blob: 7c4e18bf51d9224814486a1b8525c0a9dc253a90 [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 Tarreaube1f8a82014-10-31 10:06:03 +01005 version 1.5.8
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreaube1f8a82014-10-31 10:06:03 +01007 2014/10/31
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
1151peer <peername> <ip>:<port>
1152 Defines a peer inside a peers section.
1153 If <peername> is set to the local peer name (by default hostname, or forced
1154 using "-L" command line option), haproxy will listen for incoming remote peer
1155 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1156 to join the remote peer, and <peername> is used at the protocol level to
1157 identify and validate the remote peer on the server side.
1158
1159 During a soft restart, local peer <ip>:<port> is used by the old instance to
1160 connect the new one and initiate a complete replication (teaching process).
1161
1162 It is strongly recommended to have the exact same peers declaration on all
1163 peers and to only rely on the "-L" command line argument to change the local
1164 peer name. This makes it easier to maintain coherent configuration files
1165 across all peers.
1166
Willy Tarreaudad36a32013-03-11 01:20:04 +01001167 Any part of the address string may reference any number of environment
1168 variables by preceding their name with a dollar sign ('$') and optionally
1169 enclosing them with braces ('{}'), similarly to what is done in Bourne shell.
1170
Cyril Bontédc4d9032012-04-08 21:57:39 +02001171 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001172 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001173 peer haproxy1 192.168.0.1:1024
1174 peer haproxy2 192.168.0.2:1024
1175 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001176
1177 backend mybackend
1178 mode tcp
1179 balance roundrobin
1180 stick-table type ip size 20k peers mypeers
1181 stick on src
1182
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001183 server srv1 192.168.0.30:80
1184 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001185
1186
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011874. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001188----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001189
Willy Tarreau6a06a402007-07-15 20:15:28 +02001190Proxy configuration can be located in a set of sections :
1191 - defaults <name>
1192 - frontend <name>
1193 - backend <name>
1194 - listen <name>
1195
1196A "defaults" section sets default parameters for all other sections following
1197its declaration. Those default parameters are reset by the next "defaults"
1198section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001199section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001200
1201A "frontend" section describes a set of listening sockets accepting client
1202connections.
1203
1204A "backend" section describes a set of servers to which the proxy will connect
1205to forward incoming connections.
1206
1207A "listen" section defines a complete proxy with its frontend and backend
1208parts combined in one section. It is generally useful for TCP-only traffic.
1209
Willy Tarreau0ba27502007-12-24 16:55:16 +01001210All proxy names must be formed from upper and lower case letters, digits,
1211'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1212case-sensitive, which means that "www" and "WWW" are two different proxies.
1213
1214Historically, all proxy names could overlap, it just caused troubles in the
1215logs. Since the introduction of content switching, it is mandatory that two
1216proxies with overlapping capabilities (frontend/backend) have different names.
1217However, it is still permitted that a frontend and a backend share the same
1218name, as this configuration seems to be commonly encountered.
1219
1220Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1221and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001222bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001223protocol, and can interact with it by allowing, blocking, switching, adding,
1224modifying, or removing arbitrary contents in requests or responses, based on
1225arbitrary criteria.
1226
Willy Tarreau70dffda2014-01-30 03:07:23 +01001227In HTTP mode, the processing applied to requests and responses flowing over
1228a connection depends in the combination of the frontend's HTTP options and
1229the backend's. HAProxy supports 5 connection modes :
1230
1231 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1232 requests and responses are processed, and connections remain open but idle
1233 between responses and new requests.
1234
1235 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1236 1.0 to 1.5-dev21 : only the first request and response are processed, and
1237 everything else is forwarded with no analysis at all. This mode should not
1238 be used as it creates lots of trouble with logging and HTTP processing.
1239
1240 - PCL: passive close ("option httpclose") : exactly the same as tunnel mode,
1241 but with "Connection: close" appended in both directions to try to make
1242 both ends close after the first request/response exchange.
1243
1244 - SCL: server close ("option http-server-close") : the server-facing
1245 connection is closed after the end of the response is received, but the
1246 client-facing connection remains open.
1247
1248 - FCL: forced close ("option forceclose") : the connection is actively closed
1249 after the end of the response.
1250
1251The effective mode that will be applied to a connection passing through a
1252frontend and a backend can be determined by both proxy modes according to the
1253following matrix, but in short, the modes are symmetric, keep-alive is the
1254weakest option and force close is the strongest.
1255
1256 Backend mode
1257
1258 | KAL | TUN | PCL | SCL | FCL
1259 ----+-----+-----+-----+-----+----
1260 KAL | KAL | TUN | PCL | SCL | FCL
1261 ----+-----+-----+-----+-----+----
1262 TUN | TUN | TUN | PCL | SCL | FCL
1263 Frontend ----+-----+-----+-----+-----+----
1264 mode PCL | PCL | PCL | PCL | FCL | FCL
1265 ----+-----+-----+-----+-----+----
1266 SCL | SCL | SCL | FCL | SCL | FCL
1267 ----+-----+-----+-----+-----+----
1268 FCL | FCL | FCL | FCL | FCL | FCL
1269
Willy Tarreau0ba27502007-12-24 16:55:16 +01001270
Willy Tarreau70dffda2014-01-30 03:07:23 +01001271
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012724.1. Proxy keywords matrix
1273--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001274
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001275The following list of keywords is supported. Most of them may only be used in a
1276limited set of section types. Some of them are marked as "deprecated" because
1277they are inherited from an old syntax which may be confusing or functionally
1278limited, and there are new recommended keywords to replace them. Keywords
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001279marked with "(*)" can be optionally inverted using the "no" prefix, eg. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001280option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02001281and must be disabled for a specific instance. Such options may also be prefixed
1282with "default" in order to restore default settings regardless of what has been
1283specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001284
Willy Tarreau6a06a402007-07-15 20:15:28 +02001285
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001286 keyword defaults frontend listen backend
1287------------------------------------+----------+----------+---------+---------
1288acl - X X X
1289appsession - - X X
1290backlog X X X -
1291balance X - X X
1292bind - X X -
1293bind-process X X X X
1294block - X X X
1295capture cookie - X X -
1296capture request header - X X -
1297capture response header - X X -
1298clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02001299compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001300contimeout (deprecated) X - X X
1301cookie X - X X
1302default-server X - X X
1303default_backend X X X -
1304description - X X X
1305disabled X X X X
1306dispatch - - X X
1307enabled X X X X
1308errorfile X X X X
1309errorloc X X X X
1310errorloc302 X X X X
1311-- keyword -------------------------- defaults - frontend - listen -- backend -
1312errorloc303 X X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001313force-persist - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001314fullconn X - X X
1315grace X X X X
1316hash-type X - X X
1317http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01001318http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02001319http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001320http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02001321http-response - X X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02001322http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001323id - X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001324ignore-persist - X X X
William Lallemand0f99e342011-10-12 17:50:54 +02001325log (*) X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02001326max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001327maxconn X X X -
1328mode X X X X
1329monitor fail - X X -
1330monitor-net X X X -
1331monitor-uri X X X -
1332option abortonclose (*) X - X X
1333option accept-invalid-http-request (*) X X X -
1334option accept-invalid-http-response (*) X - X X
1335option allbackups (*) X - X X
1336option checkcache (*) X - X X
1337option clitcpka (*) X X X -
1338option contstats (*) X X X -
1339option dontlog-normal (*) X X X -
1340option dontlognull (*) X X X -
1341option forceclose (*) X X X X
1342-- keyword -------------------------- defaults - frontend - listen -- backend -
1343option forwardfor X X X X
Willy Tarreau16bfb022010-01-16 19:48:41 +01001344option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02001345option http-no-delay (*) X X X X
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02001346option http-pretend-keepalive (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001347option http-server-close (*) X X X X
Willy Tarreau02bce8b2014-01-30 00:15:28 +01001348option http-tunnel (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001349option http-use-proxy-header (*) X X X -
1350option httpchk X - X X
1351option httpclose (*) X X X X
1352option httplog X X X X
1353option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001354option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02001355option ldap-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001356option log-health-checks (*) X - X X
1357option log-separate-errors (*) X X X -
1358option logasap (*) X X X -
1359option mysql-check X - X X
Rauf Kuliyev38b41562011-01-04 15:14:13 +01001360option pgsql-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001361option nolinger (*) X X X X
1362option originalto X X X X
1363option persist (*) X - X X
1364option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02001365option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001366option smtpchk X - X X
1367option socket-stats (*) X X X -
1368option splice-auto (*) X X X X
1369option splice-request (*) X X X X
1370option splice-response (*) X X X X
1371option srvtcpka (*) X - X X
1372option ssl-hello-chk X - X X
1373-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01001374option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001375option tcp-smart-accept (*) X X X -
1376option tcp-smart-connect (*) X - X X
1377option tcpka X X X X
1378option tcplog X X X X
1379option transparent (*) X - X X
1380persist rdp-cookie X - X X
1381rate-limit sessions X X X -
1382redirect - X X X
1383redisp (deprecated) X - X X
1384redispatch (deprecated) X - X X
1385reqadd - X X X
1386reqallow - X X X
1387reqdel - X X X
1388reqdeny - X X X
1389reqiallow - X X X
1390reqidel - X X X
1391reqideny - X X X
1392reqipass - X X X
1393reqirep - X X X
1394reqisetbe - X X X
1395reqitarpit - X X X
1396reqpass - X X X
1397reqrep - X X X
1398-- keyword -------------------------- defaults - frontend - listen -- backend -
1399reqsetbe - X X X
1400reqtarpit - X X X
1401retries X - X X
1402rspadd - X X X
1403rspdel - X X X
1404rspdeny - X X X
1405rspidel - X X X
1406rspideny - X X X
1407rspirep - X X X
1408rsprep - X X X
1409server - - X X
1410source X - X X
1411srvtimeout (deprecated) X - X X
Cyril Bonté66c327d2010-10-12 00:14:37 +02001412stats admin - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001413stats auth X - X X
1414stats enable X - X X
1415stats hide-version X - X X
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02001416stats http-request - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001417stats realm X - X X
1418stats refresh X - X X
1419stats scope X - X X
1420stats show-desc X - X X
1421stats show-legends X - X X
1422stats show-node X - X X
1423stats uri X - X X
1424-- keyword -------------------------- defaults - frontend - listen -- backend -
1425stick match - - X X
1426stick on - - X X
1427stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02001428stick store-response - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001429stick-table - - X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02001430tcp-check connect - - X X
1431tcp-check expect - - X X
1432tcp-check send - - X X
1433tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02001434tcp-request connection - X X -
1435tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02001436tcp-request inspect-delay - X X X
Emeric Brun0a3b67f2010-09-24 15:34:53 +02001437tcp-response content - - X X
1438tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001439timeout check X - X X
1440timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02001441timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001442timeout clitimeout (deprecated) X X X -
1443timeout connect X - X X
1444timeout contimeout (deprecated) X - X X
1445timeout http-keep-alive X X X X
1446timeout http-request X X X X
1447timeout queue X - X X
1448timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02001449timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001450timeout srvtimeout (deprecated) X - X X
1451timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02001452timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001453transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01001454unique-id-format X X X -
1455unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001456use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02001457use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001458------------------------------------+----------+----------+---------+---------
1459 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02001460
Willy Tarreau0ba27502007-12-24 16:55:16 +01001461
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014624.2. Alphabetically sorted keywords reference
1463---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001464
1465This section provides a description of each keyword and its usage.
1466
1467
1468acl <aclname> <criterion> [flags] [operator] <value> ...
1469 Declare or complete an access list.
1470 May be used in sections : defaults | frontend | listen | backend
1471 no | yes | yes | yes
1472 Example:
1473 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1474 acl invalid_src src_port 0:1023
1475 acl local_dst hdr(host) -i localhost
1476
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001477 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001478
1479
Cyril Bontéb21570a2009-11-29 20:04:48 +01001480appsession <cookie> len <length> timeout <holdtime>
1481 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001482 Define session stickiness on an existing application cookie.
1483 May be used in sections : defaults | frontend | listen | backend
1484 no | no | yes | yes
1485 Arguments :
1486 <cookie> this is the name of the cookie used by the application and which
1487 HAProxy will have to learn for each new session.
1488
Cyril Bontéb21570a2009-11-29 20:04:48 +01001489 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01001490 checked in each cookie value.
1491
1492 <holdtime> this is the time after which the cookie will be removed from
1493 memory if unused. If no unit is specified, this time is in
1494 milliseconds.
1495
Cyril Bontébf47aeb2009-10-15 00:15:40 +02001496 request-learn
1497 If this option is specified, then haproxy will be able to learn
1498 the cookie found in the request in case the server does not
1499 specify any in response. This is typically what happens with
1500 PHPSESSID cookies, or when haproxy's session expires before
1501 the application's session and the correct server is selected.
1502 It is recommended to specify this option to improve reliability.
1503
Cyril Bontéb21570a2009-11-29 20:04:48 +01001504 prefix When this option is specified, haproxy will match on the cookie
1505 prefix (or URL parameter prefix). The appsession value is the
1506 data following this prefix.
1507
1508 Example :
1509 appsession ASPSESSIONID len 64 timeout 3h prefix
1510
1511 This will match the cookie ASPSESSIONIDXXXX=XXXXX,
1512 the appsession value will be XXXX=XXXXX.
1513
1514 mode This option allows to change the URL parser mode.
1515 2 modes are currently supported :
1516 - path-parameters :
1517 The parser looks for the appsession in the path parameters
1518 part (each parameter is separated by a semi-colon), which is
1519 convenient for JSESSIONID for example.
1520 This is the default mode if the option is not set.
1521 - query-string :
1522 In this mode, the parser will look for the appsession in the
1523 query string.
1524
Willy Tarreau0ba27502007-12-24 16:55:16 +01001525 When an application cookie is defined in a backend, HAProxy will check when
1526 the server sets such a cookie, and will store its value in a table, and
1527 associate it with the server's identifier. Up to <length> characters from
1528 the value will be retained. On each connection, haproxy will look for this
Cyril Bontéb21570a2009-11-29 20:04:48 +01001529 cookie both in the "Cookie:" headers, and as a URL parameter (depending on
1530 the mode used). If a known value is found, the client will be directed to the
1531 server associated with this value. Otherwise, the load balancing algorithm is
Willy Tarreau0ba27502007-12-24 16:55:16 +01001532 applied. Cookies are automatically removed from memory when they have been
1533 unused for a duration longer than <holdtime>.
1534
1535 The definition of an application cookie is limited to one per backend.
1536
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001537 Note : Consider not using this feature in multi-process mode (nbproc > 1)
1538 unless you know what you do : memory is not shared between the
1539 processes, which can result in random behaviours.
1540
Willy Tarreau0ba27502007-12-24 16:55:16 +01001541 Example :
1542 appsession JSESSIONID len 52 timeout 3h
1543
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001544 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
1545 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001546
1547
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001548backlog <conns>
1549 Give hints to the system about the approximate listen backlog desired size
1550 May be used in sections : defaults | frontend | listen | backend
1551 yes | yes | yes | no
1552 Arguments :
1553 <conns> is the number of pending connections. Depending on the operating
1554 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02001555 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001556
1557 In order to protect against SYN flood attacks, one solution is to increase
1558 the system's SYN backlog size. Depending on the system, sometimes it is just
1559 tunable via a system parameter, sometimes it is not adjustable at all, and
1560 sometimes the system relies on hints given by the application at the time of
1561 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
1562 to the listen() syscall. On systems which can make use of this value, it can
1563 sometimes be useful to be able to specify a different value, hence this
1564 backlog parameter.
1565
1566 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
1567 used as a hint and the system accepts up to the smallest greater power of
1568 two, and never more than some limits (usually 32768).
1569
1570 See also : "maxconn" and the target operating system's tuning guide.
1571
1572
Willy Tarreau0ba27502007-12-24 16:55:16 +01001573balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02001574balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001575 Define the load balancing algorithm to be used in a backend.
1576 May be used in sections : defaults | frontend | listen | backend
1577 yes | no | yes | yes
1578 Arguments :
1579 <algorithm> is the algorithm used to select a server when doing load
1580 balancing. This only applies when no persistence information
1581 is available, or when a connection is redispatched to another
1582 server. <algorithm> may be one of the following :
1583
1584 roundrobin Each server is used in turns, according to their weights.
1585 This is the smoothest and fairest algorithm when the server's
1586 processing time remains equally distributed. This algorithm
1587 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02001588 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08001589 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02001590 large farms, when a server becomes up after having been down
1591 for a very short time, it may sometimes take a few hundreds
1592 requests for it to be re-integrated into the farm and start
1593 receiving traffic. This is normal, though very rare. It is
1594 indicated here in case you would have the chance to observe
1595 it, so that you don't worry.
1596
1597 static-rr Each server is used in turns, according to their weights.
1598 This algorithm is as similar to roundrobin except that it is
1599 static, which means that changing a server's weight on the
1600 fly will have no effect. On the other hand, it has no design
1601 limitation on the number of servers, and when a server goes
1602 up, it is always immediately reintroduced into the farm, once
1603 the full map is recomputed. It also uses slightly less CPU to
1604 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01001605
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01001606 leastconn The server with the lowest number of connections receives the
1607 connection. Round-robin is performed within groups of servers
1608 of the same load to ensure that all servers will be used. Use
1609 of this algorithm is recommended where very long sessions are
1610 expected, such as LDAP, SQL, TSE, etc... but is not very well
1611 suited for protocols using short sessions such as HTTP. This
1612 algorithm is dynamic, which means that server weights may be
1613 adjusted on the fly for slow starts for instance.
1614
Willy Tarreauf09c6602012-02-13 17:12:08 +01001615 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001616 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01001617 identifier to the highest (see server parameter "id"), which
1618 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02001619 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01001620 not make sense to use this algorithm without setting maxconn.
1621 The purpose of this algorithm is to always use the smallest
1622 number of servers so that extra servers can be powered off
1623 during non-intensive hours. This algorithm ignores the server
1624 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02001625 or IMAP than HTTP, though it can be useful there too. In
1626 order to use this algorithm efficiently, it is recommended
1627 that a cloud controller regularly checks server usage to turn
1628 them off when unused, and regularly checks backend queue to
1629 turn new servers on when the queue inflates. Alternatively,
1630 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01001631
Willy Tarreau0ba27502007-12-24 16:55:16 +01001632 source The source IP address is hashed and divided by the total
1633 weight of the running servers to designate which server will
1634 receive the request. This ensures that the same client IP
1635 address will always reach the same server as long as no
1636 server goes down or up. If the hash result changes due to the
1637 number of running servers changing, many clients will be
1638 directed to a different server. This algorithm is generally
1639 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001640 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01001641 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001642 static by default, which means that changing a server's
1643 weight on the fly will have no effect, but this can be
1644 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001645
Oskar Stolc8dc41842012-05-19 10:19:54 +01001646 uri This algorithm hashes either the left part of the URI (before
1647 the question mark) or the whole URI (if the "whole" parameter
1648 is present) and divides the hash value by the total weight of
1649 the running servers. The result designates which server will
1650 receive the request. This ensures that the same URI will
1651 always be directed to the same server as long as no server
1652 goes up or down. This is used with proxy caches and
1653 anti-virus proxies in order to maximize the cache hit rate.
1654 Note that this algorithm may only be used in an HTTP backend.
1655 This algorithm is static by default, which means that
1656 changing a server's weight on the fly will have no effect,
1657 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001658
Oskar Stolc8dc41842012-05-19 10:19:54 +01001659 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001660 "depth", both followed by a positive integer number. These
1661 options may be helpful when it is needed to balance servers
1662 based on the beginning of the URI only. The "len" parameter
1663 indicates that the algorithm should only consider that many
1664 characters at the beginning of the URI to compute the hash.
1665 Note that having "len" set to 1 rarely makes sense since most
1666 URIs start with a leading "/".
1667
1668 The "depth" parameter indicates the maximum directory depth
1669 to be used to compute the hash. One level is counted for each
1670 slash in the request. If both parameters are specified, the
1671 evaluation stops when either is reached.
1672
Willy Tarreau0ba27502007-12-24 16:55:16 +01001673 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001674 the query string of each HTTP GET request.
1675
1676 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02001677 request entity will be searched for the parameter argument,
1678 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02001679 ('?') in the URL. The message body will only start to be
1680 analyzed once either the advertised amount of data has been
1681 received or the request buffer is full. In the unlikely event
1682 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02001683 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02001684 be randomly balanced if at all. This keyword used to support
1685 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001686
1687 If the parameter is found followed by an equal sign ('=') and
1688 a value, then the value is hashed and divided by the total
1689 weight of the running servers. The result designates which
1690 server will receive the request.
1691
1692 This is used to track user identifiers in requests and ensure
1693 that a same user ID will always be sent to the same server as
1694 long as no server goes up or down. If no value is found or if
1695 the parameter is not found, then a round robin algorithm is
1696 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001697 backend. This algorithm is static by default, which means
1698 that changing a server's weight on the fly will have no
1699 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001700
Cyril Bontédc4d9032012-04-08 21:57:39 +02001701 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
1702 request. Just as with the equivalent ACL 'hdr()' function,
1703 the header name in parenthesis is not case sensitive. If the
1704 header is absent or if it does not contain any value, the
1705 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01001706
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001707 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01001708 reducing the hash algorithm to the main domain part with some
1709 specific headers such as 'Host'. For instance, in the Host
1710 value "haproxy.1wt.eu", only "1wt" will be considered.
1711
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001712 This algorithm is static by default, which means that
1713 changing a server's weight on the fly will have no effect,
1714 but this can be changed using "hash-type".
1715
Emeric Brun736aa232009-06-30 17:56:00 +02001716 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02001717 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02001718 The RDP cookie <name> (or "mstshash" if omitted) will be
1719 looked up and hashed for each incoming TCP request. Just as
1720 with the equivalent ACL 'req_rdp_cookie()' function, the name
1721 is not case-sensitive. This mechanism is useful as a degraded
1722 persistence mode, as it makes it possible to always send the
1723 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001724 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02001725 used instead.
1726
1727 Note that for this to work, the frontend must ensure that an
1728 RDP cookie is already present in the request buffer. For this
1729 you must use 'tcp-request content accept' rule combined with
1730 a 'req_rdp_cookie_cnt' ACL.
1731
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001732 This algorithm is static by default, which means that
1733 changing a server's weight on the fly will have no effect,
1734 but this can be changed using "hash-type".
1735
Cyril Bontédc4d9032012-04-08 21:57:39 +02001736 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09001737
Willy Tarreau0ba27502007-12-24 16:55:16 +01001738 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001739 algorithms. Right now, only "url_param" and "uri" support an
1740 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001741
Willy Tarreau3cd9af22009-03-15 14:06:41 +01001742 The load balancing algorithm of a backend is set to roundrobin when no other
1743 algorithm, mode nor option have been set. The algorithm may only be set once
1744 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001745
1746 Examples :
1747 balance roundrobin
1748 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001749 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01001750 balance hdr(User-Agent)
1751 balance hdr(host)
1752 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001753
1754 Note: the following caveats and limitations on using the "check_post"
1755 extension with "url_param" must be considered :
1756
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001757 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001758 to determine if the parameters will be found in the body or entity which
1759 may contain binary data. Therefore another method may be required to
1760 restrict consideration of POST requests that have no URL parameters in
1761 the body. (see acl reqideny http_end)
1762
1763 - using a <max_wait> value larger than the request buffer size does not
1764 make sense and is useless. The buffer size is set at build time, and
1765 defaults to 16 kB.
1766
1767 - Content-Encoding is not supported, the parameter search will probably
1768 fail; and load balancing will fall back to Round Robin.
1769
1770 - Expect: 100-continue is not supported, load balancing will fall back to
1771 Round Robin.
1772
1773 - Transfer-Encoding (RFC2616 3.6.1) is only supported in the first chunk.
1774 If the entire parameter value is not present in the first chunk, the
1775 selection of server is undefined (actually, defined by how little
1776 actually appeared in the first chunk).
1777
1778 - This feature does not support generation of a 100, 411 or 501 response.
1779
1780 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001781 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001782 white space or control characters are found, indicating the end of what
1783 might be a URL parameter list. This is probably not a concern with SGML
1784 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001785
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001786 See also : "dispatch", "cookie", "appsession", "transparent", "hash-type" and
1787 "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001788
1789
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001790bind [<address>]:<port_range> [, ...] [param*]
1791bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001792 Define one or several listening addresses and/or ports in a frontend.
1793 May be used in sections : defaults | frontend | listen | backend
1794 no | yes | yes | no
1795 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001796 <address> is optional and can be a host name, an IPv4 address, an IPv6
1797 address, or '*'. It designates the address the frontend will
1798 listen on. If unset, all IPv4 addresses of the system will be
1799 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01001800 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01001801 Optionally, an address family prefix may be used before the
1802 address to force the family regardless of the address format,
1803 which can be useful to specify a path to a unix socket with
1804 no slash ('/'). Currently supported prefixes are :
1805 - 'ipv4@' -> address is always IPv4
1806 - 'ipv6@' -> address is always IPv6
1807 - 'unix@' -> address is a path to a local unix socket
Willy Tarreaub4fca5d2014-07-08 00:37:50 +02001808 - 'abns@' -> address is in abstract namespace (Linux only).
1809 Note: since abstract sockets are not "rebindable", they
1810 do not cope well with multi-process mode during
1811 soft-restart, so it is better to avoid them if
1812 nbproc is greater than 1. The effect is that if the
1813 new process fails to start, only one of the old ones
1814 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01001815 - 'fd@<n>' -> use file descriptor <n> inherited from the
1816 parent. The fd must be bound and may or may not already
1817 be listening.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001818 Any part of the address string may reference any number of
1819 environment variables by preceding their name with a dollar
1820 sign ('$') and optionally enclosing them with braces ('{}'),
1821 similarly to what is done in Bourne shell.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001822
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001823 <port_range> is either a unique TCP port, or a port range for which the
1824 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001825 above. The port is mandatory for TCP listeners. Note that in
1826 the case of an IPv6 address, the port is always the number
1827 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001828 - a numerical port (ex: '80')
1829 - a dash-delimited ports range explicitly stating the lower
1830 and upper bounds (ex: '2000-2100') which are included in
1831 the range.
1832
1833 Particular care must be taken against port ranges, because
1834 every <address:port> couple consumes one socket (= a file
1835 descriptor), so it's easy to consume lots of descriptors
1836 with a simple range, and to run out of sockets. Also, each
1837 <address:port> couple must be used only once among all
1838 instances running on a same system. Please note that binding
1839 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001840 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001841 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001842
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001843 <path> is a UNIX socket path beginning with a slash ('/'). This is
1844 alternative to the TCP listening port. Haproxy will then
1845 receive UNIX connections on the socket located at this place.
1846 The path must begin with a slash and by default is absolute.
1847 It can be relative to the prefix defined by "unix-bind" in
1848 the global section. Note that the total length of the prefix
1849 followed by the socket path cannot exceed some system limits
1850 for UNIX sockets, which commonly are set to 107 characters.
1851
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001852 <param*> is a list of parameters common to all sockets declared on the
1853 same line. These numerous parameters depend on OS and build
1854 options and have a complete section dedicated to them. Please
1855 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001856
Willy Tarreau0ba27502007-12-24 16:55:16 +01001857 It is possible to specify a list of address:port combinations delimited by
1858 commas. The frontend will then listen on all of these addresses. There is no
1859 fixed limit to the number of addresses and ports which can be listened on in
1860 a frontend, as well as there is no limit to the number of "bind" statements
1861 in a frontend.
1862
1863 Example :
1864 listen http_proxy
1865 bind :80,:443
1866 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001867 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01001868
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001869 listen http_https_proxy
1870 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02001871 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001872
Willy Tarreau24709282013-03-10 21:32:12 +01001873 listen http_https_proxy_explicit
1874 bind ipv6@:80
1875 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
1876 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
1877
Willy Tarreaudad36a32013-03-11 01:20:04 +01001878 listen external_bind_app1
1879 bind fd@${FD_APP1}
1880
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001881 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001882 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001883
1884
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001885bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001886 Limit visibility of an instance to a certain set of processes numbers.
1887 May be used in sections : defaults | frontend | listen | backend
1888 yes | yes | yes | yes
1889 Arguments :
1890 all All process will see this instance. This is the default. It
1891 may be used to override a default value.
1892
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001893 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001894 option may be combined with other numbers.
1895
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001896 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001897 option may be combined with other numbers. Do not use it
1898 with less than 2 processes otherwise some instances might be
1899 missing from all processes.
1900
Willy Tarreau110ecc12012-11-15 17:50:01 +01001901 number The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001902 whose values must all be between 1 and 32 or 64 depending on
Willy Tarreau102df612014-05-07 23:56:38 +02001903 the machine's word size. If a proxy is bound to process
1904 numbers greater than the configured global.nbproc, it will
1905 either be forced to process #1 if a single process was
1906 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001907
1908 This keyword limits binding of certain instances to certain processes. This
1909 is useful in order not to have too many processes listening to the same
1910 ports. For instance, on a dual-core machine, it might make sense to set
1911 'nbproc 2' in the global section, then distributes the listeners among 'odd'
1912 and 'even' instances.
1913
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001914 At the moment, it is not possible to reference more than 32 or 64 processes
1915 using this keyword, but this should be more than enough for most setups.
1916 Please note that 'all' really means all processes regardless of the machine's
1917 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001918
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001919 Each "bind" line may further be limited to a subset of the proxy's processes,
1920 please consult the "process" bind keyword in section 5.1.
1921
Willy Tarreaue56c4f12014-09-16 13:21:03 +02001922 When a frontend has no explicit "bind-process" line, it tries to bind to all
1923 the processes referenced by its "bind" lines. That means that frontends can
1924 easily adapt to their listeners' processes.
1925
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001926 If some backends are referenced by frontends bound to other processes, the
1927 backend automatically inherits the frontend's processes.
1928
1929 Example :
1930 listen app_ip1
1931 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001932 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001933
1934 listen app_ip2
1935 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001936 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001937
1938 listen management
1939 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001940 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001941
Willy Tarreau110ecc12012-11-15 17:50:01 +01001942 listen management
1943 bind 10.0.0.4:80
1944 bind-process 1-4
1945
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001946 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001947
1948
Willy Tarreau0ba27502007-12-24 16:55:16 +01001949block { if | unless } <condition>
1950 Block a layer 7 request if/unless a condition is matched
1951 May be used in sections : defaults | frontend | listen | backend
1952 no | yes | yes | yes
1953
1954 The HTTP request will be blocked very early in the layer 7 processing
1955 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001956 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02001957 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01001958 conditions are met or not met. There is no fixed limit to the number of
1959 "block" statements per instance.
1960
1961 Example:
1962 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1963 acl invalid_src src_port 0:1023
1964 acl local_dst hdr(host) -i localhost
1965 block if invalid_src || local_dst
1966
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001967 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001968
1969
1970capture cookie <name> len <length>
1971 Capture and log a cookie in the request and in the response.
1972 May be used in sections : defaults | frontend | listen | backend
1973 no | yes | yes | no
1974 Arguments :
1975 <name> is the beginning of the name of the cookie to capture. In order
1976 to match the exact name, simply suffix the name with an equal
1977 sign ('='). The full name will appear in the logs, which is
1978 useful with application servers which adjust both the cookie name
1979 and value (eg: ASPSESSIONXXXXX).
1980
1981 <length> is the maximum number of characters to report in the logs, which
1982 include the cookie name, the equal sign and the value, all in the
1983 standard "name=value" form. The string will be truncated on the
1984 right if it exceeds <length>.
1985
1986 Only the first cookie is captured. Both the "cookie" request headers and the
1987 "set-cookie" response headers are monitored. This is particularly useful to
1988 check for application bugs causing session crossing or stealing between
1989 users, because generally the user's cookies can only change on a login page.
1990
1991 When the cookie was not presented by the client, the associated log column
1992 will report "-". When a request does not cause a cookie to be assigned by the
1993 server, a "-" is reported in the response column.
1994
1995 The capture is performed in the frontend only because it is necessary that
1996 the log format does not change for a given frontend depending on the
1997 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01001998 "capture cookie" statement in a frontend. The maximum capture length is set
1999 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2000 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002001
2002 Example:
2003 capture cookie ASPSESSION len 32
2004
2005 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002006 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002007
2008
2009capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002010 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002011 May be used in sections : defaults | frontend | listen | backend
2012 no | yes | yes | no
2013 Arguments :
2014 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002015 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002016 appear in the requests, with the first letter of each word in
2017 upper case. The header name will not appear in the logs, only the
2018 value is reported, but the position in the logs is respected.
2019
2020 <length> is the maximum number of characters to extract from the value and
2021 report in the logs. The string will be truncated on the right if
2022 it exceeds <length>.
2023
Willy Tarreau4460d032012-11-21 23:37:37 +01002024 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002025 value will be added to the logs between braces ('{}'). If multiple headers
2026 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002027 in the same order they were declared in the configuration. Non-existent
2028 headers will be logged just as an empty string. Common uses for request
2029 header captures include the "Host" field in virtual hosting environments, the
2030 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002031 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002032 environments to find where the request came from.
2033
2034 Note that when capturing headers such as "User-agent", some spaces may be
2035 logged, making the log analysis more difficult. Thus be careful about what
2036 you log if you know your log parser is not smart enough to rely on the
2037 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002038
Willy Tarreau0900abb2012-11-22 00:21:46 +01002039 There is no limit to the number of captured request headers nor to their
2040 length, though it is wise to keep them low to limit memory usage per session.
2041 In order to keep log format consistent for a same frontend, header captures
2042 can only be declared in a frontend. It is not possible to specify a capture
2043 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002044
2045 Example:
2046 capture request header Host len 15
2047 capture request header X-Forwarded-For len 15
2048 capture request header Referrer len 15
2049
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002050 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002051 about logging.
2052
2053
2054capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002055 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002056 May be used in sections : defaults | frontend | listen | backend
2057 no | yes | yes | no
2058 Arguments :
2059 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002060 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002061 appear in the response, with the first letter of each word in
2062 upper case. The header name will not appear in the logs, only the
2063 value is reported, but the position in the logs is respected.
2064
2065 <length> is the maximum number of characters to extract from the value and
2066 report in the logs. The string will be truncated on the right if
2067 it exceeds <length>.
2068
Willy Tarreau4460d032012-11-21 23:37:37 +01002069 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002070 result will be added to the logs between braces ('{}') after the captured
2071 request headers. If multiple headers are captured, they will be delimited by
2072 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002073 the configuration. Non-existent headers will be logged just as an empty
2074 string. Common uses for response header captures include the "Content-length"
2075 header which indicates how many bytes are expected to be returned, the
2076 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002077
Willy Tarreau0900abb2012-11-22 00:21:46 +01002078 There is no limit to the number of captured response headers nor to their
2079 length, though it is wise to keep them low to limit memory usage per session.
2080 In order to keep log format consistent for a same frontend, header captures
2081 can only be declared in a frontend. It is not possible to specify a capture
2082 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002083
2084 Example:
2085 capture response header Content-length len 9
2086 capture response header Location len 15
2087
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002088 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002089 about logging.
2090
2091
Cyril Bontéf0c60612010-02-06 14:44:47 +01002092clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002093 Set the maximum inactivity time on the client side.
2094 May be used in sections : defaults | frontend | listen | backend
2095 yes | yes | yes | no
2096 Arguments :
2097 <timeout> is the timeout value is specified in milliseconds by default, but
2098 can be in any other unit if the number is suffixed by the unit,
2099 as explained at the top of this document.
2100
2101 The inactivity timeout applies when the client is expected to acknowledge or
2102 send data. In HTTP mode, this timeout is particularly important to consider
2103 during the first phase, when the client sends the request, and during the
2104 response while it is reading data sent by the server. The value is specified
2105 in milliseconds by default, but can be in any other unit if the number is
2106 suffixed by the unit, as specified at the top of this document. In TCP mode
2107 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2108 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002109 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002110 losses by specifying timeouts that are slightly above multiples of 3 seconds
2111 (eg: 4 or 5 seconds).
2112
2113 This parameter is specific to frontends, but can be specified once for all in
2114 "defaults" sections. This is in fact one of the easiest solutions not to
2115 forget about it. An unspecified timeout results in an infinite timeout, which
2116 is not recommended. Such a usage is accepted and works but reports a warning
2117 during startup because it may results in accumulation of expired sessions in
2118 the system if the system's timeouts are not configured either.
2119
2120 This parameter is provided for compatibility but is currently deprecated.
2121 Please use "timeout client" instead.
2122
Willy Tarreau036fae02008-01-06 13:24:40 +01002123 See also : "timeout client", "timeout http-request", "timeout server", and
2124 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002125
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002126compression algo <algorithm> ...
2127compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002128compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002129 Enable HTTP compression.
2130 May be used in sections : defaults | frontend | listen | backend
2131 yes | yes | yes | yes
2132 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002133 algo is followed by the list of supported compression algorithms.
2134 type is followed by the list of MIME types that will be compressed.
2135 offload makes haproxy work as a compression offloader only (see notes).
2136
2137 The currently supported algorithms are :
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002138 identity this is mostly for debugging, and it was useful for developing
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002139 the compression feature. Identity does not apply any change on
2140 data.
2141
2142 gzip applies gzip compression. This setting is only available when
2143 support for zlib was built in.
2144
2145 deflate same as gzip, but with deflate algorithm and zlib format.
2146 Note that this algorithm has ambiguous support on many browsers
2147 and no support at all from recent ones. It is strongly
2148 recommended not to use it for anything else than experimentation.
2149 This setting is only available when support for zlib was built
2150 in.
2151
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002152 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002153 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002154 If backend servers support HTTP compression, these directives
2155 will be no-op: haproxy will see the compressed response and will not
2156 compress again. If backend servers do not support HTTP compression and
2157 there is Accept-Encoding header in request, haproxy will compress the
2158 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002159
2160 The "offload" setting makes haproxy remove the Accept-Encoding header to
2161 prevent backend servers from compressing responses. It is strongly
2162 recommended not to do this because this means that all the compression work
2163 will be done on the single point where haproxy is located. However in some
2164 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002165 with broken HTTP compression implementation which can't be turned off.
2166 In that case haproxy can be used to prevent that gateway from emitting
2167 invalid payloads. In this case, simply removing the header in the
2168 configuration does not work because it applies before the header is parsed,
2169 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreau4cf57af2014-07-12 16:37:02 +02002170 then be used for such scenarios. Note: for now, the "offload" setting is
2171 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002172
William Lallemand05097442012-11-20 12:14:28 +01002173 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002174 * the request does not advertise a supported compression algorithm in the
2175 "Accept-Encoding" header
2176 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002177 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002178 * response header "Transfer-Encoding" contains "chunked" (Temporary
2179 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002180 * response contain neither a "Content-Length" header nor a
2181 "Transfer-Encoding" whose last value is "chunked"
2182 * response contains a "Content-Type" header whose first value starts with
2183 "multipart"
2184 * the response contains the "no-transform" value in the "Cache-control"
2185 header
2186 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2187 and later
2188 * The response contains a "Content-Encoding" header, indicating that the
2189 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002190
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002191 Note: The compression does not rewrite Etag headers, and does not emit the
2192 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002193
William Lallemand82fe75c2012-10-23 10:25:10 +02002194 Examples :
2195 compression algo gzip
2196 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002197
Cyril Bontéf0c60612010-02-06 14:44:47 +01002198contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002199 Set the maximum time to wait for a connection attempt to a server to succeed.
2200 May be used in sections : defaults | frontend | listen | backend
2201 yes | no | yes | yes
2202 Arguments :
2203 <timeout> is the timeout value is specified in milliseconds by default, but
2204 can be in any other unit if the number is suffixed by the unit,
2205 as explained at the top of this document.
2206
2207 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002208 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002209 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01002210 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
2211 connect timeout also presets the queue timeout to the same value if this one
2212 has not been specified. Historically, the contimeout was also used to set the
2213 tarpit timeout in a listen section, which is not possible in a pure frontend.
2214
2215 This parameter is specific to backends, but can be specified once for all in
2216 "defaults" sections. This is in fact one of the easiest solutions not to
2217 forget about it. An unspecified timeout results in an infinite timeout, which
2218 is not recommended. Such a usage is accepted and works but reports a warning
2219 during startup because it may results in accumulation of failed sessions in
2220 the system if the system's timeouts are not configured either.
2221
2222 This parameter is provided for backwards compatibility but is currently
2223 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2224 instead.
2225
2226 See also : "timeout connect", "timeout queue", "timeout tarpit",
2227 "timeout server", "contimeout".
2228
2229
Willy Tarreau55165fe2009-05-10 12:02:55 +02002230cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002231 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2232 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002233 Enable cookie-based persistence in a backend.
2234 May be used in sections : defaults | frontend | listen | backend
2235 yes | no | yes | yes
2236 Arguments :
2237 <name> is the name of the cookie which will be monitored, modified or
2238 inserted in order to bring persistence. This cookie is sent to
2239 the client via a "Set-Cookie" header in the response, and is
2240 brought back by the client in a "Cookie" header in all requests.
2241 Special care should be taken to choose a name which does not
2242 conflict with any likely application cookie. Also, if the same
2243 backends are subject to be used by the same clients (eg:
2244 HTTP/HTTPS), care should be taken to use different cookie names
2245 between all backends if persistence between them is not desired.
