" parameter. If this value is very often close or equal
12228 to the server's "maxconn", it means that traffic regulation is involved a
12229 lot, meaning that either the server's maxconn value is too low, or that
12230 there aren't enough servers to process the load with an optimal response
12231 time. When only one of the server's "srv_conn" is high, it usually means
12232 that this server has some trouble causing the requests to take longer to be
12233 processed than on other servers.
12234
12235 - "retries" is the number of connection retries experienced by this session
12236 when trying to connect to the server. It must normally be zero, unless a
12237 server is being stopped at the same moment the connection was attempted.
12238 Frequent retries generally indicate either a network problem between
12239 haproxy and the server, or a misconfigured system backlog on the server
12240 preventing new connections from being queued. This field may optionally be
12241 prefixed with a '+' sign, indicating that the session has experienced a
12242 redispatch after the maximal retry count has been reached on the initial
12243 server. In this case, the server name appearing in the log is the one the
12244 connection was redispatched to, and not the first one, though both may
12245 sometimes be the same in case of hashing for instance. So as a general rule
12246 of thumb, when a '+' is present in front of the retry count, this count
12247 should not be attributed to the logged server.
12248
12249 - "srv_queue" is the total number of requests which were processed before
12250 this one in the server queue. It is zero when the request has not gone
12251 through the server queue. It makes it possible to estimate the approximate
12252 server's response time by dividing the time spent in queue by the number of
12253 requests in the queue. It is worth noting that if a session experiences a
12254 redispatch and passes through two server queues, their positions will be
12255 cumulated. A request should not pass through both the server queue and the
12256 backend queue unless a redispatch occurs.
12257
12258 - "backend_queue" is the total number of requests which were processed before
12259 this one in the backend's global queue. It is zero when the request has not
12260 gone through the global queue. It makes it possible to estimate the average
12261 queue length, which easily translates into a number of missing servers when
12262 divided by a server's "maxconn" parameter. It is worth noting that if a
12263 session experiences a redispatch, it may pass twice in the backend's queue,
12264 and then both positions will be cumulated. A request should not pass
12265 through both the server queue and the backend queue unless a redispatch
12266 occurs.
12267
12268 - "captured_request_headers" is a list of headers captured in the request due
12269 to the presence of the "capture request header" statement in the frontend.
12270 Multiple headers can be captured, they will be delimited by a vertical bar
12271 ('|'). When no capture is enabled, the braces do not appear, causing a
12272 shift of remaining fields. It is important to note that this field may
12273 contain spaces, and that using it requires a smarter log parser than when
12274 it's not used. Please consult the section "Capturing HTTP headers and
12275 cookies" below for more details.
12276
12277 - "captured_response_headers" is a list of headers captured in the response
12278 due to the presence of the "capture response header" statement in the
12279 frontend. Multiple headers can be captured, they will be delimited by a
12280 vertical bar ('|'). When no capture is enabled, the braces do not appear,
12281 causing a shift of remaining fields. It is important to note that this
12282 field may contain spaces, and that using it requires a smarter log parser
12283 than when it's not used. Please consult the section "Capturing HTTP headers
12284 and cookies" below for more details.
12285
12286 - "http_request" is the complete HTTP request line, including the method,
12287 request and HTTP version string. Non-printable characters are encoded (see
12288 below the section "Non-printable characters"). This is always the last
12289 field, and it is always delimited by quotes and is the only one which can
12290 contain quotes. If new fields are added to the log format, they will be
12291 added before this field. This field might be truncated if the request is
12292 huge and does not fit in the standard syslog buffer (1024 characters). This
12293 is the reason why this field must always remain the last one.
12294
12295
Cyril Bontédc4d9032012-04-08 21:57:39 +0200122968.2.4. Custom log format
12297------------------------
William Lallemand48940402012-01-30 16:47:22 +010012298
Willy Tarreau2beef582012-12-20 17:22:52 +010012299The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010012300mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010012301
12302HAproxy understands some log format variables. % precedes log format variables.
12303Variables can take arguments using braces ('{}'), and multiple arguments are
12304separated by commas within the braces. Flags may be added or removed by
12305prefixing them with a '+' or '-' sign.
12306
12307Special variable "%o" may be used to propagate its flags to all other
12308variables on the same format string. This is particularly handy with quoted
12309string formats ("Q").
12310
Willy Tarreauc8368452012-12-21 00:09:23 +010012311If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020012312as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010012313less common information such as the client's SSL certificate's DN, or to log
12314the key that would be used to store an entry into a stick table.
12315
William Lallemand48940402012-01-30 16:47:22 +010012316Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012317In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010012318in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010012319
12320Flags are :
12321 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012322 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
William Lallemand48940402012-01-30 16:47:22 +010012323
12324 Example:
12325
12326 log-format %T\ %t\ Some\ Text
12327 log-format %{+Q}o\ %t\ %s\ %{-Q}r
12328
12329At the moment, the default HTTP format is defined this way :
12330
Willy Tarreau2beef582012-12-20 17:22:52 +010012331 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tq/%Tw/%Tc/%Tr/%Tt\ %ST\ %B\ %CC\ \
12332 %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r
William Lallemand48940402012-01-30 16:47:22 +010012333
William Lallemandbddd4fd2012-02-27 11:23:10 +010012334the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010012335
Willy Tarreau2beef582012-12-20 17:22:52 +010012336 log-format %{+Q}o\ %{-Q}ci\ -\ -\ [%T]\ %r\ %ST\ %B\ \"\"\ \"\"\ %cp\ \
Willy Tarreau773d65f2012-10-12 14:56:11 +020012337 %ms\ %ft\ %b\ %s\ \%Tq\ %Tw\ %Tc\ %Tr\ %Tt\ %tsc\ %ac\ %fc\ \
Willy Tarreau2beef582012-12-20 17:22:52 +010012338 %bc\ %sc\ %rc\ %sq\ %bq\ %CC\ %CS\ \%hrl\ %hsl
William Lallemand48940402012-01-30 16:47:22 +010012339
William Lallemandbddd4fd2012-02-27 11:23:10 +010012340and the default TCP format is defined this way :
12341
Willy Tarreau2beef582012-12-20 17:22:52 +010012342 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ \
William Lallemandbddd4fd2012-02-27 11:23:10 +010012343 %ac/%fc/%bc/%sc/%rc\ %sq/%bq
12344
William Lallemand48940402012-01-30 16:47:22 +010012345Please refer to the table below for currently defined variables :
12346
William Lallemandbddd4fd2012-02-27 11:23:10 +010012347 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012348 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012349 +---+------+-----------------------------------------------+-------------+
12350 | | %o | special variable, apply flags on all next var | |
12351 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010012352 | | %B | bytes_read (from server to client) | numeric |
12353 | H | %CC | captured_request_cookie | string |
12354 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020012355 | | %H | hostname | string |
William Lallemanda73203e2012-03-12 12:48:57 +010012356 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020012357 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012358 | | %T | gmt_date_time | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012359 | | %Tc | Tc | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080012360 | | %Tl | local_date_time | date |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012361 | H | %Tq | Tq | numeric |
12362 | H | %Tr | Tr | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012363 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012364 | | %Tt | Tt | numeric |
12365 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012366 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012367 | | %ac | actconn | numeric |
12368 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012369 | | %bc | beconn (backend concurrent connections) | numeric |
12370 | | %bi | backend_source_ip (connecting address) | IP |
12371 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012372 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012373 | | %ci | client_ip (accepted address) | IP |
12374 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012375 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012376 | | %fc | feconn (frontend concurrent connections) | numeric |
12377 | | %fi | frontend_ip (accepting address) | IP |
12378 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020012379 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020012380 | | %hr | captured_request_headers default style | string |
12381 | | %hrl | captured_request_headers CLF style | string list |
12382 | | %hs | captured_response_headers default style | string |
12383 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreauc3453d52015-08-09 10:56:35 +020012384 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012385 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012386 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012387 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010012388 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012389 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012390 | | %sc | srv_conn (server concurrent connections) | numeric |
12391 | | %si | server_IP (target address) | IP |
12392 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012393 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012394 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
12395 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012396 | | %t | date_time (with millisecond resolution) | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012397 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012398 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012399 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010012400
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012401 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010012402
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010012403
124048.2.5. Error log format
12405-----------------------
12406
12407When an incoming connection fails due to an SSL handshake or an invalid PROXY
12408protocol header, haproxy will log the event using a shorter, fixed line format.
12409By default, logs are emitted at the LOG_INFO level, unless the option
12410"log-separate-errors" is set in the backend, in which case the LOG_ERR level
12411will be used. Connections on which no data are exchanged (eg: probes) are not
12412logged if the "dontlognull" option is set.
12413
12414The format looks like this :
12415
12416 >>> Dec 3 18:27:14 localhost \
12417 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
12418 Connection error during SSL handshake
12419
12420 Field Format Extract from the example above
12421 1 process_name '[' pid ']:' haproxy[6103]:
12422 2 client_ip ':' client_port 127.0.0.1:56059
12423 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
12424 4 frontend_name "/" bind_name ":" frt/f1:
12425 5 message Connection error during SSL handshake
12426
12427These fields just provide minimal information to help debugging connection
12428failures.
12429
12430
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124318.3. Advanced logging options
12432-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012433
12434Some advanced logging options are often looked for but are not easy to find out
12435just by looking at the various options. Here is an entry point for the few
12436options which can enable better logging. Please refer to the keywords reference
12437for more information about their usage.
12438
12439
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124408.3.1. Disabling logging of external tests
12441------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012442
12443It is quite common to have some monitoring tools perform health checks on
12444haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
12445commercial load-balancer, and sometimes it will simply be a more complete
12446monitoring system such as Nagios. When the tests are very frequent, users often
12447ask how to disable logging for those checks. There are three possibilities :
12448
12449 - if connections come from everywhere and are just TCP probes, it is often
12450 desired to simply disable logging of connections without data exchange, by
12451 setting "option dontlognull" in the frontend. It also disables logging of
12452 port scans, which may or may not be desired.
12453
12454 - if the connection come from a known source network, use "monitor-net" to
12455 declare this network as monitoring only. Any host in this network will then
12456 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012457 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012458 such as other load-balancers.