2246
2247 rewrite This keyword indicates that the cookie will be provided by the
2248 server and that haproxy will have to modify its value to set the
2249 server's identifier in it. This mode is handy when the management
2250 of complex combinations of "Set-cookie" and "Cache-control"
2251 headers is left to the application. The application can then
2252 decide whether or not it is appropriate to emit a persistence
2253 cookie. Since all responses should be monitored, this mode only
2254 works in HTTP close mode. Unless the application behaviour is
2255 very complex and/or broken, it is advised not to start with this
2256 mode for new deployments. This keyword is incompatible with
2257 "insert" and "prefix".
2258
2259 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02002260 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002261
Willy Tarreaua79094d2010-08-31 22:54:15 +02002262 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002263 server. When used without the "preserve" option, if the server
2264 emits a cookie with the same name, it will be remove before
2265 processing. For this reason, this mode can be used to upgrade
2266 existing configurations running in the "rewrite" mode. The cookie
2267 will only be a session cookie and will not be stored on the
2268 client's disk. By default, unless the "indirect" option is added,
2269 the server will see the cookies emitted by the client. Due to
2270 caching effects, it is generally wise to add the "nocache" or
2271 "postonly" keywords (see below). The "insert" keyword is not
2272 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002273
2274 prefix This keyword indicates that instead of relying on a dedicated
2275 cookie for the persistence, an existing one will be completed.
2276 This may be needed in some specific environments where the client
2277 does not support more than one single cookie and the application
2278 already needs it. In this case, whenever the server sets a cookie
2279 named <name>, it will be prefixed with the server's identifier
2280 and a delimiter. The prefix will be removed from all client
2281 requests so that the server still finds the cookie it emitted.
2282 Since all requests and responses are subject to being modified,
2283 this mode requires the HTTP close mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02002284 not compatible with "rewrite" and "insert". Note: it is highly
2285 recommended not to use "indirect" with "prefix", otherwise server
2286 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002287
Willy Tarreaua79094d2010-08-31 22:54:15 +02002288 indirect When this option is specified, no cookie will be emitted to a
2289 client which already has a valid one for the server which has
2290 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002291 it will be removed, unless the "preserve" option is also set. In
2292 "insert" mode, this will additionally remove cookies from the
2293 requests transmitted to the server, making the persistence
2294 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02002295 Note: it is highly recommended not to use "indirect" with
2296 "prefix", otherwise server cookie updates would not be sent to
2297 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002298
2299 nocache This option is recommended in conjunction with the insert mode
2300 when there is a cache between the client and HAProxy, as it
2301 ensures that a cacheable response will be tagged non-cacheable if
2302 a cookie needs to be inserted. This is important because if all
2303 persistence cookies are added on a cacheable home page for
2304 instance, then all customers will then fetch the page from an
2305 outer cache and will all share the same persistence cookie,
2306 leading to one server receiving much more traffic than others.
2307 See also the "insert" and "postonly" options.
2308
2309 postonly This option ensures that cookie insertion will only be performed
2310 on responses to POST requests. It is an alternative to the
2311 "nocache" option, because POST responses are not cacheable, so
2312 this ensures that the persistence cookie will never get cached.
2313 Since most sites do not need any sort of persistence before the
2314 first POST which generally is a login request, this is a very
2315 efficient method to optimize caching without risking to find a
2316 persistence cookie in the cache.
2317 See also the "insert" and "nocache" options.
2318
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002319 preserve This option may only be used with "insert" and/or "indirect". It
2320 allows the server to emit the persistence cookie itself. In this
2321 case, if a cookie is found in the response, haproxy will leave it
2322 untouched. This is useful in order to end persistence after a
2323 logout request for instance. For this, the server just has to
2324 emit a cookie with an invalid value (eg: empty) or with a date in
2325 the past. By combining this mechanism with the "disable-on-404"
2326 check option, it is possible to perform a completely graceful
2327 shutdown because users will definitely leave the server after
2328 they logout.
2329
Willy Tarreau4992dd22012-05-31 21:02:17 +02002330 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
2331 when a cookie is inserted. This attribute is used so that a
2332 user agent doesn't share the cookie with non-HTTP components.
2333 Please check RFC6265 for more information on this attribute.
2334
2335 secure This option tells haproxy to add a "Secure" cookie attribute when
2336 a cookie is inserted. This attribute is used so that a user agent
2337 never emits this cookie over non-secure channels, which means
2338 that a cookie learned with this flag will be presented only over
2339 SSL/TLS connections. Please check RFC6265 for more information on
2340 this attribute.
2341
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002342 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002343 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01002344 name. If the domain begins with a dot, the browser is allowed to
2345 use it for any host ending with that name. It is also possible to
2346 specify several domain names by invoking this option multiple
2347 times. Some browsers might have small limits on the number of
2348 domains, so be careful when doing that. For the record, sending
2349 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002350
Willy Tarreau996a92c2010-10-13 19:30:47 +02002351 maxidle This option allows inserted cookies to be ignored after some idle
2352 time. It only works with insert-mode cookies. When a cookie is
2353 sent to the client, the date this cookie was emitted is sent too.
2354 Upon further presentations of this cookie, if the date is older
2355 than the delay indicated by the parameter (in seconds), it will
2356 be ignored. Otherwise, it will be refreshed if needed when the
2357 response is sent to the client. This is particularly useful to
2358 prevent users who never close their browsers from remaining for
2359 too long on the same server (eg: after a farm size change). When
2360 this option is set and a cookie has no date, it is always
2361 accepted, but gets refreshed in the response. This maintains the
2362 ability for admins to access their sites. Cookies that have a
2363 date in the future further than 24 hours are ignored. Doing so
2364 lets admins fix timezone issues without risking kicking users off
2365 the site.
2366
2367 maxlife This option allows inserted cookies to be ignored after some life
2368 time, whether they're in use or not. It only works with insert
2369 mode cookies. When a cookie is first sent to the client, the date
2370 this cookie was emitted is sent too. Upon further presentations
2371 of this cookie, if the date is older than the delay indicated by
2372 the parameter (in seconds), it will be ignored. If the cookie in
2373 the request has no date, it is accepted and a date will be set.
2374 Cookies that have a date in the future further than 24 hours are
2375 ignored. Doing so lets admins fix timezone issues without risking
2376 kicking users off the site. Contrary to maxidle, this value is
2377 not refreshed, only the first visit date counts. Both maxidle and
2378 maxlife may be used at the time. This is particularly useful to
2379 prevent users who never close their browsers from remaining for
2380 too long on the same server (eg: after a farm size change). This
2381 is stronger than the maxidle method in that it forces a
2382 redispatch after some absolute delay.
2383
Willy Tarreau0ba27502007-12-24 16:55:16 +01002384 There can be only one persistence cookie per HTTP backend, and it can be
2385 declared in a defaults section. The value of the cookie will be the value
2386 indicated after the "cookie" keyword in a "server" statement. If no cookie
2387 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002388
Willy Tarreau0ba27502007-12-24 16:55:16 +01002389 Examples :
2390 cookie JSESSIONID prefix
2391 cookie SRV insert indirect nocache
2392 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02002393 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01002394
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002395 See also : "appsession", "balance source", "capture cookie", "server"
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002396 and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002397
Willy Tarreau983e01e2010-01-11 18:42:06 +01002398
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002399default-server [param*]
2400 Change default options for a server in a backend
2401 May be used in sections : defaults | frontend | listen | backend
2402 yes | no | yes | yes
2403 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01002404 <param*> is a list of parameters for this server. The "default-server"
2405 keyword accepts an important number of options and has a complete
2406 section dedicated to it. Please refer to section 5 for more
2407 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002408
Willy Tarreau983e01e2010-01-11 18:42:06 +01002409 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002410 default-server inter 1000 weight 13
2411
2412 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01002413
Willy Tarreau983e01e2010-01-11 18:42:06 +01002414
Willy Tarreau0ba27502007-12-24 16:55:16 +01002415default_backend <backend>
2416 Specify the backend to use when no "use_backend" rule has been matched.
2417 May be used in sections : defaults | frontend | listen | backend
2418 yes | yes | yes | no
2419 Arguments :
2420 <backend> is the name of the backend to use.
2421
2422 When doing content-switching between frontend and backends using the
2423 "use_backend" keyword, it is often useful to indicate which backend will be
2424 used when no rule has matched. It generally is the dynamic backend which
2425 will catch all undetermined requests.
2426
Willy Tarreau0ba27502007-12-24 16:55:16 +01002427 Example :
2428
2429 use_backend dynamic if url_dyn
2430 use_backend static if url_css url_img extension_img
2431 default_backend dynamic
2432
Willy Tarreau2769aa02007-12-27 18:26:09 +01002433 See also : "use_backend", "reqsetbe", "reqisetbe"
2434
Willy Tarreau0ba27502007-12-24 16:55:16 +01002435
Baptiste Assmann27f51342013-10-09 06:51:49 +02002436description <string>
2437 Describe a listen, frontend or backend.
2438 May be used in sections : defaults | frontend | listen | backend
2439 no | yes | yes | yes
2440 Arguments : string
2441
2442 Allows to add a sentence to describe the related object in the HAProxy HTML
2443 stats page. The description will be printed on the right of the object name
2444 it describes.
2445 No need to backslash spaces in the <string> arguments.
2446
2447
Willy Tarreau0ba27502007-12-24 16:55:16 +01002448disabled
2449 Disable a proxy, frontend or backend.
2450 May be used in sections : defaults | frontend | listen | backend
2451 yes | yes | yes | yes
2452 Arguments : none
2453
2454 The "disabled" keyword is used to disable an instance, mainly in order to
2455 liberate a listening port or to temporarily disable a service. The instance
2456 will still be created and its configuration will be checked, but it will be
2457 created in the "stopped" state and will appear as such in the statistics. It
2458 will not receive any traffic nor will it send any health-checks or logs. It
2459 is possible to disable many instances at once by adding the "disabled"
2460 keyword in a "defaults" section.
2461
2462 See also : "enabled"
2463
2464
Willy Tarreau5ce94572010-06-07 14:35:41 +02002465dispatch <address>:<port>
2466 Set a default server address
2467 May be used in sections : defaults | frontend | listen | backend
2468 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02002469 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02002470
2471 <address> is the IPv4 address of the default server. Alternatively, a
2472 resolvable hostname is supported, but this name will be resolved
2473 during start-up.
2474
2475 <ports> is a mandatory port specification. All connections will be sent
2476 to this port, and it is not permitted to use port offsets as is
2477 possible with normal servers.
2478
Willy Tarreau787aed52011-04-15 06:45:37 +02002479 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02002480 server can take the connection. In the past it was used to forward non
2481 persistent connections to an auxiliary load balancer. Due to its simple
2482 syntax, it has also been used for simple TCP relays. It is recommended not to
2483 use it for more clarity, and to use the "server" directive instead.
2484
2485 See also : "server"
2486
2487
Willy Tarreau0ba27502007-12-24 16:55:16 +01002488enabled
2489 Enable a proxy, frontend or backend.
2490 May be used in sections : defaults | frontend | listen | backend
2491 yes | yes | yes | yes
2492 Arguments : none
2493
2494 The "enabled" keyword is used to explicitly enable an instance, when the
2495 defaults has been set to "disabled". This is very rarely used.
2496
2497 See also : "disabled"
2498
2499
2500errorfile <code> <file>
2501 Return a file contents instead of errors generated by HAProxy
2502 May be used in sections : defaults | frontend | listen | backend
2503 yes | yes | yes | yes
2504 Arguments :
2505 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002506 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002507
2508 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002509 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01002510 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01002511 error pages, and to use absolute paths, since files are read
2512 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002513
2514 It is important to understand that this keyword is not meant to rewrite
2515 errors returned by the server, but errors detected and returned by HAProxy.
2516 This is why the list of supported errors is limited to a small set.
2517
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002518 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2519
Willy Tarreau0ba27502007-12-24 16:55:16 +01002520 The files are returned verbatim on the TCP socket. This allows any trick such
2521 as redirections to another URL or site, as well as tricks to clean cookies,
2522 force enable or disable caching, etc... The package provides default error
2523 files returning the same contents as default errors.
2524
Willy Tarreau59140a22009-02-22 12:02:30 +01002525 The files should not exceed the configured buffer size (BUFSIZE), which
2526 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
2527 not to put any reference to local contents (eg: images) in order to avoid
2528 loops between the client and HAProxy when all servers are down, causing an
2529 error to be returned instead of an image. For better HTTP compliance, it is
2530 recommended that all header lines end with CR-LF and not LF alone.
2531
Willy Tarreau0ba27502007-12-24 16:55:16 +01002532 The files are read at the same time as the configuration and kept in memory.
2533 For this reason, the errors continue to be returned even when the process is
2534 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01002535 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002536 403 status code and interrogating a blocked URL.
2537
2538 See also : "errorloc", "errorloc302", "errorloc303"
2539
Willy Tarreau59140a22009-02-22 12:02:30 +01002540 Example :
2541 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau2705a612014-05-23 17:38:34 +02002542 errorfile 408 /dev/null # workaround Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01002543 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
2544 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
2545
Willy Tarreau2769aa02007-12-27 18:26:09 +01002546
2547errorloc <code> <url>
2548errorloc302 <code> <url>
2549 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2550 May be used in sections : defaults | frontend | listen | backend
2551 yes | yes | yes | yes
2552 Arguments :
2553 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002554 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002555
2556 <url> it is the exact contents of the "Location" header. It may contain
2557 either a relative URI to an error page hosted on the same site,
2558 or an absolute URI designating an error page on another site.
2559 Special care should be given to relative URIs to avoid redirect
2560 loops if the URI itself may generate the same error (eg: 500).
2561
2562 It is important to understand that this keyword is not meant to rewrite
2563 errors returned by the server, but errors detected and returned by HAProxy.
2564 This is why the list of supported errors is limited to a small set.
2565
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002566 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2567
Willy Tarreau2769aa02007-12-27 18:26:09 +01002568 Note that both keyword return the HTTP 302 status code, which tells the
2569 client to fetch the designated URL using the same HTTP method. This can be
2570 quite problematic in case of non-GET methods such as POST, because the URL
2571 sent to the client might not be allowed for something other than GET. To
2572 workaround this problem, please use "errorloc303" which send the HTTP 303
2573 status code, indicating to the client that the URL must be fetched with a GET
2574 request.
2575
2576 See also : "errorfile", "errorloc303"
2577
2578
2579errorloc303 <code> <url>
2580 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2581 May be used in sections : defaults | frontend | listen | backend
2582 yes | yes | yes | yes
2583 Arguments :
2584 <code> is the HTTP status code. Currently, HAProxy is capable of
2585 generating codes 400, 403, 408, 500, 502, 503, and 504.
2586
2587 <url> it is the exact contents of the "Location" header. It may contain
2588 either a relative URI to an error page hosted on the same site,
2589 or an absolute URI designating an error page on another site.
2590 Special care should be given to relative URIs to avoid redirect
2591 loops if the URI itself may generate the same error (eg: 500).
2592
2593 It is important to understand that this keyword is not meant to rewrite
2594 errors returned by the server, but errors detected and returned by HAProxy.
2595 This is why the list of supported errors is limited to a small set.
2596
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002597 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2598
Willy Tarreau2769aa02007-12-27 18:26:09 +01002599 Note that both keyword return the HTTP 303 status code, which tells the
2600 client to fetch the designated URL using the same HTTP GET method. This
2601 solves the usual problems associated with "errorloc" and the 302 code. It is
2602 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002603 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002604
2605 See also : "errorfile", "errorloc", "errorloc302"
2606
2607
Willy Tarreau4de91492010-01-22 19:10:05 +01002608force-persist { if | unless } <condition>
2609 Declare a condition to force persistence on down servers
2610 May be used in sections: defaults | frontend | listen | backend
2611 no | yes | yes | yes
2612
2613 By default, requests are not dispatched to down servers. It is possible to
2614 force this using "option persist", but it is unconditional and redispatches
2615 to a valid server if "option redispatch" is set. That leaves with very little
2616 possibilities to force some requests to reach a server which is artificially
2617 marked down for maintenance operations.
2618
2619 The "force-persist" statement allows one to declare various ACL-based
2620 conditions which, when met, will cause a request to ignore the down status of
2621 a server and still try to connect to it. That makes it possible to start a
2622 server, still replying an error to the health checks, and run a specially
2623 configured browser to test the service. Among the handy methods, one could
2624 use a specific source IP address, or a specific cookie. The cookie also has
2625 the advantage that it can easily be added/removed on the browser from a test
2626 page. Once the service is validated, it is then possible to open the service
2627 to the world by returning a valid response to health checks.
2628
2629 The forced persistence is enabled when an "if" condition is met, or unless an
2630 "unless" condition is met. The final redispatch is always disabled when this
2631 is used.
2632
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002633 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002634 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01002635
2636
Willy Tarreau2769aa02007-12-27 18:26:09 +01002637fullconn <conns>
2638 Specify at what backend load the servers will reach their maxconn
2639 May be used in sections : defaults | frontend | listen | backend
2640 yes | no | yes | yes
2641 Arguments :
2642 <conns> is the number of connections on the backend which will make the
2643 servers use the maximal number of connections.
2644
Willy Tarreau198a7442008-01-17 12:05:32 +01002645 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01002646 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01002647 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01002648 load. The server will then always accept at least <minconn> connections,
2649 never more than <maxconn>, and the limit will be on the ramp between both
2650 values when the backend has less than <conns> concurrent connections. This
2651 makes it possible to limit the load on the servers during normal loads, but
2652 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002653 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002654
Willy Tarreaufbb78422011-06-05 15:38:35 +02002655 Since it's hard to get this value right, haproxy automatically sets it to
2656 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01002657 backend (based on "use_backend" and "default_backend" rules). That way it's
2658 safe to leave it unset. However, "use_backend" involving dynamic names are
2659 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02002660
Willy Tarreau2769aa02007-12-27 18:26:09 +01002661 Example :
2662 # The servers will accept between 100 and 1000 concurrent connections each
2663 # and the maximum of 1000 will be reached when the backend reaches 10000
2664 # connections.
2665 backend dynamic
2666 fullconn 10000
2667 server srv1 dyn1:80 minconn 100 maxconn 1000
2668 server srv2 dyn2:80 minconn 100 maxconn 1000
2669
2670 See also : "maxconn", "server"
2671
2672
2673grace <time>
2674 Maintain a proxy operational for some time after a soft stop
2675 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01002676 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01002677 Arguments :
2678 <time> is the time (by default in milliseconds) for which the instance
2679 will remain operational with the frontend sockets still listening
2680 when a soft-stop is received via the SIGUSR1 signal.
2681
2682 This may be used to ensure that the services disappear in a certain order.
2683 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002684 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01002685 needed by the equipment to detect the failure.
2686
2687 Note that currently, there is very little benefit in using this parameter,
2688 and it may in fact complicate the soft-reconfiguration process more than
2689 simplify it.
2690
Willy Tarreau0ba27502007-12-24 16:55:16 +01002691
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002692hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002693 Specify a method to use for mapping hashes to servers
2694 May be used in sections : defaults | frontend | listen | backend
2695 yes | no | yes | yes
2696 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04002697 <method> is the method used to select a server from the hash computed by
2698 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002699
Bhaskar98634f02013-10-29 23:30:51 -04002700 map-based the hash table is a static array containing all alive servers.
2701 The hashes will be very smooth, will consider weights, but
2702 will be static in that weight changes while a server is up
2703 will be ignored. This means that there will be no slow start.
2704 Also, since a server is selected by its position in the array,
2705 most mappings are changed when the server count changes. This
2706 means that when a server goes up or down, or when a server is
2707 added to a farm, most connections will be redistributed to
2708 different servers. This can be inconvenient with caches for
2709 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01002710
Bhaskar98634f02013-10-29 23:30:51 -04002711 consistent the hash table is a tree filled with many occurrences of each
2712 server. The hash key is looked up in the tree and the closest
2713 server is chosen. This hash is dynamic, it supports changing
2714 weights while the servers are up, so it is compatible with the
2715 slow start feature. It has the advantage that when a server
2716 goes up or down, only its associations are moved. When a
2717 server is added to the farm, only a few part of the mappings
2718 are redistributed, making it an ideal method for caches.
2719 However, due to its principle, the distribution will never be
2720 very smooth and it may sometimes be necessary to adjust a
2721 server's weight or its ID to get a more balanced distribution.
2722 In order to get the same distribution on multiple load
2723 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002724 same IDs. Note: consistent hash uses sdbm and avalanche if no
2725 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04002726
2727 <function> is the hash function to be used :
2728
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002729 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04002730 reimplementation of ndbm) database library. It was found to do
2731 well in scrambling bits, causing better distribution of the keys
2732 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002733 function with good distribution, unless the total server weight
2734 is a multiple of 64, in which case applying the avalanche
2735 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04002736
2737 djb2 this function was first proposed by Dan Bernstein many years ago
2738 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002739 function provides a better distribution than sdbm. It generally
2740 works well with text-based inputs though it can perform extremely
2741 poorly with numeric-only input or when the total server weight is
2742 a multiple of 33, unless the avalanche modifier is also used.
2743
Willy Tarreaua0f42712013-11-14 14:30:35 +01002744 wt6 this function was designed for haproxy while testing other
2745 functions in the past. It is not as smooth as the other ones, but
2746 is much less sensible to the input data set or to the number of
2747 servers. It can make sense as an alternative to sdbm+avalanche or
2748 djb2+avalanche for consistent hashing or when hashing on numeric
2749 data such as a source IP address or a visitor identifier in a URL
2750 parameter.
2751
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002752 <modifier> indicates an optional method applied after hashing the key :
2753
2754 avalanche This directive indicates that the result from the hash
2755 function above should not be used in its raw form but that
2756 a 4-byte full avalanche hash must be applied first. The
2757 purpose of this step is to mix the resulting bits from the
2758 previous hash in order to avoid any undesired effect when
2759 the input contains some limited values or when the number of
2760 servers is a multiple of one of the hash's components (64
2761 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
2762 result less predictable, but it's also not as smooth as when
2763 using the original function. Some testing might be needed
2764 with some workloads. This hash is one of the many proposed
2765 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002766
Bhaskar98634f02013-10-29 23:30:51 -04002767 The default hash type is "map-based" and is recommended for most usages. The
2768 default function is "sdbm", the selection of a function should be based on
2769 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002770
2771 See also : "balance", "server"
2772
2773
Willy Tarreau0ba27502007-12-24 16:55:16 +01002774http-check disable-on-404
2775 Enable a maintenance mode upon HTTP/404 response to health-checks
2776 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01002777 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01002778 Arguments : none
2779
2780 When this option is set, a server which returns an HTTP code 404 will be
2781 excluded from further load-balancing, but will still receive persistent
2782 connections. This provides a very convenient method for Web administrators
2783 to perform a graceful shutdown of their servers. It is also important to note
2784 that a server which is detected as failed while it was in this mode will not
2785 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
2786 will immediately be reinserted into the farm. The status on the stats page
2787 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01002788 option only works in conjunction with the "httpchk" option. If this option
2789 is used with "http-check expect", then it has precedence over it so that 404
2790 responses will still be considered as soft-stop.
2791
2792 See also : "option httpchk", "http-check expect"
2793
2794
2795http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002796 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01002797 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02002798 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01002799 Arguments :
2800 <match> is a keyword indicating how to look for a specific pattern in the
2801 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002802 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01002803 exclamation mark ("!") to negate the match. Spaces are allowed
2804 between the exclamation mark and the keyword. See below for more
2805 details on the supported keywords.
2806
2807 <pattern> is the pattern to look for. It may be a string or a regular
2808 expression. If the pattern contains spaces, they must be escaped
2809 with the usual backslash ('\').
2810
2811 By default, "option httpchk" considers that response statuses 2xx and 3xx
2812 are valid, and that others are invalid. When "http-check expect" is used,
2813 it defines what is considered valid or invalid. Only one "http-check"
2814 statement is supported in a backend. If a server fails to respond or times
2815 out, the check obviously fails. The available matches are :
2816
2817 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002818 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002819 response's status code is exactly this string. If the
2820 "status" keyword is prefixed with "!", then the response
2821 will be considered invalid if the status code matches.
2822
2823 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002824 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002825 response's status code matches the expression. If the
2826 "rstatus" keyword is prefixed with "!", then the response
2827 will be considered invalid if the status code matches.
2828 This is mostly used to check for multiple codes.
2829
2830 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002831 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002832 response's body contains this exact string. If the
2833 "string" keyword is prefixed with "!", then the response
2834 will be considered invalid if the body contains this
2835 string. This can be used to look for a mandatory word at
2836 the end of a dynamic page, or to detect a failure when a
2837 specific error appears on the check page (eg: a stack
2838 trace).
2839
2840 rstring <regex> : test a regular expression on 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 matches this expression. If the "rstring"
2843 keyword is prefixed with "!", then the response will be
2844 considered invalid if the body matches the expression.
2845 This can be used to look for a mandatory word at the end
2846 of a dynamic page, or to detect a failure when a specific
2847 error appears on the check page (eg: a stack trace).
2848
2849 It is important to note that the responses will be limited to a certain size
2850 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
2851 Thus, too large responses may not contain the mandatory pattern when using
2852 "string" or "rstring". If a large response is absolutely required, it is
2853 possible to change the default max size by setting the global variable.
2854 However, it is worth keeping in mind that parsing very large responses can
2855 waste some CPU cycles, especially when regular expressions are used, and that
2856 it is always better to focus the checks on smaller resources.
2857
2858 Last, if "http-check expect" is combined with "http-check disable-on-404",
2859 then this last one has precedence when the server responds with 404.
2860
2861 Examples :
2862 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002863 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01002864
2865 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002866 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01002867
2868 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002869 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01002870
2871 # check that we have a correct hexadecimal tag before /html
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002872 http-check expect rstring <!--tag:[0-9a-f]*</html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01002873
Willy Tarreaubd741542010-03-16 18:46:54 +01002874 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01002875
2876
Willy Tarreauef781042010-01-27 11:53:01 +01002877http-check send-state
2878 Enable emission of a state header with HTTP health checks
2879 May be used in sections : defaults | frontend | listen | backend
2880 yes | no | yes | yes
2881 Arguments : none
2882
2883 When this option is set, haproxy will systematically send a special header
2884 "X-Haproxy-Server-State" with a list of parameters indicating to each server
2885 how they are seen by haproxy. This can be used for instance when a server is
2886 manipulated without access to haproxy and the operator needs to know whether
2887 haproxy still sees it up or not, or if the server is the last one in a farm.
2888
2889 The header is composed of fields delimited by semi-colons, the first of which
2890 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
2891 checks on the total number before transition, just as appears in the stats
2892 interface. Next headers are in the form "<variable>=<value>", indicating in
2893 no specific order some values available in the stats interface :
2894 - a variable "name", containing the name of the backend followed by a slash
2895 ("/") then the name of the server. This can be used when a server is
2896 checked in multiple backends.
2897
2898 - a variable "node" containing the name of the haproxy node, as set in the
2899 global "node" variable, otherwise the system's hostname if unspecified.
2900
2901 - a variable "weight" indicating the weight of the server, a slash ("/")
2902 and the total weight of the farm (just counting usable servers). This
2903 helps to know if other servers are available to handle the load when this
2904 one fails.
2905
2906 - a variable "scur" indicating the current number of concurrent connections
2907 on the server, followed by a slash ("/") then the total number of
2908 connections on all servers of the same backend.
2909
2910 - a variable "qcur" indicating the current number of requests in the
2911 server's queue.
2912
2913 Example of a header received by the application server :
2914 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
2915 scur=13/22; qcur=0
2916
2917 See also : "option httpchk", "http-check disable-on-404"
2918
Willy Tarreauccbcc372012-12-27 12:37:57 +01002919http-request { allow | deny | tarpit | auth [realm <realm>] | redirect <rule> |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02002920 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002921 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06002922 replace-header <name> <match-regex> <replace-fmt> |
2923 replace-value <name> <match-regex> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02002924 set-tos <tos> | set-mark <mark> |
2925 add-acl(<file name>) <key fmt> |
2926 del-acl(<file name>) <key fmt> |
2927 del-map(<file name>) <key fmt> |
2928 set-map(<file name>) <key fmt> <value fmt>
2929 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01002930 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002931 Access control for Layer 7 requests
2932
2933 May be used in sections: defaults | frontend | listen | backend
2934 no | yes | yes | yes
2935
Willy Tarreau20b0de52012-12-24 15:45:22 +01002936 The http-request statement defines a set of rules which apply to layer 7
2937 processing. The rules are evaluated in their declaration order when they are
2938 met in a frontend, listen or backend section. Any rule may optionally be
2939 followed by an ACL-based condition, in which case it will only be evaluated
2940 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002941
Willy Tarreau20b0de52012-12-24 15:45:22 +01002942 The first keyword is the rule's action. Currently supported actions include :
2943 - "allow" : this stops the evaluation of the rules and lets the request
2944 pass the check. No further "http-request" rules are evaluated.
2945
2946 - "deny" : this stops the evaluation of the rules and immediately rejects
2947 the request and emits an HTTP 403 error. No further "http-request" rules
2948 are evaluated.
2949
Willy Tarreauccbcc372012-12-27 12:37:57 +01002950 - "tarpit" : this stops the evaluation of the rules and immediately blocks
2951 the request without responding for a delay specified by "timeout tarpit"
2952 or "timeout connect" if the former is not set. After that delay, if the
2953 client is still connected, an HTTP error 500 is returned so that the
2954 client does not suspect it has been tarpitted. Logs will report the flags
2955 "PT". The goal of the tarpit rule is to slow down robots during an attack
2956 when they're limited on the number of concurrent requests. It can be very
2957 efficient against very dumb robots, and will significantly reduce the
2958 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002959 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreauccbcc372012-12-27 12:37:57 +01002960 front firewall to support insane number of concurrent connections.
2961
Willy Tarreau20b0de52012-12-24 15:45:22 +01002962 - "auth" : this stops the evaluation of the rules and immediately responds
2963 with an HTTP 401 or 407 error code to invite the user to present a valid
2964 user name and password. No further "http-request" rules are evaluated. An
2965 optional "realm" parameter is supported, it sets the authentication realm
2966 that is returned with the response (typically the application's name).
2967
Willy Tarreau81499eb2012-12-27 12:19:02 +01002968 - "redirect" : this performs an HTTP redirection based on a redirect rule.
2969 This is exactly the same as the "redirect" statement except that it
2970 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01002971 "http-request" rules and that these rules use the "log-format" strings.
2972 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01002973
Willy Tarreau20b0de52012-12-24 15:45:22 +01002974 - "add-header" appends an HTTP header field whose name is specified in
2975 <name> and whose value is defined by <fmt> which follows the log-format
2976 rules (see Custom Log Format in section 8.2.4). This is particularly
2977 useful to pass connection-specific information to the server (eg: the
2978 client's SSL certificate), or to combine several headers into one. This
2979 rule is not final, so it is possible to add other similar rules. Note
2980 that header addition is performed immediately, so one rule might reuse
2981 the resulting header from a previous rule.
2982
2983 - "set-header" does the same as "add-header" except that the header name
2984 is first removed if it existed. This is useful when passing security
2985 information to the server, where the header must not be manipulated by
2986 external users.
2987
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002988 - "del-header" removes all HTTP header fields whose name is specified in
2989 <name>.
2990
Sasha Pachev218f0642014-06-16 12:05:59 -06002991 - "replace-header" matches the regular expression in all occurrences of
2992 header field <name> according to <match-regex>, and replaces them with
2993 the <replace-fmt> argument. Format characters are allowed in replace-fmt
2994 and work like in <fmt> arguments in "add-header". The match is only
2995 case-sensitive. It is important to understand that this action only
2996 considers whole header lines, regardless of the number of values they
2997 may contain. This usage is suited to headers naturally containing commas
2998 in their value, such as If-Modified-Since and so on.
2999
3000 Example:
3001
3002 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
3003
3004 applied to:
3005
3006 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3007
3008 outputs:
3009
3010 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3011
3012 assuming the backend IP is 192.168.1.20
3013
3014 - "replace-value" works like "replace-header" except that it matches the
3015 regex against every comma-delimited value of the header field <name>
3016 instead of the entire header. This is suited for all headers which are
3017 allowed to carry more than one value. An example could be the Accept
3018 header.
3019
3020 Example:
3021
3022 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
3023
3024 applied to:
3025
3026 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
3027
3028 outputs:
3029
3030 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
3031
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003032 - "set-nice" sets the "nice" factor of the current request being processed.
3033 It only has effect against the other requests being processed at the same
3034 time. The default value is 0, unless altered by the "nice" setting on the
3035 "bind" line. The accepted range is -1024..1024. The higher the value, the
3036 nicest the request will be. Lower values will make the request more
3037 important than other ones. This can be useful to improve the speed of
3038 some requests, or lower the priority of non-important requests. Using
3039 this setting without prior experimentation can cause some major slowdown.
3040
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003041 - "set-log-level" is used to change the log level of the current request
3042 when a certain condition is met. Valid levels are the 8 syslog levels
3043 (see the "log" keyword) plus the special level "silent" which disables
3044 logging for this request. This rule is not final so the last matching
3045 rule wins. This rule can be useful to disable health checks coming from
3046 another equipment.
3047
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003048 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3049 the client to the value passed in <tos> on platforms which support this.
3050 This value represents the whole 8 bits of the IP TOS field, and can be
3051 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3052 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3053 bits are always 0. This can be used to adjust some routing behaviour on
3054 border routers based on some information from the request. See RFC 2474,
3055 2597, 3260 and 4594 for more information.
3056
Willy Tarreau51347ed2013-06-11 19:34:13 +02003057 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3058 client to the value passed in <mark> on platforms which support it. This
3059 value is an unsigned 32 bit value which can be matched by netfilter and
3060 by the routing table. It can be expressed both in decimal or hexadecimal
3061 format (prefixed by "0x"). This can be useful to force certain packets to
3062 take a different route (for example a cheaper network path for bulk
3063 downloads). This works on Linux kernels 2.6.32 and above and requires
3064 admin privileges.
3065
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003066 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3067 from a file (even a dummy empty file). The file name of the ACL to be
3068 updated is passed between parentheses. It takes one argument: <key fmt>,
3069 which follows log-format rules, to collect content of the new entry. It
3070 performs a lookup in the ACL before insertion, to avoid duplicated (or
3071 more) values. This lookup is done by a linear search and can be expensive
3072 with large lists! It is the equivalent of the "add acl" command from the
3073 stats socket, but can be triggered by an HTTP request.
3074
3075 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3076 from a file (even a dummy empty file). The file name of the ACL to be
3077 updated is passed between parentheses. It takes one argument: <key fmt>,
3078 which follows log-format rules, to collect content of the entry to delete.
3079 It is the equivalent of the "del acl" command from the stats socket, but
3080 can be triggered by an HTTP request.
3081
3082 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3083 from a file (even a dummy empty file). The file name of the MAP to be
3084 updated is passed between parentheses. It takes one argument: <key fmt>,
3085 which follows log-format rules, to collect content of the entry to delete.
3086 It takes one argument: "file name" It is the equivalent of the "del map"
3087 command from the stats socket, but can be triggered by an HTTP request.
3088
3089 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3090 from a file (even a dummy empty file). The file name of the MAP to be
3091 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3092 which follows log-format rules, used to collect MAP key, and <value fmt>,
3093 which follows log-format rules, used to collect content for the new entry.
3094 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3095 more) values. This lookup is done by a linear search and can be expensive
3096 with large lists! It is the equivalent of the "set map" command from the
3097 stats socket, but can be triggered by an HTTP request.
3098
Willy Tarreau20b0de52012-12-24 15:45:22 +01003099 There is no limit to the number of http-request statements per instance.
3100
3101 It is important to know that http-request rules are processed very early in
3102 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
3103 rules. That way, headers added by "add-header"/"set-header" are visible by
3104 almost all further ACL rules.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003105
3106 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01003107 acl nagios src 192.168.129.3
3108 acl local_net src 192.168.0.0/16
3109 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003110
Cyril Bonté78caf842010-03-10 22:41:43 +01003111 http-request allow if nagios
3112 http-request allow if local_net auth_ok
3113 http-request auth realm Gimme if local_net auth_ok
3114 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003115
Cyril Bonté78caf842010-03-10 22:41:43 +01003116 Example:
3117 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01003118 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003119
Willy Tarreau20b0de52012-12-24 15:45:22 +01003120 Example:
3121 http-request set-header X-Haproxy-Current-Date %T
3122 http-request set-header X-SSL %[ssl_fc]
3123 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id]
3124 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
3125 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
3126 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
3127 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
3128 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
3129 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
3130
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003131 Example:
3132 acl key req.hdr(X-Add-Acl-Key) -m found
3133 acl add path /addacl
3134 acl del path /delacl
3135
3136 acl myhost hdr(Host) -f myhost.lst
3137
3138 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
3139 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
3140
3141 Example:
3142 acl value req.hdr(X-Value) -m found
3143 acl setmap path /setmap
3144 acl delmap path /delmap
3145
3146 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3147
3148 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
3149 http-request del-map(map.lst) %[src] if delmap
3150
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02003151 See also : "stats http-request", section 3.4 about userlists and section 7
3152 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01003153
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003154http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003155 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003156 replace-header <name> <regex-match> <replace-fmt> |
3157 replace-value <name> <regex-match> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003158 set-log-level <level> | set-mark <mark> | set-tos <tos> |
3159 add-acl(<file name>) <key fmt> |
3160 del-acl(<file name>) <key fmt> |
3161 del-map(<file name>) <key fmt> |
3162 set-map(<file name>) <key fmt> <value fmt>
3163 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02003164 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003165 Access control for Layer 7 responses
3166
3167 May be used in sections: defaults | frontend | listen | backend
3168 no | yes | yes | yes
3169
3170 The http-response statement defines a set of rules which apply to layer 7
3171 processing. The rules are evaluated in their declaration order when they are
3172 met in a frontend, listen or backend section. Any rule may optionally be
3173 followed by an ACL-based condition, in which case it will only be evaluated
3174 if the condition is true. Since these rules apply on responses, the backend
3175 rules are applied first, followed by the frontend's rules.
3176
3177 The first keyword is the rule's action. Currently supported actions include :
3178 - "allow" : this stops the evaluation of the rules and lets the response
3179 pass the check. No further "http-response" rules are evaluated for the
3180 current section.
3181
3182 - "deny" : this stops the evaluation of the rules and immediately rejects
3183 the response and emits an HTTP 502 error. No further "http-response"
3184 rules are evaluated.
3185
3186 - "add-header" appends an HTTP header field whose name is specified in
3187 <name> and whose value is defined by <fmt> which follows the log-format
3188 rules (see Custom Log Format in section 8.2.4). This may be used to send
3189 a cookie to a client for example, or to pass some internal information.
3190 This rule is not final, so it is possible to add other similar rules.
3191 Note that header addition is performed immediately, so one rule might
3192 reuse the resulting header from a previous rule.
3193
3194 - "set-header" does the same as "add-header" except that the header name
3195 is first removed if it existed. This is useful when passing security
3196 information to the server, where the header must not be manipulated by
3197 external users.
3198
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003199 - "del-header" removes all HTTP header fields whose name is specified in
3200 <name>.
3201
Sasha Pachev218f0642014-06-16 12:05:59 -06003202 - "replace-header" matches the regular expression in all occurrences of
3203 header field <name> according to <match-regex>, and replaces them with
3204 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3205 and work like in <fmt> arguments in "add-header". The match is only
3206 case-sensitive. It is important to understand that this action only
3207 considers whole header lines, regardless of the number of values they
3208 may contain. This usage is suited to headers naturally containing commas
3209 in their value, such as Set-Cookie, Expires and so on.
3210
3211 Example:
3212
3213 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
3214
3215 applied to:
3216
3217 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
3218
3219 outputs:
3220
3221 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
3222
3223 assuming the backend IP is 192.168.1.20.
3224
3225 - "replace-value" works like "replace-header" except that it matches the
3226 regex against every comma-delimited value of the header field <name>
3227 instead of the entire header. This is suited for all headers which are
3228 allowed to carry more than one value. An example could be the Accept
3229 header.
3230
3231 Example:
3232
3233 http-response replace-value Cache-control ^public$ private
3234
3235 applied to:
3236
3237 Cache-Control: max-age=3600, public
3238
3239 outputs:
3240
3241 Cache-Control: max-age=3600, private
3242
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003243 - "set-nice" sets the "nice" factor of the current request being processed.
3244 It only has effect against the other requests being processed at the same
3245 time. The default value is 0, unless altered by the "nice" setting on the
3246 "bind" line. The accepted range is -1024..1024. The higher the value, the
3247 nicest the request will be. Lower values will make the request more
3248 important than other ones. This can be useful to improve the speed of
3249 some requests, or lower the priority of non-important requests. Using
3250 this setting without prior experimentation can cause some major slowdown.
3251
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003252 - "set-log-level" is used to change the log level of the current request
3253 when a certain condition is met. Valid levels are the 8 syslog levels
3254 (see the "log" keyword) plus the special level "silent" which disables
3255 logging for this request. This rule is not final so the last matching
3256 rule wins. This rule can be useful to disable health checks coming from
3257 another equipment.
3258
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003259 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3260 the client to the value passed in <tos> on platforms which support this.
3261 This value represents the whole 8 bits of the IP TOS field, and can be
3262 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3263 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3264 bits are always 0. This can be used to adjust some routing behaviour on
3265 border routers based on some information from the request. See RFC 2474,
3266 2597, 3260 and 4594 for more information.