12459
12460 - if the tests are performed on a known URI, use "monitor-uri" to declare
12461 this URI as dedicated to monitoring. Any host sending this request will
12462 only get the result of a health-check, and the request will not be logged.
12463
12464
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124658.3.2. Logging before waiting for the session to terminate
12466----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012467
12468The problem with logging at end of connection is that you have no clue about
12469what is happening during very long sessions, such as remote terminal sessions
12470or large file downloads. This problem can be worked around by specifying
12471"option logasap" in the frontend. Haproxy will then log as soon as possible,
12472just before data transfer begins. This means that in case of TCP, it will still
12473log the connection status to the server, and in case of HTTP, it will log just
12474after processing the server headers. In this case, the number of bytes reported
12475is the number of header bytes sent to the client. In order to avoid confusion
12476with normal logs, the total time field and the number of bytes are prefixed
12477with a '+' sign which means that real numbers are certainly larger.
12478
12479
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124808.3.3. Raising log level upon errors
12481------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012482
12483Sometimes it is more convenient to separate normal traffic from errors logs,
12484for instance in order to ease error monitoring from log files. When the option
12485"log-separate-errors" is used, connections which experience errors, timeouts,
12486retries, redispatches or HTTP status codes 5xx will see their syslog level
12487raised from "info" to "err". This will help a syslog daemon store the log in
12488a separate file. It is very important to keep the errors in the normal traffic
12489file too, so that log ordering is not altered. You should also be careful if
12490you already have configured your syslog daemon to store all logs higher than
12491"notice" in an "admin" file, because the "err" level is higher than "notice".
12492
12493
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124948.3.4. Disabling logging of successful connections
12495--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012496
12497Although this may sound strange at first, some large sites have to deal with
12498multiple thousands of logs per second and are experiencing difficulties keeping
12499them intact for a long time or detecting errors within them. If the option
12500"dontlog-normal" is set on the frontend, all normal connections will not be
12501logged. In this regard, a normal connection is defined as one without any
12502error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
12503and a response with a status 5xx is not considered normal and will be logged
12504too. Of course, doing is is really discouraged as it will remove most of the
12505useful information from the logs. Do this only if you have no other
12506alternative.
12507
12508
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200125098.4. Timing events
12510------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012511
12512Timers provide a great help in troubleshooting network problems. All values are
12513reported in milliseconds (ms). These timers should be used in conjunction with
12514the session termination flags. In TCP mode with "option tcplog" set on the
12515frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
12516mode, 5 control points are reported under the form "Tq/Tw/Tc/Tr/Tt" :
12517
12518 - Tq: total time to get the client request (HTTP mode only). It's the time
12519 elapsed between the moment the client connection was accepted and the
12520 moment the proxy received the last HTTP header. The value "-1" indicates
12521 that the end of headers (empty line) has never been seen. This happens when
12522 the client closes prematurely or times out.
12523
12524 - Tw: total time spent in the queues waiting for a connection slot. It
12525 accounts for backend queue as well as the server queues, and depends on the
12526 queue size, and the time needed for the server to complete previous
12527 requests. The value "-1" means that the request was killed before reaching
12528 the queue, which is generally what happens with invalid or denied requests.
12529
12530 - Tc: total time to establish the TCP connection to the server. It's the time
12531 elapsed between the moment the proxy sent the connection request, and the
12532 moment it was acknowledged by the server, or between the TCP SYN packet and
12533 the matching SYN/ACK packet in return. The value "-1" means that the
12534 connection never established.
12535
12536 - Tr: server response time (HTTP mode only). It's the time elapsed between
12537 the moment the TCP connection was established to the server and the moment
12538 the server sent its complete response headers. It purely shows its request
12539 processing time, without the network overhead due to the data transmission.
12540 It is worth noting that when the client has data to send to the server, for
12541 instance during a POST request, the time already runs, and this can distort
12542 apparent response time. For this reason, it's generally wise not to trust
12543 too much this field for POST requests initiated from clients behind an
12544 untrusted network. A value of "-1" here means that the last the response
12545 header (empty line) was never seen, most likely because the server timeout
12546 stroke before the server managed to process the request.
12547
12548 - Tt: total session duration time, between the moment the proxy accepted it
12549 and the moment both ends were closed. The exception is when the "logasap"
12550 option is specified. In this case, it only equals (Tq+Tw+Tc+Tr), and is
12551 prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012552 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012553
12554 Td = Tt - (Tq + Tw + Tc + Tr)
12555
12556 Timers with "-1" values have to be excluded from this equation. In TCP
12557 mode, "Tq" and "Tr" have to be excluded too. Note that "Tt" can never be
12558 negative.
12559
12560These timers provide precious indications on trouble causes. Since the TCP
12561protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
12562that timers close to multiples of 3s are nearly always related to lost packets
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012563due to network problems (wires, negotiation, congestion). Moreover, if "Tt" is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012564close to a timeout value specified in the configuration, it often means that a
12565session has been aborted on timeout.
12566
12567Most common cases :
12568
12569 - If "Tq" is close to 3000, a packet has probably been lost between the
12570 client and the proxy. This is very rare on local networks but might happen
12571 when clients are on far remote networks and send large requests. It may
12572 happen that values larger than usual appear here without any network cause.
12573 Sometimes, during an attack or just after a resource starvation has ended,
12574 haproxy may accept thousands of connections in a few milliseconds. The time
12575 spent accepting these connections will inevitably slightly delay processing
12576 of other connections, and it can happen that request times in the order of
12577 a few tens of milliseconds are measured after a few thousands of new
Patrick Mezard105faca2010-06-12 17:02:46 +020012578 connections have been accepted at once. Setting "option http-server-close"
12579 may display larger request times since "Tq" also measures the time spent
12580 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012581
12582 - If "Tc" is close to 3000, a packet has probably been lost between the
12583 server and the proxy during the server connection phase. This value should
12584 always be very low, such as 1 ms on local networks and less than a few tens
12585 of ms on remote networks.
12586
Willy Tarreau55165fe2009-05-10 12:02:55 +020012587 - If "Tr" is nearly always lower than 3000 except some rare values which seem
12588 to be the average majored by 3000, there are probably some packets lost
12589 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012590
12591 - If "Tt" is large even for small byte counts, it generally is because
12592 neither the client nor the server decides to close the connection, for
12593 instance because both have agreed on a keep-alive connection mode. In order
12594 to solve this issue, it will be needed to specify "option httpclose" on
12595 either the frontend or the backend. If the problem persists, it means that
12596 the server ignores the "close" connection mode and expects the client to
12597 close. Then it will be required to use "option forceclose". Having the
12598 smallest possible 'Tt' is important when connection regulation is used with
12599 the "maxconn" option on the servers, since no new connection will be sent
12600 to the server until another one is released.
12601
12602Other noticeable HTTP log cases ('xx' means any value to be ignored) :
12603
12604 Tq/Tw/Tc/Tr/+Tt The "option logasap" is present on the frontend and the log
12605 was emitted before the data phase. All the timers are valid
12606 except "Tt" which is shorter than reality.
12607
12608 -1/xx/xx/xx/Tt The client was not able to send a complete request in time
12609 or it aborted too early. Check the session termination flags
12610 then "timeout http-request" and "timeout client" settings.
12611
12612 Tq/-1/xx/xx/Tt It was not possible to process the request, maybe because
12613 servers were out of order, because the request was invalid
12614 or forbidden by ACL rules. Check the session termination
12615 flags.
12616
12617 Tq/Tw/-1/xx/Tt The connection could not establish on the server. Either it
12618 actively refused it or it timed out after Tt-(Tq+Tw) ms.
12619 Check the session termination flags, then check the
12620 "timeout connect" setting. Note that the tarpit action might
12621 return similar-looking patterns, with "Tw" equal to the time
12622 the client connection was maintained open.
12623
12624 Tq/Tw/Tc/-1/Tt The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012625 a complete response in time, or it closed its connection
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012626 unexpectedly after Tt-(Tq+Tw+Tc) ms. Check the session
12627 termination flags, then check the "timeout server" setting.
12628
12629
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126308.5. Session state at disconnection
12631-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012632
12633TCP and HTTP logs provide a session termination indicator in the
12634"termination_state" field, just before the number of active connections. It is
126352-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
12636each of which has a special meaning :
12637
12638 - On the first character, a code reporting the first event which caused the
12639 session to terminate :
12640
12641 C : the TCP session was unexpectedly aborted by the client.
12642
12643 S : the TCP session was unexpectedly aborted by the server, or the
12644 server explicitly refused it.
12645
12646 P : the session was prematurely aborted by the proxy, because of a
12647 connection limit enforcement, because a DENY filter was matched,
12648 because of a security check which detected and blocked a dangerous
12649 error in server response which might have caused information leak
Willy Tarreau570f2212013-06-10 16:42:09 +020012650 (eg: cacheable cookie).
12651
12652 L : the session was locally processed by haproxy and was not passed to
12653 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012654
12655 R : a resource on the proxy has been exhausted (memory, sockets, source
12656 ports, ...). Usually, this appears during the connection phase, and
12657 system logs should contain a copy of the precise error. If this
12658 happens, it must be considered as a very serious anomaly which
12659 should be fixed as soon as possible by any means.
12660
12661 I : an internal error was identified by the proxy during a self-check.
12662 This should NEVER happen, and you are encouraged to report any log
12663 containing this, because this would almost certainly be a bug. It
12664 would be wise to preventively restart the process after such an
12665 event too, in case it would be caused by memory corruption.
12666
Simon Horman752dc4a2011-06-21 14:34:59 +090012667 D : the session was killed by haproxy because the server was detected
12668 as down and was configured to kill all connections when going down.
12669
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012670 U : the session was killed by haproxy on this backup server because an
12671 active server was detected as up and was configured to kill all
12672 backup connections when going up.
12673
Willy Tarreaua2a64e92011-09-07 23:01:56 +020012674 K : the session was actively killed by an admin operating on haproxy.
12675
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012676 c : the client-side timeout expired while waiting for the client to
12677 send or receive data.
12678
12679 s : the server-side timeout expired while waiting for the server to
12680 send or receive data.
12681
12682 - : normal session completion, both the client and the server closed
12683 with nothing left in the buffers.
12684
12685 - on the second character, the TCP or HTTP session state when it was closed :
12686
Willy Tarreauf7b30a92010-12-06 22:59:17 +010012687 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012688 (HTTP mode only). Nothing was sent to any server.
12689
12690 Q : the proxy was waiting in the QUEUE for a connection slot. This can
12691 only happen when servers have a 'maxconn' parameter set. It can
12692 also happen in the global queue after a redispatch consecutive to
12693 a failed attempt to connect to a dying server. If no redispatch is
12694 reported, then no connection attempt was made to any server.
12695
12696 C : the proxy was waiting for the CONNECTION to establish on the
12697 server. The server might at most have noticed a connection attempt.
12698
12699 H : the proxy was waiting for complete, valid response HEADERS from the
12700 server (HTTP only).
12701
12702 D : the session was in the DATA phase.
12703
12704 L : the proxy was still transmitting LAST data to the client while the
12705 server had already finished. This one is very rare as it can only
12706 happen when the client dies while receiving the last packets.
12707
12708 T : the request was tarpitted. It has been held open with the client
12709 during the whole "timeout tarpit" duration or until the client
12710 closed, both of which will be reported in the "Tw" timer.
12711
12712 - : normal session completion after end of data transfer.
12713
12714 - the third character tells whether the persistence cookie was provided by
12715 the client (only in HTTP mode) :
12716
12717 N : the client provided NO cookie. This is usually the case for new
12718 visitors, so counting the number of occurrences of this flag in the
12719 logs generally indicate a valid trend for the site frequentation.
12720
12721 I : the client provided an INVALID cookie matching no known server.
12722 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020012723 cookies between HTTP/HTTPS sites, persistence conditionally
12724 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012725
12726 D : the client provided a cookie designating a server which was DOWN,
12727 so either "option persist" was used and the client was sent to
12728 this server, or it was not set and the client was redispatched to
12729 another server.
12730
Willy Tarreau996a92c2010-10-13 19:30:47 +020012731 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012732 server.
12733
Willy Tarreau996a92c2010-10-13 19:30:47 +020012734 E : the client provided a valid cookie, but with a last date which was
12735 older than what is allowed by the "maxidle" cookie parameter, so
12736 the cookie is consider EXPIRED and is ignored. The request will be
12737 redispatched just as if there was no cookie.
12738
12739 O : the client provided a valid cookie, but with a first date which was
12740 older than what is allowed by the "maxlife" cookie parameter, so
12741 the cookie is consider too OLD and is ignored. The request will be
12742 redispatched just as if there was no cookie.
12743
Willy Tarreauc89ccb62012-04-05 21:18:22 +020012744 U : a cookie was present but was not used to select the server because
12745 some other server selection mechanism was used instead (typically a
12746 "use-server" rule).
12747
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012748 - : does not apply (no cookie set in configuration).
12749
12750 - the last character reports what operations were performed on the persistence
12751 cookie returned by the server (only in HTTP mode) :
12752
12753 N : NO cookie was provided by the server, and none was inserted either.
12754
12755 I : no cookie was provided by the server, and the proxy INSERTED one.
12756 Note that in "cookie insert" mode, if the server provides a cookie,
12757 it will still be overwritten and reported as "I" here.
12758
Willy Tarreau996a92c2010-10-13 19:30:47 +020012759 U : the proxy UPDATED the last date in the cookie that was presented by
12760 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012761 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020012762 date indicated in the cookie. If any other change happens, such as
12763 a redispatch, then the cookie will be marked as inserted instead.
12764
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012765 P : a cookie was PROVIDED by the server and transmitted as-is.
12766
12767 R : the cookie provided by the server was REWRITTEN by the proxy, which
12768 happens in "cookie rewrite" or "cookie prefix" modes.
12769
12770 D : the cookie provided by the server was DELETED by the proxy.
12771
12772 - : does not apply (no cookie set in configuration).
12773
Willy Tarreau996a92c2010-10-13 19:30:47 +020012774The combination of the two first flags gives a lot of information about what
12775was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012776helpful to detect server saturation, network troubles, local system resource
12777starvation, attacks, etc...
12778
12779The most common termination flags combinations are indicated below. They are
12780alphabetically sorted, with the lowercase set just after the upper case for
12781easier finding and understanding.
12782
12783 Flags Reason
12784
12785 -- Normal termination.
12786
12787 CC The client aborted before the connection could be established to the
12788 server. This can happen when haproxy tries to connect to a recently
12789 dead (or unchecked) server, and the client aborts while haproxy is
12790 waiting for the server to respond or for "timeout connect" to expire.
12791
12792 CD The client unexpectedly aborted during data transfer. This can be
12793 caused by a browser crash, by an intermediate equipment between the
12794 client and haproxy which decided to actively break the connection,
12795 by network routing issues between the client and haproxy, or by a
12796 keep-alive session between the server and the client terminated first
12797 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012798
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012799 cD The client did not send nor acknowledge any data for as long as the
12800 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020012801 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012802
12803 CH The client aborted while waiting for the server to start responding.
12804 It might be the server taking too long to respond or the client
12805 clicking the 'Stop' button too fast.
12806
12807 cH The "timeout client" stroke while waiting for client data during a
12808 POST request. This is sometimes caused by too large TCP MSS values
12809 for PPPoE networks which cannot transport full-sized packets. It can
12810 also happen when client timeout is smaller than server timeout and
12811 the server takes too long to respond.
12812
12813 CQ The client aborted while its session was queued, waiting for a server
12814 with enough empty slots to accept it. It might be that either all the
12815 servers were saturated or that the assigned server was taking too
12816 long a time to respond.
12817
12818 CR The client aborted before sending a full HTTP request. Most likely
12819 the request was typed by hand using a telnet client, and aborted
12820 too early. The HTTP status code is likely a 400 here. Sometimes this
12821 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau1c3a6122015-05-01 15:37:53 +020012822 and the client. "option http-ignore-probes" can be used to ignore
12823 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012824
12825 cR The "timeout http-request" stroke before the client sent a full HTTP
12826 request. This is sometimes caused by too large TCP MSS values on the
12827 client side for PPPoE networks which cannot transport full-sized
12828 packets, or by clients sending requests by hand and not typing fast
12829 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020012830 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau1c3a6122015-05-01 15:37:53 +020012831 some browsers started to implement a "pre-connect" feature consisting
12832 in speculatively connecting to some recently visited web sites just
12833 in case the user would like to visit them. This results in many
12834 connections being established to web sites, which end up in 408
12835 Request Timeout if the timeout strikes first, or 400 Bad Request when
12836 the browser decides to close them first. These ones pollute the log
12837 and feed the error counters. Some versions of some browsers have even
12838 been reported to display the error code. It is possible to work
12839 around the undesirable effects of this behaviour by adding "option
12840 http-ignore-probes" in the frontend, resulting in connections with
12841 zero data transfer to be totally ignored. This will definitely hide
12842 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012843
12844 CT The client aborted while its session was tarpitted. It is important to
12845 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020012846 wrong tarpit rules have been written. If a lot of them happen, it
12847 might make sense to lower the "timeout tarpit" value to something
12848 closer to the average reported "Tw" timer, in order not to consume
12849 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012850
Willy Tarreau570f2212013-06-10 16:42:09 +020012851 LR The request was intercepted and locally handled by haproxy. Generally
12852 it means that this was a redirect or a stats request.
12853
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012854 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012855 the TCP connection (the proxy received a TCP RST or an ICMP message
12856 in return). Under some circumstances, it can also be the network
12857 stack telling the proxy that the server is unreachable (eg: no route,
12858 or no ARP response on local network). When this happens in HTTP mode,
12859 the status code is likely a 502 or 503 here.
12860
12861 sC The "timeout connect" stroke before a connection to the server could
12862 complete. When this happens in HTTP mode, the status code is likely a
12863 503 or 504 here.
12864
12865 SD The connection to the server died with an error during the data
12866 transfer. This usually means that haproxy has received an RST from
12867 the server or an ICMP message from an intermediate equipment while
12868 exchanging data with the server. This can be caused by a server crash
12869 or by a network issue on an intermediate equipment.
12870
12871 sD The server did not send nor acknowledge any data for as long as the
12872 "timeout server" setting during the data phase. This is often caused
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012873 by too short timeouts on L4 equipments before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012874 load-balancers, ...), as well as keep-alive sessions maintained
12875 between the client and the server expiring first on haproxy.
12876
12877 SH The server aborted before sending its full HTTP response headers, or
12878 it crashed while processing the request. Since a server aborting at
12879 this moment is very rare, it would be wise to inspect its logs to
12880 control whether it crashed and why. The logged request may indicate a
12881 small set of faulty requests, demonstrating bugs in the application.
12882 Sometimes this might also be caused by an IDS killing the connection
12883 between haproxy and the server.
12884
12885 sH The "timeout server" stroke before the server could return its
12886 response headers. This is the most common anomaly, indicating too
12887 long transactions, probably caused by server or database saturation.
12888 The immediate workaround consists in increasing the "timeout server"
12889 setting, but it is important to keep in mind that the user experience
12890 will suffer from these long response times. The only long term
12891 solution is to fix the application.
12892
12893 sQ The session spent too much time in queue and has been expired. See
12894 the "timeout queue" and "timeout connect" settings to find out how to
12895 fix this if it happens too often. If it often happens massively in
12896 short periods, it may indicate general problems on the affected
12897 servers due to I/O or database congestion, or saturation caused by
12898 external attacks.
12899
12900 PC The proxy refused to establish a connection to the server because the
12901 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020012902 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012903 so that it does not happen anymore. This status is very rare and
12904 might happen when the global "ulimit-n" parameter is forced by hand.
12905
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012906 PD The proxy blocked an incorrectly formatted chunked encoded message in
12907 a request or a response, after the server has emitted its headers. In
12908 most cases, this will indicate an invalid message from the server to
Willy Tarreauf3a3e132013-08-31 08:16:26 +020012909 the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647
12910 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012911
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012912 PH The proxy blocked the server's response, because it was invalid,
12913 incomplete, dangerous (cache control), or matched a security filter.