3267
Willy Tarreau51347ed2013-06-11 19:34:13 +02003268 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3269 client to the value passed in <mark> on platforms which support it. This
3270 value is an unsigned 32 bit value which can be matched by netfilter and
3271 by the routing table. It can be expressed both in decimal or hexadecimal
3272 format (prefixed by "0x"). This can be useful to force certain packets to
3273 take a different route (for example a cheaper network path for bulk
3274 downloads). This works on Linux kernels 2.6.32 and above and requires
3275 admin privileges.
3276
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003277 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3278 from a file (even a dummy empty file). The file name of the ACL to be
3279 updated is passed between parentheses. It takes one argument: <key fmt>,
3280 which follows log-format rules, to collect content of the new entry. It
3281 performs a lookup in the ACL before insertion, to avoid duplicated (or
3282 more) values. This lookup is done by a linear search and can be expensive
3283 with large lists! It is the equivalent of the "add acl" command from the
3284 stats socket, but can be triggered by an HTTP response.
3285
3286 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3287 from a file (even a dummy empty file). The file name of the ACL to be
3288 updated is passed between parentheses. It takes one argument: <key fmt>,
3289 which follows log-format rules, to collect content of the entry to delete.
3290 It is the equivalent of the "del acl" command from the stats socket, but
3291 can be triggered by an HTTP response.
3292
3293 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3294 from a file (even a dummy empty file). The file name of the MAP to be
3295 updated is passed between parentheses. It takes one argument: <key fmt>,
3296 which follows log-format rules, to collect content of the entry to delete.
3297 It takes one argument: "file name" It is the equivalent of the "del map"
3298 command from the stats socket, but can be triggered by an HTTP response.
3299
3300 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3301 from a file (even a dummy empty file). The file name of the MAP to be
3302 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3303 which follows log-format rules, used to collect MAP key, and <value fmt>,
3304 which follows log-format rules, used to collect content for the new entry.
3305 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3306 more) values. This lookup is done by a linear search and can be expensive
3307 with large lists! It is the equivalent of the "set map" command from the
3308 stats socket, but can be triggered by an HTTP response.
3309
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003310 There is no limit to the number of http-response statements per instance.
3311
Godbach09250262013-07-02 01:19:15 +08003312 It is important to know that http-response rules are processed very early in
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003313 the HTTP processing, before "reqdel" or "reqrep" rules. That way, headers
3314 added by "add-header"/"set-header" are visible by almost all further ACL
3315 rules.
3316
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003317 Example:
3318 acl key_acl res.hdr(X-Acl-Key) -m found
3319
3320 acl myhost hdr(Host) -f myhost.lst
3321
3322 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3323 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3324
3325 Example:
3326 acl value res.hdr(X-Value) -m found
3327
3328 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3329
3330 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
3331 http-response del-map(map.lst) %[src] if ! value
3332
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003333 See also : "http-request", section 3.4 about userlists and section 7 about
3334 ACL usage.
3335
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02003336
Mark Lamourinec2247f02012-01-04 13:02:01 -05003337http-send-name-header [<header>]
3338 Add the server name to a request. Use the header string given by <header>
3339
3340 May be used in sections: defaults | frontend | listen | backend
3341 yes | no | yes | yes
3342
3343 Arguments :
3344
3345 <header> The header string to use to send the server name
3346
3347 The "http-send-name-header" statement causes the name of the target
3348 server to be added to the headers of an HTTP request. The name
3349 is added with the header string proved.
3350
3351 See also : "server"
3352
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003353id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02003354 Set a persistent ID to a proxy.
3355 May be used in sections : defaults | frontend | listen | backend
3356 no | yes | yes | yes
3357 Arguments : none
3358
3359 Set a persistent ID for the proxy. This ID must be unique and positive.
3360 An unused ID will automatically be assigned if unset. The first assigned
3361 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003362
3363
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003364ignore-persist { if | unless } <condition>
3365 Declare a condition to ignore persistence
3366 May be used in sections: defaults | frontend | listen | backend
3367 no | yes | yes | yes
3368
3369 By default, when cookie persistence is enabled, every requests containing
3370 the cookie are unconditionally persistent (assuming the target server is up
3371 and running).
3372
3373 The "ignore-persist" statement allows one to declare various ACL-based
3374 conditions which, when met, will cause a request to ignore persistence.
3375 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003376 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003377 persistence for a specific User-Agent (for example, some web crawler bots).
3378
3379 Combined with "appsession", it can also help reduce HAProxy memory usage, as
3380 the appsession table won't grow if persistence is ignored.
3381
3382 The persistence is ignored when an "if" condition is met, or unless an
3383 "unless" condition is met.
3384
3385 See also : "force-persist", "cookie", and section 7 about ACL usage.
3386
3387
Willy Tarreau2769aa02007-12-27 18:26:09 +01003388log global
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003389log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02003390no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01003391 Enable per-instance logging of events and traffic.
3392 May be used in sections : defaults | frontend | listen | backend
3393 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02003394
3395 Prefix :
3396 no should be used when the logger list must be flushed. For example,
3397 if you don't want to inherit from the default logger list. This
3398 prefix does not allow arguments.
3399
Willy Tarreau2769aa02007-12-27 18:26:09 +01003400 Arguments :
3401 global should be used when the instance's logging parameters are the
3402 same as the global ones. This is the most common usage. "global"
3403 replaces <address>, <facility> and <level> with those of the log
3404 entries found in the "global" section. Only one "log global"
3405 statement may be used per instance, and this form takes no other
3406 parameter.
3407
3408 <address> indicates where to send the logs. It takes the same format as
3409 for the "global" section's logs, and can be one of :
3410
3411 - An IPv4 address optionally followed by a colon (':') and a UDP
3412 port. If no port is specified, 514 is used by default (the
3413 standard syslog port).
3414
David du Colombier24bb5f52011-03-17 10:40:23 +01003415 - An IPv6 address followed by a colon (':') and optionally a UDP
3416 port. If no port is specified, 514 is used by default (the
3417 standard syslog port).
3418
Willy Tarreau2769aa02007-12-27 18:26:09 +01003419 - A filesystem path to a UNIX domain socket, keeping in mind
3420 considerations for chroot (be sure the path is accessible
3421 inside the chroot) and uid/gid (be sure the path is
3422 appropriately writeable).
3423
Willy Tarreaudad36a32013-03-11 01:20:04 +01003424 Any part of the address string may reference any number of
3425 environment variables by preceding their name with a dollar
3426 sign ('$') and optionally enclosing them with braces ('{}'),
3427 similarly to what is done in Bourne shell.
3428
Willy Tarreaudc2695c2014-06-27 18:10:07 +02003429 <length> is an optional maximum line length. Log lines larger than this
3430 value will be truncated before being sent. The reason is that
3431 syslog servers act differently on log line length. All servers
3432 support the default value of 1024, but some servers simply drop
3433 larger lines while others do log them. If a server supports long
3434 lines, it may make sense to set this value here in order to avoid
3435 truncating long lines. Similarly, if a server drops long lines,
3436 it is preferable to truncate them before sending them. Accepted
3437 values are 80 to 65535 inclusive. The default value of 1024 is
3438 generally fine for all standard usages. Some specific cases of
3439 long captures or JSON-formated logs may require larger values.
3440
Willy Tarreau2769aa02007-12-27 18:26:09 +01003441 <facility> must be one of the 24 standard syslog facilities :
3442
3443 kern user mail daemon auth syslog lpr news
3444 uucp cron auth2 ftp ntp audit alert cron2
3445 local0 local1 local2 local3 local4 local5 local6 local7
3446
3447 <level> is optional and can be specified to filter outgoing messages. By
3448 default, all messages are sent. If a level is specified, only
3449 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003450 will be sent. An optional minimum level can be specified. If it
3451 is set, logs emitted with a more severe level than this one will
3452 be capped to this level. This is used to avoid sending "emerg"
3453 messages on all terminals on some default syslog configurations.
3454 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003455
3456 emerg alert crit err warning notice info debug
3457
William Lallemand0f99e342011-10-12 17:50:54 +02003458 It is important to keep in mind that it is the frontend which decides what to
3459 log from a connection, and that in case of content switching, the log entries
3460 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003461
3462 However, backend log declaration define how and where servers status changes
3463 will be logged. Level "notice" will be used to indicate a server going up,
3464 "warning" will be used for termination signals and definitive service
3465 termination, and "alert" will be used for when a server goes down.
3466
3467 Note : According to RFC3164, messages are truncated to 1024 bytes before
3468 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003469
3470 Example :
3471 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003472 log 127.0.0.1:514 local0 notice # only send important events
3473 log 127.0.0.1:514 local0 notice notice # same but limit output level
Willy Tarreaudad36a32013-03-11 01:20:04 +01003474 log ${LOCAL_SYSLOG}:514 local0 notice # send to local server
3475
Willy Tarreau2769aa02007-12-27 18:26:09 +01003476
William Lallemand48940402012-01-30 16:47:22 +01003477log-format <string>
3478 Allows you to custom a log line.
3479
3480 See also : Custom Log Format (8.2.4)
3481
Willy Tarreau2769aa02007-12-27 18:26:09 +01003482
Willy Tarreauc35362a2014-04-25 13:58:37 +02003483max-keep-alive-queue <value>
3484 Set the maximum server queue size for maintaining keep-alive connections
3485 May be used in sections: defaults | frontend | listen | backend
3486 yes | no | yes | yes
3487
3488 HTTP keep-alive tries to reuse the same server connection whenever possible,
3489 but sometimes it can be counter-productive, for example if a server has a lot
3490 of connections while other ones are idle. This is especially true for static
3491 servers.
3492
3493 The purpose of this setting is to set a threshold on the number of queued
3494 connections at which haproxy stops trying to reuse the same server and prefers
3495 to find another one. The default value, -1, means there is no limit. A value
3496 of zero means that keep-alive requests will never be queued. For very close
3497 servers which can be reached with a low latency and which are not sensible to
3498 breaking keep-alive, a low value is recommended (eg: local static server can
3499 use a value of 10 or less). For remote servers suffering from a high latency,
3500 higher values might be needed to cover for the latency and/or the cost of
3501 picking a different server.
3502
3503 Note that this has no impact on responses which are maintained to the same
3504 server consecutively to a 401 response. They will still go to the same server
3505 even if they have to be queued.
3506
3507 See also : "option http-server-close", "option prefer-last-server", server
3508 "maxconn" and cookie persistence.
3509
3510
Willy Tarreau2769aa02007-12-27 18:26:09 +01003511maxconn <conns>
3512 Fix the maximum number of concurrent connections on a frontend
3513 May be used in sections : defaults | frontend | listen | backend
3514 yes | yes | yes | no
3515 Arguments :
3516 <conns> is the maximum number of concurrent connections the frontend will
3517 accept to serve. Excess connections will be queued by the system
3518 in the socket's listen queue and will be served once a connection
3519 closes.
3520
3521 If the system supports it, it can be useful on big sites to raise this limit
3522 very high so that haproxy manages connection queues, instead of leaving the
3523 clients with unanswered connection attempts. This value should not exceed the
3524 global maxconn. Also, keep in mind that a connection contains two buffers
3525 of 8kB each, as well as some other data resulting in about 17 kB of RAM being
3526 consumed per established connection. That means that a medium system equipped
3527 with 1GB of RAM can withstand around 40000-50000 concurrent connections if
3528 properly tuned.
3529
3530 Also, when <conns> is set to large values, it is possible that the servers
3531 are not sized to accept such loads, and for this reason it is generally wise
3532 to assign them some reasonable connection limits.
3533
Vincent Bernat6341be52012-06-27 17:18:30 +02003534 By default, this value is set to 2000.
3535
Willy Tarreau2769aa02007-12-27 18:26:09 +01003536 See also : "server", global section's "maxconn", "fullconn"
3537
3538
3539mode { tcp|http|health }
3540 Set the running mode or protocol of the instance
3541 May be used in sections : defaults | frontend | listen | backend
3542 yes | yes | yes | yes
3543 Arguments :
3544 tcp The instance will work in pure TCP mode. A full-duplex connection
3545 will be established between clients and servers, and no layer 7
3546 examination will be performed. This is the default mode. It
3547 should be used for SSL, SSH, SMTP, ...
3548
3549 http The instance will work in HTTP mode. The client request will be
3550 analyzed in depth before connecting to any server. Any request
3551 which is not RFC-compliant will be rejected. Layer 7 filtering,
3552 processing and switching will be possible. This is the mode which
3553 brings HAProxy most of its value.
3554
3555 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02003556 to incoming connections and close the connection. Alternatively,
3557 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
3558 instead. Nothing will be logged in either case. This mode is used
3559 to reply to external components health checks. This mode is
3560 deprecated and should not be used anymore as it is possible to do
3561 the same and even better by combining TCP or HTTP modes with the
3562 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003563
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003564 When doing content switching, it is mandatory that the frontend and the
3565 backend are in the same mode (generally HTTP), otherwise the configuration
3566 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003567
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003568 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003569 defaults http_instances
3570 mode http
3571
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003572 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003573
Willy Tarreau0ba27502007-12-24 16:55:16 +01003574
Cyril Bontéf0c60612010-02-06 14:44:47 +01003575monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01003576 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003577 May be used in sections : defaults | frontend | listen | backend
3578 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01003579 Arguments :
3580 if <cond> the monitor request will fail if the condition is satisfied,
3581 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003582 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01003583 are met, for instance a low number of servers both in a
3584 backend and its backup.
3585
3586 unless <cond> the monitor request will succeed only if the condition is
3587 satisfied, and will fail otherwise. Such a condition may be
3588 based on a test on the presence of a minimum number of active
3589 servers in a list of backends.
3590
3591 This statement adds a condition which can force the response to a monitor
3592 request to report a failure. By default, when an external component queries
3593 the URI dedicated to monitoring, a 200 response is returned. When one of the
3594 conditions above is met, haproxy will return 503 instead of 200. This is
3595 very useful to report a site failure to an external component which may base
3596 routing advertisements between multiple sites on the availability reported by
3597 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003598 criterion. Note that "monitor fail" only works in HTTP mode. Both status
3599 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003600
3601 Example:
3602 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01003603 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01003604 acl site_dead nbsrv(dynamic) lt 2
3605 acl site_dead nbsrv(static) lt 2
3606 monitor-uri /site_alive
3607 monitor fail if site_dead
3608
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003609 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003610
3611
3612monitor-net <source>
3613 Declare a source network which is limited to monitor requests
3614 May be used in sections : defaults | frontend | listen | backend
3615 yes | yes | yes | no
3616 Arguments :
3617 <source> is the source IPv4 address or network which will only be able to
3618 get monitor responses to any request. It can be either an IPv4
3619 address, a host name, or an address followed by a slash ('/')
3620 followed by a mask.
3621
3622 In TCP mode, any connection coming from a source matching <source> will cause
3623 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003624 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01003625 forwarding the connection to a remote server.
3626
3627 In HTTP mode, a connection coming from a source matching <source> will be
3628 accepted, the following response will be sent without waiting for a request,
3629 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
3630 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02003631 running without forwarding the request to a backend server. Note that this
3632 response is sent in raw format, without any transformation. This is important
3633 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003634
Willy Tarreau82569f92012-09-27 23:48:56 +02003635 Monitor requests are processed very early, just after tcp-request connection
3636 ACLs which are the only ones able to block them. These connections are short
3637 lived and never wait for any data from the client. They cannot be logged, and
3638 it is the intended purpose. They are only used to report HAProxy's health to
3639 an upper component, nothing more. Please note that "monitor fail" rules do
3640 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01003641
Willy Tarreau95cd2832010-03-04 23:36:33 +01003642 Last, please note that only one "monitor-net" statement can be specified in
3643 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003644
Willy Tarreau2769aa02007-12-27 18:26:09 +01003645 Example :
3646 # addresses .252 and .253 are just probing us.
3647 frontend www
3648 monitor-net 192.168.0.252/31
3649
3650 See also : "monitor fail", "monitor-uri"
3651
3652
3653monitor-uri <uri>
3654 Intercept a URI used by external components' monitor requests
3655 May be used in sections : defaults | frontend | listen | backend
3656 yes | yes | yes | no
3657 Arguments :
3658 <uri> is the exact URI which we want to intercept to return HAProxy's
3659 health status instead of forwarding the request.
3660
3661 When an HTTP request referencing <uri> will be received on a frontend,
3662 HAProxy will not forward it nor log it, but instead will return either
3663 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
3664 conditions defined with "monitor fail". This is normally enough for any
3665 front-end HTTP probe to detect that the service is UP and running without
3666 forwarding the request to a backend server. Note that the HTTP method, the
3667 version and all headers are ignored, but the request must at least be valid
3668 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
3669
3670 Monitor requests are processed very early. It is not possible to block nor
3671 divert them using ACLs. They cannot be logged either, and it is the intended
3672 purpose. They are only used to report HAProxy's health to an upper component,
3673 nothing more. However, it is possible to add any number of conditions using
3674 "monitor fail" and ACLs so that the result can be adjusted to whatever check
3675 can be imagined (most often the number of available servers in a backend).
3676
3677 Example :
3678 # Use /haproxy_test to report haproxy's status
3679 frontend www
3680 mode http
3681 monitor-uri /haproxy_test
3682
3683 See also : "monitor fail", "monitor-net"
3684
Willy Tarreau0ba27502007-12-24 16:55:16 +01003685
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003686option abortonclose
3687no option abortonclose
3688 Enable or disable early dropping of aborted requests pending in queues.
3689 May be used in sections : defaults | frontend | listen | backend
3690 yes | no | yes | yes
3691 Arguments : none
3692
3693 In presence of very high loads, the servers will take some time to respond.
3694 The per-instance connection queue will inflate, and the response time will
3695 increase respective to the size of the queue times the average per-session
3696 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01003697 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003698 the queue, and slowing down other users, and the servers as well, because the
3699 request will eventually be served, then aborted at the first error
3700 encountered while delivering the response.
3701
3702 As there is no way to distinguish between a full STOP and a simple output
3703 close on the client side, HTTP agents should be conservative and consider
3704 that the client might only have closed its output channel while waiting for
3705 the response. However, this introduces risks of congestion when lots of users
3706 do the same, and is completely useless nowadays because probably no client at
3707 all will close the session while waiting for the response. Some HTTP agents
Willy Tarreaud72758d2010-01-12 10:42:19 +01003708 support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003709 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01003710 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003711 of being the single component to break rare but valid traffic is extremely
3712 low, which adds to the temptation to be able to abort a session early while
3713 still not served and not pollute the servers.
3714
3715 In HAProxy, the user can choose the desired behaviour using the option
3716 "abortonclose". By default (without the option) the behaviour is HTTP
3717 compliant and aborted requests will be served. But when the option is
3718 specified, a session with an incoming channel closed will be aborted while
3719 it is still possible, either pending in the queue for a connection slot, or
3720 during the connection establishment if the server has not yet acknowledged
3721 the connection request. This considerably reduces the queue size and the load
3722 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01003723 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003724
3725 If this option has been enabled in a "defaults" section, it can be disabled
3726 in a specific instance by prepending the "no" keyword before it.
3727
3728 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
3729
3730
Willy Tarreau4076a152009-04-02 15:18:36 +02003731option accept-invalid-http-request
3732no option accept-invalid-http-request
3733 Enable or disable relaxing of HTTP request parsing
3734 May be used in sections : defaults | frontend | listen | backend
3735 yes | yes | yes | no
3736 Arguments : none
3737
3738 By default, HAProxy complies with RFC2616 in terms of message parsing. This
3739 means that invalid characters in header names are not permitted and cause an
3740 error to be returned to the client. This is the desired behaviour as such
3741 forbidden characters are essentially used to build attacks exploiting server
3742 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3743 server will emit invalid header names for whatever reason (configuration,
3744 implementation) and the issue will not be immediately fixed. In such a case,
3745 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01003746 even if that does not make sense, by specifying this option. Similarly, the
3747 list of characters allowed to appear in a URI is well defined by RFC3986, and
3748 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
3749 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
3750 not allowed at all. Haproxy always blocks a number of them (0..32, 127). The
3751 remaining ones are blocked by default unless this option is enabled.
Willy Tarreau4076a152009-04-02 15:18:36 +02003752
3753 This option should never be enabled by default as it hides application bugs
3754 and open security breaches. It should only be deployed after a problem has
3755 been confirmed.
3756
3757 When this option is enabled, erroneous header names will still be accepted in
3758 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01003759 analysis using the "show errors" request on the UNIX stats socket. Similarly,
3760 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02003761 also helps confirming that the issue has been solved.
3762
3763 If this option has been enabled in a "defaults" section, it can be disabled
3764 in a specific instance by prepending the "no" keyword before it.
3765
3766 See also : "option accept-invalid-http-response" and "show errors" on the
3767 stats socket.
3768
3769
3770option accept-invalid-http-response
3771no option accept-invalid-http-response
3772 Enable or disable relaxing of HTTP response parsing
3773 May be used in sections : defaults | frontend | listen | backend
3774 yes | no | yes | yes
3775 Arguments : none
3776
3777 By default, HAProxy complies with RFC2616 in terms of message parsing. This
3778 means that invalid characters in header names are not permitted and cause an
3779 error to be returned to the client. This is the desired behaviour as such
3780 forbidden characters are essentially used to build attacks exploiting server
3781 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3782 server will emit invalid header names for whatever reason (configuration,
3783 implementation) and the issue will not be immediately fixed. In such a case,
3784 it is possible to relax HAProxy's header name parser to accept any character
3785 even if that does not make sense, by specifying this option.
3786
3787 This option should never be enabled by default as it hides application bugs
3788 and open security breaches. It should only be deployed after a problem has
3789 been confirmed.
3790
3791 When this option is enabled, erroneous header names will still be accepted in
3792 responses, but the complete response will be captured in order to permit
3793 later analysis using the "show errors" request on the UNIX stats socket.
3794 Doing this also helps confirming that the issue has been solved.
3795
3796 If this option has been enabled in a "defaults" section, it can be disabled
3797 in a specific instance by prepending the "no" keyword before it.
3798
3799 See also : "option accept-invalid-http-request" and "show errors" on the
3800 stats socket.
3801
3802
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003803option allbackups
3804no option allbackups
3805 Use either all backup servers at a time or only the first one
3806 May be used in sections : defaults | frontend | listen | backend
3807 yes | no | yes | yes
3808 Arguments : none
3809
3810 By default, the first operational backup server gets all traffic when normal
3811 servers are all down. Sometimes, it may be preferred to use multiple backups
3812 at once, because one will not be enough. When "option allbackups" is enabled,
3813 the load balancing will be performed among all backup servers when all normal
3814 ones are unavailable. The same load balancing algorithm will be used and the
3815 servers' weights will be respected. Thus, there will not be any priority
3816 order between the backup servers anymore.
3817
3818 This option is mostly used with static server farms dedicated to return a
3819 "sorry" page when an application is completely offline.
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
3825option checkcache
3826no option checkcache
Godbach7056a352013-12-11 20:01:07 +08003827 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003828 May be used in sections : defaults | frontend | listen | backend
3829 yes | no | yes | yes
3830 Arguments : none
3831
3832 Some high-level frameworks set application cookies everywhere and do not
3833 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003834 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003835 high risk of session crossing or stealing between users traversing the same
3836 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003837 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003838
3839 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003840 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01003841 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003842 response to check if there's a risk of caching a cookie on a client-side
3843 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01003844 to the client are :
3845 - all those without "Set-Cookie" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003846 - all those with a return code other than 200, 203, 206, 300, 301, 410,
Willy Tarreau198a7442008-01-17 12:05:32 +01003847 provided that the server has not set a "Cache-control: public" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003848 - all those that come from a POST request, provided that the server has not
3849 set a 'Cache-Control: public' header ;
3850 - those with a 'Pragma: no-cache' header
3851 - those with a 'Cache-control: private' header
3852 - those with a 'Cache-control: no-store' header
3853 - those with a 'Cache-control: max-age=0' header
3854 - those with a 'Cache-control: s-maxage=0' header
3855 - those with a 'Cache-control: no-cache' header
3856 - those with a 'Cache-control: no-cache="set-cookie"' header
3857 - those with a 'Cache-control: no-cache="set-cookie,' header
3858 (allowing other fields after set-cookie)
3859
3860 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01003861 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003862 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003863 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003864 that admins are informed that there's something to be fixed.
3865
3866 Due to the high impact on the application, the application should be tested
3867 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003868 good practice to always activate it during tests, even if it is not used in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003869 production, as it will report potentially dangerous application behaviours.
3870
3871 If this option has been enabled in a "defaults" section, it can be disabled
3872 in a specific instance by prepending the "no" keyword before it.
3873
3874
3875option clitcpka
3876no option clitcpka
3877 Enable or disable the sending of TCP keepalive packets on the client side
3878 May be used in sections : defaults | frontend | listen | backend
3879 yes | yes | yes | no
3880 Arguments : none
3881
3882 When there is a firewall or any session-aware component between a client and
3883 a server, and when the protocol involves very long sessions with long idle
3884 periods (eg: remote desktops), there is a risk that one of the intermediate
3885 components decides to expire a session which has remained idle for too long.
3886
3887 Enabling socket-level TCP keep-alives makes the system regularly send packets
3888 to the other end of the connection, leaving it active. The delay between
3889 keep-alive probes is controlled by the system only and depends both on the
3890 operating system and its tuning parameters.
3891
3892 It is important to understand that keep-alive packets are neither emitted nor
3893 received at the application level. It is only the network stacks which sees
3894 them. For this reason, even if one side of the proxy already uses keep-alives
3895 to maintain its connection alive, those keep-alive packets will not be
3896 forwarded to the other side of the proxy.
3897
3898 Please note that this has nothing to do with HTTP keep-alive.
3899
3900 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
3901 client side of a connection, which should help when session expirations are
3902 noticed between HAProxy and a client.
3903
3904 If this option has been enabled in a "defaults" section, it can be disabled
3905 in a specific instance by prepending the "no" keyword before it.
3906
3907 See also : "option srvtcpka", "option tcpka"
3908
3909
Willy Tarreau0ba27502007-12-24 16:55:16 +01003910option contstats
3911 Enable continuous traffic statistics updates
3912 May be used in sections : defaults | frontend | listen | backend
3913 yes | yes | yes | no
3914 Arguments : none
3915
3916 By default, counters used for statistics calculation are incremented
3917 only when a session finishes. It works quite well when serving small
3918 objects, but with big ones (for example large images or archives) or
3919 with A/V streaming, a graph generated from haproxy counters looks like
3920 a hedgehog. With this option enabled counters get incremented continuously,
3921 during a whole session. Recounting touches a hotpath directly so
3922 it is not enabled by default, as it has small performance impact (~0.5%).
3923
3924
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003925option dontlog-normal
3926no option dontlog-normal
3927 Enable or disable logging of normal, successful connections
3928 May be used in sections : defaults | frontend | listen | backend
3929 yes | yes | yes | no
3930 Arguments : none
3931
3932 There are large sites dealing with several thousand connections per second
3933 and for which logging is a major pain. Some of them are even forced to turn
3934 logs off and cannot debug production issues. Setting this option ensures that
3935 normal connections, those which experience no error, no timeout, no retry nor
3936 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
3937 mode, the response status code is checked and return codes 5xx will still be
3938 logged.
3939
3940 It is strongly discouraged to use this option as most of the time, the key to
3941 complex issues is in the normal logs which will not be logged here. If you
3942 need to separate logs, see the "log-separate-errors" option instead.
3943
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003944 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003945 logging.
3946
3947
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003948option dontlognull
3949no option dontlognull
3950 Enable or disable logging of null connections
3951 May be used in sections : defaults | frontend | listen | backend
3952 yes | yes | yes | no
3953 Arguments : none
3954
3955 In certain environments, there are components which will regularly connect to
3956 various systems to ensure that they are still alive. It can be the case from
3957 another load balancer as well as from monitoring systems. By default, even a
3958 simple port probe or scan will produce a log. If those connections pollute
3959 the logs too much, it is possible to enable option "dontlognull" to indicate
3960 that a connection on which no data has been transferred will not be logged,
3961 which typically corresponds to those probes.
3962
3963 It is generally recommended not to use this option in uncontrolled
3964 environments (eg: internet), otherwise scans and other malicious activities
3965 would not be logged.
3966
3967 If this option has been enabled in a "defaults" section, it can be disabled
3968 in a specific instance by prepending the "no" keyword before it.
3969
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003970 See also : "log", "monitor-net", "monitor-uri" and section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003971
3972
3973option forceclose
3974no option forceclose
3975 Enable or disable active connection closing after response is transferred.
3976 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01003977 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003978 Arguments : none
3979
3980 Some HTTP servers do not necessarily close the connections when they receive
3981 the "Connection: close" set by "option httpclose", and if the client does not
3982 close either, then the connection remains open till the timeout expires. This
3983 causes high number of simultaneous connections on the servers and shows high
3984 global session times in the logs.
3985
3986 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01003987 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01003988 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003989
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02003990 This option may also be combined with "option http-pretend-keepalive", which
3991 will disable sending of the "Connection: close" header, but will still cause
3992 the connection to be closed once the whole response is received.
3993
Cyril Bonté653dcd62014-02-20 00:13:15 +01003994 This option disables and replaces any previous "option httpclose", "option
3995 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01003996
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003997 If this option has been enabled in a "defaults" section, it can be disabled
3998 in a specific instance by prepending the "no" keyword before it.
3999
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004000 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004001
4002
Willy Tarreau87cf5142011-08-19 22:57:24 +02004003option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004004 Enable insertion of the X-Forwarded-For header to requests sent to servers
4005 May be used in sections : defaults | frontend | listen | backend
4006 yes | yes | yes | yes
4007 Arguments :
4008 <network> is an optional argument used to disable this option for sources
4009 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02004010 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01004011 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004012
4013 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
4014 their client address. This is sometimes annoying when the client's IP address
4015 is expected in server logs. To solve this problem, the well-known HTTP header
4016 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
4017 This header contains a value representing the client's IP address. Since this
4018 header is always appended at the end of the existing header list, the server
4019 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02004020 the server's manual to find how to enable use of this standard header. Note
4021 that only the last occurrence of the header must be used, since it is really
4022 possible that the client has already brought one.
4023
Willy Tarreaud72758d2010-01-12 10:42:19 +01004024 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02004025 the default "X-Forwarded-For". This can be useful where you might already
Willy Tarreaud72758d2010-01-12 10:42:19 +01004026 have a "X-Forwarded-For" header from a different application (eg: stunnel),
4027 and you need preserve it. Also if your backend server doesn't use the
Ross Westaf72a1d2008-08-03 10:51:45 +02004028 "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers
4029 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01004030
4031 Sometimes, a same HAProxy instance may be shared between a direct client
4032 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4033 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4034 header for a known source address or network by adding the "except" keyword
4035 followed by the network address. In this case, any source IP matching the
4036 network will not cause an addition of this header. Most common uses are with
4037 private networks or 127.0.0.1.
4038
Willy Tarreau87cf5142011-08-19 22:57:24 +02004039 Alternatively, the keyword "if-none" states that the header will only be
4040 added if it is not present. This should only be used in perfectly trusted
4041 environment, as this might cause a security issue if headers reaching haproxy
4042 are under the control of the end-user.
4043
Willy Tarreauc27debf2008-01-06 08:57:02 +01004044 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02004045 least one of them uses it, the header will be added. Note that the backend's
4046 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02004047 both are defined. In the case of the "if-none" argument, if at least one of
4048 the frontend or the backend does not specify it, it wants the addition to be
4049 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004050
Ross Westaf72a1d2008-08-03 10:51:45 +02004051 Examples :
Willy Tarreauc27debf2008-01-06 08:57:02 +01004052 # Public HTTP address also used by stunnel on the same machine
4053 frontend www
4054 mode http
4055 option forwardfor except 127.0.0.1 # stunnel already adds the header
4056
Ross Westaf72a1d2008-08-03 10:51:45 +02004057 # Those servers want the IP Address in X-Client
4058 backend www
4059 mode http
4060 option forwardfor header X-Client
4061
Willy Tarreau87cf5142011-08-19 22:57:24 +02004062 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01004063 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01004064
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004065
Willy Tarreau16bfb022010-01-16 19:48:41 +01004066option http-keep-alive
4067no option http-keep-alive
4068 Enable or disable HTTP keep-alive from client to server
4069 May be used in sections : defaults | frontend | listen | backend
4070 yes | yes | yes | yes
4071 Arguments : none
4072
Willy Tarreau70dffda2014-01-30 03:07:23 +01004073 By default HAProxy operates in keep-alive mode with regards to persistent
4074 connections: for each connection it processes each request and response, and
4075 leaves the connection idle on both sides between the end of a response and the
4076 start of a new request. This mode may be changed by several options such as
4077 "option http-server-close", "option forceclose", "option httpclose" or
4078 "option http-tunnel". This option allows to set back the keep-alive mode,
4079 which can be useful when another mode was used in a defaults section.
4080
4081 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
4082 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01004083 network) and the fastest session reuse on the server side at the expense
4084 of maintaining idle connections to the servers. In general, it is possible
4085 with this option to achieve approximately twice the request rate that the
4086 "http-server-close" option achieves on small objects. There are mainly two
4087 situations where this option may be useful :
4088
4089 - when the server is non-HTTP compliant and authenticates the connection
4090 instead of requests (eg: NTLM authentication)
4091
4092 - when the cost of establishing the connection to the server is significant
4093 compared to the cost of retrieving the associated object from the server.
4094
4095 This last case can happen when the server is a fast static server of cache.
4096 In this case, the server will need to be properly tuned to support high enough
4097 connection counts because connections will last until the client sends another
4098 request.
4099
4100 If the client request has to go to another backend or another server due to
4101 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01004102 immediately be closed and a new one re-opened. Option "prefer-last-server" is
4103 available to try optimize server selection so that if the server currently
4104 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01004105
4106 In general it is preferred to use "option http-server-close" with application
4107 servers, and some static servers might benefit from "option http-keep-alive".
4108
4109 At the moment, logs will not indicate whether requests came from the same
4110 session or not. The accept date reported in the logs corresponds to the end
4111 of the previous request, and the request time corresponds to the time spent
4112 waiting for a new request. The keep-alive request time is still bound to the
4113 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4114 not set.
4115
Cyril Bonté653dcd62014-02-20 00:13:15 +01004116 This option disables and replaces any previous "option httpclose", "option
4117 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01004118 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01004119 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004120
4121 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01004122 "option prefer-last-server", "option http-pretend-keepalive",
4123 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004124
4125
Willy Tarreau96e31212011-05-30 18:10:30 +02004126option http-no-delay
4127no option http-no-delay
4128 Instruct the system to favor low interactive delays over performance in HTTP
4129 May be used in sections : defaults | frontend | listen | backend
4130 yes | yes | yes | yes
4131 Arguments : none
4132
4133 In HTTP, each payload is unidirectional and has no notion of interactivity.
4134 Any agent is expected to queue data somewhat for a reasonably low delay.
4135 There are some very rare server-to-server applications that abuse the HTTP
4136 protocol and expect the payload phase to be highly interactive, with many
4137 interleaved data chunks in both directions within a single request. This is
4138 absolutely not supported by the HTTP specification and will not work across
4139 most proxies or servers. When such applications attempt to do this through
4140 haproxy, it works but they will experience high delays due to the network
4141 optimizations which favor performance by instructing the system to wait for
4142 enough data to be available in order to only send full packets. Typical
4143 delays are around 200 ms per round trip. Note that this only happens with
4144 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
4145 affected.
4146
4147 When "option http-no-delay" is present in either the frontend or the backend
4148 used by a connection, all such optimizations will be disabled in order to
4149 make the exchanges as fast as possible. Of course this offers no guarantee on
4150 the functionality, as it may break at any other place. But if it works via
4151 HAProxy, it will work as fast as possible. This option should never be used
4152 by default, and should never be used at all unless such a buggy application
4153 is discovered. The impact of using this option is an increase of bandwidth
4154 usage and CPU usage, which may significantly lower performance in high
4155 latency environments.
4156
4157
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004158option http-pretend-keepalive
4159no option http-pretend-keepalive
4160 Define whether haproxy will announce keepalive to the server or not
4161 May be used in sections : defaults | frontend | listen | backend
4162 yes | yes | yes | yes
4163 Arguments : none
4164
4165 When running with "option http-server-close" or "option forceclose", haproxy
4166 adds a "Connection: close" header to the request forwarded to the server.
4167 Unfortunately, when some servers see this header, they automatically refrain
4168 from using the chunked encoding for responses of unknown length, while this
4169 is totally unrelated. The immediate effect is that this prevents haproxy from
4170 maintaining the client connection alive. A second effect is that a client or
4171 a cache could receive an incomplete response without being aware of it, and
4172 consider the response complete.
4173
4174 By setting "option http-pretend-keepalive", haproxy will make the server
4175 believe it will keep the connection alive. The server will then not fall back
4176 to the abnormal undesired above. When haproxy gets the whole response, it
4177 will close the connection with the server just as it would do with the
4178 "forceclose" option. That way the client gets a normal response and the
4179 connection is correctly closed on the server side.
4180
4181 It is recommended not to enable this option by default, because most servers
4182 will more efficiently close the connection themselves after the last packet,
4183 and release its buffers slightly earlier. Also, the added packet on the
4184 network could slightly reduce the overall peak performance. However it is
4185 worth noting that when this option is enabled, haproxy will have slightly
4186 less work to do. So if haproxy is the bottleneck on the whole architecture,
4187 enabling this option might save a few CPU cycles.
4188
4189 This option may be set both in a frontend and in a backend. It is enabled if
4190 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004191 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02004192 keepalive to be announced to the server and close to be announced to the
4193 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004194
4195 If this option has been enabled in a "defaults" section, it can be disabled
4196 in a specific instance by prepending the "no" keyword before it.
4197
Willy Tarreau16bfb022010-01-16 19:48:41 +01004198 See also : "option forceclose", "option http-server-close", and
4199 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004200
Willy Tarreauc27debf2008-01-06 08:57:02 +01004201
Willy Tarreaub608feb2010-01-02 22:47:18 +01004202option http-server-close
4203no option http-server-close
4204 Enable or disable HTTP connection closing on the server side
4205 May be used in sections : defaults | frontend | listen | backend
4206 yes | yes | yes | yes
4207 Arguments : none
4208
Willy Tarreau70dffda2014-01-30 03:07:23 +01004209 By default HAProxy operates in keep-alive mode with regards to persistent
4210 connections: for each connection it processes each request and response, and
4211 leaves the connection idle on both sides between the end of a response and
4212 the start of a new request. This mode may be changed by several options such
4213 as "option http-server-close", "option forceclose", "option httpclose" or
4214 "option http-tunnel". Setting "option http-server-close" enables HTTP
4215 connection-close mode on the server side while keeping the ability to support
4216 HTTP keep-alive and pipelining on the client side. This provides the lowest
4217 latency on the client side (slow network) and the fastest session reuse on
4218 the server side to save server resources, similarly to "option forceclose".
4219 It also permits non-keepalive capable servers to be served in keep-alive mode
4220 to the clients if they conform to the requirements of RFC2616. Please note
4221 that some servers do not always conform to those requirements when they see
4222 "Connection: close" in the request. The effect will be that keep-alive will
4223 never be used. A workaround consists in enabling "option
4224 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004225
4226 At the moment, logs will not indicate whether requests came from the same
4227 session or not. The accept date reported in the logs corresponds to the end
4228 of the previous request, and the request time corresponds to the time spent
4229 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01004230 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4231 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004232
4233 This option may be set both in a frontend and in a backend. It is enabled if
4234 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004235 It disables and replaces any previous "option httpclose", "option forceclose",
4236 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01004237 ("Proxies") to see how this option combines with others when frontend and
4238 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004239
4240 If this option has been enabled in a "defaults" section, it can be disabled
4241 in a specific instance by prepending the "no" keyword before it.
4242
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004243 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01004244 "option httpclose", "option http-keep-alive", and
4245 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004246
4247
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004248option http-tunnel
4249no option http-tunnel
4250 Disable or enable HTTP connection processing after first transaction
4251 May be used in sections : defaults | frontend | listen | backend
4252 yes | yes | yes | yes
4253 Arguments : none
4254
Willy Tarreau70dffda2014-01-30 03:07:23 +01004255 By default HAProxy operates in keep-alive mode with regards to persistent
4256 connections: for each connection it processes each request and response, and
4257 leaves the connection idle on both sides between the end of a response and
4258 the start of a new request. This mode may be changed by several options such
4259 as "option http-server-close", "option forceclose", "option httpclose" or
4260 "option http-tunnel".
4261
4262 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004263 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01004264 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
4265 is normally not needed anymore unless in very specific cases such as when
4266 using an in-house protocol that looks like HTTP but is not compatible, or
4267 just to log one request per client in order to reduce log size. Note that
4268 everything which works at the HTTP level, including header parsing/addition,
4269 cookie processing or content switching will only work for the first request
4270 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004271
4272 If this option has been enabled in a "defaults" section, it can be disabled
4273 in a specific instance by prepending the "no" keyword before it.
4274
4275 See also : "option forceclose", "option http-server-close",
4276 "option httpclose", "option http-keep-alive", and
4277 "1.1. The HTTP transaction model".