12914 In any case, an HTTP 502 error is sent to the client. One possible
12915 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012916 containing unauthorized characters. It is also possible but quite
12917 rare, that the proxy blocked a chunked-encoding request from the
12918 client due to an invalid syntax, before the server responded. In this
12919 case, an HTTP 400 error is sent to the client and reported in the
12920 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012921
12922 PR The proxy blocked the client's HTTP request, either because of an
12923 invalid HTTP syntax, in which case it returned an HTTP 400 error to
12924 the client, or because a deny filter matched, in which case it
12925 returned an HTTP 403 error.
12926
12927 PT The proxy blocked the client's request and has tarpitted its
12928 connection before returning it a 500 server error. Nothing was sent
12929 to the server. The connection was maintained open for as long as
12930 reported by the "Tw" timer field.
12931
12932 RC A local resource has been exhausted (memory, sockets, source ports)
12933 preventing the connection to the server from establishing. The error
12934 logs will tell precisely what was missing. This is very rare and can
12935 only be solved by proper system tuning.
12936
Willy Tarreau996a92c2010-10-13 19:30:47 +020012937The combination of the two last flags gives a lot of information about how
12938persistence was handled by the client, the server and by haproxy. This is very
12939important to troubleshoot disconnections, when users complain they have to
12940re-authenticate. The commonly encountered flags are :
12941
12942 -- Persistence cookie is not enabled.
12943
12944 NN No cookie was provided by the client, none was inserted in the
12945 response. For instance, this can be in insert mode with "postonly"
12946 set on a GET request.
12947
12948 II A cookie designating an invalid server was provided by the client,
12949 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012950 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020012951 value can be presented by a client when no other server knows it.
12952
12953 NI No cookie was provided by the client, one was inserted in the
12954 response. This typically happens for first requests from every user
12955 in "insert" mode, which makes it an easy way to count real users.
12956
12957 VN A cookie was provided by the client, none was inserted in the
12958 response. This happens for most responses for which the client has
12959 already got a cookie.
12960
12961 VU A cookie was provided by the client, with a last visit date which is
12962 not completely up-to-date, so an updated cookie was provided in
12963 response. This can also happen if there was no date at all, or if
12964 there was a date but the "maxidle" parameter was not set, so that the
12965 cookie can be switched to unlimited time.
12966
12967 EI A cookie was provided by the client, with a last visit date which is
12968 too old for the "maxidle" parameter, so the cookie was ignored and a
12969 new cookie was inserted in the response.
12970
12971 OI A cookie was provided by the client, with a first visit date which is
12972 too old for the "maxlife" parameter, so the cookie was ignored and a
12973 new cookie was inserted in the response.
12974
12975 DI The server designated by the cookie was down, a new server was
12976 selected and a new cookie was emitted in the response.
12977
12978 VI The server designated by the cookie was not marked dead but could not
12979 be reached. A redispatch happened and selected another one, which was
12980 then advertised in the response.
12981
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012982
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129838.6. Non-printable characters
12984-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012985
12986In order not to cause trouble to log analysis tools or terminals during log
12987consulting, non-printable characters are not sent as-is into log files, but are
12988converted to the two-digits hexadecimal representation of their ASCII code,
12989prefixed by the character '#'. The only characters that can be logged without
12990being escaped are comprised between 32 and 126 (inclusive). Obviously, the
12991escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
12992is the same for the character '"' which becomes "#22", as well as '{', '|' and
12993'}' when logging headers.
12994
12995Note that the space character (' ') is not encoded in headers, which can cause
12996issues for tools relying on space count to locate fields. A typical header
12997containing spaces is "User-Agent".
12998
12999Last, it has been observed that some syslog daemons such as syslog-ng escape
13000the quote ('"') with a backslash ('\'). The reverse operation can safely be
13001performed since no quote may appear anywhere else in the logs.
13002
13003
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130048.7. Capturing HTTP cookies
13005---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013006
13007Cookie capture simplifies the tracking a complete user session. This can be
13008achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013009section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013010cookie will simultaneously be checked in the request ("Cookie:" header) and in
13011the response ("Set-Cookie:" header). The respective values will be reported in
13012the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013013locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013014not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
13015user switches to a new session for example, because the server will reassign it
13016a new cookie. It is also possible to detect if a server unexpectedly sets a
13017wrong cookie to a client, leading to session crossing.
13018
13019 Examples :
13020 # capture the first cookie whose name starts with "ASPSESSION"
13021 capture cookie ASPSESSION len 32
13022
13023 # capture the first cookie whose name is exactly "vgnvisitor"
13024 capture cookie vgnvisitor= len 32
13025
13026
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130278.8. Capturing HTTP headers
13028---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013029
13030Header captures are useful to track unique request identifiers set by an upper
13031proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
13032the response, one can search for information about the response length, how the
13033server asked the cache to behave, or an object location during a redirection.
13034
13035Header captures are performed using the "capture request header" and "capture
13036response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013037section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013038
13039It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013040time. Non-existent headers are logged as empty strings, and if one header
13041appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013042are grouped within braces '{' and '}' in the same order as they were declared,
13043and delimited with a vertical bar '|' without any space. Response headers
13044follow the same representation, but are displayed after a space following the
13045request headers block. These blocks are displayed just before the HTTP request
13046in the logs.
13047
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020013048As a special case, it is possible to specify an HTTP header capture in a TCP
13049frontend. The purpose is to enable logging of headers which will be parsed in
13050an HTTP backend if the request is then switched to this HTTP backend.
13051
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013052 Example :
13053 # This instance chains to the outgoing proxy
13054 listen proxy-out
13055 mode http
13056 option httplog
13057 option logasap
13058 log global
13059 server cache1 192.168.1.1:3128
13060
13061 # log the name of the virtual server
13062 capture request header Host len 20
13063
13064 # log the amount of data uploaded during a POST
13065 capture request header Content-Length len 10
13066
13067 # log the beginning of the referrer
13068 capture request header Referer len 20
13069
13070 # server name (useful for outgoing proxies only)
13071 capture response header Server len 20
13072
13073 # logging the content-length is useful with "option logasap"
13074 capture response header Content-Length len 10
13075
13076 # log the expected cache behaviour on the response
13077 capture response header Cache-Control len 8
13078
13079 # the Via header will report the next proxy's name
13080 capture response header Via len 20
13081
13082 # log the URL location during a redirection
13083 capture response header Location len 20
13084
13085 >>> Aug 9 20:26:09 localhost \
13086 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
13087 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
13088 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
13089 "GET http://fr.adserver.yahoo.com/"
13090
13091 >>> Aug 9 20:30:46 localhost \
13092 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
13093 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
13094 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013095 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013096
13097 >>> Aug 9 20:30:46 localhost \
13098 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
13099 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
13100 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
13101 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013102 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013103
13104
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131058.9. Examples of logs
13106---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013107
13108These are real-world examples of logs accompanied with an explanation. Some of
13109them have been made up by hand. The syslog part has been removed for better
13110reading. Their sole purpose is to explain how to decipher them.
13111
13112 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
13113 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
13114 "HEAD / HTTP/1.0"
13115
13116 => long request (6.5s) entered by hand through 'telnet'. The server replied
13117 in 147 ms, and the session ended normally ('----')
13118
13119 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
13120 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
13121 0/9 "HEAD / HTTP/1.0"
13122
13123 => Idem, but the request was queued in the global queue behind 9 other
13124 requests, and waited there for 1230 ms.
13125
13126 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
13127 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
13128 "GET /image.iso HTTP/1.0"
13129
13130 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013131 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013132 14 ms, 243 bytes of headers were sent to the client, and total time from
13133 accept to first data byte is 30 ms.
13134
13135 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
13136 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
13137 "GET /cgi-bin/bug.cgi? HTTP/1.0"
13138
13139 => the proxy blocked a server response either because of an "rspdeny" or
13140 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020013141 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013142 risked being cached. In this case, the response is replaced with a "502
13143 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
13144 to return the 502 and not the server.
13145
13146 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013147 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 +010013148
13149 => the client never completed its request and aborted itself ("C---") after
13150 8.5s, while the proxy was waiting for the request headers ("-R--").
13151 Nothing was sent to any server.
13152
13153 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
13154 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
13155
13156 => The client never completed its request, which was aborted by the
13157 time-out ("c---") after 50s, while the proxy was waiting for the request
13158 headers ("-R--"). Nothing was sent to any server, but the proxy could
13159 send a 408 return code to the client.
13160
13161 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
13162 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
13163
13164 => This log was produced with "option tcplog". The client timed out after
13165 5 seconds ("c----").
13166
13167 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
13168 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013169 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013170
13171 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013172 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013173 (config says 'retries 3'), and no redispatch (otherwise we would have
13174 seen "/+3"). Status code 503 was returned to the client. There were 115
13175 connections on this server, 202 connections on this proxy, and 205 on
13176 the global process. It is possible that the server refused the
13177 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013178
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013179
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131809. Statistics and monitoring
13181----------------------------
13182
13183It is possible to query HAProxy about its status. The most commonly used
13184mechanism is the HTTP statistics page. This page also exposes an alternative
13185CSV output format for monitoring tools. The same format is provided on the
13186Unix socket.
13187
13188
131899.1. CSV format
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013190---------------
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +010013191
Willy Tarreau7f062c42009-03-05 18:43:00 +010013192The statistics may be consulted either from the unix socket or from the HTTP
Willy Tarreaua3310dc2014-06-16 15:43:21 +020013193page. Both means provide a CSV format whose fields follow. The first line
13194begins with a sharp ('#') and has one word per comma-delimited field which
13195represents the title of the column. All other lines starting at the second one
13196use a classical CSV format using a comma as the delimiter, and the double quote
13197('"') as an optional text delimiter, but only if the enclosed text is ambiguous
13198(if it contains a quote or a comma). The double-quote character ('"') in the
13199text is doubled ('""'), which is the format that most tools recognize. Please
13200do not insert any column before these ones in order not to break tools which
13201use hard-coded column positions.
Willy Tarreau7f062c42009-03-05 18:43:00 +010013202
James Westby75c98ad2014-07-08 10:14:57 -040013203In brackets after each field name are the types which may have a value for
13204that field. The types are L (Listeners), F (Frontends), B (Backends), and
13205S (Servers).
13206
13207 0. pxname [LFBS]: proxy name
13208 1. svname [LFBS]: service name (FRONTEND for frontend, BACKEND for backend,
13209 any name for server/listener)
13210 2. qcur [..BS]: current queued requests. For the backend this reports the
13211 number queued without a server assigned.
13212 3. qmax [..BS]: max value of qcur
13213 4. scur [LFBS]: current sessions
13214 5. smax [LFBS]: max sessions
13215 6. slim [LFBS]: configured session limit
13216 7. stot [LFBS]: cumulative number of connections
13217 8. bin [LFBS]: bytes in
13218 9. bout [LFBS]: bytes out
13219 10. dreq [LFB.]: requests denied because of security concerns.
13220 - For tcp this is because of a matched tcp-request content rule.
13221 - For http this is because of a matched http-request or tarpit rule.
13222 11. dresp [LFBS]: responses denied because of security concerns.
13223 - For http this is because of a matched http-request rule, or
13224 "option checkcache".
13225 12. ereq [LF..]: request errors. Some of the possible causes are:
13226 - early termination from the client, before the request has been sent.
13227 - read error from the client
13228 - client timeout
13229 - client closed connection
13230 - various bad requests from the client.
13231 - request was tarpitted.
13232 13. econ [..BS]: number of requests that encountered an error trying to
13233 connect to a backend server. The backend stat is the sum of the stat
13234 for all servers of that backend, plus any connection errors not
13235 associated with a particular server (such as the backend having no
13236 active servers).
13237 14. eresp [..BS]: response errors. srv_abrt will be counted here also.
13238 Some other errors are:
13239 - write error on the client socket (won't be counted for the server stat)
13240 - failure applying filters to the response.
13241 15. wretr [..BS]: number of times a connection to a server was retried.
13242 16. wredis [..BS]: number of times a request was redispatched to another
13243 server. The server value counts the number of times that server was
13244 switched away from.
13245 17. status [LFBS]: status (UP/DOWN/NOLB/MAINT/MAINT(via)...)
Pavlos Parissisac372e12015-05-02 20:30:44 +020013246 18. weight [..BS]: total weight (backend), server weight (server)
13247 19. act [..BS]: number of active servers (backend), server is active (server)
13248 20. bck [..BS]: number of backup servers (backend), server is backup (server)
James Westby75c98ad2014-07-08 10:14:57 -040013249 21. chkfail [...S]: number of failed checks. (Only counts checks failed when
13250 the server is up.)
13251 22. chkdown [..BS]: number of UP->DOWN transitions. The backend counter counts
13252 transitions to the whole backend being down, rather than the sum of the
13253 counters for each server.
13254 23. lastchg [..BS]: number of seconds since the last UP<->DOWN transition
13255 24. downtime [..BS]: total downtime (in seconds). The value for the backend
13256 is the downtime for the whole backend, not the sum of the server downtime.
13257 25. qlimit [...S]: configured maxqueue for the server, or nothing in the
13258 value is 0 (default, meaning no limit)
13259 26. pid [LFBS]: process id (0 for first instance, 1 for second, ...)
13260 27. iid [LFBS]: unique proxy id
13261 28. sid [L..S]: server id (unique inside a proxy)
13262 29. throttle [...S]: current throttle percentage for the server, when
13263 slowstart is active, or no value if not in slowstart.
13264 30. lbtot [..BS]: total number of times a server was selected, either for new
13265 sessions, or when re-dispatching. The server counter is the number
13266 of times that server was selected.
13267 31. tracked [...S]: id of proxy/server if tracking is enabled.
13268 32. type [LFBS]: (0=frontend, 1=backend, 2=server, 3=socket/listener)
13269 33. rate [.FBS]: number of sessions per second over last elapsed second
13270 34. rate_lim [.F..]: configured limit on new sessions per second
13271 35. rate_max [.FBS]: max number of new sessions per second
13272 36. check_status [...S]: status of last health check, one of:
Cyril Bontéf0c60612010-02-06 14:44:47 +010013273 UNK -> unknown
13274 INI -> initializing
13275 SOCKERR -> socket error
13276 L4OK -> check passed on layer 4, no upper layers testing enabled
Jason Harveycac41222015-04-16 11:13:21 -080013277 L4TOUT -> layer 1-4 timeout
Cyril Bontéf0c60612010-02-06 14:44:47 +010013278 L4CON -> layer 1-4 connection problem, for example
13279 "Connection refused" (tcp rst) or "No route to host" (icmp)
13280 L6OK -> check passed on layer 6
13281 L6TOUT -> layer 6 (SSL) timeout
13282 L6RSP -> layer 6 invalid response - protocol error
13283 L7OK -> check passed on layer 7
13284 L7OKC -> check conditionally passed on layer 7, for example 404 with
13285 disable-on-404
13286 L7TOUT -> layer 7 (HTTP/SMTP) timeout
13287 L7RSP -> layer 7 invalid response - protocol error
13288 L7STS -> layer 7 response error, for example HTTP 5xx
James Westby75c98ad2014-07-08 10:14:57 -040013289 37. check_code [...S]: layer5-7 code, if available
13290 38. check_duration [...S]: time in ms took to finish last health check
13291 39. hrsp_1xx [.FBS]: http responses with 1xx code
13292 40. hrsp_2xx [.FBS]: http responses with 2xx code
13293 41. hrsp_3xx [.FBS]: http responses with 3xx code
13294 42. hrsp_4xx [.FBS]: http responses with 4xx code
13295 43. hrsp_5xx [.FBS]: http responses with 5xx code
13296 44. hrsp_other [.FBS]: http responses with other codes (protocol error)
13297 45. hanafail [...S]: failed health checks details
13298 46. req_rate [.F..]: HTTP requests per second over last elapsed second
13299 47. req_rate_max [.F..]: max number of HTTP requests per second observed
13300 48. req_tot [.F..]: total number of HTTP requests received
13301 49. cli_abrt [..BS]: number of data transfers aborted by the client
13302 50. srv_abrt [..BS]: number of data transfers aborted by the server
13303 (inc. in eresp)
13304 51. comp_in [.FB.]: number of HTTP response bytes fed to the compressor
13305 52. comp_out [.FB.]: number of HTTP response bytes emitted by the compressor
13306 53. comp_byp [.FB.]: number of bytes that bypassed the HTTP compressor
13307 (CPU/BW limit)
13308 54. comp_rsp [.FB.]: number of HTTP responses that were compressed
13309 55. lastsess [..BS]: number of seconds since last session assigned to
13310 server/backend
13311 56. last_chk [...S]: last health check contents or textual error
13312 57. last_agt [...S]: last agent check contents or textual error
13313 58. qtime [..BS]: the average queue time in ms over the 1024 last requests
13314 59. ctime [..BS]: the average connect time in ms over the 1024 last requests
13315 60. rtime [..BS]: the average response time in ms over the 1024 last requests
13316 (0 for TCP)
13317 61. ttime [..BS]: the average total session time in ms over the 1024 last
13318 requests
Willy Tarreau844e3c52008-01-11 16:28:18 +010013319
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013320
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200133219.2. Unix Socket commands
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013322-------------------------
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013323
Willy Tarreau468f4932013-08-01 16:50:16 +020013324The stats socket is not enabled by default. In order to enable it, it is
13325necessary to add one line in the global section of the haproxy configuration.
13326A second line is recommended to set a larger timeout, always appreciated when
13327issuing commands by hand :
13328
13329 global
13330 stats socket /var/run/haproxy.sock mode 600 level admin
13331 stats timeout 2m
13332
13333It is also possible to add multiple instances of the stats socket by repeating
13334the line, and make them listen to a TCP port instead of a UNIX socket. This is
13335never done by default because this is dangerous, but can be handy in some
13336situations :
13337
13338 global
13339 stats socket /var/run/haproxy.sock mode 600 level admin
13340 stats socket ipv4@192.168.0.1:9999 level admin
13341 stats timeout 2m
13342
13343To access the socket, an external utility such as "socat" is required. Socat is a
13344swiss-army knife to connect anything to anything. We use it to connect terminals
13345to the socket, or a couple of stdin/stdout pipes to it for scripts. The two main
13346syntaxes we'll use are the following :
13347
13348 # socat /var/run/haproxy.sock stdio
13349 # socat /var/run/haproxy.sock readline
13350
13351The first one is used with scripts. It is possible to send the output of a
13352script to haproxy, and pass haproxy's output to another script. That's useful
13353for retrieving counters or attack traces for example.
13354
13355The second one is only useful for issuing commands by hand. It has the benefit
13356that the terminal is handled by the readline library which supports line
13357editing and history, which is very convenient when issuing repeated commands
13358(eg: watch a counter).
13359
13360The socket supports two operation modes :
13361 - interactive
13362 - non-interactive
13363
13364The non-interactive mode is the default when socat connects to the socket. In
13365this mode, a single line may be sent. It is processed as a whole, responses are
13366sent back, and the connection closes after the end of the response. This is the
13367mode that scripts and monitoring tools use. It is possible to send multiple
13368commands in this mode, they need to be delimited by a semi-colon (';'). For
13369example :
13370
13371 # echo "show info;show stat;show table" | socat /var/run/haproxy stdio
13372
13373The interactive mode displays a prompt ('>') and waits for commands to be
13374entered on the line, then processes them, and displays the prompt again to wait
13375for a new command. This mode is entered via the "prompt" command which must be
13376sent on the first line in non-interactive mode. The mode is a flip switch, if
13377"prompt" is sent in interactive mode, it is disabled and the connection closes
13378after processing the last command of the same line.
13379
13380For this reason, when debugging by hand, it's quite common to start with the
13381"prompt" command :
13382
13383 # socat /var/run/haproxy readline
13384 prompt
13385 > show info
13386 ...