4278
4279
Willy Tarreau88d349d2010-01-25 12:15:43 +01004280option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01004281no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01004282 Make use of non-standard Proxy-Connection header instead of Connection
4283 May be used in sections : defaults | frontend | listen | backend
4284 yes | yes | yes | no
4285 Arguments : none
4286
4287 While RFC2616 explicitly states that HTTP/1.1 agents must use the
4288 Connection header to indicate their wish of persistent or non-persistent
4289 connections, both browsers and proxies ignore this header for proxied
4290 connections and make use of the undocumented, non-standard Proxy-Connection
4291 header instead. The issue begins when trying to put a load balancer between
4292 browsers and such proxies, because there will be a difference between what
4293 haproxy understands and what the client and the proxy agree on.
4294
4295 By setting this option in a frontend, haproxy can automatically switch to use
4296 that non-standard header if it sees proxied requests. A proxied request is
4297 defined here as one where the URI begins with neither a '/' nor a '*'. The
4298 choice of header only affects requests passing through proxies making use of
4299 one of the "httpclose", "forceclose" and "http-server-close" options. Note
4300 that this option can only be specified in a frontend and will affect the
4301 request along its whole life.
4302
Willy Tarreau844a7e72010-01-31 21:46:18 +01004303 Also, when this option is set, a request which requires authentication will
4304 automatically switch to use proxy authentication headers if it is itself a
4305 proxied request. That makes it possible to check or enforce authentication in
4306 front of an existing proxy.
4307
Willy Tarreau88d349d2010-01-25 12:15:43 +01004308 This option should normally never be used, except in front of a proxy.
4309
4310 See also : "option httpclose", "option forceclose" and "option
4311 http-server-close".
4312
4313
Willy Tarreaud63335a2010-02-26 12:56:52 +01004314option httpchk
4315option httpchk <uri>
4316option httpchk <method> <uri>
4317option httpchk <method> <uri> <version>
4318 Enable HTTP protocol to check on the servers health
4319 May be used in sections : defaults | frontend | listen | backend
4320 yes | no | yes | yes
4321 Arguments :
4322 <method> is the optional HTTP method used with the requests. When not set,
4323 the "OPTIONS" method is used, as it generally requires low server
4324 processing and is easy to filter out from the logs. Any method
4325 may be used, though it is not recommended to invent non-standard
4326 ones.
4327
4328 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
4329 which is accessible by default on almost any server, but may be
4330 changed to any other URI. Query strings are permitted.
4331
4332 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
4333 but some servers might behave incorrectly in HTTP 1.0, so turning
4334 it to HTTP/1.1 may sometimes help. Note that the Host field is
4335 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
4336 after "\r\n" following the version string.
4337
4338 By default, server health checks only consist in trying to establish a TCP
4339 connection. When "option httpchk" is specified, a complete HTTP request is
4340 sent once the TCP connection is established, and responses 2xx and 3xx are
4341 considered valid, while all other ones indicate a server failure, including
4342 the lack of any response.
4343
4344 The port and interval are specified in the server configuration.
4345
4346 This option does not necessarily require an HTTP backend, it also works with
4347 plain TCP backends. This is particularly useful to check simple scripts bound
4348 to some dedicated ports using the inetd daemon.
4349
4350 Examples :
4351 # Relay HTTPS traffic to Apache instance and check service availability
4352 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
4353 backend https_relay
4354 mode tcp
4355 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
4356 server apache1 192.168.1.1:443 check port 80
4357
Simon Hormanafc47ee2013-11-25 10:46:35 +09004358 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
4359 "option pgsql-check", "http-check" and the "check", "port" and
4360 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01004361
4362
Willy Tarreauc27debf2008-01-06 08:57:02 +01004363option httpclose
4364no option httpclose
4365 Enable or disable passive HTTP connection closing
4366 May be used in sections : defaults | frontend | listen | backend
4367 yes | yes | yes | yes
4368 Arguments : none
4369
Willy Tarreau70dffda2014-01-30 03:07:23 +01004370 By default HAProxy operates in keep-alive mode with regards to persistent
4371 connections: for each connection it processes each request and response, and
4372 leaves the connection idle on both sides between the end of a response and
4373 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01004374 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01004375 "option http-tunnel".
4376
4377 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
4378 if a "Connection: close" header is already set in each direction, and will
4379 add one if missing. Each end should react to this by actively closing the TCP
4380 connection after each transfer, thus resulting in a switch to the HTTP close
4381 mode. Any "Connection" header different from "close" will also be removed.
4382 Note that this option is deprecated since what it does is very cheap but not
4383 reliable. Using "option http-server-close" or "option forceclose" is strongly
4384 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004385
4386 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004387 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01004388 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
4389 it is possible to use the "option forceclose" which actively closes the
4390 request connection once the server responds. Option "forceclose" also
4391 releases the server connection earlier because it does not have to wait for
4392 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004393
4394 This option may be set both in a frontend and in a backend. It is enabled if
4395 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004396 It disables and replaces any previous "option http-server-close",
4397 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01004398 check section 4 ("Proxies") to see how this option combines with others when
4399 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004400
4401 If this option has been enabled in a "defaults" section, it can be disabled
4402 in a specific instance by prepending the "no" keyword before it.
4403
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004404 See also : "option forceclose", "option http-server-close" and
4405 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01004406
4407
Emeric Brun3a058f32009-06-30 18:26:00 +02004408option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004409 Enable logging of HTTP request, session state and timers
4410 May be used in sections : defaults | frontend | listen | backend
4411 yes | yes | yes | yes
Emeric Brun3a058f32009-06-30 18:26:00 +02004412 Arguments :
4413 clf if the "clf" argument is added, then the output format will be
4414 the CLF format instead of HAProxy's default HTTP format. You can
4415 use this when you need to feed HAProxy's logs through a specific
4416 log analyser which only support the CLF format and which is not
4417 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004418
4419 By default, the log output format is very poor, as it only contains the
4420 source and destination addresses, and the instance name. By specifying
4421 "option httplog", each log line turns into a much richer format including,
4422 but not limited to, the HTTP request, the connection timers, the session
4423 status, the connections numbers, the captured headers and cookies, the
4424 frontend, backend and server name, and of course the source address and
4425 ports.
4426
4427 This option may be set either in the frontend or the backend.
4428
Emeric Brun3a058f32009-06-30 18:26:00 +02004429 If this option has been enabled in a "defaults" section, it can be disabled
4430 in a specific instance by prepending the "no" keyword before it. Specifying
4431 only "option httplog" will automatically clear the 'clf' mode if it was set
4432 by default.
4433
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004434 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004435
Willy Tarreau55165fe2009-05-10 12:02:55 +02004436
4437option http_proxy
4438no option http_proxy
4439 Enable or disable plain HTTP proxy mode
4440 May be used in sections : defaults | frontend | listen | backend
4441 yes | yes | yes | yes
4442 Arguments : none
4443
4444 It sometimes happens that people need a pure HTTP proxy which understands
4445 basic proxy requests without caching nor any fancy feature. In this case,
4446 it may be worth setting up an HAProxy instance with the "option http_proxy"
4447 set. In this mode, no server is declared, and the connection is forwarded to
4448 the IP address and port found in the URL after the "http://" scheme.
4449
4450 No host address resolution is performed, so this only works when pure IP
4451 addresses are passed. Since this option's usage perimeter is rather limited,
4452 it will probably be used only by experts who know they need exactly it. Last,
4453 if the clients are susceptible of sending keep-alive requests, it will be
Cyril Bonté2409e682010-12-14 22:47:51 +01004454 needed to add "option httpclose" to ensure that all requests will correctly
Willy Tarreau55165fe2009-05-10 12:02:55 +02004455 be analyzed.
4456
4457 If this option has been enabled in a "defaults" section, it can be disabled
4458 in a specific instance by prepending the "no" keyword before it.
4459
4460 Example :
4461 # this backend understands HTTP proxy requests and forwards them directly.
4462 backend direct_forward
4463 option httpclose
4464 option http_proxy
4465
4466 See also : "option httpclose"
4467
Willy Tarreau211ad242009-10-03 21:45:07 +02004468
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004469option independent-streams
4470no option independent-streams
4471 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004472 May be used in sections : defaults | frontend | listen | backend
4473 yes | yes | yes | yes
4474 Arguments : none
4475
4476 By default, when data is sent over a socket, both the write timeout and the
4477 read timeout for that socket are refreshed, because we consider that there is
4478 activity on that socket, and we have no other means of guessing if we should
4479 receive data or not.
4480
4481 While this default behaviour is desirable for almost all applications, there
4482 exists a situation where it is desirable to disable it, and only refresh the
4483 read timeout if there are incoming data. This happens on sessions with large
4484 timeouts and low amounts of exchanged data such as telnet session. If the
4485 server suddenly disappears, the output data accumulates in the system's
4486 socket buffers, both timeouts are correctly refreshed, and there is no way
4487 to know the server does not receive them, so we don't timeout. However, when
4488 the underlying protocol always echoes sent data, it would be enough by itself
4489 to detect the issue using the read timeout. Note that this problem does not
4490 happen with more verbose protocols because data won't accumulate long in the
4491 socket buffers.
4492
4493 When this option is set on the frontend, it will disable read timeout updates
4494 on data sent to the client. There probably is little use of this case. When
4495 the option is set on the backend, it will disable read timeout updates on
4496 data sent to the server. Doing so will typically break large HTTP posts from
4497 slow lines, so use it with caution.
4498
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004499 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004500 with a spelling mistake. This spelling is still supported but
4501 deprecated.
4502
Willy Tarreauce887fd2012-05-12 12:50:00 +02004503 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004504
4505
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02004506option ldap-check
4507 Use LDAPv3 health checks for server testing
4508 May be used in sections : defaults | frontend | listen | backend
4509 yes | no | yes | yes
4510 Arguments : none
4511
4512 It is possible to test that the server correctly talks LDAPv3 instead of just
4513 testing that it accepts the TCP connection. When this option is set, an
4514 LDAPv3 anonymous simple bind message is sent to the server, and the response
4515 is analyzed to find an LDAPv3 bind response message.
4516
4517 The server is considered valid only when the LDAP response contains success
4518 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
4519
4520 Logging of bind requests is server dependent see your documentation how to
4521 configure it.
4522
4523 Example :
4524 option ldap-check
4525
4526 See also : "option httpchk"
4527
4528
Willy Tarreau211ad242009-10-03 21:45:07 +02004529option log-health-checks
4530no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02004531 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02004532 May be used in sections : defaults | frontend | listen | backend
4533 yes | no | yes | yes
4534 Arguments : none
4535
Willy Tarreaubef1b322014-05-13 21:01:39 +02004536 By default, failed health check are logged if server is UP and successful
4537 health checks are logged if server is DOWN, so the amount of additional
4538 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02004539
Willy Tarreaubef1b322014-05-13 21:01:39 +02004540 When this option is enabled, any change of the health check status or to
4541 the server's health will be logged, so that it becomes possible to know
4542 that a server was failing occasional checks before crashing, or exactly when
4543 it failed to respond a valid HTTP status, then when the port started to
4544 reject connections, then when the server stopped responding at all.
4545
4546 Note that status changes not caused by health checks (eg: enable/disable on
4547 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02004548
Willy Tarreaubef1b322014-05-13 21:01:39 +02004549 See also: "option httpchk", "option ldap-check", "option mysql-check",
4550 "option pgsql-check", "option redis-check", "option smtpchk",
4551 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02004552
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004553
4554option log-separate-errors
4555no option log-separate-errors
4556 Change log level for non-completely successful connections
4557 May be used in sections : defaults | frontend | listen | backend
4558 yes | yes | yes | no
4559 Arguments : none
4560
4561 Sometimes looking for errors in logs is not easy. This option makes haproxy
4562 raise the level of logs containing potentially interesting information such
4563 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
4564 level changes from "info" to "err". This makes it possible to log them
4565 separately to a different file with most syslog daemons. Be careful not to
4566 remove them from the original file, otherwise you would lose ordering which
4567 provides very important information.
4568
4569 Using this option, large sites dealing with several thousand connections per
4570 second may log normal traffic to a rotating buffer and only archive smaller
4571 error logs.
4572
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004573 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004574 logging.
4575
Willy Tarreauc27debf2008-01-06 08:57:02 +01004576
4577option logasap
4578no option logasap
4579 Enable or disable early logging of HTTP requests
4580 May be used in sections : defaults | frontend | listen | backend
4581 yes | yes | yes | no
4582 Arguments : none
4583
4584 By default, HTTP requests are logged upon termination so that the total
4585 transfer time and the number of bytes appear in the logs. When large objects
4586 are being transferred, it may take a while before the request appears in the
4587 logs. Using "option logasap", the request gets logged as soon as the server
4588 sends the complete headers. The only missing information in the logs will be
4589 the total number of bytes which will indicate everything except the amount
4590 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004591 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01004592 "Content-Length" response header so that the logs at least indicate how many
4593 bytes are expected to be transferred.
4594
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004595 Examples :
4596 listen http_proxy 0.0.0.0:80
4597 mode http
4598 option httplog
4599 option logasap
4600 log 192.168.2.200 local3
4601
4602 >>> Feb 6 12:14:14 localhost \
4603 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
4604 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
4605 "GET /image.iso HTTP/1.0"
4606
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004607 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01004608 logging.
4609
4610
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004611option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004612 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004613 May be used in sections : defaults | frontend | listen | backend
4614 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004615 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004616 <username> This is the username which will be used when connecting to MySQL
4617 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004618 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004619
4620 If you specify a username, the check consists of sending two MySQL packet,
4621 one Client Authentication packet, and one QUIT packet, to correctly close
4622 MySQL session. We then parse the MySQL Handshake Initialisation packet and/or
4623 Error packet. It is a basic but useful test which does not produce error nor
4624 aborted connect on the server. However, it requires adding an authorization
4625 in the MySQL table, like this :
4626
4627 USE mysql;
4628 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
4629 FLUSH PRIVILEGES;
4630
4631 If you don't specify a username (it is deprecated and not recommended), the
4632 check only consists in parsing the Mysql Handshake Initialisation packet or
4633 Error packet, we don't send anything in this mode. It was reported that it
4634 can generate lockout if check is too frequent and/or if there is not enough
4635 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
4636 value as if a connection is established successfully within fewer than MySQL
4637 "max_connect_errors" attempts after a previous connection was interrupted,
4638 the error count for the host is cleared to zero. If HAProxy's server get
4639 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
4640
4641 Remember that this does not check database presence nor database consistency.
4642 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004643
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02004644 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004645
4646 Most often, an incoming MySQL server needs to see the client's IP address for
4647 various purposes, including IP privilege matching and connection logging.
4648 When possible, it is often wise to masquerade the client's IP address when
4649 connecting to the server using the "usesrc" argument of the "source" keyword,
4650 which requires the cttproxy feature to be compiled in, and the MySQL server
4651 to route the client via the machine hosting haproxy.
4652
4653 See also: "option httpchk"
4654
4655
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004656option nolinger
4657no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004658 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004659 May be used in sections: defaults | frontend | listen | backend
4660 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004661 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004662
4663 When clients or servers abort connections in a dirty way (eg: they are
4664 physically disconnected), the session timeouts triggers and the session is
4665 closed. But it will remain in FIN_WAIT1 state for some time in the system,
4666 using some resources and possibly limiting the ability to establish newer
4667 connections.
4668
4669 When this happens, it is possible to activate "option nolinger" which forces
4670 the system to immediately remove any socket's pending data on close. Thus,
4671 the session is instantly purged from the system's tables. This usually has
4672 side effects such as increased number of TCP resets due to old retransmits
4673 getting immediately rejected. Some firewalls may sometimes complain about
4674 this too.
4675
4676 For this reason, it is not recommended to use this option when not absolutely
4677 needed. You know that you need it when you have thousands of FIN_WAIT1
4678 sessions on your system (TIME_WAIT ones do not count).
4679
4680 This option may be used both on frontends and backends, depending on the side
4681 where it is required. Use it on the frontend for clients, and on the backend
4682 for servers.
4683
4684 If this option has been enabled in a "defaults" section, it can be disabled
4685 in a specific instance by prepending the "no" keyword before it.
4686
4687
Willy Tarreau55165fe2009-05-10 12:02:55 +02004688option originalto [ except <network> ] [ header <name> ]
4689 Enable insertion of the X-Original-To header to requests sent to servers
4690 May be used in sections : defaults | frontend | listen | backend
4691 yes | yes | yes | yes
4692 Arguments :
4693 <network> is an optional argument used to disable this option for sources
4694 matching <network>
4695 <name> an optional argument to specify a different "X-Original-To"
4696 header name.
4697
4698 Since HAProxy can work in transparent mode, every request from a client can
4699 be redirected to the proxy and HAProxy itself can proxy every request to a
4700 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
4701 be lost. This is annoying when you want access rules based on destination ip
4702 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
4703 added by HAProxy to all requests sent to the server. This header contains a
4704 value representing the original destination IP address. Since this must be
4705 configured to always use the last occurrence of this header only. Note that
4706 only the last occurrence of the header must be used, since it is really
4707 possible that the client has already brought one.
4708
4709 The keyword "header" may be used to supply a different header name to replace
4710 the default "X-Original-To". This can be useful where you might already
4711 have a "X-Original-To" header from a different application, and you need
4712 preserve it. Also if your backend server doesn't use the "X-Original-To"
4713 header and requires different one.
4714
4715 Sometimes, a same HAProxy instance may be shared between a direct client
4716 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4717 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4718 header for a known source address or network by adding the "except" keyword
4719 followed by the network address. In this case, any source IP matching the
4720 network will not cause an addition of this header. Most common uses are with
4721 private networks or 127.0.0.1.
4722
4723 This option may be specified either in the frontend or in the backend. If at
4724 least one of them uses it, the header will be added. Note that the backend's
4725 setting of the header subargument takes precedence over the frontend's if
4726 both are defined.
4727
Willy Tarreau55165fe2009-05-10 12:02:55 +02004728 Examples :
4729 # Original Destination address
4730 frontend www
4731 mode http
4732 option originalto except 127.0.0.1
4733
4734 # Those servers want the IP Address in X-Client-Dst
4735 backend www
4736 mode http
4737 option originalto header X-Client-Dst
4738
Willy Tarreau87cf5142011-08-19 22:57:24 +02004739 See also : "option httpclose", "option http-server-close",
4740 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02004741
4742
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004743option persist
4744no option persist
4745 Enable or disable forced persistence on down servers
4746 May be used in sections: defaults | frontend | listen | backend
4747 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004748 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004749
4750 When an HTTP request reaches a backend with a cookie which references a dead
4751 server, by default it is redispatched to another server. It is possible to
4752 force the request to be sent to the dead server first using "option persist"
4753 if absolutely needed. A common use case is when servers are under extreme
4754 load and spend their time flapping. In this case, the users would still be
4755 directed to the server they opened the session on, in the hope they would be
4756 correctly served. It is recommended to use "option redispatch" in conjunction
4757 with this option so that in the event it would not be possible to connect to
4758 the server at all (server definitely dead), the client would finally be
4759 redirected to another valid server.
4760
4761 If this option has been enabled in a "defaults" section, it can be disabled
4762 in a specific instance by prepending the "no" keyword before it.
4763
Willy Tarreau4de91492010-01-22 19:10:05 +01004764 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004765
4766
Willy Tarreau0c122822013-12-15 18:49:01 +01004767option pgsql-check [ user <username> ]
4768 Use PostgreSQL health checks for server testing
4769 May be used in sections : defaults | frontend | listen | backend
4770 yes | no | yes | yes
4771 Arguments :
4772 <username> This is the username which will be used when connecting to
4773 PostgreSQL server.
4774
4775 The check sends a PostgreSQL StartupMessage and waits for either
4776 Authentication request or ErrorResponse message. It is a basic but useful
4777 test which does not produce error nor aborted connect on the server.
4778 This check is identical with the "mysql-check".
4779
4780 See also: "option httpchk"
4781
4782
Willy Tarreau9420b122013-12-15 18:58:25 +01004783option prefer-last-server
4784no option prefer-last-server
4785 Allow multiple load balanced requests to remain on the same server
4786 May be used in sections: defaults | frontend | listen | backend
4787 yes | no | yes | yes
4788 Arguments : none
4789
4790 When the load balancing algorithm in use is not deterministic, and a previous
4791 request was sent to a server to which haproxy still holds a connection, it is
4792 sometimes desirable that subsequent requests on a same session go to the same
4793 server as much as possible. Note that this is different from persistence, as
4794 we only indicate a preference which haproxy tries to apply without any form
4795 of warranty. The real use is for keep-alive connections sent to servers. When
4796 this option is used, haproxy will try to reuse the same connection that is
4797 attached to the server instead of rebalancing to another server, causing a
4798 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01004799 not make much sense to use this in combination with hashing algorithms. Note,
4800 haproxy already automatically tries to stick to a server which sends a 401 or
4801 to a proxy which sends a 407 (authentication required). This is mandatory for
4802 use with the broken NTLM authentication challenge, and significantly helps in
4803 troubleshooting some faulty applications. Option prefer-last-server might be
4804 desirable in these environments as well, to avoid redistributing the traffic
4805 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01004806
4807 If this option has been enabled in a "defaults" section, it can be disabled
4808 in a specific instance by prepending the "no" keyword before it.
4809
4810 See also: "option http-keep-alive"
4811
4812
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004813option redispatch
4814no option redispatch
4815 Enable or disable session redistribution in case of connection failure
4816 May be used in sections: defaults | frontend | listen | backend
4817 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004818 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004819
4820 In HTTP mode, if a server designated by a cookie is down, clients may
4821 definitely stick to it because they cannot flush the cookie, so they will not
4822 be able to access the service anymore.
4823
4824 Specifying "option redispatch" will allow the proxy to break their
4825 persistence and redistribute them to a working server.
4826
4827 It also allows to retry last connection to another server in case of multiple
4828 connection failures. Of course, it requires having "retries" set to a nonzero
4829 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01004830
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004831 This form is the preferred form, which replaces both the "redispatch" and
4832 "redisp" keywords.
4833
4834 If this option has been enabled in a "defaults" section, it can be disabled
4835 in a specific instance by prepending the "no" keyword before it.
4836
Willy Tarreau4de91492010-01-22 19:10:05 +01004837 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004838
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004839
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02004840option redis-check
4841 Use redis health checks for server testing
4842 May be used in sections : defaults | frontend | listen | backend
4843 yes | no | yes | yes
4844 Arguments : none
4845
4846 It is possible to test that the server correctly talks REDIS protocol instead
4847 of just testing that it accepts the TCP connection. When this option is set,
4848 a PING redis command is sent to the server, and the response is analyzed to
4849 find the "+PONG" response message.
4850
4851 Example :
4852 option redis-check
4853
4854 See also : "option httpchk"
4855
4856
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004857option smtpchk
4858option smtpchk <hello> <domain>
4859 Use SMTP health checks for server testing
4860 May be used in sections : defaults | frontend | listen | backend
4861 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01004862 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004863 <hello> is an optional argument. It is the "hello" command to use. It can
4864 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
4865 values will be turned into the default command ("HELO").
4866
4867 <domain> is the domain name to present to the server. It may only be
4868 specified (and is mandatory) if the hello command has been
4869 specified. By default, "localhost" is used.
4870
4871 When "option smtpchk" is set, the health checks will consist in TCP
4872 connections followed by an SMTP command. By default, this command is
4873 "HELO localhost". The server's return code is analyzed and only return codes
4874 starting with a "2" will be considered as valid. All other responses,
4875 including a lack of response will constitute an error and will indicate a
4876 dead server.
4877
4878 This test is meant to be used with SMTP servers or relays. Depending on the
4879 request, it is possible that some servers do not log each connection attempt,
4880 so you may want to experiment to improve the behaviour. Using telnet on port
4881 25 is often easier than adjusting the configuration.
4882
4883 Most often, an incoming SMTP server needs to see the client's IP address for
4884 various purposes, including spam filtering, anti-spoofing and logging. When
4885 possible, it is often wise to masquerade the client's IP address when
4886 connecting to the server using the "usesrc" argument of the "source" keyword,
4887 which requires the cttproxy feature to be compiled in.
4888
4889 Example :
4890 option smtpchk HELO mydomain.org
4891
4892 See also : "option httpchk", "source"
4893
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004894
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02004895option socket-stats
4896no option socket-stats
4897
4898 Enable or disable collecting & providing separate statistics for each socket.
4899 May be used in sections : defaults | frontend | listen | backend
4900 yes | yes | yes | no
4901
4902 Arguments : none
4903
4904
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004905option splice-auto
4906no option splice-auto
4907 Enable or disable automatic kernel acceleration on sockets in both directions
4908 May be used in sections : defaults | frontend | listen | backend
4909 yes | yes | yes | yes
4910 Arguments : none
4911
4912 When this option is enabled either on a frontend or on a backend, haproxy
4913 will automatically evaluate the opportunity to use kernel tcp splicing to
4914 forward data between the client and the server, in either direction. Haproxy
4915 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004916 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004917 are not much aggressive in order to limit excessive use of splicing. This
4918 option requires splicing to be enabled at compile time, and may be globally
4919 disabled with the global option "nosplice". Since splice uses pipes, using it
4920 requires that there are enough spare pipes.
4921
4922 Important note: kernel-based TCP splicing is a Linux-specific feature which
4923 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
4924 transfer data between sockets without copying these data to user-space, thus
4925 providing noticeable performance gains and CPU cycles savings. Since many
4926 early implementations are buggy, corrupt data and/or are inefficient, this
4927 feature is not enabled by default, and it should be used with extreme care.
4928 While it is not possible to detect the correctness of an implementation,
4929 2.6.29 is the first version offering a properly working implementation. In
4930 case of doubt, splicing may be globally disabled using the global "nosplice"
4931 keyword.
4932
4933 Example :
4934 option splice-auto
4935
4936 If this option has been enabled in a "defaults" section, it can be disabled
4937 in a specific instance by prepending the "no" keyword before it.
4938
4939 See also : "option splice-request", "option splice-response", and global
4940 options "nosplice" and "maxpipes"
4941
4942
4943option splice-request
4944no option splice-request
4945 Enable or disable automatic kernel acceleration on sockets for requests
4946 May be used in sections : defaults | frontend | listen | backend
4947 yes | yes | yes | yes
4948 Arguments : none
4949
4950 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004951 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004952 the client to the server. It might still use the recv/send scheme if there
4953 are no spare pipes left. This option requires splicing to be enabled at
4954 compile time, and may be globally disabled with the global option "nosplice".
4955 Since splice uses pipes, using it requires that there are enough spare pipes.
4956
4957 Important note: see "option splice-auto" for usage limitations.
4958
4959 Example :
4960 option splice-request
4961
4962 If this option has been enabled in a "defaults" section, it can be disabled
4963 in a specific instance by prepending the "no" keyword before it.
4964
4965 See also : "option splice-auto", "option splice-response", and global options
4966 "nosplice" and "maxpipes"
4967
4968
4969option splice-response
4970no option splice-response
4971 Enable or disable automatic kernel acceleration on sockets for responses
4972 May be used in sections : defaults | frontend | listen | backend
4973 yes | yes | yes | yes
4974 Arguments : none
4975
4976 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004977 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004978 the server to the client. It might still use the recv/send scheme if there
4979 are no spare pipes left. This option requires splicing to be enabled at
4980 compile time, and may be globally disabled with the global option "nosplice".
4981 Since splice uses pipes, using it requires that there are enough spare pipes.
4982
4983 Important note: see "option splice-auto" for usage limitations.
4984
4985 Example :
4986 option splice-response
4987
4988 If this option has been enabled in a "defaults" section, it can be disabled
4989 in a specific instance by prepending the "no" keyword before it.
4990
4991 See also : "option splice-auto", "option splice-request", and global options
4992 "nosplice" and "maxpipes"
4993
4994
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004995option srvtcpka
4996no option srvtcpka
4997 Enable or disable the sending of TCP keepalive packets on the server side
4998 May be used in sections : defaults | frontend | listen | backend
4999 yes | no | yes | yes
5000 Arguments : none
5001
5002 When there is a firewall or any session-aware component between a client and
5003 a server, and when the protocol involves very long sessions with long idle
5004 periods (eg: remote desktops), there is a risk that one of the intermediate
5005 components decides to expire a session which has remained idle for too long.
5006
5007 Enabling socket-level TCP keep-alives makes the system regularly send packets
5008 to the other end of the connection, leaving it active. The delay between
5009 keep-alive probes is controlled by the system only and depends both on the
5010 operating system and its tuning parameters.
5011
5012 It is important to understand that keep-alive packets are neither emitted nor
5013 received at the application level. It is only the network stacks which sees
5014 them. For this reason, even if one side of the proxy already uses keep-alives
5015 to maintain its connection alive, those keep-alive packets will not be
5016 forwarded to the other side of the proxy.
5017
5018 Please note that this has nothing to do with HTTP keep-alive.
5019
5020 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
5021 server side of a connection, which should help when session expirations are
5022 noticed between HAProxy and a server.
5023
5024 If this option has been enabled in a "defaults" section, it can be disabled
5025 in a specific instance by prepending the "no" keyword before it.
5026
5027 See also : "option clitcpka", "option tcpka"
5028
5029
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005030option ssl-hello-chk
5031 Use SSLv3 client hello health checks for server testing
5032 May be used in sections : defaults | frontend | listen | backend
5033 yes | no | yes | yes
5034 Arguments : none
5035
5036 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
5037 possible to test that the server correctly talks SSL instead of just testing
5038 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
5039 SSLv3 client hello messages are sent once the connection is established to
5040 the server, and the response is analyzed to find an SSL server hello message.
5041 The server is considered valid only when the response contains this server
5042 hello message.
5043
5044 All servers tested till there correctly reply to SSLv3 client hello messages,
5045 and most servers tested do not even log the requests containing only hello
5046 messages, which is appreciable.
5047
Willy Tarreau763a95b2012-10-04 23:15:39 +02005048 Note that this check works even when SSL support was not built into haproxy
5049 because it forges the SSL message. When SSL support is available, it is best
5050 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005051
Willy Tarreau763a95b2012-10-04 23:15:39 +02005052 See also: "option httpchk", "check-ssl"
5053
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005054
Willy Tarreaued179852013-12-16 01:07:00 +01005055option tcp-check
5056 Perform health checks using tcp-check send/expect sequences
5057 May be used in sections: defaults | frontend | listen | backend
5058 yes | no | yes | yes
5059
5060 This health check method is intended to be combined with "tcp-check" command
5061 lists in order to support send/expect types of health check sequences.
5062
5063 TCP checks currently support 4 modes of operations :
5064 - no "tcp-check" directive : the health check only consists in a connection
5065 attempt, which remains the default mode.
5066
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005067 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005068 used to send a string along with a connection opening. With some
5069 protocols, it helps sending a "QUIT" message for example that prevents
5070 the server from logging a connection error for each health check. The
5071 check result will still be based on the ability to open the connection
5072 only.
5073
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005074 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01005075 The connection is opened and haproxy waits for the server to present some
5076 contents which must validate some rules. The check result will be based
5077 on the matching between the contents and the rules. This is suited for
5078 POP, IMAP, SMTP, FTP, SSH, TELNET.
5079
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005080 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005081 used to test a hello-type protocol. Haproxy sends a message, the server
5082 responds and its response is analysed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005083 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01005084 suited for protocols which require a binding or a request/response model.
5085 LDAP, MySQL, Redis and SSL are example of such protocols, though they
5086 already all have their dedicated checks with a deeper understanding of
5087 the respective protocols.
5088 In this mode, many questions may be sent and many answers may be
5089 analysed.
5090
5091 Examples :
5092 # perform a POP check (analyse only server's banner)
5093 option tcp-check
5094 tcp-check expect string +OK\ POP3\ ready
5095
5096 # perform an IMAP check (analyse only server's banner)
5097 option tcp-check
5098 tcp-check expect string *\ OK\ IMAP4\ ready
5099
5100 # look for the redis master server after ensuring it speaks well
5101 # redis protocol, then it exits properly.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005102 # (send a command then analyse the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01005103 option tcp-check
5104 tcp-check send PING\r\n
5105 tcp-check expect +PONG
5106 tcp-check send info\ replication\r\n
5107 tcp-check expect string role:master
5108 tcp-check send QUIT\r\n
5109 tcp-check expect string +OK
5110
5111 forge a HTTP request, then analyse the response
5112 (send many headers before analyzing)
5113 option tcp-check
5114 tcp-check send HEAD\ /\ HTTP/1.1\r\n
5115 tcp-check send Host:\ www.mydomain.com\r\n
5116 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
5117 tcp-check send \r\n
5118 tcp-check expect rstring HTTP/1\..\ (2..|3..)
5119
5120
5121 See also : "tcp-check expect", "tcp-check send"
5122
5123
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005124option tcp-smart-accept
5125no option tcp-smart-accept
5126 Enable or disable the saving of one ACK packet during the accept sequence
5127 May be used in sections : defaults | frontend | listen | backend
5128 yes | yes | yes | no
5129 Arguments : none
5130
5131 When an HTTP connection request comes in, the system acknowledges it on
5132 behalf of HAProxy, then the client immediately sends its request, and the
5133 system acknowledges it too while it is notifying HAProxy about the new
5134 connection. HAProxy then reads the request and responds. This means that we
5135 have one TCP ACK sent by the system for nothing, because the request could
5136 very well be acknowledged by HAProxy when it sends its response.
5137
5138 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
5139 sending this useless ACK on platforms which support it (currently at least
5140 Linux). It must not cause any problem, because the system will send it anyway
5141 after 40 ms if the response takes more time than expected to come.
5142
5143 During complex network debugging sessions, it may be desirable to disable
5144 this optimization because delayed ACKs can make troubleshooting more complex
5145 when trying to identify where packets are delayed. It is then possible to
5146 fall back to normal behaviour by specifying "no option tcp-smart-accept".
5147
5148 It is also possible to force it for non-HTTP proxies by simply specifying
5149 "option tcp-smart-accept". For instance, it can make sense with some services
5150 such as SMTP where the server speaks first.
5151
5152 It is recommended to avoid forcing this option in a defaults section. In case
5153 of doubt, consider setting it back to automatic values by prepending the
5154 "default" keyword before it, or disabling it using the "no" keyword.
5155
Willy Tarreaud88edf22009-06-14 15:48:17 +02005156 See also : "option tcp-smart-connect"
5157
5158
5159option tcp-smart-connect
5160no option tcp-smart-connect
5161 Enable or disable the saving of one ACK packet during the connect sequence
5162 May be used in sections : defaults | frontend | listen | backend
5163 yes | no | yes | yes
5164 Arguments : none
5165
5166 On certain systems (at least Linux), HAProxy can ask the kernel not to
5167 immediately send an empty ACK upon a connection request, but to directly
5168 send the buffer request instead. This saves one packet on the network and
5169 thus boosts performance. It can also be useful for some servers, because they
5170 immediately get the request along with the incoming connection.
5171
5172 This feature is enabled when "option tcp-smart-connect" is set in a backend.
5173 It is not enabled by default because it makes network troubleshooting more
5174 complex.
5175
5176 It only makes sense to enable it with protocols where the client speaks first
5177 such as HTTP. In other situations, if there is no data to send in place of
5178 the ACK, a normal ACK is sent.
5179
5180 If this option has been enabled in a "defaults" section, it can be disabled
5181 in a specific instance by prepending the "no" keyword before it.
5182
5183 See also : "option tcp-smart-accept"
5184
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005185
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005186option tcpka
5187 Enable or disable the sending of TCP keepalive packets on both sides
5188 May be used in sections : defaults | frontend | listen | backend
5189 yes | yes | yes | yes
5190 Arguments : none
5191
5192 When there is a firewall or any session-aware component between a client and
5193 a server, and when the protocol involves very long sessions with long idle
5194 periods (eg: remote desktops), there is a risk that one of the intermediate
5195 components decides to expire a session which has remained idle for too long.
5196
5197 Enabling socket-level TCP keep-alives makes the system regularly send packets
5198 to the other end of the connection, leaving it active. The delay between
5199 keep-alive probes is controlled by the system only and depends both on the
5200 operating system and its tuning parameters.
5201
5202 It is important to understand that keep-alive packets are neither emitted nor
5203 received at the application level. It is only the network stacks which sees
5204 them. For this reason, even if one side of the proxy already uses keep-alives
5205 to maintain its connection alive, those keep-alive packets will not be
5206 forwarded to the other side of the proxy.
5207
5208 Please note that this has nothing to do with HTTP keep-alive.
5209
5210 Using option "tcpka" enables the emission of TCP keep-alive probes on both
5211 the client and server sides of a connection. Note that this is meaningful
5212 only in "defaults" or "listen" sections. If this option is used in a
5213 frontend, only the client side will get keep-alives, and if this option is
5214 used in a backend, only the server side will get keep-alives. For this
5215 reason, it is strongly recommended to explicitly use "option clitcpka" and
5216 "option srvtcpka" when the configuration is split between frontends and
5217 backends.
5218
5219 See also : "option clitcpka", "option srvtcpka"
5220
Willy Tarreau844e3c52008-01-11 16:28:18 +01005221
5222option tcplog
5223 Enable advanced logging of TCP connections with session state and timers
5224 May be used in sections : defaults | frontend | listen | backend
5225 yes | yes | yes | yes
5226 Arguments : none
5227
5228 By default, the log output format is very poor, as it only contains the
5229 source and destination addresses, and the instance name. By specifying
5230 "option tcplog", each log line turns into a much richer format including, but
5231 not limited to, the connection timers, the session status, the connections
5232 numbers, the frontend, backend and server name, and of course the source
5233 address and ports. This option is useful for pure TCP proxies in order to
5234 find which of the client or server disconnects or times out. For normal HTTP
5235 proxies, it's better to use "option httplog" which is even more complete.
5236
5237 This option may be set either in the frontend or the backend.
5238
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005239 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005240
5241
Willy Tarreau844e3c52008-01-11 16:28:18 +01005242option transparent
5243no option transparent
5244 Enable client-side transparent proxying
5245 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01005246 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01005247 Arguments : none
5248
5249 This option was introduced in order to provide layer 7 persistence to layer 3
5250 load balancers. The idea is to use the OS's ability to redirect an incoming
5251 connection for a remote address to a local process (here HAProxy), and let
5252 this process know what address was initially requested. When this option is
5253 used, sessions without cookies will be forwarded to the original destination
5254 IP address of the incoming request (which should match that of another
5255 equipment), while requests with cookies will still be forwarded to the
5256 appropriate server.
5257
5258 Note that contrary to a common belief, this option does NOT make HAProxy
5259 present the client's IP to the server when establishing the connection.
5260
Willy Tarreaua1146052011-03-01 09:51:54 +01005261 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005262 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005263
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005264
Emeric Brun647caf12009-06-30 17:57:00 +02005265persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02005266persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02005267 Enable RDP cookie-based persistence
5268 May be used in sections : defaults | frontend | listen | backend
5269 yes | no | yes | yes
5270 Arguments :
5271 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02005272 default cookie name "msts" will be used. There currently is no
5273 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02005274
5275 This statement enables persistence based on an RDP cookie. The RDP cookie
5276 contains all information required to find the server in the list of known
5277 servers. So when this option is set in the backend, the request is analysed
5278 and if an RDP cookie is found, it is decoded. If it matches a known server
5279 which is still UP (or if "option persist" is set), then the connection is
5280 forwarded to this server.
5281
5282 Note that this only makes sense in a TCP backend, but for this to work, the
5283 frontend must have waited long enough to ensure that an RDP cookie is present
5284 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005285 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02005286 a single "listen" section.
5287
Willy Tarreau61e28f22010-05-16 22:31:05 +02005288 Also, it is important to understand that the terminal server will emit this
5289 RDP cookie only if it is configured for "token redirection mode", which means
5290 that the "IP address redirection" option is disabled.
5291
Emeric Brun647caf12009-06-30 17:57:00 +02005292 Example :
5293 listen tse-farm
5294 bind :3389
5295 # wait up to 5s for an RDP cookie in the request
5296 tcp-request inspect-delay 5s
5297 tcp-request content accept if RDP_COOKIE
5298 # apply RDP cookie persistence
5299 persist rdp-cookie
5300 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02005301 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02005302 balance rdp-cookie
5303 server srv1 1.1.1.1:3389
5304 server srv2 1.1.1.2:3389
5305
Simon Hormanab814e02011-06-24 14:50:20 +09005306 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
5307 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02005308
5309
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005310rate-limit sessions <rate>
5311 Set a limit on the number of new sessions accepted per second on a frontend
5312 May be used in sections : defaults | frontend | listen | backend
5313 yes | yes | yes | no
5314 Arguments :
5315 <rate> The <rate> parameter is an integer designating the maximum number
5316 of new sessions per second to accept on the frontend.
5317
5318 When the frontend reaches the specified number of new sessions per second, it
5319 stops accepting new connections until the rate drops below the limit again.
5320 During this time, the pending sessions will be kept in the socket's backlog
5321 (in system buffers) and haproxy will not even be aware that sessions are
5322 pending. When applying very low limit on a highly loaded service, it may make
5323 sense to increase the socket's backlog using the "backlog" keyword.
5324
5325 This feature is particularly efficient at blocking connection-based attacks
5326 or service abuse on fragile servers. Since the session rate is measured every
5327 millisecond, it is extremely accurate. Also, the limit applies immediately,
5328 no delay is needed at all to detect the threshold.