13387 >
13388
13389Since multiple commands may be issued at once, haproxy uses the empty line as a
13390delimiter to mark an end of output for each command, and takes care of ensuring
13391that no command can emit an empty line on output. A script can thus easily
13392parse the output even when multiple commands were pipelined on a single line.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013393
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013394It is important to understand that when multiple haproxy processes are started
13395on the same sockets, any process may pick up the request and will output its
13396own stats.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013397
Willy Tarreau468f4932013-08-01 16:50:16 +020013398The list of commands currently supported on the stats socket is provided below.
13399If an unknown command is sent, haproxy displays the usage message which reminds
13400all supported commands. Some commands support a more complex syntax, generally
13401it will explain what part of the command is invalid when this happens.
13402
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013403add acl <acl> <pattern>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013404 Add an entry into the acl <acl>. <acl> is the #<id> or the <file> returned by
13405 "show acl". This command does not verify if the entry already exists. This
13406 command cannot be used if the reference <acl> is a file also used with a map.
13407 In this case, you must use the command "add map" in place of "add acl".
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013408
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013409add map <map> <key> <value>
13410 Add an entry into the map <map> to associate the value <value> to the key
13411 <key>. This command does not verify if the entry already exists. It is
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013412 mainly used to fill a map after a clear operation. Note that if the reference
13413 <map> is a file and is shared with a map, this map will contain also a new
13414 pattern entry.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013415
Willy Tarreaud63335a2010-02-26 12:56:52 +010013416clear counters
13417 Clear the max values of the statistics counters in each proxy (frontend &
13418 backend) and in each server. The cumulated counters are not affected. This
13419 can be used to get clean counters after an incident, without having to
13420 restart nor to clear traffic counters. This command is restricted and can
13421 only be issued on sockets configured for levels "operator" or "admin".
13422
13423clear counters all
13424 Clear all statistics counters in each proxy (frontend & backend) and in each
13425 server. This has the same effect as restarting. This command is restricted
13426 and can only be issued on sockets configured for level "admin".
13427
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013428clear acl <acl>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013429 Remove all entries from the acl <acl>. <acl> is the #<id> or the <file>
13430 returned by "show acl". Note that if the reference <acl> is a file and is
13431 shared with a map, this map will be also cleared.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013432
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013433clear map <map>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013434 Remove all entries from the map <map>. <map> is the #<id> or the <file>
13435 returned by "show map". Note that if the reference <map> is a file and is
13436 shared with a acl, this acl will be also cleared.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013437
Simon Hormanc88b8872011-06-15 15:18:49 +090013438clear table <table> [ data.<type> <operator> <value> ] | [ key <key> ]
13439 Remove entries from the stick-table <table>.
13440
13441 This is typically used to unblock some users complaining they have been
13442 abusively denied access to a service, but this can also be used to clear some
13443 stickiness entries matching a server that is going to be replaced (see "show
13444 table" below for details). Note that sometimes, removal of an entry will be
13445 refused because it is currently tracked by a session. Retrying a few seconds
13446 later after the session ends is usual enough.
13447
13448 In the case where no options arguments are given all entries will be removed.
13449
13450 When the "data." form is used entries matching a filter applied using the
13451 stored data (see "stick-table" in section 4.2) are removed. A stored data
13452 type must be specified in <type>, and this data type must be stored in the
13453 table otherwise an error is reported. The data is compared according to
13454 <operator> with the 64-bit integer <value>. Operators are the same as with
13455 the ACLs :
13456
13457 - eq : match entries whose data is equal to this value
13458 - ne : match entries whose data is not equal to this value
13459 - le : match entries whose data is less than or equal to this value
13460 - ge : match entries whose data is greater than or equal to this value
13461 - lt : match entries whose data is less than this value
13462 - gt : match entries whose data is greater than this value
13463
13464 When the key form is used the entry <key> is removed. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013465 same type as the table, which currently is limited to IPv4, IPv6, integer and
13466 string.
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013467
13468 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013469 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013470 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013471 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13472 bytes_out_rate(60000)=187
13473 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13474 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013475
13476 $ echo "clear table http_proxy key 127.0.0.1" | socat stdio /tmp/sock1
13477
13478 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013479 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau62a36c42010-08-17 15:53:10 +020013480 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13481 bytes_out_rate(60000)=191
Simon Hormanc88b8872011-06-15 15:18:49 +090013482 $ echo "clear table http_proxy data.gpc0 eq 1" | socat stdio /tmp/sock1
13483 $ echo "show table http_proxy" | socat stdio /tmp/sock1
13484 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013485
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013486del acl <acl> [<key>|#<ref>]
13487 Delete all the acl entries from the acl <acl> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013488 <acl> is the #<id> or the <file> returned by "show acl". If the <ref> is used,
13489 this command delete only the listed reference. The reference can be found with
13490 listing the content of the acl. Note that if the reference <acl> is a file and
13491 is shared with a map, the entry will be also deleted in the map.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013492
13493del map <map> [<key>|#<ref>]
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013494 Delete all the map entries from the map <map> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013495 <map> is the #<id> or the <file> returned by "show map". If the <ref> is used,
13496 this command delete only the listed reference. The reference can be found with
13497 listing the content of the map. Note that if the reference <map> is a file and
13498 is shared with a acl, the entry will be also deleted in the map.
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013499
13500disable agent <backend>/<server>
Simon Horman671b6f02013-11-25 10:46:39 +090013501 Mark the auxiliary agent check as temporarily stopped.
13502
13503 In the case where an agent check is being run as a auxiliary check, due
13504 to the agent-check parameter of a server directive, new checks are only
13505 initialised when the agent is in the enabled. Thus, disable agent will
13506 prevent any new agent checks from begin initiated until the agent
13507 re-enabled using enable agent.
13508
13509 When an agent is disabled the processing of an auxiliary agent check that
13510 was initiated while the agent was set as enabled is as follows: All
13511 results that would alter the weight, specifically "drain" or a weight
13512 returned by the agent, are ignored. The processing of agent check is
13513 otherwise unchanged.
13514
13515 The motivation for this feature is to allow the weight changing effects
13516 of the agent checks to be paused to allow the weight of a server to be
13517 configured using set weight without being overridden by the agent.
13518
13519 This command is restricted and can only be issued on sockets configured for
13520 level "admin".
13521
Willy Tarreau532a4502011-09-07 22:37:44 +020013522disable frontend <frontend>
13523 Mark the frontend as temporarily stopped. This corresponds to the mode which
13524 is used during a soft restart : the frontend releases the port but can be
13525 enabled again if needed. This should be used with care as some non-Linux OSes
13526 are unable to enable it back. This is intended to be used in environments
13527 where stopping a proxy is not even imaginable but a misconfigured proxy must
13528 be fixed. That way it's possible to release the port and bind it into another
13529 process to restore operations. The frontend will appear with status "STOP"
13530 on the stats page.
13531
13532 The frontend may be specified either by its name or by its numeric ID,
13533 prefixed with a sharp ('#').
13534
13535 This command is restricted and can only be issued on sockets configured for
13536 level "admin".
13537
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013538disable health <backend>/<server>
13539 Mark the primary health check as temporarily stopped. This will disable
13540 sending of health checks, and the last health check result will be ignored.
13541 The server will be in unchecked state and considered UP unless an auxiliary
13542 agent check forces it down.
13543
13544 This command is restricted and can only be issued on sockets configured for
13545 level "admin".
13546
Willy Tarreaud63335a2010-02-26 12:56:52 +010013547disable server <backend>/<server>
13548 Mark the server DOWN for maintenance. In this mode, no more checks will be
13549 performed on the server until it leaves maintenance.
13550 If the server is tracked by other servers, those servers will be set to DOWN
13551 during the maintenance.
13552
13553 In the statistics page, a server DOWN for maintenance will appear with a
13554 "MAINT" status, its tracking servers with the "MAINT(via)" one.
13555
13556 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013557 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013558
13559 This command is restricted and can only be issued on sockets configured for
13560 level "admin".
13561
Simon Horman671b6f02013-11-25 10:46:39 +090013562enable agent <backend>/<server>
13563 Resume auxiliary agent check that was temporarily stopped.
13564
13565 See "disable agent" for details of the effect of temporarily starting
13566 and stopping an auxiliary agent.
13567
13568 This command is restricted and can only be issued on sockets configured for
13569 level "admin".
13570
Willy Tarreau532a4502011-09-07 22:37:44 +020013571enable frontend <frontend>
13572 Resume a frontend which was temporarily stopped. It is possible that some of
13573 the listening ports won't be able to bind anymore (eg: if another process
13574 took them since the 'disable frontend' operation). If this happens, an error
13575 is displayed. Some operating systems might not be able to resume a frontend
13576 which was disabled.
13577
13578 The frontend may be specified either by its name or by its numeric ID,
13579 prefixed with a sharp ('#').
13580
13581 This command is restricted and can only be issued on sockets configured for
13582 level "admin".
13583
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013584enable health <backend>/<server>
13585 Resume a primary health check that was temporarily stopped. This will enable
13586 sending of health checks again. Please see "disable health" for details.
13587
13588 This command is restricted and can only be issued on sockets configured for
13589 level "admin".
13590
Willy Tarreaud63335a2010-02-26 12:56:52 +010013591enable server <backend>/<server>
13592 If the server was previously marked as DOWN for maintenance, this marks the
13593 server UP and checks are re-enabled.
13594
13595 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013596 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013597
13598 This command is restricted and can only be issued on sockets configured for
13599 level "admin".
13600
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013601get map <map> <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013602get acl <acl> <value>
13603 Lookup the value <value> in the map <map> or in the ACL <acl>. <map> or <acl>
13604 are the #<id> or the <file> returned by "show map" or "show acl". This command
13605 returns all the matching patterns associated with this map. This is useful for
13606 debugging maps and ACLs. The output format is composed by one line par
13607 matching type. Each line is composed by space-delimited series of words.
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013608
13609 The first two words are:
13610
13611 <match method>: The match method applied. It can be "found", "bool",
13612 "int", "ip", "bin", "len", "str", "beg", "sub", "dir",
13613 "dom", "end" or "reg".
13614
13615 <match result>: The result. Can be "match" or "no-match".
13616
13617 The following words are returned only if the pattern matches an entry.
13618
13619 <index type>: "tree" or "list". The internal lookup algorithm.
13620
13621 <case>: "case-insensitive" or "case-sensitive". The
13622 interpretation of the case.
13623
13624 <entry matched>: match="<entry>". Return the matched pattern. It is
13625 useful with regular expressions.
13626
13627 The two last word are used to show the returned value and its type. With the
13628 "acl" case, the pattern doesn't exist.
13629
13630 return=nothing: No return because there are no "map".
13631 return="<value>": The value returned in the string format.
13632 return=cannot-display: The value cannot be converted as string.
13633
13634 type="<type>": The type of the returned sample.
13635
Willy Tarreaud63335a2010-02-26 12:56:52 +010013636get weight <backend>/<server>
13637 Report the current weight and the initial weight of server <server> in
13638 backend <backend> or an error if either doesn't exist. The initial weight is
13639 the one that appears in the configuration file. Both are normally equal
13640 unless the current weight has been changed. Both the backend and the server
13641 may be specified either by their name or by their numeric ID, prefixed with a
Willy Tarreauf5f31922011-08-02 11:32:07 +020013642 sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013643
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013644help
13645 Print the list of known keywords and their basic usage. The same help screen
13646 is also displayed for unknown commands.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013647
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013648prompt
13649 Toggle the prompt at the beginning of the line and enter or leave interactive
13650 mode. In interactive mode, the connection is not closed after a command
13651 completes. Instead, the prompt will appear again, indicating the user that
13652 the interpreter is waiting for a new command. The prompt consists in a right
13653 angle bracket followed by a space "> ". This mode is particularly convenient
13654 when one wants to periodically check information such as stats or errors.
13655 It is also a good idea to enter interactive mode before issuing a "help"
13656 command.
13657
13658quit
13659 Close the connection when in interactive mode.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013660
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013661set map <map> [<key>|#<ref>] <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013662 Modify the value corresponding to each key <key> in a map <map>. <map> is the
13663 #<id> or <file> returned by "show map". If the <ref> is used in place of
13664 <key>, only the entry pointed by <ref> is changed. The new value is <value>.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013665
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013666set maxconn frontend <frontend> <value>
Willy Tarreau3c7a79d2012-09-26 21:07:15 +020013667 Dynamically change the specified frontend's maxconn setting. Any positive
13668 value is allowed including zero, but setting values larger than the global
13669 maxconn does not make much sense. If the limit is increased and connections
13670 were pending, they will immediately be accepted. If it is lowered to a value
13671 below the current number of connections, new connections acceptation will be
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013672 delayed until the threshold is reached. The frontend might be specified by
13673 either its name or its numeric ID prefixed with a sharp ('#').
13674
Willy Tarreau91886b62011-09-07 14:38:31 +020013675set maxconn global <maxconn>
13676 Dynamically change the global maxconn setting within the range defined by the
13677 initial global maxconn setting. If it is increased and connections were
13678 pending, they will immediately be accepted. If it is lowered to a value below
13679 the current number of connections, new connections acceptation will be
13680 delayed until the threshold is reached. A value of zero restores the initial
13681 setting.
13682
Willy Tarreauf5b22872011-09-07 16:13:44 +020013683set rate-limit connections global <value>
13684 Change the process-wide connection rate limit, which is set by the global
13685 'maxconnrate' setting. A value of zero disables the limitation. This limit
13686 applies to all frontends and the change has an immediate effect. The value
13687 is passed in number of connections per second.
13688
William Lallemandd85f9172012-11-09 17:05:39 +010013689set rate-limit http-compression global <value>
13690 Change the maximum input compression rate, which is set by the global
13691 'maxcomprate' setting. A value of zero disables the limitation. The value is
William Lallemand096f5542012-11-19 17:26:05 +010013692 passed in number of kilobytes per second. The value is available in the "show
13693 info" on the line "CompressBpsRateLim" in bytes.
William Lallemandd85f9172012-11-09 17:05:39 +010013694
Willy Tarreau93e7c002013-10-07 18:51:07 +020013695set rate-limit sessions global <value>
13696 Change the process-wide session rate limit, which is set by the global
13697 'maxsessrate' setting. A value of zero disables the limitation. This limit
13698 applies to all frontends and the change has an immediate effect. The value
13699 is passed in number of sessions per second.
13700
Willy Tarreaue43d5322013-10-07 20:01:52 +020013701set rate-limit ssl-sessions global <value>
13702 Change the process-wide SSL session rate limit, which is set by the global
13703 'maxsslrate' setting. A value of zero disables the limitation. This limit
13704 applies to all frontends and the change has an immediate effect. The value
13705 is passed in number of sessions per second sent to the SSL stack. It applies
13706 before the handshake in order to protect the stack against handshake abuses.
13707
Willy Tarreau2a4b70f2014-05-22 18:42:35 +020013708set server <backend>/<server> agent [ up | down ]
13709 Force a server's agent to a new state. This can be useful to immediately
13710 switch a server's state regardless of some slow agent checks for example.
13711 Note that the change is propagated to tracking servers if any.
13712
13713set server <backend>/<server> health [ up | stopping | down ]
13714 Force a server's health to a new state. This can be useful to immediately
13715 switch a server's state regardless of some slow health checks for example.
13716 Note that the change is propagated to tracking servers if any.
13717
13718set server <backend>/<server> state [ ready | drain | maint ]
13719 Force a server's administrative state to a new state. This can be useful to
13720 disable load balancing and/or any traffic to a server. Setting the state to
13721 "ready" puts the server in normal mode, and the command is the equivalent of
13722 the "enable server" command. Setting the state to "maint" disables any traffic
13723 to the server as well as any health checks. This is the equivalent of the
13724 "disable server" command. Setting the mode to "drain" only removes the server
13725 from load balancing but still allows it to be checked and to accept new
13726 persistent connections. Changes are propagated to tracking servers if any.
13727
13728set server <backend>/<server> weight <weight>[%]
13729 Change a server's weight to the value passed in argument. This is the exact
13730 equivalent of the "set weight" command below.
13731
Emeric Brun4147b2e2014-06-16 18:36:30 +020013732set ssl ocsp-response <response>
13733 This command is used to update an OCSP Response for a certificate (see "crt"
13734 on "bind" lines). Same controls are performed as during the initial loading of
13735 the response. The <response> must be passed as a base64 encoded string of the
13736 DER encoded response from the OCSP server.
13737
13738 Example:
13739 openssl ocsp -issuer issuer.pem -cert server.pem \
13740 -host ocsp.issuer.com:80 -respout resp.der
13741 echo "set ssl ocsp-response $(base64 -w 10000 resp.der)" | \
13742 socat stdio /var/run/haproxy.stat
13743
Willy Tarreau47060b62013-08-01 21:11:42 +020013744set table <table> key <key> [data.<data_type> <value>]*
Willy Tarreau654694e2012-06-07 01:03:16 +020013745 Create or update a stick-table entry in the table. If the key is not present,
13746 an entry is inserted. See stick-table in section 4.2 to find all possible
13747 values for <data_type>. The most likely use consists in dynamically entering
13748 entries for source IP addresses, with a flag in gpc0 to dynamically block an
Willy Tarreau47060b62013-08-01 21:11:42 +020013749 IP address or affect its quality of service. It is possible to pass multiple
13750 data_types in a single call.
Willy Tarreau654694e2012-06-07 01:03:16 +020013751
Willy Tarreaud63335a2010-02-26 12:56:52 +010013752set timeout cli <delay>
13753 Change the CLI interface timeout for current connection. This can be useful
13754 during long debugging sessions where the user needs to constantly inspect
13755 some indicators without being disconnected. The delay is passed in seconds.
13756
13757set weight <backend>/<server> <weight>[%]
13758 Change a server's weight to the value passed in argument. If the value ends
13759 with the '%' sign, then the new weight will be relative to the initially
Simon Horman58b5d292013-02-12 10:45:52 +090013760 configured weight. Absolute weights are permitted between 0 and 256.
13761 Relative weights must be positive with the resulting absolute weight is
13762 capped at 256. Servers which are part of a farm running a static
13763 load-balancing algorithm have stricter limitations because the weight
13764 cannot change once set. Thus for these servers, the only accepted values
13765 are 0 and 100% (or 0 and the initial weight). Changes take effect
13766 immediately, though certain LB algorithms require a certain amount of
13767 requests to consider changes. A typical usage of this command is to
13768 disable a server during an update by setting its weight to zero, then to
13769 enable it again after the update by setting it back to 100%. This command
13770 is restricted and can only be issued on sockets configured for level
13771 "admin". Both the backend and the server may be specified either by their
13772 name or by their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013773
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013774show errors [<iid>]
13775 Dump last known request and response errors collected by frontends and
13776 backends. If <iid> is specified, the limit the dump to errors concerning
Willy Tarreau6162db22009-10-10 17:13:00 +020013777 either frontend or backend whose ID is <iid>. This command is restricted
13778 and can only be issued on sockets configured for levels "operator" or
13779 "admin".
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013780
13781 The errors which may be collected are the last request and response errors
13782 caused by protocol violations, often due to invalid characters in header
13783 names. The report precisely indicates what exact character violated the
13784 protocol. Other important information such as the exact date the error was
13785 detected, frontend and backend names, the server name (when known), the
13786 internal session ID and the source address which has initiated the session
13787 are reported too.
13788
13789 All characters are returned, and non-printable characters are encoded. The
13790 most common ones (\t = 9, \n = 10, \r = 13 and \e = 27) are encoded as one
13791 letter following a backslash. The backslash itself is encoded as '\\' to
13792 avoid confusion. Other non-printable characters are encoded '\xNN' where
13793 NN is the two-digits hexadecimal representation of the character's ASCII
13794 code.
13795
13796 Lines are prefixed with the position of their first character, starting at 0
13797 for the beginning of the buffer. At most one input line is printed per line,
13798 and large lines will be broken into multiple consecutive output lines so that
13799 the output never goes beyond 79 characters wide. It is easy to detect if a
13800 line was broken, because it will not end with '\n' and the next line's offset
13801 will be followed by a '+' sign, indicating it is a continuation of previous
13802 line.