5329
5330 Example : limit the connection rate on SMTP to 10 per second max
5331 listen smtp
5332 mode tcp
5333 bind :25
5334 rate-limit sessions 10
5335 server 127.0.0.1:1025
5336
Willy Tarreaua17c2d92011-07-25 08:16:20 +02005337 Note : when the maximum rate is reached, the frontend's status is not changed
5338 but its sockets appear as "WAITING" in the statistics if the
5339 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005340
5341 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
5342
5343
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005344redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
5345redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
5346redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005347 Return an HTTP redirection if/unless a condition is matched
5348 May be used in sections : defaults | frontend | listen | backend
5349 no | yes | yes | yes
5350
5351 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01005352 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005353
Willy Tarreau0140f252008-11-19 21:07:09 +01005354 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005355 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005356 the HTTP "Location" header. When used in an "http-request" rule,
5357 <loc> value follows the log-format rules and can include some
5358 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005359
5360 <pfx> With "redirect prefix", the "Location" header is built from the
5361 concatenation of <pfx> and the complete URI path, including the
5362 query string, unless the "drop-query" option is specified (see
5363 below). As a special case, if <pfx> equals exactly "/", then
5364 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005365 redirect to the same URL (for instance, to insert a cookie). When
5366 used in an "http-request" rule, <pfx> value follows the log-format
5367 rules and can include some dynamic values (see Custom Log Format
5368 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005369
5370 <sch> With "redirect scheme", then the "Location" header is built by
5371 concatenating <sch> with "://" then the first occurrence of the
5372 "Host" header, and then the URI path, including the query string
5373 unless the "drop-query" option is specified (see below). If no
5374 path is found or if the path is "*", then "/" is used instead. If
5375 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005376 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005377 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005378 HTTPS. When used in an "http-request" rule, <sch> value follows
5379 the log-format rules and can include some dynamic values (see
5380 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01005381
5382 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005383 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
5384 with 302 used by default if no code is specified. 301 means
5385 "Moved permanently", and a browser may cache the Location. 302
5386 means "Moved permanently" and means that the browser should not
5387 cache the redirection. 303 is equivalent to 302 except that the
5388 browser will fetch the location with a GET method. 307 is just
5389 like 302 but makes it clear that the same method must be reused.
5390 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01005391
5392 <option> There are several options which can be specified to adjust the
5393 expected behaviour of a redirection :
5394
5395 - "drop-query"
5396 When this keyword is used in a prefix-based redirection, then the
5397 location will be set without any possible query-string, which is useful
5398 for directing users to a non-secure page for instance. It has no effect
5399 with a location-type redirect.
5400
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005401 - "append-slash"
5402 This keyword may be used in conjunction with "drop-query" to redirect
5403 users who use a URL not ending with a '/' to the same one with the '/'.
5404 It can be useful to ensure that search engines will only see one URL.
5405 For this, a return code 301 is preferred.
5406
Willy Tarreau0140f252008-11-19 21:07:09 +01005407 - "set-cookie NAME[=value]"
5408 A "Set-Cookie" header will be added with NAME (and optionally "=value")
5409 to the response. This is sometimes used to indicate that a user has
5410 been seen, for instance to protect against some types of DoS. No other
5411 cookie option is added, so the cookie will be a session cookie. Note
5412 that for a browser, a sole cookie name without an equal sign is
5413 different from a cookie with an equal sign.
5414
5415 - "clear-cookie NAME[=]"
5416 A "Set-Cookie" header will be added with NAME (and optionally "="), but
5417 with the "Max-Age" attribute set to zero. This will tell the browser to
5418 delete this cookie. It is useful for instance on logout pages. It is
5419 important to note that clearing the cookie "NAME" will not remove a
5420 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
5421 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005422
5423 Example: move the login URL only to HTTPS.
5424 acl clear dst_port 80
5425 acl secure dst_port 8080
5426 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01005427 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01005428 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01005429 acl cookie_set hdr_sub(cookie) SEEN=1
5430
5431 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01005432 redirect prefix https://mysite.com if login_page !secure
5433 redirect prefix http://mysite.com drop-query if login_page !uid_given
5434 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01005435 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005436
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005437 Example: send redirects for request for articles without a '/'.
5438 acl missing_slash path_reg ^/article/[^/]*$
5439 redirect code 301 prefix / drop-query append-slash if missing_slash
5440
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005441 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01005442 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005443
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005444 Example: append 'www.' prefix in front of all hosts not having it
5445 http-request redirect code 301 location www.%[hdr(host)]%[req.uri] \
5446 unless { hdr_beg(host) -i www }
5447
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005448 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005449
5450
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005451redisp (deprecated)
5452redispatch (deprecated)
5453 Enable or disable session redistribution in case of connection failure
5454 May be used in sections: defaults | frontend | listen | backend
5455 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005456 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005457
5458 In HTTP mode, if a server designated by a cookie is down, clients may
5459 definitely stick to it because they cannot flush the cookie, so they will not
5460 be able to access the service anymore.
5461
5462 Specifying "redispatch" will allow the proxy to break their persistence and
5463 redistribute them to a working server.
5464
5465 It also allows to retry last connection to another server in case of multiple
5466 connection failures. Of course, it requires having "retries" set to a nonzero
5467 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01005468
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005469 This form is deprecated, do not use it in any new configuration, use the new
5470 "option redispatch" instead.
5471
5472 See also : "option redispatch"
5473
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005474
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005475reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005476 Add a header at the end of the HTTP request
5477 May be used in sections : defaults | frontend | listen | backend
5478 no | yes | yes | yes
5479 Arguments :
5480 <string> is the complete line to be added. Any space or known delimiter
5481 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005482 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005483
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005484 <cond> is an optional matching condition built from ACLs. It makes it
5485 possible to ignore this rule when other conditions are not met.
5486
Willy Tarreau303c0352008-01-17 19:01:39 +01005487 A new line consisting in <string> followed by a line feed will be added after
5488 the last header of an HTTP request.
5489
5490 Header transformations only apply to traffic which passes through HAProxy,
5491 and not to traffic generated by HAProxy, such as health-checks or error
5492 responses.
5493
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005494 Example : add "X-Proto: SSL" to requests coming via port 81
5495 acl is-ssl dst_port 81
5496 reqadd X-Proto:\ SSL if is-ssl
5497
5498 See also: "rspadd", section 6 about HTTP header manipulation, and section 7
5499 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005500
5501
Willy Tarreau5321c422010-01-28 20:35:13 +01005502reqallow <search> [{if | unless} <cond>]
5503reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005504 Definitely allow an HTTP request if a line matches a regular expression
5505 May be used in sections : defaults | frontend | listen | backend
5506 no | yes | yes | yes
5507 Arguments :
5508 <search> is the regular expression applied to HTTP headers and to the
5509 request line. This is an extended regular expression. Parenthesis
5510 grouping is supported and no preliminary backslash is required.
5511 Any space or known delimiter must be escaped using a backslash
5512 ('\'). The pattern applies to a full line at a time. The
5513 "reqallow" keyword strictly matches case while "reqiallow"
5514 ignores case.
5515
Willy Tarreau5321c422010-01-28 20:35:13 +01005516 <cond> is an optional matching condition built from ACLs. It makes it
5517 possible to ignore this rule when other conditions are not met.
5518
Willy Tarreau303c0352008-01-17 19:01:39 +01005519 A request containing any line which matches extended regular expression
5520 <search> will mark the request as allowed, even if any later test would
5521 result in a deny. The test applies both to the request line and to request
5522 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005523 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005524
5525 It is easier, faster and more powerful to use ACLs to write access policies.
5526 Reqdeny, reqallow and reqpass should be avoided in new designs.
5527
5528 Example :
5529 # allow www.* but refuse *.local
5530 reqiallow ^Host:\ www\.
5531 reqideny ^Host:\ .*\.local
5532
Willy Tarreau5321c422010-01-28 20:35:13 +01005533 See also: "reqdeny", "block", section 6 about HTTP header manipulation, and
5534 section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005535
5536
Willy Tarreau5321c422010-01-28 20:35:13 +01005537reqdel <search> [{if | unless} <cond>]
5538reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005539 Delete all headers matching a regular expression in an HTTP request
5540 May be used in sections : defaults | frontend | listen | backend
5541 no | yes | yes | yes
5542 Arguments :
5543 <search> is the regular expression applied to HTTP headers and to the
5544 request line. This is an extended regular expression. Parenthesis
5545 grouping is supported and no preliminary backslash is required.
5546 Any space or known delimiter must be escaped using a backslash
5547 ('\'). The pattern applies to a full line at a time. The "reqdel"
5548 keyword strictly matches case while "reqidel" ignores case.
5549
Willy Tarreau5321c422010-01-28 20:35:13 +01005550 <cond> is an optional matching condition built from ACLs. It makes it
5551 possible to ignore this rule when other conditions are not met.
5552
Willy Tarreau303c0352008-01-17 19:01:39 +01005553 Any header line matching extended regular expression <search> in the request
5554 will be completely deleted. Most common use of this is to remove unwanted
5555 and/or dangerous headers or cookies from a request before passing it to the
5556 next servers.
5557
5558 Header transformations only apply to traffic which passes through HAProxy,
5559 and not to traffic generated by HAProxy, such as health-checks or error
5560 responses. Keep in mind that header names are not case-sensitive.
5561
5562 Example :
5563 # remove X-Forwarded-For header and SERVER cookie
5564 reqidel ^X-Forwarded-For:.*
5565 reqidel ^Cookie:.*SERVER=
5566
Willy Tarreau5321c422010-01-28 20:35:13 +01005567 See also: "reqadd", "reqrep", "rspdel", section 6 about HTTP header
5568 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005569
5570
Willy Tarreau5321c422010-01-28 20:35:13 +01005571reqdeny <search> [{if | unless} <cond>]
5572reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005573 Deny an HTTP request if a line matches a regular expression
5574 May be used in sections : defaults | frontend | listen | backend
5575 no | yes | yes | yes
5576 Arguments :
5577 <search> is the regular expression applied to HTTP headers and to the
5578 request line. This is an extended regular expression. Parenthesis
5579 grouping is supported and no preliminary backslash is required.
5580 Any space or known delimiter must be escaped using a backslash
5581 ('\'). The pattern applies to a full line at a time. The
5582 "reqdeny" keyword strictly matches case while "reqideny" ignores
5583 case.
5584
Willy Tarreau5321c422010-01-28 20:35:13 +01005585 <cond> is an optional matching condition built from ACLs. It makes it
5586 possible to ignore this rule when other conditions are not met.
5587
Willy Tarreau303c0352008-01-17 19:01:39 +01005588 A request containing any line which matches extended regular expression
5589 <search> will mark the request as denied, even if any later test would
5590 result in an allow. The test applies both to the request line and to request
5591 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005592 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005593
Willy Tarreauced27012008-01-17 20:35:34 +01005594 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005595 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01005596 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01005597
Willy Tarreau303c0352008-01-17 19:01:39 +01005598 It is easier, faster and more powerful to use ACLs to write access policies.
5599 Reqdeny, reqallow and reqpass should be avoided in new designs.
5600
5601 Example :
5602 # refuse *.local, then allow www.*
5603 reqideny ^Host:\ .*\.local
5604 reqiallow ^Host:\ www\.
5605
Willy Tarreau5321c422010-01-28 20:35:13 +01005606 See also: "reqallow", "rspdeny", "block", section 6 about HTTP header
5607 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005608
5609
Willy Tarreau5321c422010-01-28 20:35:13 +01005610reqpass <search> [{if | unless} <cond>]
5611reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005612 Ignore any HTTP request line matching a regular expression in next rules
5613 May be used in sections : defaults | frontend | listen | backend
5614 no | yes | yes | yes
5615 Arguments :
5616 <search> is the regular expression applied to HTTP headers and to the
5617 request line. This is an extended regular expression. Parenthesis
5618 grouping is supported and no preliminary backslash is required.
5619 Any space or known delimiter must be escaped using a backslash
5620 ('\'). The pattern applies to a full line at a time. The
5621 "reqpass" keyword strictly matches case while "reqipass" ignores
5622 case.
5623
Willy Tarreau5321c422010-01-28 20:35:13 +01005624 <cond> is an optional matching condition built from ACLs. It makes it
5625 possible to ignore this rule when other conditions are not met.
5626
Willy Tarreau303c0352008-01-17 19:01:39 +01005627 A request containing any line which matches extended regular expression
5628 <search> will skip next rules, without assigning any deny or allow verdict.
5629 The test applies both to the request line and to request headers. Keep in
5630 mind that URLs in request line are case-sensitive while header names are not.
5631
5632 It is easier, faster and more powerful to use ACLs to write access policies.
5633 Reqdeny, reqallow and reqpass should be avoided in new designs.
5634
5635 Example :
5636 # refuse *.local, then allow www.*, but ignore "www.private.local"
5637 reqipass ^Host:\ www.private\.local
5638 reqideny ^Host:\ .*\.local
5639 reqiallow ^Host:\ www\.
5640
Willy Tarreau5321c422010-01-28 20:35:13 +01005641 See also: "reqallow", "reqdeny", "block", section 6 about HTTP header
5642 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005643
5644
Willy Tarreau5321c422010-01-28 20:35:13 +01005645reqrep <search> <string> [{if | unless} <cond>]
5646reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005647 Replace a regular expression with a string in an HTTP request line
5648 May be used in sections : defaults | frontend | listen | backend
5649 no | yes | yes | yes
5650 Arguments :
5651 <search> is the regular expression applied to HTTP headers and to the
5652 request line. This is an extended regular expression. Parenthesis
5653 grouping is supported and no preliminary backslash is required.
5654 Any space or known delimiter must be escaped using a backslash
5655 ('\'). The pattern applies to a full line at a time. The "reqrep"
5656 keyword strictly matches case while "reqirep" ignores case.
5657
5658 <string> is the complete line to be added. Any space or known delimiter
5659 must be escaped using a backslash ('\'). References to matched
5660 pattern groups are possible using the common \N form, with N
5661 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005662 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005663
Willy Tarreau5321c422010-01-28 20:35:13 +01005664 <cond> is an optional matching condition built from ACLs. It makes it
5665 possible to ignore this rule when other conditions are not met.
5666
Willy Tarreau303c0352008-01-17 19:01:39 +01005667 Any line matching extended regular expression <search> in the request (both
5668 the request line and header lines) will be completely replaced with <string>.
5669 Most common use of this is to rewrite URLs or domain names in "Host" headers.
5670
5671 Header transformations only apply to traffic which passes through HAProxy,
5672 and not to traffic generated by HAProxy, such as health-checks or error
5673 responses. Note that for increased readability, it is suggested to add enough
5674 spaces between the request and the response. Keep in mind that URLs in
5675 request line are case-sensitive while header names are not.
5676
5677 Example :
5678 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04005679 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01005680 # replace "www.mydomain.com" with "www" in the host name.
5681 reqirep ^Host:\ www.mydomain.com Host:\ www
5682
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04005683 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", section 6 about
5684 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005685
5686
Willy Tarreau5321c422010-01-28 20:35:13 +01005687reqtarpit <search> [{if | unless} <cond>]
5688reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005689 Tarpit an HTTP request containing a line matching a regular expression
5690 May be used in sections : defaults | frontend | listen | backend
5691 no | yes | yes | yes
5692 Arguments :
5693 <search> is the regular expression applied to HTTP headers and to the
5694 request line. This is an extended regular expression. Parenthesis
5695 grouping is supported and no preliminary backslash is required.
5696 Any space or known delimiter must be escaped using a backslash
5697 ('\'). The pattern applies to a full line at a time. The
5698 "reqtarpit" keyword strictly matches case while "reqitarpit"
5699 ignores case.
5700
Willy Tarreau5321c422010-01-28 20:35:13 +01005701 <cond> is an optional matching condition built from ACLs. It makes it
5702 possible to ignore this rule when other conditions are not met.
5703
Willy Tarreau303c0352008-01-17 19:01:39 +01005704 A request containing any line which matches extended regular expression
5705 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01005706 be kept open for a pre-defined time, then will return an HTTP error 500 so
5707 that the attacker does not suspect it has been tarpitted. The status 500 will
5708 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01005709 delay is defined by "timeout tarpit", or "timeout connect" if the former is
5710 not set.
5711
5712 The goal of the tarpit is to slow down robots attacking servers with
5713 identifiable requests. Many robots limit their outgoing number of connections
5714 and stay connected waiting for a reply which can take several minutes to
5715 come. Depending on the environment and attack, it may be particularly
5716 efficient at reducing the load on the network and firewalls.
5717
Willy Tarreau5321c422010-01-28 20:35:13 +01005718 Examples :
Willy Tarreau303c0352008-01-17 19:01:39 +01005719 # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but
5720 # block all others.
5721 reqipass ^User-Agent:\.*(Mozilla|MSIE)
5722 reqitarpit ^User-Agent:
5723
Willy Tarreau5321c422010-01-28 20:35:13 +01005724 # block bad guys
5725 acl badguys src 10.1.0.3 172.16.13.20/28
5726 reqitarpit . if badguys
5727
5728 See also: "reqallow", "reqdeny", "reqpass", section 6 about HTTP header
5729 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005730
5731
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02005732retries <value>
5733 Set the number of retries to perform on a server after a connection failure
5734 May be used in sections: defaults | frontend | listen | backend
5735 yes | no | yes | yes
5736 Arguments :
5737 <value> is the number of times a connection attempt should be retried on
5738 a server when a connection either is refused or times out. The
5739 default value is 3.
5740
5741 It is important to understand that this value applies to the number of
5742 connection attempts, not full requests. When a connection has effectively
5743 been established to a server, there will be no more retry.
5744
5745 In order to avoid immediate reconnections to a server which is restarting,
5746 a turn-around timer of 1 second is applied before a retry occurs.
5747
5748 When "option redispatch" is set, the last retry may be performed on another
5749 server even if a cookie references a different server.
5750
5751 See also : "option redispatch"
5752
5753
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005754rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005755 Add a header at the end of the HTTP response
5756 May be used in sections : defaults | frontend | listen | backend
5757 no | yes | yes | yes
5758 Arguments :
5759 <string> is the complete line to be added. Any space or known delimiter
5760 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005761 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005762
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005763 <cond> is an optional matching condition built from ACLs. It makes it
5764 possible to ignore this rule when other conditions are not met.
5765
Willy Tarreau303c0352008-01-17 19:01:39 +01005766 A new line consisting in <string> followed by a line feed will be added after
5767 the last header of an HTTP response.
5768
5769 Header transformations only apply to traffic which passes through HAProxy,
5770 and not to traffic generated by HAProxy, such as health-checks or error
5771 responses.
5772
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005773 See also: "reqadd", section 6 about HTTP header manipulation, and section 7
5774 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005775
5776
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005777rspdel <search> [{if | unless} <cond>]
5778rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005779 Delete all headers matching a regular expression in an HTTP response
5780 May be used in sections : defaults | frontend | listen | backend
5781 no | yes | yes | yes
5782 Arguments :
5783 <search> is the regular expression applied to HTTP headers and to the
5784 response line. This is an extended regular expression, so
5785 parenthesis grouping is supported and no preliminary backslash
5786 is required. Any space or known delimiter must be escaped using
5787 a backslash ('\'). The pattern applies to a full line at a time.
5788 The "rspdel" keyword strictly matches case while "rspidel"
5789 ignores case.
5790
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005791 <cond> is an optional matching condition built from ACLs. It makes it
5792 possible to ignore this rule when other conditions are not met.
5793
Willy Tarreau303c0352008-01-17 19:01:39 +01005794 Any header line matching extended regular expression <search> in the response
5795 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005796 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01005797 client.
5798
5799 Header transformations only apply to traffic which passes through HAProxy,
5800 and not to traffic generated by HAProxy, such as health-checks or error
5801 responses. Keep in mind that header names are not case-sensitive.
5802
5803 Example :
5804 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02005805 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01005806
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005807 See also: "rspadd", "rsprep", "reqdel", section 6 about HTTP header
5808 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005809
5810
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005811rspdeny <search> [{if | unless} <cond>]
5812rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005813 Block an HTTP response if a line matches a regular expression
5814 May be used in sections : defaults | frontend | listen | backend
5815 no | yes | yes | yes
5816 Arguments :
5817 <search> is the regular expression applied to HTTP headers and to the
5818 response line. This is an extended regular expression, so
5819 parenthesis grouping is supported and no preliminary backslash
5820 is required. Any space or known delimiter must be escaped using
5821 a backslash ('\'). The pattern applies to a full line at a time.
5822 The "rspdeny" keyword strictly matches case while "rspideny"
5823 ignores case.
5824
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005825 <cond> is an optional matching condition built from ACLs. It makes it
5826 possible to ignore this rule when other conditions are not met.
5827
Willy Tarreau303c0352008-01-17 19:01:39 +01005828 A response containing any line which matches extended regular expression
5829 <search> will mark the request as denied. The test applies both to the
5830 response line and to response headers. Keep in mind that header names are not
5831 case-sensitive.
5832
5833 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01005834 block the response before it reaches the client. If a response is denied, it
5835 will be replaced with an HTTP 502 error so that the client never retrieves
5836 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01005837
5838 It is easier, faster and more powerful to use ACLs to write access policies.
5839 Rspdeny should be avoided in new designs.
5840
5841 Example :
5842 # Ensure that no content type matching ms-word will leak
5843 rspideny ^Content-type:\.*/ms-word
5844
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005845 See also: "reqdeny", "acl", "block", section 6 about HTTP header manipulation
5846 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005847
5848
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005849rsprep <search> <string> [{if | unless} <cond>]
5850rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005851 Replace a regular expression with a string in an HTTP response line
5852 May be used in sections : defaults | frontend | listen | backend
5853 no | yes | yes | yes
5854 Arguments :
5855 <search> is the regular expression applied to HTTP headers and to the
5856 response line. This is an extended regular expression, so
5857 parenthesis grouping is supported and no preliminary backslash
5858 is required. Any space or known delimiter must be escaped using
5859 a backslash ('\'). The pattern applies to a full line at a time.
5860 The "rsprep" keyword strictly matches case while "rspirep"
5861 ignores case.
5862
5863 <string> is the complete line to be added. Any space or known delimiter
5864 must be escaped using a backslash ('\'). References to matched
5865 pattern groups are possible using the common \N form, with N
5866 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005867 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005868
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005869 <cond> is an optional matching condition built from ACLs. It makes it
5870 possible to ignore this rule when other conditions are not met.
5871
Willy Tarreau303c0352008-01-17 19:01:39 +01005872 Any line matching extended regular expression <search> in the response (both
5873 the response line and header lines) will be completely replaced with
5874 <string>. Most common use of this is to rewrite Location headers.
5875
5876 Header transformations only apply to traffic which passes through HAProxy,
5877 and not to traffic generated by HAProxy, such as health-checks or error
5878 responses. Note that for increased readability, it is suggested to add enough
5879 spaces between the request and the response. Keep in mind that header names
5880 are not case-sensitive.
5881
5882 Example :
5883 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
5884 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
5885
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005886 See also: "rspadd", "rspdel", "reqrep", section 6 about HTTP header
5887 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005888
5889
David du Colombier486df472011-03-17 10:40:26 +01005890server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005891 Declare a server in a backend
5892 May be used in sections : defaults | frontend | listen | backend
5893 no | no | yes | yes
5894 Arguments :
5895 <name> is the internal name assigned to this server. This name will
Cyril Bonté941a0c62012-10-15 19:44:24 +02005896 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05005897 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005898
David du Colombier486df472011-03-17 10:40:26 +01005899 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
5900 resolvable hostname is supported, but this name will be resolved
5901 during start-up. Address "0.0.0.0" or "*" has a special meaning.
5902 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02005903 address as the one from the client connection. This is useful in
5904 transparent proxy architectures where the client's connection is
5905 intercepted and haproxy must forward to the original destination
5906 address. This is more or less what the "transparent" keyword does
5907 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01005908 to report statistics. Optionally, an address family prefix may be
5909 used before the address to force the family regardless of the
5910 address format, which can be useful to specify a path to a unix
5911 socket with no slash ('/'). Currently supported prefixes are :
5912 - 'ipv4@' -> address is always IPv4
5913 - 'ipv6@' -> address is always IPv6
5914 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02005915 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01005916 Any part of the address string may reference any number of
5917 environment variables by preceding their name with a dollar
5918 sign ('$') and optionally enclosing them with braces ('{}'),
5919 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005920
Willy Tarreaub6205fd2012-09-24 12:27:33 +02005921 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005922 be sent to this port. If unset, the same port the client
5923 connected to will be used. The port may also be prefixed by a "+"
5924 or a "-". In this case, the server's port will be determined by
5925 adding this value to the client's port.
5926
5927 <param*> is a list of parameters for this server. The "server" keywords
5928 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005929 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005930
5931 Examples :
5932 server first 10.1.1.1:1080 cookie first check inter 1000
5933 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01005934 server transp ipv4@
Willy Tarreaudad36a32013-03-11 01:20:04 +01005935 server backup ${SRV_BACKUP}:1080 backup
5936 server www1_dc1 ${LAN_DC1}.101:80
5937 server www1_dc2 ${LAN_DC2}.101:80
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005938
Mark Lamourinec2247f02012-01-04 13:02:01 -05005939 See also: "default-server", "http-send-name-header" and section 5 about
5940 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005941
5942
5943source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02005944source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01005945source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005946 Set the source address for outgoing connections
5947 May be used in sections : defaults | frontend | listen | backend
5948 yes | no | yes | yes
5949 Arguments :
5950 <addr> is the IPv4 address HAProxy will bind to before connecting to a
5951 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01005952
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005953 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01005954 the most appropriate address to reach its destination. Optionally
5955 an address family prefix may be used before the address to force
5956 the family regardless of the address format, which can be useful
5957 to specify a path to a unix socket with no slash ('/'). Currently
5958 supported prefixes are :
5959 - 'ipv4@' -> address is always IPv4
5960 - 'ipv6@' -> address is always IPv6
5961 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02005962 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01005963 Any part of the address string may reference any number of
5964 environment variables by preceding their name with a dollar
5965 sign ('$') and optionally enclosing them with braces ('{}'),
5966 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005967
5968 <port> is an optional port. It is normally not needed but may be useful
5969 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02005970 the system will select a free port. Note that port ranges are not
5971 supported in the backend. If you want to force port ranges, you
5972 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005973
5974 <addr2> is the IP address to present to the server when connections are
5975 forwarded in full transparent proxy mode. This is currently only
5976 supported on some patched Linux kernels. When this address is
5977 specified, clients connecting to the server will be presented
5978 with this address, while health checks will still use the address
5979 <addr>.
5980
5981 <port2> is the optional port to present to the server when connections
5982 are forwarded in full transparent proxy mode (see <addr2> above).
5983 The default value of zero means the system will select a free
5984 port.
5985
Willy Tarreaubce70882009-09-07 11:51:47 +02005986 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
5987 This is the name of a comma-separated header list which can
5988 contain multiple IP addresses. By default, the last occurrence is
5989 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01005990 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02005991 by previous proxy, typically Stunnel. In order to use another
5992 occurrence from the last one, please see the <occ> parameter
5993 below. When the header (or occurrence) is not found, no binding
5994 is performed so that the proxy's default IP address is used. Also
5995 keep in mind that the header name is case insensitive, as for any
5996 HTTP header.
5997
5998 <occ> is the occurrence number of a value to be used in a multi-value
5999 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006000 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02006001 address. Positive values indicate a position from the first
6002 occurrence, 1 being the first one. Negative values indicate
6003 positions relative to the last one, -1 being the last one. This
6004 is helpful for situations where an X-Forwarded-For header is set
6005 at the entry point of an infrastructure and must be used several
6006 proxy layers away. When this value is not specified, -1 is
6007 assumed. Passing a zero here disables the feature.
6008
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006009 <name> is an optional interface name to which to bind to for outgoing
6010 traffic. On systems supporting this features (currently, only
6011 Linux), this allows one to bind all traffic to the server to
6012 this interface even if it is not the one the system would select
6013 based on routing tables. This should be used with extreme care.
6014 Note that using this option requires root privileges.
6015
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006016 The "source" keyword is useful in complex environments where a specific
6017 address only is allowed to connect to the servers. It may be needed when a
6018 private address must be used through a public gateway for instance, and it is
6019 known that the system cannot determine the adequate source address by itself.
6020
6021 An extension which is available on certain patched Linux kernels may be used
6022 through the "usesrc" optional keyword. It makes it possible to connect to the
6023 servers with an IP address which does not belong to the system itself. This
6024 is called "full transparent proxy mode". For this to work, the destination
6025 servers have to route their traffic back to this address through the machine
6026 running HAProxy, and IP forwarding must generally be enabled on this machine.
6027
6028 In this "full transparent proxy" mode, it is possible to force a specific IP
6029 address to be presented to the servers. This is not much used in fact. A more
6030 common use is to tell HAProxy to present the client's IP address. For this,
6031 there are two methods :
6032
6033 - present the client's IP and port addresses. This is the most transparent
6034 mode, but it can cause problems when IP connection tracking is enabled on
6035 the machine, because a same connection may be seen twice with different
6036 states. However, this solution presents the huge advantage of not
6037 limiting the system to the 64k outgoing address+port couples, because all
6038 of the client ranges may be used.
6039
6040 - present only the client's IP address and select a spare port. This
6041 solution is still quite elegant but slightly less transparent (downstream
6042 firewalls logs will not match upstream's). It also presents the downside
6043 of limiting the number of concurrent connections to the usual 64k ports.
6044 However, since the upstream and downstream ports are different, local IP
6045 connection tracking on the machine will not be upset by the reuse of the
6046 same session.
6047
6048 Note that depending on the transparent proxy technology used, it may be
6049 required to force the source address. In fact, cttproxy version 2 requires an
6050 IP address in <addr> above, and does not support setting of "0.0.0.0" as the
6051 IP address because it creates NAT entries which much match the exact outgoing
6052 address. Tproxy version 4 and some other kernel patches which work in pure
6053 forwarding mode generally will not have this limitation.
6054
6055 This option sets the default source for all servers in the backend. It may
6056 also be specified in a "defaults" section. Finer source address specification
6057 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006058 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006059
6060 Examples :
6061 backend private
6062 # Connect to the servers using our 192.168.1.200 source address
6063 source 192.168.1.200
6064
6065 backend transparent_ssl1
6066 # Connect to the SSL farm from the client's source address
6067 source 192.168.1.200 usesrc clientip
6068
6069 backend transparent_ssl2
6070 # Connect to the SSL farm from the client's source address and port
6071 # not recommended if IP conntrack is present on the local machine.
6072 source 192.168.1.200 usesrc client
6073
6074 backend transparent_ssl3
6075 # Connect to the SSL farm from the client's source address. It
6076 # is more conntrack-friendly.
6077 source 192.168.1.200 usesrc clientip
6078
6079 backend transparent_smtp
6080 # Connect to the SMTP farm from the client's source address/port
6081 # with Tproxy version 4.
6082 source 0.0.0.0 usesrc clientip
6083
Willy Tarreaubce70882009-09-07 11:51:47 +02006084 backend transparent_http
6085 # Connect to the servers using the client's IP as seen by previous
6086 # proxy.
6087 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
6088
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006089 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006090 the Linux kernel on www.balabit.com, the "bind" keyword.
6091
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006092
Willy Tarreau844e3c52008-01-11 16:28:18 +01006093srvtimeout <timeout> (deprecated)
6094 Set the maximum inactivity time on the server side.
6095 May be used in sections : defaults | frontend | listen | backend
6096 yes | no | yes | yes
6097 Arguments :
6098 <timeout> is the timeout value specified in milliseconds by default, but
6099 can be in any other unit if the number is suffixed by the unit,
6100 as explained at the top of this document.
6101
6102 The inactivity timeout applies when the server is expected to acknowledge or
6103 send data. In HTTP mode, this timeout is particularly important to consider
6104 during the first phase of the server's response, when it has to send the
6105 headers, as it directly represents the server's processing time for the
6106 request. To find out what value to put there, it's often good to start with
6107 what would be considered as unacceptable response times, then check the logs
6108 to observe the response time distribution, and adjust the value accordingly.
6109
6110 The value is specified in milliseconds by default, but can be in any other
6111 unit if the number is suffixed by the unit, as specified at the top of this
6112 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
6113 recommended that the client timeout remains equal to the server timeout in
6114 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006115 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01006116 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreaud72758d2010-01-12 10:42:19 +01006117 seconds (eg: 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01006118
6119 This parameter is specific to backends, but can be specified once for all in
6120 "defaults" sections. This is in fact one of the easiest solutions not to
6121 forget about it. An unspecified timeout results in an infinite timeout, which
6122 is not recommended. Such a usage is accepted and works but reports a warning
6123 during startup because it may results in accumulation of expired sessions in
6124 the system if the system's timeouts are not configured either.
6125
6126 This parameter is provided for compatibility but is currently deprecated.
6127 Please use "timeout server" instead.
6128
Willy Tarreauce887fd2012-05-12 12:50:00 +02006129 See also : "timeout server", "timeout tunnel", "timeout client" and
6130 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01006131
6132
Cyril Bonté66c327d2010-10-12 00:14:37 +02006133stats admin { if | unless } <cond>
6134 Enable statistics admin level if/unless a condition is matched
6135 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006136 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02006137
6138 This statement enables the statistics admin level if/unless a condition is
6139 matched.
6140
6141 The admin level allows to enable/disable servers from the web interface. By
6142 default, statistics page is read-only for security reasons.
6143
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006144 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6145 unless you know what you do : memory is not shared between the
6146 processes, which can result in random behaviours.
6147
Cyril Bonté23b39d92011-02-10 22:54:44 +01006148 Currently, the POST request is limited to the buffer size minus the reserved
6149 buffer space, which means that if the list of servers is too long, the
6150 request won't be processed. It is recommended to alter few servers at a
6151 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006152
6153 Example :
6154 # statistics admin level only for localhost
6155 backend stats_localhost
6156 stats enable
6157 stats admin if LOCALHOST
6158
6159 Example :
6160 # statistics admin level always enabled because of the authentication
6161 backend stats_auth
6162 stats enable
6163 stats auth admin:AdMiN123
6164 stats admin if TRUE
6165
6166 Example :
6167 # statistics admin level depends on the authenticated user
6168 userlist stats-auth
6169 group admin users admin
6170 user admin insecure-password AdMiN123
6171 group readonly users haproxy
6172 user haproxy insecure-password haproxy
6173
6174 backend stats_auth
6175 stats enable
6176 acl AUTH http_auth(stats-auth)
6177 acl AUTH_ADMIN http_auth_group(stats-auth) admin
6178 stats http-request auth unless AUTH
6179 stats admin if AUTH_ADMIN
6180
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006181 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
6182 "bind-process", section 3.4 about userlists and section 7 about
6183 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006184
6185
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006186stats auth <user>:<passwd>
6187 Enable statistics with authentication and grant access to an account
6188 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006189 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006190 Arguments :
6191 <user> is a user name to grant access to
6192
6193 <passwd> is the cleartext password associated to this user
6194
6195 This statement enables statistics with default settings, and restricts access
6196 to declared users only. It may be repeated as many times as necessary to
6197 allow as many users as desired. When a user tries to access the statistics
6198 without a valid account, a "401 Forbidden" response will be returned so that
6199 the browser asks the user to provide a valid user and password. The real
6200 which will be returned to the browser is configurable using "stats realm".
6201
6202 Since the authentication method is HTTP Basic Authentication, the passwords
6203 circulate in cleartext on the network. Thus, it was decided that the
6204 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006205 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006206
6207 It is also possible to reduce the scope of the proxies which appear in the
6208 report using "stats scope".
6209
6210 Though this statement alone is enough to enable statistics reporting, it is
6211 recommended to set all other settings in order to avoid relying on default
6212 unobvious parameters.
6213
6214 Example :
6215 # public access (limited to this backend only)
6216 backend public_www
6217 server srv1 192.168.0.1:80
6218 stats enable
6219 stats hide-version
6220 stats scope .
6221 stats uri /admin?stats
6222 stats realm Haproxy\ Statistics
6223 stats auth admin1:AdMiN123
6224 stats auth admin2:AdMiN321
6225
6226 # internal monitoring access (unlimited)
6227 backend private_monitoring
6228 stats enable
6229 stats uri /admin?stats
6230 stats refresh 5s
6231
6232 See also : "stats enable", "stats realm", "stats scope", "stats uri"
6233
6234
6235stats enable
6236 Enable statistics reporting with default settings
6237 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006238 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006239 Arguments : none
6240
6241 This statement enables statistics reporting with default settings defined
6242 at build time. Unless stated otherwise, these settings are used :
6243 - stats uri : /haproxy?stats
6244 - stats realm : "HAProxy Statistics"
6245 - stats auth : no authentication
6246 - stats scope : no restriction
6247
6248 Though this statement alone is enough to enable statistics reporting, it is
6249 recommended to set all other settings in order to avoid relying on default
6250 unobvious parameters.
6251
6252 Example :
6253 # public access (limited to this backend only)
6254 backend public_www
6255 server srv1 192.168.0.1:80
6256 stats enable
6257 stats hide-version
6258 stats scope .
6259 stats uri /admin?stats
6260 stats realm Haproxy\ Statistics
6261 stats auth admin1:AdMiN123
6262 stats auth admin2:AdMiN321
6263
6264 # internal monitoring access (unlimited)
6265 backend private_monitoring
6266 stats enable
6267 stats uri /admin?stats
6268 stats refresh 5s
6269
6270 See also : "stats auth", "stats realm", "stats uri"
6271
6272
Willy Tarreaud63335a2010-02-26 12:56:52 +01006273stats hide-version
6274 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006275 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006276 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006277 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006278
Willy Tarreaud63335a2010-02-26 12:56:52 +01006279 By default, the stats page reports some useful status information along with
6280 the statistics. Among them is HAProxy's version. However, it is generally
6281 considered dangerous to report precise version to anyone, as it can help them
6282 target known weaknesses with specific attacks. The "stats hide-version"
6283 statement removes the version from the statistics report. This is recommended
6284 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006285
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006286 Though this statement alone is enough to enable statistics reporting, it is
6287 recommended to set all other settings in order to avoid relying on default
6288 unobvious parameters.
6289
Willy Tarreaud63335a2010-02-26 12:56:52 +01006290 Example :
6291 # public access (limited to this backend only)
6292 backend public_www
6293 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006294 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006295 stats hide-version
6296 stats scope .
6297 stats uri /admin?stats
6298 stats realm Haproxy\ Statistics
6299 stats auth admin1:AdMiN123
6300 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006301
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006302 # internal monitoring access (unlimited)
6303 backend private_monitoring
6304 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006305 stats uri /admin?stats
6306 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01006307
Willy Tarreaud63335a2010-02-26 12:56:52 +01006308 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006309
Willy Tarreau983e01e2010-01-11 18:42:06 +01006310
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02006311stats http-request { allow | deny | auth [realm <realm>] }
6312 [ { if | unless } <condition> ]
6313 Access control for statistics
6314
6315 May be used in sections: defaults | frontend | listen | backend
6316 no | no | yes | yes
6317
6318 As "http-request", these set of options allow to fine control access to
6319 statistics. Each option may be followed by if/unless and acl.
6320 First option with matched condition (or option without condition) is final.
6321 For "deny" a 403 error will be returned, for "allow" normal processing is
6322 performed, for "auth" a 401/407 error code is returned so the client
6323 should be asked to enter a username and password.
6324
6325 There is no fixed limit to the number of http-request statements per
6326 instance.
6327
6328 See also : "http-request", section 3.4 about userlists and section 7
6329 about ACL usage.
6330
6331
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006332stats realm <realm>
6333 Enable statistics and set authentication realm
6334 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006335 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006336 Arguments :
6337 <realm> is the name of the HTTP Basic Authentication realm reported to
6338 the browser. The browser uses it to display it in the pop-up
6339 inviting the user to enter a valid username and password.
6340
6341 The realm is read as a single word, so any spaces in it should be escaped
6342 using a backslash ('\').
6343
6344 This statement is useful only in conjunction with "stats auth" since it is
6345 only related to authentication.
6346
6347 Though this statement alone is enough to enable statistics reporting, it is
6348 recommended to set all other settings in order to avoid relying on default
6349 unobvious parameters.
6350
6351 Example :
6352 # public access (limited to this backend only)
6353 backend public_www
6354 server srv1 192.168.0.1:80
6355 stats enable
6356 stats hide-version
6357 stats scope .
6358 stats uri /admin?stats
6359 stats realm Haproxy\ Statistics
6360 stats auth admin1:AdMiN123
6361 stats auth admin2:AdMiN321
6362
6363 # internal monitoring access (unlimited)
6364 backend private_monitoring
6365 stats enable
6366 stats uri /admin?stats
6367 stats refresh 5s
6368
6369 See also : "stats auth", "stats enable", "stats uri"
6370
6371
6372stats refresh <delay>
6373 Enable statistics with automatic refresh
6374 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006375 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006376 Arguments :
6377 <delay> is the suggested refresh delay, specified in seconds, which will
6378 be returned to the browser consulting the report page. While the
6379 browser is free to apply any delay, it will generally respect it
6380 and refresh the page this every seconds. The refresh interval may
6381 be specified in any other non-default time unit, by suffixing the
6382 unit after the value, as explained at the top of this document.
6383
6384 This statement is useful on monitoring displays with a permanent page
6385 reporting the load balancer's activity. When set, the HTML report page will
6386 include a link "refresh"/"stop refresh" so that the user can select whether
6387 he wants automatic refresh of the page or not.
6388
6389 Though this statement alone is enough to enable statistics reporting, it is
6390 recommended to set all other settings in order to avoid relying on default
6391 unobvious parameters.
6392
6393 Example :
6394 # public access (limited to this backend only)
6395 backend public_www
6396 server srv1 192.168.0.1:80
6397 stats enable
6398 stats hide-version
6399 stats scope .
6400 stats uri /admin?stats
6401 stats realm Haproxy\ Statistics
6402 stats auth admin1:AdMiN123
6403 stats auth admin2:AdMiN321
6404
6405 # internal monitoring access (unlimited)
6406 backend private_monitoring
6407 stats enable
6408 stats uri /admin?stats
6409 stats refresh 5s
6410
6411 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6412
6413
6414stats scope { <name> | "." }
6415 Enable statistics and limit access scope
6416 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006417 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006418 Arguments :
6419 <name> is the name of a listen, frontend or backend section to be
6420 reported. The special name "." (a single dot) designates the
6421 section in which the statement appears.
6422
6423 When this statement is specified, only the sections enumerated with this
6424 statement will appear in the report. All other ones will be hidden. This
6425 statement may appear as many times as needed if multiple sections need to be
6426 reported. Please note that the name checking is performed as simple string
6427 comparisons, and that it is never checked that a give section name really
6428 exists.
6429
6430 Though this statement alone is enough to enable statistics reporting, it is
6431 recommended to set all other settings in order to avoid relying on default
6432 unobvious parameters.
6433
6434 Example :
6435 # public access (limited to this backend only)
6436 backend public_www
6437 server srv1 192.168.0.1:80
6438 stats enable
6439 stats hide-version
6440 stats scope .
6441 stats uri /admin?stats
6442 stats realm Haproxy\ Statistics
6443 stats auth admin1:AdMiN123
6444 stats auth admin2:AdMiN321
6445
6446 # internal monitoring access (unlimited)
6447 backend private_monitoring
6448 stats enable
6449 stats uri /admin?stats
6450 stats refresh 5s
6451
6452 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6453
Willy Tarreaud63335a2010-02-26 12:56:52 +01006454
Willy Tarreauc9705a12010-07-27 20:05:50 +02006455stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01006456 Enable reporting of a description on the statistics page.
6457 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006458 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006459
Willy Tarreauc9705a12010-07-27 20:05:50 +02006460 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01006461 description from global section is automatically used instead.
6462
6463 This statement is useful for users that offer shared services to their
6464 customers, where node or description should be different for each customer.
6465
6466 Though this statement alone is enough to enable statistics reporting, it is
6467 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006468 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006469
6470 Example :
6471 # internal monitoring access (unlimited)
6472 backend private_monitoring
6473 stats enable
6474 stats show-desc Master node for Europe, Asia, Africa
6475 stats uri /admin?stats
6476 stats refresh 5s
6477
6478 See also: "show-node", "stats enable", "stats uri" and "description" in
6479 global section.
6480
6481
6482stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02006483 Enable reporting additional information on the statistics page
6484 May be used in sections : defaults | frontend | listen | backend
6485 yes | yes | yes | yes
6486 Arguments : none
6487
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006488 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01006489 - cap: capabilities (proxy)
6490 - mode: one of tcp, http or health (proxy)
6491 - id: SNMP ID (proxy, socket, server)
6492 - IP (socket, server)
6493 - cookie (backend, server)
6494
6495 Though this statement alone is enough to enable statistics reporting, it is
6496 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006497 unobvious parameters. Default behaviour is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006498
6499 See also: "stats enable", "stats uri".
6500
6501
6502stats show-node [ <name> ]
6503 Enable reporting of a host name on the statistics page.
6504 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006505 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006506 Arguments:
6507 <name> is an optional name to be reported. If unspecified, the
6508 node name from global section is automatically used instead.
6509
6510 This statement is useful for users that offer shared services to their
6511 customers, where node or description might be different on a stats page
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006512 provided for each customer. Default behaviour is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006513
6514 Though this statement alone is enough to enable statistics reporting, it is
6515 recommended to set all other settings in order to avoid relying on default
6516 unobvious parameters.
6517
6518 Example:
6519 # internal monitoring access (unlimited)
6520 backend private_monitoring
6521 stats enable
6522 stats show-node Europe-1
6523 stats uri /admin?stats
6524 stats refresh 5s
6525
6526 See also: "show-desc", "stats enable", "stats uri", and "node" in global
6527 section.
6528
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006529
6530stats uri <prefix>
6531 Enable statistics and define the URI prefix to access them
6532 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006533 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006534 Arguments :
6535 <prefix> is the prefix of any URI which will be redirected to stats. This
6536 prefix may contain a question mark ('?') to indicate part of a
6537 query string.
6538
6539 The statistics URI is intercepted on the relayed traffic, so it appears as a
6540 page within the normal application. It is strongly advised to ensure that the
6541 selected URI will never appear in the application, otherwise it will never be
6542 possible to reach it in the application.
6543
6544 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006545 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006546 It is generally a good idea to include a question mark in the URI so that
6547 intermediate proxies refrain from caching the results. Also, since any string
6548 beginning with the prefix will be accepted as a stats request, the question
6549 mark helps ensuring that no valid URI will begin with the same words.
6550
6551 It is sometimes very convenient to use "/" as the URI prefix, and put that
6552 statement in a "listen" instance of its own. That makes it easy to dedicate
6553 an address or a port to statistics only.
6554
6555 Though this statement alone is enough to enable statistics reporting, it is
6556 recommended to set all other settings in order to avoid relying on default
6557 unobvious parameters.
6558
6559 Example :
6560 # public access (limited to this backend only)
6561 backend public_www
6562 server srv1 192.168.0.1:80
6563 stats enable
6564 stats hide-version
6565 stats scope .
6566 stats uri /admin?stats
6567 stats realm Haproxy\ Statistics
6568 stats auth admin1:AdMiN123
6569 stats auth admin2:AdMiN321
6570
6571 # internal monitoring access (unlimited)
6572 backend private_monitoring
6573 stats enable
6574 stats uri /admin?stats
6575 stats refresh 5s
6576
6577 See also : "stats auth", "stats enable", "stats realm"
6578
6579
Willy Tarreaud63335a2010-02-26 12:56:52 +01006580stick match <pattern> [table <table>] [{if | unless} <cond>]
6581 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006582 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01006583 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006584
6585 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006586 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006587 describes what elements of the incoming request or connection
6588 will be analysed in the hope to find a matching entry in a
6589 stickiness table. This rule is mandatory.
6590
6591 <table> is an optional stickiness table name. If unspecified, the same
6592 backend's table is used. A stickiness table is declared using
6593 the "stick-table" statement.
6594
6595 <cond> is an optional matching condition. It makes it possible to match
6596 on a certain criterion only when other conditions are met (or
6597 not met). For instance, it could be used to match on a source IP
6598 address except when a request passes through a known proxy, in
6599 which case we'd match on a header containing that IP address.
6600
6601 Some protocols or applications require complex stickiness rules and cannot
6602 always simply rely on cookies nor hashing. The "stick match" statement
6603 describes a rule to extract the stickiness criterion from an incoming request
6604 or connection. See section 7 for a complete list of possible patterns and
6605 transformation rules.
6606
6607 The table has to be declared using the "stick-table" statement. It must be of
6608 a type compatible with the pattern. By default it is the one which is present
6609 in the same backend. It is possible to share a table with other backends by
6610 referencing it using the "table" keyword. If another table is referenced,
6611 the server's ID inside the backends are used. By default, all server IDs
6612 start at 1 in each backend, so the server ordering is enough. But in case of
6613 doubt, it is highly recommended to force server IDs using their "id" setting.
6614
6615 It is possible to restrict the conditions where a "stick match" statement
6616 will apply, using "if" or "unless" followed by a condition. See section 7 for
6617 ACL based conditions.
6618
6619 There is no limit on the number of "stick match" statements. The first that
6620 applies and matches will cause the request to be directed to the same server
6621 as was used for the request which created the entry. That way, multiple
6622 matches can be used as fallbacks.
6623
6624 The stick rules are checked after the persistence cookies, so they will not
6625 affect stickiness if a cookie has already been used to select a server. That
6626 way, it becomes very easy to insert cookies and match on IP addresses in
6627 order to maintain stickiness between HTTP and HTTPS.
6628
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006629 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6630 unless you know what you do : memory is not shared between the
6631 processes, which can result in random behaviours.
6632
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006633 Example :
6634 # forward SMTP users to the same server they just used for POP in the
6635 # last 30 minutes
6636 backend pop
6637 mode tcp
6638 balance roundrobin
6639 stick store-request src
6640 stick-table type ip size 200k expire 30m
6641 server s1 192.168.1.1:110
6642 server s2 192.168.1.1:110
6643
6644 backend smtp
6645 mode tcp
6646 balance roundrobin
6647 stick match src table pop
6648 server s1 192.168.1.1:25
6649 server s2 192.168.1.1:25
6650
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006651 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006652 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006653
6654
6655stick on <pattern> [table <table>] [{if | unless} <condition>]
6656 Define a request pattern to associate a user to a server
6657 May be used in sections : defaults | frontend | listen | backend
6658 no | no | yes | yes
6659
6660 Note : This form is exactly equivalent to "stick match" followed by
6661 "stick store-request", all with the same arguments. Please refer
6662 to both keywords for details. It is only provided as a convenience
6663 for writing more maintainable configurations.
6664
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006665 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6666 unless you know what you do : memory is not shared between the
6667 processes, which can result in random behaviours.
6668
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006669 Examples :
6670 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01006671 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006672
6673 # ...is strictly equivalent to this one :
6674 stick match src table pop if !localhost
6675 stick store-request src table pop if !localhost
6676
6677
6678 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
6679 # well as HTTP without cookie. Share the same table between both accesses.
6680 backend http
6681 mode http
6682 balance roundrobin
6683 stick on src table https
6684 cookie SRV insert indirect nocache
6685 server s1 192.168.1.1:80 cookie s1
6686 server s2 192.168.1.1:80 cookie s2
6687
6688 backend https
6689 mode tcp
6690 balance roundrobin
6691 stick-table type ip size 200k expire 30m
6692 stick on src
6693 server s1 192.168.1.1:443
6694 server s2 192.168.1.1:443
6695
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006696 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006697
6698
6699stick store-request <pattern> [table <table>] [{if | unless} <condition>]
6700 Define a request pattern used to create an entry in a stickiness table
6701 May be used in sections : defaults | frontend | listen | backend
6702 no | no | yes | yes
6703
6704 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006705 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006706 describes what elements of the incoming request or connection
6707 will be analysed, extracted and stored in the table once a
6708 server is selected.
6709
6710 <table> is an optional stickiness table name. If unspecified, the same
6711 backend's table is used. A stickiness table is declared using
6712 the "stick-table" statement.
6713
6714 <cond> is an optional storage condition. It makes it possible to store
6715 certain criteria only when some conditions are met (or not met).
6716 For instance, it could be used to store the source IP address
6717 except when the request passes through a known proxy, in which
6718 case we'd store a converted form of a header containing that IP
6719 address.
6720
6721 Some protocols or applications require complex stickiness rules and cannot
6722 always simply rely on cookies nor hashing. The "stick store-request" statement
6723 describes a rule to decide what to extract from the request and when to do
6724 it, in order to store it into a stickiness table for further requests to
6725 match it using the "stick match" statement. Obviously the extracted part must
6726 make sense and have a chance to be matched in a further request. Storing a
6727 client's IP address for instance often makes sense. Storing an ID found in a
6728 URL parameter also makes sense. Storing a source port will almost never make
6729 any sense because it will be randomly matched. See section 7 for a complete
6730 list of possible patterns and transformation rules.
6731
6732 The table has to be declared using the "stick-table" statement. It must be of
6733 a type compatible with the pattern. By default it is the one which is present
6734 in the same backend. It is possible to share a table with other backends by
6735 referencing it using the "table" keyword. If another table is referenced,
6736 the server's ID inside the backends are used. By default, all server IDs
6737 start at 1 in each backend, so the server ordering is enough. But in case of
6738 doubt, it is highly recommended to force server IDs using their "id" setting.
6739
6740 It is possible to restrict the conditions where a "stick store-request"
6741 statement will apply, using "if" or "unless" followed by a condition. This
6742 condition will be evaluated while parsing the request, so any criteria can be
6743 used. See section 7 for ACL based conditions.
6744
6745 There is no limit on the number of "stick store-request" statements, but
6746 there is a limit of 8 simultaneous stores per request or response. This
6747 makes it possible to store up to 8 criteria, all extracted from either the
6748 request or the response, regardless of the number of rules. Only the 8 first
6749 ones which match will be kept. Using this, it is possible to feed multiple
6750 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01006751 another protocol or access method. Using multiple store-request rules with
6752 the same table is possible and may be used to find the best criterion to rely
6753 on, by arranging the rules by decreasing preference order. Only the first
6754 extracted criterion for a given table will be stored. All subsequent store-
6755 request rules referencing the same table will be skipped and their ACLs will
6756 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006757
6758 The "store-request" rules are evaluated once the server connection has been
6759 established, so that the table will contain the real server that processed
6760 the request.
6761
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006762 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6763 unless you know what you do : memory is not shared between the
6764 processes, which can result in random behaviours.
6765
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006766 Example :
6767 # forward SMTP users to the same server they just used for POP in the
6768 # last 30 minutes
6769 backend pop
6770 mode tcp
6771 balance roundrobin
6772 stick store-request src
6773 stick-table type ip size 200k expire 30m
6774 server s1 192.168.1.1:110
6775 server s2 192.168.1.1:110
6776
6777 backend smtp
6778 mode tcp
6779 balance roundrobin
6780 stick match src table pop
6781 server s1 192.168.1.1:25
6782 server s2 192.168.1.1:25
6783
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006784 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006785 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006786
6787
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006788stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02006789 size <size> [expire <expire>] [nopurge] [peers <peersect>]
6790 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08006791 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006792 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02006793 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006794
6795 Arguments :
6796 ip a table declared with "type ip" will only store IPv4 addresses.
6797 This form is very compact (about 50 bytes per entry) and allows
6798 very fast entry lookup and stores with almost no overhead. This
6799 is mainly used to store client source IP addresses.
6800
David du Colombier9a6d3c92011-03-17 10:40:24 +01006801 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
6802 This form is very compact (about 60 bytes per entry) and allows
6803 very fast entry lookup and stores with almost no overhead. This
6804 is mainly used to store client source IP addresses.
6805
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006806 integer a table declared with "type integer" will store 32bit integers
6807 which can represent a client identifier found in a request for
6808 instance.
6809
6810 string a table declared with "type string" will store substrings of up
6811 to <len> characters. If the string provided by the pattern
6812 extractor is larger than <len>, it will be truncated before
6813 being stored. During matching, at most <len> characters will be
6814 compared between the string in the table and the extracted
6815 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006816 to 32 characters.
6817
6818 binary a table declared with "type binary" will store binary blocks
6819 of <len> bytes. If the block provided by the pattern
6820 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02006821 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006822 is shorter than <len>, it will be padded by 0. When not
6823 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006824
6825 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006826 "string" type table (See type "string" above). Or the number
6827 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006828 changing this parameter as memory usage will proportionally
6829 increase.
6830
6831 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01006832 value directly impacts memory usage. Count approximately
6833 50 bytes per entry, plus the size of a string if any. The size
6834 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006835
6836 [nopurge] indicates that we refuse to purge older entries when the table
6837 is full. When not specified and the table is full when haproxy
6838 wants to store an entry in it, it will flush a few of the oldest
6839 entries in order to release some space for the new ones. This is
6840 most often the desired behaviour. In some specific cases, it
6841 be desirable to refuse new entries instead of purging the older
6842 ones. That may be the case when the amount of data to store is
6843 far above the hardware limits and we prefer not to offer access
6844 to new clients than to reject the ones already connected. When
6845 using this parameter, be sure to properly set the "expire"
6846 parameter (see below).
6847
Emeric Brunf099e792010-09-27 12:05:28 +02006848 <peersect> is the name of the peers section to use for replication. Entries
6849 which associate keys to server IDs are kept synchronized with
6850 the remote peers declared in this section. All entries are also
6851 automatically learned from the local peer (old process) during a
6852 soft restart.
6853
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006854 NOTE : peers can't be used in multi-process mode.
6855
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006856 <expire> defines the maximum duration of an entry in the table since it
6857 was last created, refreshed or matched. The expiration delay is
6858 defined using the standard time format, similarly as the various
6859 timeouts. The maximum duration is slightly above 24 days. See
6860 section 2.2 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02006861 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006862 be removed once full. Be sure not to use the "nopurge" parameter
6863 if not expiration delay is specified.
6864
Willy Tarreau08d5f982010-06-06 13:34:54 +02006865 <data_type> is used to store additional information in the stick-table. This
6866 may be used by ACLs in order to control various criteria related
6867 to the activity of the client matching the stick-table. For each
6868 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02006869 that the additional data can fit. Several data types may be
6870 stored with an entry. Multiple data types may be specified after
6871 the "store" keyword, as a comma-separated list. Alternatively,
6872 it is possible to repeat the "store" keyword followed by one or
6873 several data types. Except for the "server_id" type which is
6874 automatically detected and enabled, all data types must be
6875 explicitly declared to be stored. If an ACL references a data
6876 type which is not stored, the ACL will simply not match. Some
6877 data types require an argument which must be passed just after
6878 the type between parenthesis. See below for the supported data
6879 types and their arguments.
6880
6881 The data types that can be stored with an entry are the following :
6882 - server_id : this is an integer which holds the numeric ID of the server a
6883 request was assigned to. It is used by the "stick match", "stick store",
6884 and "stick on" rules. It is automatically enabled when referenced.
6885
6886 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
6887 integer which may be used for anything. Most of the time it will be used
6888 to put a special tag on some entries, for instance to note that a
6889 specific behaviour was detected and must be known for future matches.
6890
Willy Tarreauba2ffd12013-05-29 15:54:14 +02006891 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
6892 over a period. It is a positive 32-bit integer integer which may be used
6893 for anything. Just like <gpc0>, it counts events, but instead of keeping
6894 a cumulative count, it maintains the rate at which the counter is
6895 incremented. Most of the time it will be used to measure the frequency of
6896 occurrence of certain events (eg: requests to a specific URL).
6897
Willy Tarreauc9705a12010-07-27 20:05:50 +02006898 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
6899 the absolute number of connections received from clients which matched
6900 this entry. It does not mean the connections were accepted, just that
6901 they were received.
6902
6903 - conn_cur : Current Connections. It is a positive 32-bit integer which
6904 stores the concurrent connection counts for the entry. It is incremented
6905 once an incoming connection matches the entry, and decremented once the
6906 connection leaves. That way it is possible to know at any time the exact
6907 number of concurrent connections for an entry.
6908
6909 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6910 integer parameter <period> which indicates in milliseconds the length
6911 of the period over which the average is measured. It reports the average
6912 incoming connection rate over that period, in connections per period. The
6913 result is an integer which can be matched using ACLs.
6914
6915 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
6916 the absolute number of sessions received from clients which matched this
6917 entry. A session is a connection that was accepted by the layer 4 rules.
6918
6919 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6920 integer parameter <period> which indicates in milliseconds the length
6921 of the period over which the average is measured. It reports the average
6922 incoming session rate over that period, in sessions per period. The
6923 result is an integer which can be matched using ACLs.
6924
6925 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
6926 counts the absolute number of HTTP requests received from clients which
6927 matched this entry. It does not matter whether they are valid requests or
6928 not. Note that this is different from sessions when keep-alive is used on
6929 the client side.
6930
6931 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6932 integer parameter <period> which indicates in milliseconds the length
6933 of the period over which the average is measured. It reports the average
6934 HTTP request rate over that period, in requests per period. The result is
6935 an integer which can be matched using ACLs. It does not matter whether
6936 they are valid requests or not. Note that this is different from sessions
6937 when keep-alive is used on the client side.
6938
6939 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
6940 counts the absolute number of HTTP requests errors induced by clients
6941 which matched this entry. Errors are counted on invalid and truncated
6942 requests, as well as on denied or tarpitted requests, and on failed
6943 authentications. If the server responds with 4xx, then the request is
6944 also counted as an error since it's an error triggered by the client
6945 (eg: vulnerability scan).
6946
6947 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6948 integer parameter <period> which indicates in milliseconds the length
6949 of the period over which the average is measured. It reports the average
6950 HTTP request error rate over that period, in requests per period (see
6951 http_err_cnt above for what is accounted as an error). The result is an
6952 integer which can be matched using ACLs.
6953
6954 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
6955 integer which counts the cumulated amount of bytes received from clients
6956 which matched this entry. Headers are included in the count. This may be
6957 used to limit abuse of upload features on photo or video servers.
6958
6959 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
6960 integer parameter <period> which indicates in milliseconds the length
6961 of the period over which the average is measured. It reports the average
6962 incoming bytes rate over that period, in bytes per period. It may be used
6963 to detect users which upload too much and too fast. Warning: with large
6964 uploads, it is possible that the amount of uploaded data will be counted
6965 once upon termination, thus causing spikes in the average transfer speed
6966 instead of having a smooth one. This may partially be smoothed with
6967 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
6968 recommended for better fairness.
6969
6970 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
6971 integer which counts the cumulated amount of bytes sent to clients which
6972 matched this entry. Headers are included in the count. This may be used
6973 to limit abuse of bots sucking the whole site.
6974
6975 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
6976 an integer parameter <period> which indicates in milliseconds the length
6977 of the period over which the average is measured. It reports the average
6978 outgoing bytes rate over that period, in bytes per period. It may be used
6979 to detect users which download too much and too fast. Warning: with large
6980 transfers, it is possible that the amount of transferred data will be
6981 counted once upon termination, thus causing spikes in the average
6982 transfer speed instead of having a smooth one. This may partially be
6983 smoothed with "option contstats" though this is not perfect yet. Use of
6984 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02006985
Willy Tarreauc00cdc22010-06-06 16:48:26 +02006986 There is only one stick-table per proxy. At the moment of writing this doc,
6987 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006988 to be required, simply create a dummy backend with a stick-table in it and
6989 reference it.
6990
6991 It is important to understand that stickiness based on learning information
6992 has some limitations, including the fact that all learned associations are
6993 lost upon restart. In general it can be good as a complement but not always
6994 as an exclusive stickiness.
6995
Willy Tarreauc9705a12010-07-27 20:05:50 +02006996 Last, memory requirements may be important when storing many data types.
6997 Indeed, storing all indicators above at once in each entry requires 116 bytes
6998 per entry, or 116 MB for a 1-million entries table. This is definitely not
6999 something that can be ignored.
7000
7001 Example:
7002 # Keep track of counters of up to 1 million IP addresses over 5 minutes
7003 # and store a general purpose counter and the average connection rate
7004 # computed over a sliding window of 30 seconds.
7005 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
7006
7007 See also : "stick match", "stick on", "stick store-request", section 2.2
David du Colombiera13d1b92011-03-17 10:40:22 +01007008 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007009
7010
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007011stick store-response <pattern> [table <table>] [{if | unless} <condition>]
7012 Define a request pattern used to create an entry in a stickiness table
7013 May be used in sections : defaults | frontend | listen | backend
7014 no | no | yes | yes
7015
7016 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007017 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007018 describes what elements of the response or connection will
7019 be analysed, extracted and stored in the table once a
7020 server is selected.
7021
7022 <table> is an optional stickiness table name. If unspecified, the same
7023 backend's table is used. A stickiness table is declared using
7024 the "stick-table" statement.
7025
7026 <cond> is an optional storage condition. It makes it possible to store
7027 certain criteria only when some conditions are met (or not met).
7028 For instance, it could be used to store the SSL session ID only
7029 when the response is a SSL server hello.
7030
7031 Some protocols or applications require complex stickiness rules and cannot
7032 always simply rely on cookies nor hashing. The "stick store-response"
7033 statement describes a rule to decide what to extract from the response and
7034 when to do it, in order to store it into a stickiness table for further
7035 requests to match it using the "stick match" statement. Obviously the
7036 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007037 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007038 See section 7 for a complete list of possible patterns and transformation
7039 rules.
7040
7041 The table has to be declared using the "stick-table" statement. It must be of
7042 a type compatible with the pattern. By default it is the one which is present
7043 in the same backend. It is possible to share a table with other backends by
7044 referencing it using the "table" keyword. If another table is referenced,
7045 the server's ID inside the backends are used. By default, all server IDs
7046 start at 1 in each backend, so the server ordering is enough. But in case of
7047 doubt, it is highly recommended to force server IDs using their "id" setting.
7048
7049 It is possible to restrict the conditions where a "stick store-response"
7050 statement will apply, using "if" or "unless" followed by a condition. This
7051 condition will be evaluated while parsing the response, so any criteria can
7052 be used. See section 7 for ACL based conditions.
7053
7054 There is no limit on the number of "stick store-response" statements, but
7055 there is a limit of 8 simultaneous stores per request or response. This
7056 makes it possible to store up to 8 criteria, all extracted from either the
7057 request or the response, regardless of the number of rules. Only the 8 first
7058 ones which match will be kept. Using this, it is possible to feed multiple
7059 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01007060 another protocol or access method. Using multiple store-response rules with
7061 the same table is possible and may be used to find the best criterion to rely
7062 on, by arranging the rules by decreasing preference order. Only the first
7063 extracted criterion for a given table will be stored. All subsequent store-
7064 response rules referencing the same table will be skipped and their ACLs will
7065 not be evaluated. However, even if a store-request rule references a table, a
7066 store-response rule may also use the same table. This means that each table
7067 may learn exactly one element from the request and one element from the
7068 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007069
7070 The table will contain the real server that processed the request.
7071
7072 Example :
7073 # Learn SSL session ID from both request and response and create affinity.
7074 backend https
7075 mode tcp
7076 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02007077 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007078 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007079
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007080 acl clienthello req_ssl_hello_type 1
7081 acl serverhello rep_ssl_hello_type 2
7082
7083 # use tcp content accepts to detects ssl client and server hello.
7084 tcp-request inspect-delay 5s
7085 tcp-request content accept if clienthello
7086
7087 # no timeout on response inspect delay by default.
7088 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007089
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007090 # SSL session ID (SSLID) may be present on a client or server hello.
7091 # Its length is coded on 1 byte at offset 43 and its value starts
7092 # at offset 44.
7093
7094 # Match and learn on request if client hello.
7095 stick on payload_lv(43,1) if clienthello
7096
7097 # Learn on response if server hello.
7098 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02007099
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007100 server s1 192.168.1.1:443
7101 server s2 192.168.1.1:443
7102
7103 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
7104 extraction.
7105
7106
Willy Tarreau938c7fe2014-04-25 14:21:39 +02007107tcp-check connect [params*]
7108 Opens a new connection
7109 May be used in sections: defaults | frontend | listen | backend
7110 no | no | yes | yes
7111
7112 When an application lies on more than a single TCP port or when HAProxy
7113 load-balance many services in a single backend, it makes sense to probe all
7114 the services individually before considering a server as operational.
7115
7116 When there are no TCP port configured on the server line neither server port
7117 directive, then the 'tcp-check connect port <port>' must be the first step
7118 of the sequence.
7119
7120 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
7121 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
7122 do.
7123
7124 Parameters :
7125 They are optional and can be used to describe how HAProxy should open and
7126 use the TCP connection.
7127
7128 port if not set, check port or server port is used.
7129 It tells HAProxy where to open the connection to.
7130 <port> must be a valid TCP port source integer, from 1 to 65535.
7131
7132 send-proxy send a PROXY protocol string
7133
7134 ssl opens a ciphered connection
7135
7136 Examples:
7137 # check HTTP and HTTPs services on a server.
7138 # first open port 80 thanks to server line port directive, then
7139 # tcp-check opens port 443, ciphered and run a request on it:
7140 option tcp-check
7141 tcp-check connect
7142 tcp-check send GET\ /\ HTTP/1.0\r\n
7143 tcp-check send Host:\ haproxy.1wt.eu\r\n
7144 tcp-check send \r\n
7145 tcp-check expect rstring (2..|3..)
7146 tcp-check connect port 443 ssl
7147 tcp-check send GET\ /\ HTTP/1.0\r\n
7148 tcp-check send Host:\ haproxy.1wt.eu\r\n
7149 tcp-check send \r\n
7150 tcp-check expect rstring (2..|3..)
7151 server www 10.0.0.1 check port 80
7152
7153 # check both POP and IMAP from a single server:
7154 option tcp-check
7155 tcp-check connect port 110
7156 tcp-check expect string +OK\ POP3\ ready
7157 tcp-check connect port 143
7158 tcp-check expect string *\ OK\ IMAP4\ ready
7159 server mail 10.0.0.1 check
7160
7161 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
7162
7163
7164tcp-check expect [!] <match> <pattern>
7165 Specify data to be collected and analysed during a generic health check
7166 May be used in sections: defaults | frontend | listen | backend
7167 no | no | yes | yes
7168
7169 Arguments :
7170 <match> is a keyword indicating how to look for a specific pattern in the
7171 response. The keyword may be one of "string", "rstring" or
7172 binary.
7173 The keyword may be preceded by an exclamation mark ("!") to negate
7174 the match. Spaces are allowed between the exclamation mark and the
7175 keyword. See below for more details on the supported keywords.
7176
7177 <pattern> is the pattern to look for. It may be a string or a regular
7178 expression. If the pattern contains spaces, they must be escaped
7179 with the usual backslash ('\').
7180 If the match is set to binary, then the pattern must be passed as
7181 a serie of hexadecimal digits in an even number. Each sequence of
7182 two digits will represent a byte. The hexadecimal digits may be
7183 used upper or lower case.
7184
7185
7186 The available matches are intentionally similar to their http-check cousins :
7187
7188 string <string> : test the exact string matches in the response buffer.
7189 A health check response will be considered valid if the
7190 response's buffer contains this exact string. If the
7191 "string" keyword is prefixed with "!", then the response
7192 will be considered invalid if the body contains this
7193 string. This can be used to look for a mandatory pattern
7194 in a protocol response, or to detect a failure when a
7195 specific error appears in a protocol banner.
7196
7197 rstring <regex> : test a regular expression on the response buffer.
7198 A health check response will be considered valid if the
7199 response's buffer matches this expression. If the
7200 "rstring" keyword is prefixed with "!", then the response
7201 will be considered invalid if the body matches the
7202 expression.
7203
7204 binary <hexstring> : test the exact string in its hexadecimal form matches
7205 in the response buffer. A health check response will
7206 be considered valid if the response's buffer contains
7207 this exact hexadecimal string.
7208 Purpose is to match data on binary protocols.
7209
7210 It is important to note that the responses will be limited to a certain size
7211 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
7212 Thus, too large responses may not contain the mandatory pattern when using
7213 "string", "rstring" or binary. If a large response is absolutely required, it
7214 is possible to change the default max size by setting the global variable.
7215 However, it is worth keeping in mind that parsing very large responses can
7216 waste some CPU cycles, especially when regular expressions are used, and that
7217 it is always better to focus the checks on smaller resources. Also, in its
7218 current state, the check will not find any string nor regex past a null
7219 character in the response. Similarly it is not possible to request matching
7220 the null character.
7221
7222 Examples :
7223 # perform a POP check
7224 option tcp-check
7225 tcp-check expect string +OK\ POP3\ ready
7226
7227 # perform an IMAP check
7228 option tcp-check
7229 tcp-check expect string *\ OK\ IMAP4\ ready
7230
7231 # look for the redis master server
7232 option tcp-check
7233 tcp-check send PING\r\n
7234 tcp-check expect +PONG
7235 tcp-check send info\ replication\r\n
7236 tcp-check expect string role:master
7237 tcp-check send QUIT\r\n
7238 tcp-check expect string +OK
7239
7240
7241 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
7242 "tcp-check send-binary", "http-check expect", tune.chksize
7243
7244
7245tcp-check send <data>
7246 Specify a string to be sent as a question during a generic health check
7247 May be used in sections: defaults | frontend | listen | backend
7248 no | no | yes | yes
7249
7250 <data> : the data to be sent as a question during a generic health check
7251 session. For now, <data> must be a string.
7252
7253 Examples :
7254 # look for the redis master server
7255 option tcp-check
7256 tcp-check send info\ replication\r\n
7257 tcp-check expect string role:master
7258
7259 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7260 "tcp-check send-binary", tune.chksize
7261
7262
7263tcp-check send-binary <hexastring>
7264 Specify an hexa digits string to be sent as a binary question during a raw
7265 tcp health check
7266 May be used in sections: defaults | frontend | listen | backend
7267 no | no | yes | yes
7268
7269 <data> : the data to be sent as a question during a generic health check
7270 session. For now, <data> must be a string.
7271 <hexastring> : test the exact string in its hexadecimal form matches in the
7272 response buffer. A health check response will be considered
7273 valid if the response's buffer contains this exact
7274 hexadecimal string.
7275 Purpose is to send binary data to ask on binary protocols.
7276
7277 Examples :
7278 # redis check in binary
7279 option tcp-check
7280 tcp-check send-binary 50494e470d0a # PING\r\n
7281 tcp-check expect binary 2b504F4e47 # +PONG
7282
7283
7284 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7285 "tcp-check send", tune.chksize
7286
7287
Willy Tarreaue9656522010-08-17 15:40:09 +02007288tcp-request connection <action> [{if | unless} <condition>]
7289 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02007290 May be used in sections : defaults | frontend | listen | backend
7291 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02007292 Arguments :
7293 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007294 actions include : "accept", "reject", "track-sc0", "track-sc1",
7295 "track-sc2", and "expect-proxy". See below for more details.
Willy Tarreau1a687942010-05-23 22:40:30 +02007296
Willy Tarreaue9656522010-08-17 15:40:09 +02007297 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007298
7299 Immediately after acceptance of a new incoming connection, it is possible to
7300 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02007301 or dropped or have its counters tracked. Those conditions cannot make use of
7302 any data contents because the connection has not been read from yet, and the
7303 buffers are not yet allocated. This is used to selectively and very quickly
7304 accept or drop connections from various sources with a very low overhead. If
7305 some contents need to be inspected in order to take the decision, the
7306 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007307
Willy Tarreaue9656522010-08-17 15:40:09 +02007308 The "tcp-request connection" rules are evaluated in their exact declaration
7309 order. If no rule matches or if there is no rule, the default action is to
7310 accept the incoming connection. There is no specific limit to the number of
7311 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007312
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007313 Five types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02007314 - accept :
7315 accepts the connection if the condition is true (when used with "if")
7316 or false (when used with "unless"). The first such rule executed ends
7317 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007318
Willy Tarreaue9656522010-08-17 15:40:09 +02007319 - reject :
7320 rejects the connection if the condition is true (when used with "if")
7321 or false (when used with "unless"). The first such rule executed ends
7322 the rules evaluation. Rejected connections do not even become a
7323 session, which is why they are accounted separately for in the stats,
7324 as "denied connections". They are not considered for the session
7325 rate-limit and are not logged either. The reason is that these rules
7326 should only be used to filter extremely high connection rates such as
7327 the ones encountered during a massive DDoS attack. Under these extreme
7328 conditions, the simple action of logging each event would make the
7329 system collapse and would considerably lower the filtering capacity. If
7330 logging is absolutely desired, then "tcp-request content" rules should
7331 be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007332
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007333 - expect-proxy layer4 :
7334 configures the client-facing connection to receive a PROXY protocol
7335 header before any byte is read from the socket. This is equivalent to
7336 having the "accept-proxy" keyword on the "bind" line, except that using
7337 the TCP rule allows the PROXY protocol to be accepted only for certain
7338 IP address ranges using an ACL. This is convenient when multiple layers
7339 of load balancers are passed through by traffic coming from public
7340 hosts.
7341
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007342 - capture <sample> len <length> :
7343 This only applies to "tcp-request content" rules. It captures sample
7344 expression <sample> from the request buffer, and converts it to a
7345 string of at most <len> characters. The resulting string is stored into
7346 the next request "capture" slot, so it will possibly appear next to
7347 some captured HTTP headers. It will then automatically appear in the
7348 logs, and it will be possible to extract it using sample fetch rules to
7349 feed it into headers or anything. The length should be limited given
7350 that this size will be allocated for each capture during the whole
7351 session life. Since it applies to Please check section 7.3 (Fetching
7352 samples) and "capture request header" for more information.
7353
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007354 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02007355 enables tracking of sticky counters from current connection. These
7356 rules do not stop evaluation and do not change default action. Two sets
7357 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007358 first "track-sc0" rule executed enables tracking of the counters of the
7359 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02007360 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007361 set. The first "track-sc2" rule executed enables tracking of the
7362 counters of the specified table as the third set. It is a recommended
7363 practice to use the first set of counters for the per-frontend counters
7364 and the second set for the per-backend ones. But this is just a
7365 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007366
Willy Tarreaue9656522010-08-17 15:40:09 +02007367 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007368 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02007369 in section 7.3. It describes what elements of the incoming
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007370 request or connection will be analysed, extracted, combined,
7371 and used to select which table entry to update the counters.
7372 Note that "tcp-request connection" cannot use content-based
7373 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007374
Willy Tarreaue9656522010-08-17 15:40:09 +02007375 <table> is an optional table to be used instead of the default one,
7376 which is the stick-table declared in the current proxy. All
7377 the counters for the matches and updates for the key will
7378 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007379
Willy Tarreaue9656522010-08-17 15:40:09 +02007380 Once a "track-sc*" rule is executed, the key is looked up in the table
7381 and if it is not found, an entry is allocated for it. Then a pointer to
7382 that entry is kept during all the session's life, and this entry's
7383 counters are updated as often as possible, every time the session's
7384 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007385 Counters are only updated for events that happen after the tracking has
7386 been started. For example, connection counters will not be updated when
7387 tracking layer 7 information, since the connection event happens before
7388 layer7 information is extracted.
7389
Willy Tarreaue9656522010-08-17 15:40:09 +02007390 If the entry tracks concurrent connection counters, one connection is
7391 counted for as long as the entry is tracked, and the entry will not
7392 expire during that time. Tracking counters also provides a performance
7393 advantage over just checking the keys, because only one table lookup is
7394 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007395
Willy Tarreaue9656522010-08-17 15:40:09 +02007396 Note that the "if/unless" condition is optional. If no condition is set on
7397 the action, it is simply performed unconditionally. That can be useful for
7398 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007399
Willy Tarreaue9656522010-08-17 15:40:09 +02007400 Example: accept all connections from white-listed hosts, reject too fast
7401 connection without counting them, and track accepted connections.
7402 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007403
Willy Tarreaue9656522010-08-17 15:40:09 +02007404 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007405 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007406 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007407
Willy Tarreaue9656522010-08-17 15:40:09 +02007408 Example: accept all connections from white-listed hosts, count all other
7409 connections and reject too fast ones. This results in abusive ones
7410 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007411
Willy Tarreaue9656522010-08-17 15:40:09 +02007412 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007413 tcp-request connection track-sc0 src
7414 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007415
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007416 Example: enable the PROXY protocol for traffic coming from all known proxies.
7417
7418 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
7419
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007420 See section 7 about ACL usage.
7421
Willy Tarreaue9656522010-08-17 15:40:09 +02007422 See also : "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007423
7424
Willy Tarreaue9656522010-08-17 15:40:09 +02007425tcp-request content <action> [{if | unless} <condition>]
7426 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02007427 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007428 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02007429 Arguments :
7430 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007431 actions include : "accept", "reject", "track-sc0", "track-sc1",
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007432 "track-sc2" and "capture". See "tcp-request connection" above
7433 for their signification.
Willy Tarreau62644772008-07-16 18:36:06 +02007434
Willy Tarreaue9656522010-08-17 15:40:09 +02007435 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02007436
Willy Tarreaue9656522010-08-17 15:40:09 +02007437 A request's contents can be analysed at an early stage of request processing
7438 called "TCP content inspection". During this stage, ACL-based rules are
7439 evaluated every time the request contents are updated, until either an
7440 "accept" or a "reject" rule matches, or the TCP request inspection delay
7441 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02007442
Willy Tarreaue9656522010-08-17 15:40:09 +02007443 The first difference between these rules and "tcp-request connection" rules
7444 is that "tcp-request content" rules can make use of contents to take a
7445 decision. Most often, these decisions will consider a protocol recognition or
7446 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01007447 both frontends and backends. In case of HTTP keep-alive with the client, all
7448 tcp-request content rules are evaluated again, so haproxy keeps a record of
7449 what sticky counters were assigned by a "tcp-request connection" versus a
7450 "tcp-request content" rule, and flushes all the content-related ones after
7451 processing an HTTP request, so that they may be evaluated again by the rules
7452 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007453 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01007454 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02007455
Willy Tarreaue9656522010-08-17 15:40:09 +02007456 Content-based rules are evaluated in their exact declaration order. If no
7457 rule matches or if there is no rule, the default action is to accept the
7458 contents. There is no specific limit to the number of rules which may be
7459 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02007460
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007461 Four types of actions are supported :
7462 - accept : the request is accepted
7463 - reject : the request is rejected and the connection is closed
7464 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007465 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Willy Tarreau62644772008-07-16 18:36:06 +02007466
Willy Tarreaue9656522010-08-17 15:40:09 +02007467 They have the same meaning as their counter-parts in "tcp-request connection"
7468 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02007469
Willy Tarreauf3338342014-01-28 21:40:28 +01007470 While there is nothing mandatory about it, it is recommended to use the
7471 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
7472 content" rules in the frontend, and track-sc2 for "tcp-request content"
7473 rules in the backend, because that makes the configuration more readable
7474 and easier to troubleshoot, but this is just a guideline and all counters
7475 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02007476
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007477 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02007478 the action, it is simply performed unconditionally. That can be useful for
7479 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02007480
Willy Tarreaue9656522010-08-17 15:40:09 +02007481 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02007482 rules, since HTTP-specific ACL matches are able to preliminarily parse the
7483 contents of a buffer before extracting the required data. If the buffered
7484 contents do not parse as a valid HTTP message, then the ACL does not match.