13803
13804 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013805 $ echo "show errors" | socat stdio /tmp/sock1
13806 >>> [04/Mar/2009:15:46:56.081] backend http-in (#2) : invalid response
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013807 src 127.0.0.1, session #54, frontend fe-eth0 (#1), server s2 (#1)
13808 response length 213 bytes, error at position 23:
13809
13810 00000 HTTP/1.0 200 OK\r\n
13811 00017 header/bizarre:blah\r\n
13812 00038 Location: blah\r\n
13813 00054 Long-line: this is a very long line which should b
13814 00104+ e broken into multiple lines on the output buffer,
13815 00154+ otherwise it would be too large to print in a ter
13816 00204+ minal\r\n
13817 00211 \r\n
13818
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013819 In the example above, we see that the backend "http-in" which has internal
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013820 ID 2 has blocked an invalid response from its server s2 which has internal
13821 ID 1. The request was on session 54 initiated by source 127.0.0.1 and
13822 received by frontend fe-eth0 whose ID is 1. The total response length was
13823 213 bytes when the error was detected, and the error was at byte 23. This
13824 is the slash ('/') in header name "header/bizarre", which is not a valid
13825 HTTP character for a header name.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013826
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013827show info
13828 Dump info about haproxy status on current process.
13829
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013830show map [<map>]
13831 Dump info about map converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013832 maps is returned. If a <map> is specified, its contents are dumped. <map> is
13833 the #<id> or <file>. The first column is a unique identifier. It can be used
13834 as reference for the operation "del map" and "set map". The second column is
13835 the pattern and the third column is the sample if available. The data returned
13836 are not directly a list of available maps, but are the list of all patterns
13837 composing any map. Many of these patterns can be shared with ACL.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013838
13839show acl [<acl>]
13840 Dump info about acl converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013841 acls is returned. If a <acl> is specified, its contents are dumped. <acl> if
13842 the #<id> or <file>. The dump format is the same than the map even for the
13843 sample value. The data returned are not a list of available ACL, but are the
13844 list of all patterns composing any ACL. Many of these patterns can be shared
13845 with maps.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013846
Willy Tarreau12833bb2014-01-28 16:49:56 +010013847show pools
13848 Dump the status of internal memory pools. This is useful to track memory
13849 usage when suspecting a memory leak for example. It does exactly the same
13850 as the SIGQUIT when running in foreground except that it does not flush
13851 the pools.
13852
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013853show sess
13854 Dump all known sessions. Avoid doing this on slow connections as this can
Willy Tarreau6162db22009-10-10 17:13:00 +020013855 be huge. This command is restricted and can only be issued on sockets
13856 configured for levels "operator" or "admin".
13857
Willy Tarreau66dc20a2010-03-05 17:53:32 +010013858show sess <id>
13859 Display a lot of internal information about the specified session identifier.
13860 This identifier is the first field at the beginning of the lines in the dumps
13861 of "show sess" (it corresponds to the session pointer). Those information are
13862 useless to most users but may be used by haproxy developers to troubleshoot a
13863 complex bug. The output format is intentionally not documented so that it can
Oliviere99d44d2014-09-05 18:49:10 +020013864 freely evolve depending on demands. You may find a description of all fields
13865 returned in src/dumpstats.c
13866
13867 The special id "all" dumps the states of all sessions, which must be avoided
13868 as much as possible as it is highly CPU intensive and can take a lot of time.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013869
13870show stat [<iid> <type> <sid>]
13871 Dump statistics in the CSV format. By passing <id>, <type> and <sid>, it is
13872 possible to dump only selected items :
13873 - <iid> is a proxy ID, -1 to dump everything
13874 - <type> selects the type of dumpable objects : 1 for frontends, 2 for
13875 backends, 4 for servers, -1 for everything. These values can be ORed,
13876 for example:
13877 1 + 2 = 3 -> frontend + backend.
13878 1 + 2 + 4 = 7 -> frontend + backend + server.
13879 - <sid> is a server ID, -1 to dump everything from the selected proxy.
13880
13881 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013882 $ echo "show info;show stat" | socat stdio unix-connect:/tmp/sock1
13883 >>> Name: HAProxy
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013884 Version: 1.4-dev2-49
13885 Release_date: 2009/09/23
13886 Nbproc: 1
13887 Process_num: 1
13888 (...)
13889
13890 # pxname,svname,qcur,qmax,scur,smax,slim,stot,bin,bout,dreq, (...)
13891 stats,FRONTEND,,,0,0,1000,0,0,0,0,0,0,,,,,OPEN,,,,,,,,,1,1,0, (...)
13892 stats,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,0,0,0,,0,250,(...)
13893 (...)
13894 www1,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,1,1,0,,0,250, (...)
13895
13896 $
13897
13898 Here, two commands have been issued at once. That way it's easy to find
13899 which process the stats apply to in multi-process mode. Notice the empty
13900 line after the information output which marks the end of the first block.
13901 A similar empty line appears at the end of the second block (stats) so that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013902 the reader knows the output has not been truncated.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013903
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013904show table
13905 Dump general information on all known stick-tables. Their name is returned
13906 (the name of the proxy which holds them), their type (currently zero, always
13907 IP), their size in maximum possible number of entries, and the number of
13908 entries currently in use.
13909
13910 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013911 $ echo "show table" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013912 >>> # table: front_pub, type: ip, size:204800, used:171454
13913 >>> # table: back_rdp, type: ip, size:204800, used:0
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013914
Simon Horman17bce342011-06-15 15:18:47 +090013915show table <name> [ data.<type> <operator> <value> ] | [ key <key> ]
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013916 Dump contents of stick-table <name>. In this mode, a first line of generic
13917 information about the table is reported as with "show table", then all
13918 entries are dumped. Since this can be quite heavy, it is possible to specify
Simon Horman17bce342011-06-15 15:18:47 +090013919 a filter in order to specify what entries to display.
13920
13921 When the "data." form is used the filter applies to the stored data (see
13922 "stick-table" in section 4.2). A stored data type must be specified
13923 in <type>, and this data type must be stored in the table otherwise an
13924 error is reported. The data is compared according to <operator> with the
13925 64-bit integer <value>. Operators are the same as with the ACLs :
13926
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013927 - eq : match entries whose data is equal to this value
13928 - ne : match entries whose data is not equal to this value
13929 - le : match entries whose data is less than or equal to this value
13930 - ge : match entries whose data is greater than or equal to this value
13931 - lt : match entries whose data is less than this value
13932 - gt : match entries whose data is greater than this value
13933
Simon Hormanc88b8872011-06-15 15:18:49 +090013934
13935 When the key form is used the entry <key> is shown. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013936 same type as the table, which currently is limited to IPv4, IPv6, integer,
13937 and string.
Simon Horman17bce342011-06-15 15:18:47 +090013938
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013939 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013940 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013941 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013942 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13943 bytes_out_rate(60000)=187
13944 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13945 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013946
Willy Tarreau62a36c42010-08-17 15:53:10 +020013947 $ echo "show table http_proxy data.gpc0 gt 0" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013948 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013949 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13950 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013951
Willy Tarreau62a36c42010-08-17 15:53:10 +020013952 $ echo "show table http_proxy data.conn_rate gt 5" | \
13953 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 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13956 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013957
Simon Horman17bce342011-06-15 15:18:47 +090013958 $ echo "show table http_proxy key 127.0.0.2" | \
13959 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013960 >>> # table: http_proxy, type: ip, size:204800, used:2
Simon Horman17bce342011-06-15 15:18:47 +090013961 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13962 bytes_out_rate(60000)=191
13963
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013964 When the data criterion applies to a dynamic value dependent on time such as
13965 a bytes rate, the value is dynamically computed during the evaluation of the
13966 entry in order to decide whether it has to be dumped or not. This means that
13967 such a filter could match for some time then not match anymore because as
13968 time goes, the average event rate drops.
13969
13970 It is possible to use this to extract lists of IP addresses abusing the
13971 service, in order to monitor them or even blacklist them in a firewall.
13972 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013973 $ echo "show table http_proxy data.gpc0 gt 0" \
13974 | socat stdio /tmp/sock1 \
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013975 | fgrep 'key=' | cut -d' ' -f2 | cut -d= -f2 > abusers-ip.txt
13976 ( or | awk '/key/{ print a[split($2,a,"=")]; }' )
Krzysztof Piotr Oledzki719e7262009-10-04 15:02:46 +020013977
Willy Tarreau532a4502011-09-07 22:37:44 +020013978shutdown frontend <frontend>
13979 Completely delete the specified frontend. All the ports it was bound to will
13980 be released. It will not be possible to enable the frontend anymore after
13981 this operation. This is intended to be used in environments where stopping a
13982 proxy is not even imaginable but a misconfigured proxy must be fixed. That
13983 way it's possible to release the port and bind it into another process to
13984 restore operations. The frontend will not appear at all on the stats page
13985 once it is terminated.
13986
13987 The frontend may be specified either by its name or by its numeric ID,
13988 prefixed with a sharp ('#').
13989
13990 This command is restricted and can only be issued on sockets configured for
13991 level "admin".
13992
Willy Tarreaua295edc2011-09-07 23:21:03 +020013993shutdown session <id>
13994 Immediately terminate the session matching the specified session identifier.
13995 This identifier is the first field at the beginning of the lines in the dumps
13996 of "show sess" (it corresponds to the session pointer). This can be used to
13997 terminate a long-running session without waiting for a timeout or when an
13998 endless transfer is ongoing. Such terminated sessions are reported with a 'K'
13999 flag in the logs.
14000
Cyril Bontéde9789b2014-07-12 18:22:42 +020014001shutdown sessions server <backend>/<server>
Willy Tarreau52b2d222011-09-07 23:48:48 +020014002 Immediately terminate all the sessions attached to the specified server. This
14003 can be used to terminate long-running sessions after a server is put into
14004 maintenance mode, for instance. Such terminated sessions are reported with a
14005 'K' flag in the logs.
14006
Willy Tarreau0ba27502007-12-24 16:55:16 +010014007/*
14008 * Local variables:
14009 * fill-column: 79
14010 * End:
14011 */