7485 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01007486 processing, so there is no risk of parsing something differently. In an HTTP
7487 backend connected to from an HTTP frontend, it is guaranteed that HTTP
7488 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02007489
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007490 Tracking layer7 information is also possible provided that the information
Willy Tarreau531485c2014-09-16 15:48:15 +02007491 are present when the rule is processed. The rule processing engine is able to
7492 wait until the inspect delay expires when the data to be tracked is not yet
7493 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007494
Willy Tarreau62644772008-07-16 18:36:06 +02007495 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02007496 # Accept HTTP requests containing a Host header saying "example.com"
7497 # and reject everything else.
7498 acl is_host_com hdr(Host) -i example.com
7499 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02007500 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02007501 tcp-request content reject
7502
7503 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02007504 # reject SMTP connection if client speaks first
7505 tcp-request inspect-delay 30s
7506 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007507 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02007508
7509 # Forward HTTPS connection only if client speaks
7510 tcp-request inspect-delay 30s
7511 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007512 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02007513 tcp-request content reject
7514
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007515 Example:
7516 # Track the last IP from X-Forwarded-For
7517 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007518 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007519
7520 Example:
7521 # track request counts per "base" (concatenation of Host+URL)
7522 tcp-request inspect-delay 10s
Willy Tarreau531485c2014-09-16 15:48:15 +02007523 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007524
Willy Tarreaue9656522010-08-17 15:40:09 +02007525 Example: track per-frontend and per-backend counters, block abusers at the
7526 frontend when the backend detects abuse.
7527
7528 frontend http
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007529 # Use General Purpose Couter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02007530 # protecting all our sites
7531 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007532 tcp-request connection track-sc0 src
7533 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02007534 ...
7535 use_backend http_dynamic if { path_end .php }
7536
7537 backend http_dynamic
7538 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007539 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02007540 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007541 acl click_too_fast sc1_http_req_rate gt 10
7542 acl mark_as_abuser sc0_inc_gpc0 gt 0
7543 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02007544 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02007545
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007546 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02007547
Willy Tarreaue9656522010-08-17 15:40:09 +02007548 See also : "tcp-request connection", "tcp-request inspect-delay"
Willy Tarreau62644772008-07-16 18:36:06 +02007549
7550
7551tcp-request inspect-delay <timeout>
7552 Set the maximum allowed time to wait for data during content inspection
7553 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007554 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02007555 Arguments :
7556 <timeout> is the timeout value specified in milliseconds by default, but
7557 can be in any other unit if the number is suffixed by the unit,
7558 as explained at the top of this document.
7559
7560 People using haproxy primarily as a TCP relay are often worried about the
7561 risk of passing any type of protocol to a server without any analysis. In
7562 order to be able to analyze the request contents, we must first withhold
7563 the data then analyze them. This statement simply enables withholding of
7564 data for at most the specified amount of time.
7565
Willy Tarreaufb356202010-08-03 14:02:05 +02007566 TCP content inspection applies very early when a connection reaches a
7567 frontend, then very early when the connection is forwarded to a backend. This
7568 means that a connection may experience a first delay in the frontend and a
7569 second delay in the backend if both have tcp-request rules.
7570
Willy Tarreau62644772008-07-16 18:36:06 +02007571 Note that when performing content inspection, haproxy will evaluate the whole
7572 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007573 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02007574 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01007575 contents are definitive. If no delay is set, haproxy will not wait at all
7576 and will immediately apply a verdict based on the available information.
7577 Obviously this is unlikely to be very useful and might even be racy, so such
7578 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02007579
7580 As soon as a rule matches, the request is released and continues as usual. If
7581 the timeout is reached and no rule matches, the default policy will be to let
7582 it pass through unaffected.
7583
7584 For most protocols, it is enough to set it to a few seconds, as most clients
7585 send the full request immediately upon connection. Add 3 or more seconds to
7586 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01007587 to use large values, for instance to ensure that the client never talks
Willy Tarreau62644772008-07-16 18:36:06 +02007588 before the server (eg: SMTP), or to wait for a client to talk before passing
7589 data to the server (eg: SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02007590 least the inspection delay, otherwise it will expire first. If the client
7591 closes the connection or if the buffer is full, the delay immediately expires
7592 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02007593
Willy Tarreau55165fe2009-05-10 12:02:55 +02007594 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02007595 "timeout client".
7596
7597
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007598tcp-response content <action> [{if | unless} <condition>]
7599 Perform an action on a session response depending on a layer 4-7 condition
7600 May be used in sections : defaults | frontend | listen | backend
7601 no | no | yes | yes
7602 Arguments :
7603 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007604 actions include : "accept", "close", "reject".
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007605
7606 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
7607
7608 Response contents can be analysed at an early stage of response processing
7609 called "TCP content inspection". During this stage, ACL-based rules are
7610 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007611 "accept", "close" or a "reject" rule matches, or a TCP response inspection
7612 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007613
7614 Most often, these decisions will consider a protocol recognition or validity.
7615
7616 Content-based rules are evaluated in their exact declaration order. If no
7617 rule matches or if there is no rule, the default action is to accept the
7618 contents. There is no specific limit to the number of rules which may be
7619 inserted.
7620
7621 Two types of actions are supported :
7622 - accept :
7623 accepts the response if the condition is true (when used with "if")
7624 or false (when used with "unless"). The first such rule executed ends
7625 the rules evaluation.
7626
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007627 - close :
7628 immediately closes the connection with the server if the condition is
7629 true (when used with "if"), or false (when used with "unless"). The
7630 first such rule executed ends the rules evaluation. The main purpose of
7631 this action is to force a connection to be finished between a client
7632 and a server after an exchange when the application protocol expects
7633 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007634 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007635 protocols.
7636
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007637 - reject :
7638 rejects the response if the condition is true (when used with "if")
7639 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007640 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007641
7642 Note that the "if/unless" condition is optional. If no condition is set on
7643 the action, it is simply performed unconditionally. That can be useful for
7644 for changing the default action to a reject.
7645
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007646 It is perfectly possible to match layer 7 contents with "tcp-response
7647 content" rules, but then it is important to ensure that a full response has
7648 been buffered, otherwise no contents will match. In order to achieve this,
7649 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007650 period.
7651
7652 See section 7 about ACL usage.
7653
7654 See also : "tcp-request content", "tcp-response inspect-delay"
7655
7656
7657tcp-response inspect-delay <timeout>
7658 Set the maximum allowed time to wait for a response during content inspection
7659 May be used in sections : defaults | frontend | listen | backend
7660 no | no | yes | yes
7661 Arguments :
7662 <timeout> is the timeout value specified in milliseconds by default, but
7663 can be in any other unit if the number is suffixed by the unit,
7664 as explained at the top of this document.
7665
7666 See also : "tcp-response content", "tcp-request inspect-delay".
7667
7668
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007669timeout check <timeout>
7670 Set additional check timeout, but only after a connection has been already
7671 established.
7672
7673 May be used in sections: defaults | frontend | listen | backend
7674 yes | no | yes | yes
7675 Arguments:
7676 <timeout> is the timeout value specified in milliseconds by default, but
7677 can be in any other unit if the number is suffixed by the unit,
7678 as explained at the top of this document.
7679
7680 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
7681 for check and "timeout check" as an additional read timeout. The "min" is
7682 used so that people running with *very* long "timeout connect" (eg. those
7683 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01007684 (Please also note that there is no valid reason to have such long connect
7685 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
7686 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007687
7688 If "timeout check" is not set haproxy uses "inter" for complete check
7689 timeout (connect + read) exactly like all <1.3.15 version.
7690
7691 In most cases check request is much simpler and faster to handle than normal
7692 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01007693 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007694
7695 This parameter is specific to backends, but can be specified once for all in
7696 "defaults" sections. This is in fact one of the easiest solutions not to
7697 forget about it.
7698
Willy Tarreau41a340d2008-01-22 12:25:31 +01007699 See also: "timeout connect", "timeout queue", "timeout server",
7700 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007701
7702
Willy Tarreau0ba27502007-12-24 16:55:16 +01007703timeout client <timeout>
7704timeout clitimeout <timeout> (deprecated)
7705 Set the maximum inactivity time on the client side.
7706 May be used in sections : defaults | frontend | listen | backend
7707 yes | yes | yes | no
7708 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007709 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007710 can be in any other unit if the number is suffixed by the unit,
7711 as explained at the top of this document.
7712
7713 The inactivity timeout applies when the client is expected to acknowledge or
7714 send data. In HTTP mode, this timeout is particularly important to consider
7715 during the first phase, when the client sends the request, and during the
7716 response while it is reading data sent by the server. The value is specified
7717 in milliseconds by default, but can be in any other unit if the number is
7718 suffixed by the unit, as specified at the top of this document. In TCP mode
7719 (and to a lesser extent, in HTTP mode), it is highly recommended that the
7720 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007721 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01007722 losses by specifying timeouts that are slightly above multiples of 3 seconds
Willy Tarreauce887fd2012-05-12 12:50:00 +02007723 (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
7724 sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02007725 which overrides "timeout client" and "timeout server" for tunnels, as well as
7726 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007727
7728 This parameter is specific to frontends, but can be specified once for all in
7729 "defaults" sections. This is in fact one of the easiest solutions not to
7730 forget about it. An unspecified timeout results in an infinite timeout, which
7731 is not recommended. Such a usage is accepted and works but reports a warning
7732 during startup because it may results in accumulation of expired sessions in
7733 the system if the system's timeouts are not configured either.
7734
7735 This parameter replaces the old, deprecated "clitimeout". It is recommended
7736 to use it to write new configurations. The form "timeout clitimeout" is
7737 provided only by backwards compatibility but its use is strongly discouraged.
7738
Willy Tarreauce887fd2012-05-12 12:50:00 +02007739 See also : "clitimeout", "timeout server", "timeout tunnel".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007740
7741
Willy Tarreau05cdd962014-05-10 14:30:07 +02007742timeout client-fin <timeout>
7743 Set the inactivity timeout on the client side for half-closed connections.
7744 May be used in sections : defaults | frontend | listen | backend
7745 yes | yes | yes | no
7746 Arguments :
7747 <timeout> is the timeout value specified in milliseconds by default, but
7748 can be in any other unit if the number is suffixed by the unit,
7749 as explained at the top of this document.
7750
7751 The inactivity timeout applies when the client is expected to acknowledge or
7752 send data while one direction is already shut down. This timeout is different
7753 from "timeout client" in that it only applies to connections which are closed
7754 in one direction. This is particularly useful to avoid keeping connections in
7755 FIN_WAIT state for too long when clients do not disconnect cleanly. This
7756 problem is particularly common long connections such as RDP or WebSocket.
7757 Note that this timeout can override "timeout tunnel" when a connection shuts
7758 down in one direction.
7759
7760 This parameter is specific to frontends, but can be specified once for all in
7761 "defaults" sections. By default it is not set, so half-closed connections
7762 will use the other timeouts (timeout.client or timeout.tunnel).
7763
7764 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
7765
7766
Willy Tarreau0ba27502007-12-24 16:55:16 +01007767timeout connect <timeout>
7768timeout contimeout <timeout> (deprecated)
7769 Set the maximum time to wait for a connection attempt to a server to succeed.
7770 May be used in sections : defaults | frontend | listen | backend
7771 yes | no | yes | yes
7772 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007773 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007774 can be in any other unit if the number is suffixed by the unit,
7775 as explained at the top of this document.
7776
7777 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007778 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01007779 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01007780 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007781 connect timeout also presets both queue and tarpit timeouts to the same value
7782 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007783
7784 This parameter is specific to backends, but can be specified once for all in
7785 "defaults" sections. This is in fact one of the easiest solutions not to
7786 forget about it. An unspecified timeout results in an infinite timeout, which
7787 is not recommended. Such a usage is accepted and works but reports a warning
7788 during startup because it may results in accumulation of failed sessions in
7789 the system if the system's timeouts are not configured either.
7790
7791 This parameter replaces the old, deprecated "contimeout". It is recommended
7792 to use it to write new configurations. The form "timeout contimeout" is
7793 provided only by backwards compatibility but its use is strongly discouraged.
7794
Willy Tarreau41a340d2008-01-22 12:25:31 +01007795 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
7796 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007797
7798
Willy Tarreaub16a5742010-01-10 14:46:16 +01007799timeout http-keep-alive <timeout>
7800 Set the maximum allowed time to wait for a new HTTP request to appear
7801 May be used in sections : defaults | frontend | listen | backend
7802 yes | yes | yes | yes
7803 Arguments :
7804 <timeout> is the timeout value specified in milliseconds by default, but
7805 can be in any other unit if the number is suffixed by the unit,
7806 as explained at the top of this document.
7807
7808 By default, the time to wait for a new request in case of keep-alive is set
7809 by "timeout http-request". However this is not always convenient because some
7810 people want very short keep-alive timeouts in order to release connections
7811 faster, and others prefer to have larger ones but still have short timeouts
7812 once the request has started to present itself.
7813
7814 The "http-keep-alive" timeout covers these needs. It will define how long to
7815 wait for a new HTTP request to start coming after a response was sent. Once
7816 the first byte of request has been seen, the "http-request" timeout is used
7817 to wait for the complete request to come. Note that empty lines prior to a
7818 new request do not refresh the timeout and are not counted as a new request.
7819
7820 There is also another difference between the two timeouts : when a connection
7821 expires during timeout http-keep-alive, no error is returned, the connection
7822 just closes. If the connection expires in "http-request" while waiting for a
7823 connection to complete, a HTTP 408 error is returned.
7824
7825 In general it is optimal to set this value to a few tens to hundreds of
7826 milliseconds, to allow users to fetch all objects of a page at once but
7827 without waiting for further clicks. Also, if set to a very small value (eg:
7828 1 millisecond) it will probably only accept pipelined requests but not the
7829 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +02007830 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +01007831
7832 If this parameter is not set, the "http-request" timeout applies, and if both
7833 are not set, "timeout client" still applies at the lower level. It should be
7834 set in the frontend to take effect, unless the frontend is in TCP mode, in
7835 which case the HTTP backend's timeout will be used.
7836
7837 See also : "timeout http-request", "timeout client".
7838
7839
Willy Tarreau036fae02008-01-06 13:24:40 +01007840timeout http-request <timeout>
7841 Set the maximum allowed time to wait for a complete HTTP request
7842 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007843 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +01007844 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007845 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +01007846 can be in any other unit if the number is suffixed by the unit,
7847 as explained at the top of this document.
7848
7849 In order to offer DoS protection, it may be required to lower the maximum
7850 accepted time to receive a complete HTTP request without affecting the client
7851 timeout. This helps protecting against established connections on which
7852 nothing is sent. The client timeout cannot offer a good protection against
7853 this abuse because it is an inactivity timeout, which means that if the
7854 attacker sends one character every now and then, the timeout will not
7855 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +02007856 types, the request will be aborted if it does not complete in time. When the
7857 timeout expires, an HTTP 408 response is sent to the client to inform it
7858 about the problem, and the connection is closed. The logs will report
7859 termination codes "cR". Some recent browsers are having problems with this
7860 standard, well-documented behaviour, so it might be needed to hide the 408
7861 code using "errorfile 408 /dev/null". See more details in the explanations of
7862 the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +01007863
7864 Note that this timeout only applies to the header part of the request, and
7865 not to any data. As soon as the empty line is received, this timeout is not
Willy Tarreaub16a5742010-01-10 14:46:16 +01007866 used anymore. It is used again on keep-alive connections to wait for a second
7867 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +01007868
7869 Generally it is enough to set it to a few seconds, as most clients send the
7870 full request immediately upon connection. Add 3 or more seconds to cover TCP
7871 retransmits but that's all. Setting it to very low values (eg: 50 ms) will
7872 generally work on local networks as long as there are no packet losses. This
7873 will prevent people from sending bare HTTP requests using telnet.
7874
7875 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007876 chunk of the incoming request. It should be set in the frontend to take
7877 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
7878 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +01007879
Willy Tarreau2705a612014-05-23 17:38:34 +02007880 See also : "errorfile", "timeout http-keep-alive", "timeout client".
Willy Tarreau036fae02008-01-06 13:24:40 +01007881
Willy Tarreau844e3c52008-01-11 16:28:18 +01007882
7883timeout queue <timeout>
7884 Set the maximum time to wait in the queue for a connection slot to be free
7885 May be used in sections : defaults | frontend | listen | backend
7886 yes | no | yes | yes
7887 Arguments :
7888 <timeout> is the timeout value specified in milliseconds by default, but
7889 can be in any other unit if the number is suffixed by the unit,
7890 as explained at the top of this document.
7891
7892 When a server's maxconn is reached, connections are left pending in a queue
7893 which may be server-specific or global to the backend. In order not to wait
7894 indefinitely, a timeout is applied to requests pending in the queue. If the
7895 timeout is reached, it is considered that the request will almost never be
7896 served, so it is dropped and a 503 error is returned to the client.
7897
7898 The "timeout queue" statement allows to fix the maximum time for a request to
7899 be left pending in a queue. If unspecified, the same value as the backend's
7900 connection timeout ("timeout connect") is used, for backwards compatibility
7901 with older versions with no "timeout queue" parameter.
7902
7903 See also : "timeout connect", "contimeout".
7904
7905
7906timeout server <timeout>
7907timeout srvtimeout <timeout> (deprecated)
7908 Set the maximum inactivity time on the server side.
7909 May be used in sections : defaults | frontend | listen | backend
7910 yes | no | yes | yes
7911 Arguments :
7912 <timeout> is the timeout value specified in milliseconds by default, but
7913 can be in any other unit if the number is suffixed by the unit,
7914 as explained at the top of this document.
7915
7916 The inactivity timeout applies when the server is expected to acknowledge or
7917 send data. In HTTP mode, this timeout is particularly important to consider
7918 during the first phase of the server's response, when it has to send the
7919 headers, as it directly represents the server's processing time for the
7920 request. To find out what value to put there, it's often good to start with
7921 what would be considered as unacceptable response times, then check the logs
7922 to observe the response time distribution, and adjust the value accordingly.
7923
7924 The value is specified in milliseconds by default, but can be in any other
7925 unit if the number is suffixed by the unit, as specified at the top of this
7926 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
7927 recommended that the client timeout remains equal to the server timeout in
7928 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007929 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01007930 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreauce887fd2012-05-12 12:50:00 +02007931 seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed
7932 with short-lived sessions (eg: WebSocket and HTTP), it's worth considering
7933 "timeout tunnel", which overrides "timeout client" and "timeout server" for
7934 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007935
7936 This parameter is specific to backends, but can be specified once for all in
7937 "defaults" sections. This is in fact one of the easiest solutions not to
7938 forget about it. An unspecified timeout results in an infinite timeout, which
7939 is not recommended. Such a usage is accepted and works but reports a warning
7940 during startup because it may results in accumulation of expired sessions in
7941 the system if the system's timeouts are not configured either.
7942
7943 This parameter replaces the old, deprecated "srvtimeout". It is recommended
7944 to use it to write new configurations. The form "timeout srvtimeout" is
7945 provided only by backwards compatibility but its use is strongly discouraged.
7946
Willy Tarreauce887fd2012-05-12 12:50:00 +02007947 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +01007948
Willy Tarreau05cdd962014-05-10 14:30:07 +02007949
7950timeout server-fin <timeout>
7951 Set the inactivity timeout on the server side for half-closed connections.
7952 May be used in sections : defaults | frontend | listen | backend
7953 yes | no | yes | yes
7954 Arguments :
7955 <timeout> is the timeout value specified in milliseconds by default, but
7956 can be in any other unit if the number is suffixed by the unit,
7957 as explained at the top of this document.
7958
7959 The inactivity timeout applies when the server is expected to acknowledge or
7960 send data while one direction is already shut down. This timeout is different
7961 from "timeout server" in that it only applies to connections which are closed
7962 in one direction. This is particularly useful to avoid keeping connections in
7963 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
7964 This problem is particularly common long connections such as RDP or WebSocket.
7965 Note that this timeout can override "timeout tunnel" when a connection shuts
7966 down in one direction. This setting was provided for completeness, but in most
7967 situations, it should not be needed.
7968
7969 This parameter is specific to backends, but can be specified once for all in
7970 "defaults" sections. By default it is not set, so half-closed connections
7971 will use the other timeouts (timeout.server or timeout.tunnel).
7972
7973 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
7974
Willy Tarreau844e3c52008-01-11 16:28:18 +01007975
7976timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +01007977 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +01007978 May be used in sections : defaults | frontend | listen | backend
7979 yes | yes | yes | yes
7980 Arguments :
7981 <timeout> is the tarpit duration specified in milliseconds by default, but
7982 can be in any other unit if the number is suffixed by the unit,
7983 as explained at the top of this document.
7984
7985 When a connection is tarpitted using "reqtarpit", it is maintained open with
7986 no activity for a certain amount of time, then closed. "timeout tarpit"
7987 defines how long it will be maintained open.
7988
7989 The value is specified in milliseconds by default, but can be in any other
7990 unit if the number is suffixed by the unit, as specified at the top of this
7991 document. If unspecified, the same value as the backend's connection timeout
7992 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +01007993 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007994
7995 See also : "timeout connect", "contimeout".
7996
7997
Willy Tarreauce887fd2012-05-12 12:50:00 +02007998timeout tunnel <timeout>
7999 Set the maximum inactivity time on the client and server side for tunnels.
8000 May be used in sections : defaults | frontend | listen | backend
8001 yes | no | yes | yes
8002 Arguments :
8003 <timeout> is the timeout value specified in milliseconds by default, but
8004 can be in any other unit if the number is suffixed by the unit,
8005 as explained at the top of this document.
8006
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008007 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +02008008 between a client and a server, and the connection remains inactive in both
8009 directions. This timeout supersedes both the client and server timeouts once
8010 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
8011 analyser remains attached to either connection (eg: tcp content rules are
8012 accepted). In HTTP, this timeout is used when a connection is upgraded (eg:
8013 when switching to the WebSocket protocol, or forwarding a CONNECT request
8014 to a proxy), or after the first response when no keepalive/close option is
8015 specified.
8016
Willy Tarreau05cdd962014-05-10 14:30:07 +02008017 Since this timeout is usually used in conjunction with long-lived connections,
8018 it usually is a good idea to also set "timeout client-fin" to handle the
8019 situation where a client suddenly disappears from the net and does not
8020 acknowledge a close, or sends a shutdown and does not acknowledge pending
8021 data anymore. This can happen in lossy networks where firewalls are present,
8022 and is detected by the presence of large amounts of sessions in a FIN_WAIT
8023 state.
8024
Willy Tarreauce887fd2012-05-12 12:50:00 +02008025 The value is specified in milliseconds by default, but can be in any other
8026 unit if the number is suffixed by the unit, as specified at the top of this
8027 document. Whatever the expected normal idle time, it is a good practice to
8028 cover at least one or several TCP packet losses by specifying timeouts that
8029 are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum).
8030
8031 This parameter is specific to backends, but can be specified once for all in
8032 "defaults" sections. This is in fact one of the easiest solutions not to
8033 forget about it.
8034
8035 Example :
8036 defaults http
8037 option http-server-close
8038 timeout connect 5s
8039 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +02008040 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +02008041 timeout server 30s
8042 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
8043
Willy Tarreau05cdd962014-05-10 14:30:07 +02008044 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +02008045
8046
Willy Tarreau844e3c52008-01-11 16:28:18 +01008047transparent (deprecated)
8048 Enable client-side transparent proxying
8049 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008050 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008051 Arguments : none
8052
8053 This keyword was introduced in order to provide layer 7 persistence to layer
8054 3 load balancers. The idea is to use the OS's ability to redirect an incoming
8055 connection for a remote address to a local process (here HAProxy), and let
8056 this process know what address was initially requested. When this option is
8057 used, sessions without cookies will be forwarded to the original destination
8058 IP address of the incoming request (which should match that of another
8059 equipment), while requests with cookies will still be forwarded to the
8060 appropriate server.
8061
8062 The "transparent" keyword is deprecated, use "option transparent" instead.
8063
8064 Note that contrary to a common belief, this option does NOT make HAProxy
8065 present the client's IP to the server when establishing the connection.
8066
Willy Tarreau844e3c52008-01-11 16:28:18 +01008067 See also: "option transparent"
8068
William Lallemanda73203e2012-03-12 12:48:57 +01008069unique-id-format <string>
8070 Generate a unique ID for each request.
8071 May be used in sections : defaults | frontend | listen | backend
8072 yes | yes | yes | no
8073 Arguments :
8074 <string> is a log-format string.
8075
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008076 This keyword creates a ID for each request using the custom log format. A
8077 unique ID is useful to trace a request passing through many components of
8078 a complex infrastructure. The newly created ID may also be logged using the
8079 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +01008080
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008081 The format should be composed from elements that are guaranteed to be
8082 unique when combined together. For instance, if multiple haproxy instances
8083 are involved, it might be important to include the node name. It is often
8084 needed to log the incoming connection's source and destination addresses
8085 and ports. Note that since multiple requests may be performed over the same
8086 connection, including a request counter may help differentiate them.
8087 Similarly, a timestamp may protect against a rollover of the counter.
8088 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +01008089
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008090 It is recommended to use hexadecimal notation for many fields since it
8091 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +01008092
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008093 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008094
Julien Vehentf21be322014-03-07 08:27:34 -05008095 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008096
8097 will generate:
8098
8099 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8100
8101 See also: "unique-id-header"
8102
8103unique-id-header <name>
8104 Add a unique ID header in the HTTP request.
8105 May be used in sections : defaults | frontend | listen | backend
8106 yes | yes | yes | no
8107 Arguments :
8108 <name> is the name of the header.
8109
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008110 Add a unique-id header in the HTTP request sent to the server, using the
8111 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +01008112
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008113 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008114
Julien Vehentf21be322014-03-07 08:27:34 -05008115 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008116 unique-id-header X-Unique-ID
8117
8118 will generate:
8119
8120 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8121
8122 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +01008123
Willy Tarreauf51658d2014-04-23 01:21:56 +02008124use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008125 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008126 May be used in sections : defaults | frontend | listen | backend
8127 no | yes | yes | no
8128 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008129 <backend> is the name of a valid backend or "listen" section, or a
8130 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008131
Willy Tarreauf51658d2014-04-23 01:21:56 +02008132 <condition> is a condition composed of ACLs, as described in section 7. If
8133 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008134
8135 When doing content-switching, connections arrive on a frontend and are then
8136 dispatched to various backends depending on a number of conditions. The
8137 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008138 "use_backend" keyword. While it is normally used with HTTP processing, it can
8139 also be used in pure TCP, either without content using stateless ACLs (eg:
8140 source address validation) or combined with a "tcp-request" rule to wait for
8141 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008142
8143 There may be as many "use_backend" rules as desired. All of these rules are
8144 evaluated in their declaration order, and the first one which matches will
8145 assign the backend.
8146
8147 In the first form, the backend will be used if the condition is met. In the
8148 second form, the backend will be used if the condition is not met. If no
8149 condition is valid, the backend defined with "default_backend" will be used.
8150 If no default backend is defined, either the servers in the same section are
8151 used (in case of a "listen" section) or, in case of a frontend, no server is
8152 used and a 503 service unavailable response is returned.
8153
Willy Tarreau51aecc72009-07-12 09:47:04 +02008154 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008155 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +02008156 and backend processing will immediately follow, or the backend will wait for
8157 a complete HTTP request to get in. This feature is useful when a frontend
8158 must decode several protocols on a unique port, one of them being HTTP.
8159
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008160 When <backend> is a simple name, it is resolved at configuration time, and an
8161 error is reported if the specified backend does not exist. If <backend> is
8162 a log-format string instead, no check may be done at configuration time, so
8163 the backend name is resolved dynamically at run time. If the resulting
8164 backend name does not correspond to any valid backend, no other rule is
8165 evaluated, and the default_backend directive is applied instead. Note that
8166 when using dynamic backend names, it is highly recommended to use a prefix
8167 that no other backend uses in order to ensure that an unauthorized backend
8168 cannot be forced from the request.
8169
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008170 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008171 used to detect the association between frontends and backends to compute the
8172 backend's "fullconn" setting. This cannot be done for dynamic names.
8173
8174 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
8175 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008176
Willy Tarreau036fae02008-01-06 13:24:40 +01008177
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008178use-server <server> if <condition>
8179use-server <server> unless <condition>
8180 Only use a specific server if/unless an ACL-based condition is matched.
8181 May be used in sections : defaults | frontend | listen | backend
8182 no | no | yes | yes
8183 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008184 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008185
8186 <condition> is a condition composed of ACLs, as described in section 7.
8187
8188 By default, connections which arrive to a backend are load-balanced across
8189 the available servers according to the configured algorithm, unless a
8190 persistence mechanism such as a cookie is used and found in the request.
8191
8192 Sometimes it is desirable to forward a particular request to a specific
8193 server without having to declare a dedicated backend for this server. This
8194 can be achieved using the "use-server" rules. These rules are evaluated after
8195 the "redirect" rules and before evaluating cookies, and they have precedence
8196 on them. There may be as many "use-server" rules as desired. All of these
8197 rules are evaluated in their declaration order, and the first one which
8198 matches will assign the server.
8199
8200 If a rule designates a server which is down, and "option persist" is not used
8201 and no force-persist rule was validated, it is ignored and evaluation goes on
8202 with the next rules until one matches.
8203
8204 In the first form, the server will be used if the condition is met. In the
8205 second form, the server will be used if the condition is not met. If no
8206 condition is valid, the processing continues and the server will be assigned
8207 according to other persistence mechanisms.
8208
8209 Note that even if a rule is matched, cookie processing is still performed but
8210 does not assign the server. This allows prefixed cookies to have their prefix
8211 stripped.
8212
8213 The "use-server" statement works both in HTTP and TCP mode. This makes it
8214 suitable for use with content-based inspection. For instance, a server could
8215 be selected in a farm according to the TLS SNI field. And if these servers
8216 have their weight set to zero, they will not be used for other traffic.
8217
8218 Example :
8219 # intercept incoming TLS requests based on the SNI field
8220 use-server www if { req_ssl_sni -i www.example.com }
8221 server www 192.168.0.1:443 weight 0
8222 use-server mail if { req_ssl_sni -i mail.example.com }
8223 server mail 192.168.0.1:587 weight 0
8224 use-server imap if { req_ssl_sni -i imap.example.com }
8225 server mail 192.168.0.1:993 weight 0
8226 # all the rest is forwarded to this server
8227 server default 192.168.0.2:443 check
8228
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008229 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008230
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008231
82325. Bind and Server options
8233--------------------------
8234
8235The "bind", "server" and "default-server" keywords support a number of settings
8236depending on some build options and on the system HAProxy was built on. These
8237settings generally each consist in one word sometimes followed by a value,
8238written on the same line as the "bind" or "server" line. All these options are
8239described in this section.
8240
8241
82425.1. Bind options
8243-----------------
8244
8245The "bind" keyword supports a certain number of settings which are all passed
8246as arguments on the same line. The order in which those arguments appear makes
8247no importance, provided that they appear after the bind address. All of these
8248parameters are optional. Some of them consist in a single words (booleans),
8249while other ones expect a value after them. In this case, the value must be
8250provided immediately after the setting name.
8251
8252The currently supported settings are the following ones.
8253
8254accept-proxy
8255 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +02008256 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
8257 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008258 3/4 addresses of the incoming connection to be used everywhere an address is
8259 used, with the only exception of "tcp-request connection" rules which will
8260 only see the real connection address. Logs will reflect the addresses
8261 indicated in the protocol, unless it is violated, in which case the real
8262 address will still be used. This keyword combined with support from external
8263 components can be used as an efficient and reliable alternative to the
8264 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +02008265 usable. See also "tcp-request connection expect-proxy" for a finer-grained
8266 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008267
Willy Tarreauab861d32013-04-02 02:30:41 +02008268alpn <protocols>
8269 This enables the TLS ALPN extension and advertises the specified protocol
8270 list as supported on top of ALPN. The protocol list consists in a comma-
8271 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
8272 quotes). This requires that the SSL library is build with support for TLS
8273 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
8274 initial NPN extension.
8275
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008276backlog <backlog>
8277 Sets the socket's backlog to this value. If unspecified, the frontend's
8278 backlog is used instead, which generally defaults to the maxconn value.
8279
Emeric Brun7fb34422012-09-28 15:26:15 +02008280ecdhe <named curve>
8281 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +01008282 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
8283 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +02008284
Emeric Brunfd33a262012-10-11 16:28:27 +02008285ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008286 This setting is only available when support for OpenSSL was built in. It
8287 designates a PEM file from which to load CA certificates used to verify
8288 client's certificate.
8289
Emeric Brunb6dc9342012-09-28 17:55:37 +02008290ca-ignore-err [all|<errorID>,...]
8291 This setting is only available when support for OpenSSL was built in.
8292 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
8293 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
8294 error is ignored.
8295
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008296ciphers <ciphers>
8297 This setting is only available when support for OpenSSL was built in. It sets
8298 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008299 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008300 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
8301 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
8302
Emeric Brunfd33a262012-10-11 16:28:27 +02008303crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008304 This setting is only available when support for OpenSSL was built in. It
8305 designates a PEM file from which to load certificate revocation list used
8306 to verify client's certificate.
8307
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008308crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +00008309 This setting is only available when support for OpenSSL was built in. It
8310 designates a PEM file containing both the required certificates and any
8311 associated private keys. This file can be built by concatenating multiple
8312 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
8313 requires an intermediate certificate, this can also be concatenated into this
8314 file.
8315
8316 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
8317 are loaded.
8318
8319 If a directory name is used instead of a PEM file, then all files found in
Emeric Brun4147b2e2014-06-16 18:36:30 +02008320 that directory will be loaded unless their name ends with '.issuer' or
8321 '.ocsp' (reserved extensions). This directive may be specified multiple times
Alex Davies0fbf0162013-03-02 16:04:50 +00008322 in order to load certificates from multiple files or directories. The
8323 certificates will be presented to clients who provide a valid TLS Server Name
8324 Indication field matching one of their CN or alt subjects. Wildcards are
8325 supported, where a wildcard character '*' is used instead of the first
8326 hostname component (eg: *.example.org matches www.example.org but not
8327 www.sub.example.org).
8328
8329 If no SNI is provided by the client or if the SSL library does not support
8330 TLS extensions, or if the client provides an SNI hostname which does not
8331 match any certificate, then the first loaded certificate will be presented.
8332 This means that when loading certificates from a directory, it is highly
8333 recommended to load the default one first as a file.
8334
Emeric Brune032bfa2012-09-28 13:01:45 +02008335 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008336
Alex Davies0fbf0162013-03-02 16:04:50 +00008337 Some CAs (such as Godaddy) offer a drop down list of server types that do not
8338 include HAProxy when obtaining a certificate. If this happens be sure to
Godbach8bf60a12014-04-21 21:42:41 +08008339 choose a webserver that the CA believes requires an intermediate CA (for
Alex Davies0fbf0162013-03-02 16:04:50 +00008340 Godaddy, selection Apache Tomcat will get the correct bundle, but many
8341 others, e.g. nginx, result in a wrong bundle that will not work for some
8342 clients).
8343
Emeric Brun4147b2e2014-06-16 18:36:30 +02008344 For each PEM file, haproxy checks for the presence of file at the same path
8345 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
8346 Status Request extension (also known as "OCSP stapling") is automatically
8347 enabled. The content of this file is optional. If not empty, it must contain
8348 a valid OCSP Response in DER format. In order to be valid an OCSP Response
8349 must comply with the following rules: it has to indicate a good status,
8350 it has to be a single response for the certificate of the PEM file, and it
8351 has to be valid at the moment of addition. If these rules are not respected
8352 the OCSP Response is ignored and a warning is emitted. In order to identify
8353 which certificate an OCSP Response applies to, the issuer's certificate is
8354 necessary. If the issuer's certificate is not found in the PEM file, it will
8355 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
8356 if it exists otherwise it will fail with an error.
8357
Emeric Brunb6dc9342012-09-28 17:55:37 +02008358crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +00008359 This setting is only available when support for OpenSSL was built in. Sets a
8360 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008361 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +00008362 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +02008363
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008364crt-list <file>
8365 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008366 designates a list of PEM file with an optional list of SNI filter per
8367 certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008368
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008369 <crtfile> [[!]<snifilter> ...]
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008370
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008371 Wildcards are supported in the SNI filter. Negative filter are also supported,
8372 only useful in combination with a wildcard filter to exclude a particular SNI.
8373 The certificates will be presented to clients who provide a valid TLS Server
8374 Name Indication field matching one of the SNI filters. If no SNI filter is
8375 specified, the CN and alt subjects are used. This directive may be specified
8376 multiple times. See the "crt" option for more information. The default
8377 certificate is still needed to meet OpenSSL expectations. If it is not used,
8378 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008379
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008380defer-accept
8381 Is an optional keyword which is supported only on certain Linux kernels. It
8382 states that a connection will only be accepted once some data arrive on it,
8383 or at worst after the first retransmit. This should be used only on protocols
8384 for which the client talks first (eg: HTTP). It can slightly improve
8385 performance by ensuring that most of the request is already available when
8386 the connection is accepted. On the other hand, it will not be able to detect
8387 connections which don't talk. It is important to note that this option is
8388 broken in all kernels up to 2.6.31, as the connection is never accepted until
8389 the client talks. This can cause issues with front firewalls which would see
8390 an established connection while the proxy will only see it in SYN_RECV. This
8391 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
8392
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008393force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008394 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008395 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun42a3e202014-10-30 15:56:50 +01008396 for high connection rates. This option is also available on global statement
8397 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008398
8399force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008400 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008401 this listener. This option is also available on global statement
8402 "ssl-default-bind-options". See also "no-tlsv*" and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008403
8404force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008405 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008406 this listener. This option is also available on global statement
8407 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008408
8409force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008410 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun42a3e202014-10-30 15:56:50 +01008411 this listener. This option is also available on global statement
8412 "ssl-default-bind-options". See also "no-tlsv*", and "no-sslv3".
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008413
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008414gid <gid>
8415 Sets the group of the UNIX sockets to the designated system gid. It can also
8416 be set by default in the global section's "unix-bind" statement. Note that
8417 some platforms simply ignore this. This setting is equivalent to the "group"
8418 setting except that the group ID is used instead of its name. This setting is
8419 ignored by non UNIX sockets.
8420
8421group <group>
8422 Sets the group of the UNIX sockets to the designated system group. It can
8423 also be set by default in the global section's "unix-bind" statement. Note
8424 that some platforms simply ignore this. This setting is equivalent to the
8425 "gid" setting except that the group name is used instead of its gid. This
8426 setting is ignored by non UNIX sockets.
8427
8428id <id>
8429 Fixes the socket ID. By default, socket IDs are automatically assigned, but
8430 sometimes it is more convenient to fix them to ease monitoring. This value
8431 must be strictly positive and unique within the listener/frontend. This
8432 option can only be used when defining only a single socket.
8433
8434interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +01008435 Restricts the socket to a specific interface. When specified, only packets
8436 received from that particular interface are processed by the socket. This is
8437 currently only supported on Linux. The interface must be a primary system
8438 interface, not an aliased interface. It is also possible to bind multiple
8439 frontends to the same address if they are bound to different interfaces. Note
8440 that binding to a network interface requires root privileges. This parameter
8441 is only compatible with TCPv4/TCPv6 sockets.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008442
Willy Tarreauabb175f2012-09-24 12:43:26 +02008443level <level>
8444 This setting is used with the stats sockets only to restrict the nature of
8445 the commands that can be issued on the socket. It is ignored by other
8446 sockets. <level> can be one of :
8447 - "user" is the least privileged level ; only non-sensitive stats can be
8448 read, and no change is allowed. It would make sense on systems where it
8449 is not easy to restrict access to the socket.
8450 - "operator" is the default level and fits most common uses. All data can
8451 be read, and only non-sensitive changes are permitted (eg: clear max
8452 counters).
8453 - "admin" should be used with care, as everything is permitted (eg: clear
8454 all counters).
8455
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008456maxconn <maxconn>
8457 Limits the sockets to this number of concurrent connections. Extraneous
8458 connections will remain in the system's backlog until a connection is
8459 released. If unspecified, the limit will be the same as the frontend's
8460 maxconn. Note that in case of port ranges or multiple addresses, the same
8461 value will be applied to each socket. This setting enables different
8462 limitations on expensive sockets, for instance SSL entries which may easily
8463 eat all memory.
8464
8465mode <mode>
8466 Sets the octal mode used to define access permissions on the UNIX socket. It
8467 can also be set by default in the global section's "unix-bind" statement.
8468 Note that some platforms simply ignore this. This setting is ignored by non
8469 UNIX sockets.
8470
8471mss <maxseg>
8472 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
8473 connections. This can be used to force a lower MSS for certain specific
8474 ports, for instance for connections passing through a VPN. Note that this
8475 relies on a kernel feature which is theoretically supported under Linux but
8476 was buggy in all versions prior to 2.6.28. It may or may not work on other
8477 operating systems. It may also not change the advertised value but change the
8478 effective size of outgoing segments. The commonly advertised value for TCPv4
8479 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
8480 positive, it will be used as the advertised MSS. If it is negative, it will
8481 indicate by how much to reduce the incoming connection's advertised MSS for
8482 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
8483
8484name <name>
8485 Sets an optional name for these sockets, which will be reported on the stats
8486 page.
8487
8488nice <nice>
8489 Sets the 'niceness' of connections initiated from the socket. Value must be
8490 in the range -1024..1024 inclusive, and defaults to zero. Positive values
8491 means that such connections are more friendly to others and easily offer
8492 their place in the scheduler. On the opposite, negative values mean that
8493 connections want to run with a higher priority than others. The difference
8494 only happens under high loads when the system is close to saturation.
8495 Negative values are appropriate for low-latency or administration services,
8496 and high values are generally recommended for CPU intensive tasks such as SSL
8497 processing or bulk transfers which are less sensible to latency. For example,
8498 it may make sense to use a positive value for an SMTP socket and a negative
8499 one for an RDP socket.
8500
Emeric Brun9b3009b2012-10-05 11:55:06 +02008501no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008502 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008503 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008504 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun42a3e202014-10-30 15:56:50 +01008505 be enabled using any configuration option. This option is also available on
8506 global statement "ssl-default-bind-options". See also "force-tls*",
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008507 and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008508
Emeric Brun90ad8722012-10-02 14:00:59 +02008509no-tls-tickets
8510 This setting is only available when support for OpenSSL was built in. It
8511 disables the stateless session resumption (RFC 5077 TLS Ticket
8512 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01008513 session resumption is more expensive in CPU usage. This option is also
8514 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +02008515
Emeric Brun9b3009b2012-10-05 11:55:06 +02008516no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008517 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008518 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008519 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008520 cannot be enabled using any configuration option. This option is also
8521 available on global statement "ssl-default-bind-options". See also
8522 "force-tlsv*", and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008523
Emeric Brun9b3009b2012-10-05 11:55:06 +02008524no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +02008525 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008526 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008527 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008528 cannot be enabled using any configuration option. This option is also
8529 available on global statement "ssl-default-bind-options". See also
8530 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008531
Emeric Brun9b3009b2012-10-05 11:55:06 +02008532no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +02008533 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008534 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008535 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun42a3e202014-10-30 15:56:50 +01008536 cannot be enabled using any configuration option. This option is also
8537 available on global statement "ssl-default-bind-options". See also
8538 "force-tlsv*", and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008539
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008540npn <protocols>
8541 This enables the NPN TLS extension and advertises the specified protocol list
8542 as supported on top of NPN. The protocol list consists in a comma-delimited
8543 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
8544 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +02008545 enabled (check with haproxy -vv). Note that the NPN extension has been
8546 replaced with the ALPN extension (see the "alpn" keyword).
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008547
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008548process [ all | odd | even | <number 1-64>[-<number 1-64>] ]
8549 This restricts the list of processes on which this listener is allowed to
8550 run. It does not enforce any process but eliminates those which do not match.
8551 If the frontend uses a "bind-process" setting, the intersection between the
8552 two is applied. If in the end the listener is not allowed to run on any
8553 remaining process, a warning is emitted, and the listener will either run on
8554 the first process of the listener if a single process was specified, or on
8555 all of its processes if multiple processes were specified. For the unlikely
Willy Tarreauae302532014-05-07 19:22:24 +02008556 case where several ranges are needed, this directive may be repeated. The
8557 main purpose of this directive is to be used with the stats sockets and have
8558 one different socket per process. The second purpose is to have multiple bind
8559 lines sharing the same IP:port but not the same process in a listener, so
8560 that the system can distribute the incoming connections into multiple queues
8561 and allow a smoother inter-process load balancing. Currently Linux 3.9 and
8562 above is known for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008563
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008564ssl
8565 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008566 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008567 certificate is necessary (see "crt" above). All contents in the buffers will
8568 appear in clear text, so that ACLs and HTTP processing will only have access
8569 to deciphered contents.
8570
Emmanuel Hocdet65623372013-01-24 17:17:15 +01008571strict-sni
8572 This setting is only available when support for OpenSSL was built in. The
8573 SSL/TLS negotiation is allow only if the client provided an SNI which match
8574 a certificate. The default certificate is not used.
8575 See the "crt" option for more information.
8576
Willy Tarreau1c862c52012-10-05 16:21:00 +02008577tfo
Lukas Tribus0defb902013-02-13 23:35:39 +01008578 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +02008579 enables TCP Fast Open on the listening socket, which means that clients which
8580 support this feature will be able to send a request and receive a response
8581 during the 3-way handshake starting from second connection, thus saving one
8582 round-trip after the first connection. This only makes sense with protocols
8583 that use high connection rates and where each round trip matters. This can
8584 possibly cause issues with many firewalls which do not accept data on SYN
8585 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +02008586 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
8587 need to build HAProxy with USE_TFO=1 if your libc doesn't define
8588 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +02008589
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008590transparent
8591 Is an optional keyword which is supported only on certain Linux kernels. It
8592 indicates that the addresses will be bound even if they do not belong to the
8593 local machine, and that packets targeting any of these addresses will be
8594 intercepted just as if the addresses were locally configured. This normally
8595 requires that IP forwarding is enabled. Caution! do not use this with the
8596 default address '*', as it would redirect any traffic for the specified port.
8597 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
8598 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
8599 kernel version. Some distribution kernels include backports of the feature,
8600 so check for support with your vendor.
8601
Willy Tarreau77e3af92012-11-24 15:07:23 +01008602v4v6
8603 Is an optional keyword which is supported only on most recent systems
8604 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
8605 and IPv6 when it uses the default address. Doing so is sometimes necessary
8606 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008607 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +01008608
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008609v6only
8610 Is an optional keyword which is supported only on most recent systems
8611 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
8612 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +01008613 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
8614 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008615
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008616uid <uid>
8617 Sets the owner of the UNIX sockets to the designated system uid. It can also
8618 be set by default in the global section's "unix-bind" statement. Note that
8619 some platforms simply ignore this. This setting is equivalent to the "user"
8620 setting except that the user numeric ID is used instead of its name. This
8621 setting is ignored by non UNIX sockets.
8622
8623user <user>
8624 Sets the owner of the UNIX sockets to the designated system user. It can also
8625 be set by default in the global section's "unix-bind" statement. Note that
8626 some platforms simply ignore this. This setting is equivalent to the "uid"
8627 setting except that the user name is used instead of its uid. This setting is
8628 ignored by non UNIX sockets.
8629
Emeric Brun1a073b42012-09-28 17:07:34 +02008630verify [none|optional|required]
8631 This setting is only available when support for OpenSSL was built in. If set
8632 to 'none', client certificate is not requested. This is the default. In other
8633 cases, a client certificate is requested. If the client does not provide a
8634 certificate after the request and if 'verify' is set to 'required', then the
8635 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +02008636 certificate provided by the client is always verified using CAs from
8637 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
8638 is aborted, regardless of the 'verify' option, unless the error code exactly
8639 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008640
Willy Tarreaub6205fd2012-09-24 12:27:33 +020086415.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +01008642------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02008643
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008644The "server" and "default-server" keywords support a certain number of settings
8645which are all passed as arguments on the server line. The order in which those
8646arguments appear does not count, and they are all optional. Some of those
8647settings are single words (booleans) while others expect one or several values
8648after them. In this case, the values must immediately follow the setting name.
8649Except default-server, all those settings must be specified after the server's
8650address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +02008651
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008652 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008653 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +02008654
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008655The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008656
Willy Tarreauceb4ac92012-04-28 00:41:46 +02008657addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008658 Using the "addr" parameter, it becomes possible to use a different IP address
8659 to send health-checks. On some servers, it may be desirable to dedicate an IP
8660 address to specific component able to perform complex tests which are more
8661 suitable to health-checks than the application. This parameter is ignored if
8662 the "check" parameter is not set. See also the "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008663
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008664 Supported in default-server: No
8665
Simon Hormand60d6912013-11-25 10:46:36 +09008666agent-check
8667 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +01008668 health check. An agent health check is performed by making a TCP connection
8669 to the port set by the "agent-port" parameter and reading an ASCII string.
8670 The string is made of a series of words delimited by spaces, tabs or commas
8671 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +09008672
Willy Tarreau81f5d942013-12-09 20:51:51 +01008673 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +09008674 Values in this format will set the weight proportional to the initial
Willy Tarreau47067ca2014-10-07 15:27:33 +02008675 weight of a server as configured when haproxy starts. Note that a zero
8676 weight is reported on the stats page as "DRAIN" since it has the same
8677 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +09008678
Willy Tarreau81f5d942013-12-09 20:51:51 +01008679 - The word "ready". This will turn the server's administrative state to the
8680 READY mode, thus cancelling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +09008681
Willy Tarreau81f5d942013-12-09 20:51:51 +01008682 - The word "drain". This will turn the server's administrative state to the
8683 DRAIN mode, thus it will not accept any new connections other than those
8684 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +09008685
Willy Tarreau81f5d942013-12-09 20:51:51 +01008686 - The word "maint". This will turn the server's administrative state to the
8687 MAINT mode, thus it will not accept any new connections at all, and health
8688 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +09008689
Willy Tarreau81f5d942013-12-09 20:51:51 +01008690 - The words "down", "failed", or "stopped", optionally followed by a
8691 description string after a sharp ('#'). All of these mark the server's
8692 operating state as DOWN, but since the word itself is reported on the stats
8693 page, the difference allows an administrator to know if the situation was
8694 expected or not : the service may intentionally be stopped, may appear up
8695 but fail some validity tests, or may be seen as down (eg: missing process,
8696 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +09008697
Willy Tarreau81f5d942013-12-09 20:51:51 +01008698 - The word "up" sets back the server's operating state as UP if health checks
8699 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +09008700
Willy Tarreau81f5d942013-12-09 20:51:51 +01008701 Parameters which are not advertised by the agent are not changed. For
8702 example, an agent might be designed to monitor CPU usage and only report a
8703 relative weight and never interact with the operating status. Similarly, an
8704 agent could be designed as an end-user interface with 3 radio buttons
8705 allowing an administrator to change only the administrative state. However,
8706 it is important to consider that only the agent may revert its own actions,
8707 so if a server is set to DRAIN mode or to DOWN state using the agent, the
8708 agent must implement the other equivalent actions to bring the service into
8709 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +09008710
Simon Horman2f1f9552013-11-25 10:46:37 +09008711 Failure to connect to the agent is not considered an error as connectivity
8712 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +01008713 parameter. Warning though, it is not a good idea to stop an agent after it
8714 reports "down", since only an agent reporting "up" will be able to turn the
8715 server up again. Note that the CLI on the Unix stats socket is also able to
8716 force an agent's result in order to workaround a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +09008717
Willy Tarreau81f5d942013-12-09 20:51:51 +01008718 Requires the "agent-port" parameter to be set. See also the "agent-inter"
8719 parameter.
Simon Hormand60d6912013-11-25 10:46:36 +09008720
8721 Supported in default-server: No
8722
8723agent-inter <delay>
8724 The "agent-inter" parameter sets the interval between two agent checks
8725 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8726
8727 Just as with every other time-based parameter, it may be entered in any
8728 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
8729 parameter also serves as a timeout for agent checks "timeout check" is
8730 not set. In order to reduce "resonance" effects when multiple servers are
8731 hosted on the same hardware, the agent and health checks of all servers
8732 are started with a small time offset between them. It is also possible to
8733 add some random noise in the agent and health checks interval using the
8734 global "spread-checks" keyword. This makes sense for instance when a lot
8735 of backends use the same servers.
8736
8737 See also the "agent-check" and "agent-port" parameters.
8738
8739 Supported in default-server: Yes
8740
8741agent-port <port>
8742 The "agent-port" parameter sets the TCP port used for agent checks.
8743
8744 See also the "agent-check" and "agent-inter" parameters.
8745
8746 Supported in default-server: Yes
8747
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008748backup
8749 When "backup" is present on a server line, the server is only used in load
8750 balancing when all other non-backup servers are unavailable. Requests coming
8751 with a persistence cookie referencing the server will always be served
8752 though. By default, only the first operational backup server is used, unless
8753 the "allbackups" option is set in the backend. See also the "allbackups"
8754 option.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008755
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008756 Supported in default-server: No
8757
Emeric Brunef42d922012-10-11 16:11:36 +02008758ca-file <cafile>
8759 This setting is only available when support for OpenSSL was built in. It
8760 designates a PEM file from which to load CA certificates used to verify
8761 server's certificate.
8762
8763 Supported in default-server: No
8764
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008765check
8766 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +01008767 always considered available. If "check" is set, the server is available when
8768 accepting periodic TCP connections, to ensure that it is really able to serve
8769 requests. The default address and port to send the tests to are those of the
8770 server, and the default source is the same as the one defined in the
8771 backend. It is possible to change the address using the "addr" parameter, the
8772 port using the "port" parameter, the source address using the "source"
8773 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +09008774 parameters. The request method is define in the backend using the "httpchk",
8775 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
8776 refer to those options and parameters for more information.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008777
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008778 Supported in default-server: No
8779
Willy Tarreau6c16adc2012-10-05 00:04:16 +02008780check-send-proxy
8781 This option forces emission of a PROXY protocol line with outgoing health
8782 checks, regardless of whether the server uses send-proxy or not for the
8783 normal traffic. By default, the PROXY protocol is enabled for health checks
8784 if it is already enabled for normal traffic and if no "port" nor "addr"
8785 directive is present. However, if such a directive is present, the
8786 "check-send-proxy" option needs to be used to force the use of the
8787 protocol. See also the "send-proxy" option for more information.
8788
8789 Supported in default-server: No
8790
Willy Tarreau763a95b2012-10-04 23:15:39 +02008791check-ssl
8792 This option forces encryption of all health checks over SSL, regardless of
8793 whether the server uses SSL or not for the normal traffic. This is generally
8794 used when an explicit "port" or "addr" directive is specified and SSL health
8795 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008796 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +02008797 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
8798 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
8799 All SSL settings are common to health checks and traffic (eg: ciphers).
8800 See the "ssl" option for more information.
8801
8802 Supported in default-server: No
8803
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008804ciphers <ciphers>
8805 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008806 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008807 string is defined in "man 1 ciphers". When SSL is used to communicate with
8808 servers on the local network, it is common to see a weaker set of algorithms
8809 than what is used over the internet. Doing so reduces CPU usage on both the
8810 server and haproxy while still keeping it compatible with deployed software.
8811 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
8812 is needed and just connectivity, using DES can be appropriate.
8813
Willy Tarreau763a95b2012-10-04 23:15:39 +02008814 Supported in default-server: No
8815
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008816cookie <value>
8817 The "cookie" parameter sets the cookie value assigned to the server to
8818 <value>. This value will be checked in incoming requests, and the first
8819 operational server possessing the same value will be selected. In return, in
8820 cookie insertion or rewrite modes, this value will be assigned to the cookie
8821 sent to the client. There is nothing wrong in having several servers sharing
8822 the same cookie value, and it is in fact somewhat common between normal and
8823 backup servers. See also the "cookie" keyword in backend section.
8824
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008825 Supported in default-server: No
8826
Emeric Brunef42d922012-10-11 16:11:36 +02008827crl-file <crlfile>
8828 This setting is only available when support for OpenSSL was built in. It
8829 designates a PEM file from which to load certificate revocation list used
8830 to verify server's certificate.
8831
8832 Supported in default-server: No
8833
Emeric Bruna7aa3092012-10-26 12:58:00 +02008834crt <cert>
8835 This setting is only available when support for OpenSSL was built in.
8836 It designates a PEM file from which to load both a certificate and the
8837 associated private key. This file can be built by concatenating both PEM
8838 files into one. This certificate will be sent if the server send a client
8839 certificate request.
8840
8841 Supported in default-server: No
8842
Willy Tarreau96839092010-03-29 10:02:24 +02008843disabled
8844 The "disabled" keyword starts the server in the "disabled" state. That means
8845 that it is marked down in maintenance mode, and no connection other than the
8846 ones allowed by persist mode will reach it. It is very well suited to setup
8847 new servers, because normal traffic will never reach them, while it is still
8848 possible to test the service by making use of the force-persist mechanism.
8849
8850 Supported in default-server: No
8851
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008852error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +01008853 If health observing is enabled, the "error-limit" parameter specifies the
8854 number of consecutive errors that triggers event selected by the "on-error"
8855 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008856
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008857 Supported in default-server: Yes
8858
8859 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008860
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008861fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008862 The "fall" parameter states that a server will be considered as dead after
8863 <count> consecutive unsuccessful health checks. This value defaults to 3 if
8864 unspecified. See also the "check", "inter" and "rise" parameters.
8865
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008866 Supported in default-server: Yes
8867
Emeric Brun8694b9a2012-10-05 14:39:07 +02008868force-sslv3
8869 This option enforces use of SSLv3 only when SSL is used to communicate with
8870 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun42a3e202014-10-30 15:56:50 +01008871 high connection rates. This option is also available on global statement
8872 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008873
8874 Supported in default-server: No
8875
8876force-tlsv10
8877 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008878 the server. This option is also available on global statement
8879 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008880
8881 Supported in default-server: No
8882
8883force-tlsv11
8884 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008885 the server. This option is also available on global statement
8886 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008887
8888 Supported in default-server: No
8889
8890force-tlsv12
8891 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun42a3e202014-10-30 15:56:50 +01008892 the server. This option is also available on global statement
8893 "ssl-default-server-options". See also "no-tlsv*", "no-sslv3".
Emeric Brun8694b9a2012-10-05 14:39:07 +02008894
8895 Supported in default-server: No
8896
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008897id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02008898 Set a persistent ID for the server. This ID must be positive and unique for
8899 the proxy. An unused ID will automatically be assigned if unset. The first
8900 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008901
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008902 Supported in default-server: No
8903
8904inter <delay>
8905fastinter <delay>
8906downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008907 The "inter" parameter sets the interval between two consecutive health checks
8908 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8909 It is also possible to use "fastinter" and "downinter" to optimize delays
8910 between checks depending on the server state :
8911
8912 Server state | Interval used
8913 ---------------------------------+-----------------------------------------
8914 UP 100% (non-transitional) | "inter"
8915 ---------------------------------+-----------------------------------------
8916 Transitionally UP (going down), |
8917 Transitionally DOWN (going up), | "fastinter" if set, "inter" otherwise.
8918 or yet unchecked. |
8919 ---------------------------------+-----------------------------------------
8920 DOWN 100% (non-transitional) | "downinter" if set, "inter" otherwise.
8921 ---------------------------------+-----------------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +01008922
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008923 Just as with every other time-based parameter, they can be entered in any
8924 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
8925 serves as a timeout for health checks sent to servers if "timeout check" is
8926 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +09008927 hosted on the same hardware, the agent and health checks of all servers
8928 are started with a small time offset between them. It is also possible to
8929 add some random noise in the agent and health checks interval using the
8930 global "spread-checks" keyword. This makes sense for instance when a lot
8931 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008932
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008933 Supported in default-server: Yes
8934
8935maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008936 The "maxconn" parameter specifies the maximal number of concurrent
8937 connections that will be sent to this server. If the number of incoming
8938 concurrent requests goes higher than this value, they will be queued, waiting
8939 for a connection to be released. This parameter is very important as it can
8940 save fragile servers from going down under extreme loads. If a "minconn"
8941 parameter is specified, the limit becomes dynamic. The default value is "0"
8942 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
8943 the backend's "fullconn" keyword.
8944
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008945 Supported in default-server: Yes
8946
8947maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008948 The "maxqueue" parameter specifies the maximal number of connections which
8949 will wait in the queue for this server. If this limit is reached, next
8950 requests will be redispatched to other servers instead of indefinitely
8951 waiting to be served. This will break persistence but may allow people to
8952 quickly re-log in when the server they try to connect to is dying. The
8953 default value is "0" which means the queue is unlimited. See also the
8954 "maxconn" and "minconn" parameters.
8955
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008956 Supported in default-server: Yes
8957
8958minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008959 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
8960 limit following the backend's load. The server will always accept at least
8961 <minconn> connections, never more than <maxconn>, and the limit will be on
8962 the ramp between both values when the backend has less than <fullconn>
8963 concurrent connections. This makes it possible to limit the load on the
8964 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008965 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008966 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008967
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008968 Supported in default-server: Yes
8969
Emeric Brun9b3009b2012-10-05 11:55:06 +02008970no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008971 This option disables support for SSLv3 when SSL is used to communicate with
8972 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emeric Brun8694b9a2012-10-05 14:39:07 +02008973 using any configuration option. See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008974
Willy Tarreau763a95b2012-10-04 23:15:39 +02008975 Supported in default-server: No
8976
Emeric Brunf9c5c472012-10-11 15:28:34 +02008977no-tls-tickets
8978 This setting is only available when support for OpenSSL was built in. It
8979 disables the stateless session resumption (RFC 5077 TLS Ticket
8980 extension) and force to use stateful session resumption. Stateless
Emeric Brun42a3e202014-10-30 15:56:50 +01008981 session resumption is more expensive in CPU usage for servers. This option
8982 is also available on global statement "ssl-default-server-options".
Emeric Brunf9c5c472012-10-11 15:28:34 +02008983
8984 Supported in default-server: No
8985
Emeric Brun9b3009b2012-10-05 11:55:06 +02008986no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +02008987 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02008988 the server. Note that SSLv2 is disabled in the code and cannot be enabled
8989 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01008990 often makes sense to disable it when communicating with local servers. This
8991 option is also available on global statement "ssl-default-server-options".
8992 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02008993
Willy Tarreau763a95b2012-10-04 23:15:39 +02008994 Supported in default-server: No
8995
Emeric Brun9b3009b2012-10-05 11:55:06 +02008996no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +02008997 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02008998 the server. Note that SSLv2 is disabled in the code and cannot be enabled
8999 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009000 often makes sense to disable it when communicating with local servers. This
9001 option is also available on global statement "ssl-default-server-options".
9002 See also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009003
Willy Tarreau763a95b2012-10-04 23:15:39 +02009004 Supported in default-server: No
9005
Emeric Brun9b3009b2012-10-05 11:55:06 +02009006no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +02009007 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009008 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9009 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun42a3e202014-10-30 15:56:50 +01009010 often makes sense to disable it when communicating with local servers. This
9011 option is also available on global statement "ssl-default-server-options".
9012 See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009013
Willy Tarreau763a95b2012-10-04 23:15:39 +02009014 Supported in default-server: No
9015
Simon Hormanfa461682011-06-25 09:39:49 +09009016non-stick
9017 Never add connections allocated to this sever to a stick-table.
9018 This may be used in conjunction with backup to ensure that
9019 stick-table persistence is disabled for backup servers.
9020
Willy Tarreau763a95b2012-10-04 23:15:39 +02009021 Supported in default-server: No
9022
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009023observe <mode>
9024 This option enables health adjusting based on observing communication with
9025 the server. By default this functionality is disabled and enabling it also
9026 requires to enable health checks. There are two supported modes: "layer4" and
9027 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
9028 significant. In layer7, which is only allowed for http proxies, responses
9029 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +01009030 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009031
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009032 Supported in default-server: No
9033
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009034 See also the "check", "on-error" and "error-limit".
9035
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009036on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009037 Select what should happen when enough consecutive errors are detected.
9038 Currently, four modes are available:
9039 - fastinter: force fastinter
9040 - fail-check: simulate a failed check, also forces fastinter (default)
9041 - sudden-death: simulate a pre-fatal failed health check, one more failed
9042 check will mark a server down, forces fastinter
9043 - mark-down: mark the server immediately down and force fastinter
9044
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009045 Supported in default-server: Yes
9046
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009047 See also the "check", "observe" and "error-limit".
9048
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009049on-marked-down <action>
9050 Modify what occurs when a server is marked down.
9051 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009052 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
9053 all connections to the server are immediately terminated when the server
9054 goes down. It might be used if the health check detects more complex cases
9055 than a simple connection status, and long timeouts would cause the service
9056 to remain unresponsive for too long a time. For instance, a health check
9057 might detect that a database is stuck and that there's no chance to reuse
9058 existing connections anymore. Connections killed this way are logged with
9059 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009060
9061 Actions are disabled by default
9062
9063 Supported in default-server: Yes
9064
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009065on-marked-up <action>
9066 Modify what occurs when a server is marked up.
9067 Currently one action is available:
9068 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
9069 done only if the server is not in backup state and if it is not disabled
9070 (it must have an effective weight > 0). This can be used sometimes to force
9071 an active server to take all the traffic back after recovery when dealing
9072 with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble
9073 than it tries to solve (eg: incomplete transactions), so use this feature
9074 with extreme care. Sessions killed because a server comes up are logged
9075 with an 'U' termination code (for "Up").
9076
9077 Actions are disabled by default
9078
9079 Supported in default-server: Yes
9080
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009081port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009082 Using the "port" parameter, it becomes possible to use a different port to
9083 send health-checks. On some servers, it may be desirable to dedicate a port
9084 to a specific component able to perform complex tests which are more suitable
9085 to health-checks than the application. It is common to run a simple script in
9086 inetd for instance. This parameter is ignored if the "check" parameter is not
9087 set. See also the "addr" parameter.
9088
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009089 Supported in default-server: Yes
9090
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009091redir <prefix>
9092 The "redir" parameter enables the redirection mode for all GET and HEAD
9093 requests addressing this server. This means that instead of having HAProxy
9094 forward the request to the server, it will send an "HTTP 302" response with
9095 the "Location" header composed of this prefix immediately followed by the
9096 requested URI beginning at the leading '/' of the path component. That means
9097 that no trailing slash should be used after <prefix>. All invalid requests
9098 will be rejected, and all non-GET or HEAD requests will be normally served by
9099 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009100 mangling nor cookie insertion is possible in the response. However, cookies in
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009101 requests are still analysed, making this solution completely usable to direct
9102 users to a remote location in case of local disaster. Main use consists in
9103 increasing bandwidth for static servers by having the clients directly
9104 connect to them. Note: never use a relative location here, it would cause a
9105 loop between the client and HAProxy!
9106
9107 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
9108
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009109 Supported in default-server: No
9110
9111rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009112 The "rise" parameter states that a server will be considered as operational
9113 after <count> consecutive successful health checks. This value defaults to 2
9114 if unspecified. See also the "check", "inter" and "fall" parameters.
9115
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009116 Supported in default-server: Yes
9117
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009118send-proxy
9119 The "send-proxy" parameter enforces use of the PROXY protocol over any
9120 connection established to this server. The PROXY protocol informs the other
9121 end about the layer 3/4 addresses of the incoming connection, so that it can
9122 know the client's address or the public address it accessed to, whatever the
9123 upper layer protocol. For connections accepted by an "accept-proxy" listener,
9124 the advertised address will be used. Only TCPv4 and TCPv6 address families
9125 are supported. Other families such as Unix sockets, will report an UNKNOWN
9126 family. Servers using this option can fully be chained to another instance of
9127 haproxy listening with an "accept-proxy" setting. This setting must not be
Willy Tarreau6c16adc2012-10-05 00:04:16 +02009128 used if the server isn't aware of the protocol. When health checks are sent
9129 to the server, the PROXY protocol is automatically used when this option is
9130 set, unless there is an explicit "port" or "addr" directive, in which case an
9131 explicit "check-send-proxy" directive would also be needed to use the PROXY
9132 protocol. See also the "accept-proxy" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009133
9134 Supported in default-server: No
9135
David Safb76832014-05-08 23:42:08 -04009136send-proxy-v2
9137 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
9138 over any connection established to this server. The PROXY protocol informs
9139 the other end about the layer 3/4 addresses of the incoming connection, so
9140 that it can know the client's address or the public address it accessed to,
9141 whatever the upper layer protocol. This setting must not be used if the
9142 server isn't aware of this version of the protocol. See also the "send-proxy"
9143 option of the "bind" keyword.
9144
9145 Supported in default-server: No
9146
9147send-proxy-v2-ssl
9148 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9149 2 over any connection established to this server. The PROXY protocol informs
9150 the other end about the layer 3/4 addresses of the incoming connection, so
9151 that it can know the client's address or the public address it accessed to,
9152 whatever the upper layer protocol. In addition, the SSL information extension
9153 of the PROXY protocol is added to the PROXY protocol header. This setting
9154 must not be used if the server isn't aware of this version of the protocol.
9155 See also the "send-proxy-v2" option of the "bind" keyword.
9156
9157 Supported in default-server: No
9158
9159send-proxy-v2-ssl-cn
9160 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9161 2 over any connection established to this server. The PROXY protocol informs
9162 the other end about the layer 3/4 addresses of the incoming connection, so
9163 that it can know the client's address or the public address it accessed to,
9164 whatever the upper layer protocol. In addition, the SSL information extension
9165 of the PROXY protocol, along along with the Common Name from the subject of
9166 the client certificate (if any), is added to the PROXY protocol header. This
9167 setting must not be used if the server isn't aware of this version of the
9168 protocol. See also the "send-proxy-v2" option of the "bind" keyword.
9169
9170 Supported in default-server: No
9171
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009172slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009173 The "slowstart" parameter for a server accepts a value in milliseconds which
9174 indicates after how long a server which has just come back up will run at
9175 full speed. Just as with every other time-based parameter, it can be entered
9176 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
9177 linearly from 0 to 100% during this time. The limitation applies to two
9178 parameters :
9179
9180 - maxconn: the number of connections accepted by the server will grow from 1
9181 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
9182
9183 - weight: when the backend uses a dynamic weighted algorithm, the weight
9184 grows linearly from 1 to 100%. In this case, the weight is updated at every
9185 health-check. For this reason, it is important that the "inter" parameter
9186 is smaller than the "slowstart", in order to maximize the number of steps.
9187
9188 The slowstart never applies when haproxy starts, otherwise it would cause
9189 trouble to running servers. It only applies when a server has been previously
9190 seen as failed.
9191
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009192 Supported in default-server: Yes
9193
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009194source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009195source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009196source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009197 The "source" parameter sets the source address which will be used when
9198 connecting to the server. It follows the exact same parameters and principle
9199 as the backend "source" keyword, except that it only applies to the server
9200 referencing it. Please consult the "source" keyword for details.
9201
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009202 Additionally, the "source" statement on a server line allows one to specify a
9203 source port range by indicating the lower and higher bounds delimited by a
9204 dash ('-'). Some operating systems might require a valid IP address when a
9205 source port range is specified. It is permitted to have the same IP/range for
9206 several servers. Doing so makes it possible to bypass the maximum of 64k
9207 total concurrent connections. The limit will then reach 64k connections per
9208 server.
9209
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009210 Supported in default-server: No
9211
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009212ssl
Willy Tarreau44f65392013-06-25 07:56:20 +02009213 This option enables SSL ciphering on outgoing connections to the server. It
9214 is critical to verify server certificates using "verify" when using SSL to
9215 connect to servers, otherwise the communication is prone to trivial man in
9216 the-middle attacks rendering SSL useless. When this option is used, health
9217 checks are automatically sent in SSL too unless there is a "port" or an
9218 "addr" directive indicating the check should be sent to a different location.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009219 See the "check-ssl" option to force SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +02009220
9221 Supported in default-server: No
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009222
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009223track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +02009224 This option enables ability to set the current state of the server by tracking
9225 another one. It is possible to track a server which itself tracks another
9226 server, provided that at the end of the chain, a server has health checks
9227 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009228 used, it has to be enabled on both proxies.
9229
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009230 Supported in default-server: No
9231
Emeric Brunef42d922012-10-11 16:11:36 +02009232verify [none|required]
9233 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +01009234 to 'none', server certificate is not verified. In the other case, The
9235 certificate provided by the server is verified using CAs from 'ca-file'
9236 and optional CRLs from 'crl-file'. If 'ssl_server_verify' is not specified
9237 in global section, this is the default. On verify failure the handshake
Willy Tarreau44f65392013-06-25 07:56:20 +02009238 is aborted. It is critically important to verify server certificates when
9239 using SSL to connect to servers, otherwise the communication is prone to
9240 trivial man-in-the-middle attacks rendering SSL totally useless.
Emeric Brunef42d922012-10-11 16:11:36 +02009241
9242 Supported in default-server: No
9243
Evan Broderbe554312013-06-27 00:05:25 -07009244verifyhost <hostname>
9245 This setting is only available when support for OpenSSL was built in, and
9246 only takes effect if 'verify required' is also specified. When set, the
9247 hostnames in the subject and subjectAlternateNames of the certificate
9248 provided by the server are checked. If none of the hostnames in the
9249 certificate match the specified hostname, the handshake is aborted. The
9250 hostnames in the server-provided certificate may include wildcards.
9251
9252 Supported in default-server: No
9253
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009254weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009255 The "weight" parameter is used to adjust the server's weight relative to
9256 other servers. All servers will receive a load proportional to their weight
9257 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +02009258 load. The default weight is 1, and the maximal value is 256. A value of 0
9259 means the server will not participate in load-balancing but will still accept
9260 persistent connections. If this parameter is used to distribute the load
9261 according to server's capacity, it is recommended to start with values which
9262 can both grow and shrink, for instance between 10 and 100 to leave enough
9263 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009264
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009265 Supported in default-server: Yes
9266
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009267
92686. HTTP header manipulation
9269---------------------------
9270
9271In HTTP mode, it is possible to rewrite, add or delete some of the request and
9272response headers based on regular expressions. It is also possible to block a
9273request or a response if a particular header matches a regular expression,
9274which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +01009275against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009276
Willy Tarreau70dffda2014-01-30 03:07:23 +01009277If HAProxy encounters an "Informational Response" (status code 1xx), it is able
9278to process all rsp* rules which can allow, deny, rewrite or delete a header,
9279but it will refuse to add a header to any such messages as this is not
9280HTTP-compliant. The reason for still processing headers in such responses is to
9281stop and/or fix any possible information leak which may happen, for instance
9282because another downstream equipment would unconditionally add a header, or if
9283a server name appears there. When such messages are seen, normal processing
9284still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +02009285
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009286This section covers common usage of the following keywords, described in detail
9287in section 4.2 :
9288
9289 - reqadd <string>
9290 - reqallow <search>
9291 - reqiallow <search>
9292 - reqdel <search>
9293 - reqidel <search>
9294 - reqdeny <search>
9295 - reqideny <search>
9296 - reqpass <search>
9297 - reqipass <search>
9298 - reqrep <search> <replace>
9299 - reqirep <search> <replace>
9300 - reqtarpit <search>
9301 - reqitarpit <search>
9302 - rspadd <string>
9303 - rspdel <search>
9304 - rspidel <search>
9305 - rspdeny <search>
9306 - rspideny <search>
9307 - rsprep <search> <replace>
9308 - rspirep <search> <replace>
9309
9310With all these keywords, the same conventions are used. The <search> parameter
9311is a POSIX extended regular expression (regex) which supports grouping through
9312parenthesis (without the backslash). Spaces and other delimiters must be
9313prefixed with a backslash ('\') to avoid confusion with a field delimiter.
9314Other characters may be prefixed with a backslash to change their meaning :
9315
9316 \t for a tab
9317 \r for a carriage return (CR)
9318 \n for a new line (LF)
9319 \ to mark a space and differentiate it from a delimiter
9320 \# to mark a sharp and differentiate it from a comment
9321 \\ to use a backslash in a regex
9322 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
9323 \xXX to write the ASCII hex code XX as in the C language
9324
9325The <replace> parameter contains the string to be used to replace the largest
9326portion of text matching the regex. It can make use of the special characters
9327above, and can reference a substring which is delimited by parenthesis in the
9328regex, by writing a backslash ('\') immediately followed by one digit from 0 to
93299 indicating the group position (0 designating the entire line). This practice
9330is very common to users of the "sed" program.
9331
9332The <string> parameter represents the string which will systematically be added
9333after the last header line. It can also use special character sequences above.
9334
9335Notes related to these keywords :
9336---------------------------------
9337 - these keywords are not always convenient to allow/deny based on header
9338 contents. It is strongly recommended to use ACLs with the "block" keyword
9339 instead, resulting in far more flexible and manageable rules.
9340
9341 - lines are always considered as a whole. It is not possible to reference
9342 a header name only or a value only. This is important because of the way
9343 headers are written (notably the number of spaces after the colon).
9344
9345 - the first line is always considered as a header, which makes it possible to
9346 rewrite or filter HTTP requests URIs or response codes, but in turn makes
9347 it harder to distinguish between headers and request line. The regex prefix
9348 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
9349 ^[^ \t:]*: matches any header name followed by a colon.
9350
9351 - for performances reasons, the number of characters added to a request or to
9352 a response is limited at build time to values between 1 and 4 kB. This
9353 should normally be far more than enough for most usages. If it is too short
9354 on occasional usages, it is possible to gain some space by removing some
9355 useless headers before adding new ones.
9356
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009357 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009358 without the 'i' letter except that they ignore case when matching patterns.
9359
9360 - when a request passes through a frontend then a backend, all req* rules
9361 from the frontend will be evaluated, then all req* rules from the backend
9362 will be evaluated. The reverse path is applied to responses.
9363
9364 - req* statements are applied after "block" statements, so that "block" is
9365 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +01009366 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009367
9368
Willy Tarreau74ca5042013-06-11 23:12:07 +020093697. Using ACLs and fetching samples
9370----------------------------------
9371
9372Haproxy is capable of extracting data from request or response streams, from
9373client or server information, from tables, environmental information etc...
9374The action of extracting such data is called fetching a sample. Once retrieved,
9375these samples may be used for various purposes such as a key to a stick-table,
9376but most common usages consist in matching them against predefined constant
9377data called patterns.
9378
9379
93807.1. ACL basics
9381---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009382
9383The use of Access Control Lists (ACL) provides a flexible solution to perform
9384content switching and generally to take decisions based on content extracted
9385from the request, the response or any environmental status. The principle is
9386simple :
9387
Willy Tarreau74ca5042013-06-11 23:12:07 +02009388 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009389 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +02009390 - apply one or multiple pattern matching methods on this sample
9391 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009392
Willy Tarreau74ca5042013-06-11 23:12:07 +02009393The actions generally consist in blocking a request, selecting a backend, or
9394adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009395
9396In order to define a test, the "acl" keyword is used. The syntax is :
9397
Willy Tarreau74ca5042013-06-11 23:12:07 +02009398 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009399
9400This creates a new ACL <aclname> or completes an existing one with new tests.
9401Those tests apply to the portion of request/response specified in <criterion>
9402and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009403an operator which may be specified before the set of values. Optionally some
9404conversion operators may be applied to the sample, and they will be specified
9405as a comma-delimited list of keywords just after the first keyword. The values
9406are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009407
9408ACL names must be formed from upper and lower case letters, digits, '-' (dash),
9409'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
9410which means that "my_acl" and "My_Acl" are two different ACLs.
9411
9412There is no enforced limit to the number of ACLs. The unused ones do not affect
9413performance, they just consume a small amount of memory.
9414
Willy Tarreau74ca5042013-06-11 23:12:07 +02009415The criterion generally is the name of a sample fetch method, or one of its ACL
9416specific declinations. The default test method is implied by the output type of
9417this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009418methods of a same sample fetch method. The sample fetch methods are the only
9419ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009420
9421Sample fetch methods return data which can be of the following types :
9422 - boolean
9423 - integer (signed or unsigned)
9424 - IPv4 or IPv6 address
9425 - string
9426 - data block
9427
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009428Converters transform any of these data into any of these. For example, some
9429converters might convert a string to a lower-case string while other ones
9430would turn a string to an IPv4 address, or apply a netmask to an IP address.
9431The resulting sample is of the type of the last converter applied to the list,
9432which defaults to the type of the sample fetch method.
9433
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009434Each sample or converter returns data of a specific type, specified with its
9435keyword in this documentation. When an ACL is declared using a standard sample
9436fetch method, certain types automatically involved a default matching method
9437which are summarized in the table below :
9438
9439 +---------------------+-----------------+
9440 | Sample or converter | Default |
9441 | output type | matching method |
9442 +---------------------+-----------------+
9443 | boolean | bool |
9444 +---------------------+-----------------+
9445 | integer | int |
9446 +---------------------+-----------------+
9447 | ip | ip |
9448 +---------------------+-----------------+
9449 | string | str |
9450 +---------------------+-----------------+
9451 | binary | none, use "-m" |
9452 +---------------------+-----------------+
9453
9454Note that in order to match a binary samples, it is mandatory to specify a
9455matching method, see below.
9456
Willy Tarreau74ca5042013-06-11 23:12:07 +02009457The ACL engine can match these types against patterns of the following types :
9458 - boolean
9459 - integer or integer range
9460 - IP address / network
9461 - string (exact, substring, suffix, prefix, subdir, domain)
9462 - regular expression
9463 - hex block
9464
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009465The following ACL flags are currently supported :
9466
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009467 -i : ignore case during matching of all subsequent patterns.
9468 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009469 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009470 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009471 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009472 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +02009473 -- : force end of flags. Useful when a string looks like one of the flags.
9474
Willy Tarreau74ca5042013-06-11 23:12:07 +02009475The "-f" flag is followed by the name of a file from which all lines will be
9476read as individual values. It is even possible to pass multiple "-f" arguments
9477if the patterns are to be loaded from multiple files. Empty lines as well as
9478lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
9479will be stripped. If it is absolutely necessary to insert a valid pattern
9480beginning with a sharp, just prefix it with a space so that it is not taken for
9481a comment. Depending on the data type and match method, haproxy may load the
9482lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
9483exact string matching. In this case, duplicates will automatically be removed.
9484
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009485The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
9486parsed as two column file. The first column contains the patterns used by the
9487ACL, and the second column contain the samples. The sample can be used later by
9488a map. This can be useful in some rare cases where an ACL would just be used to
9489check for the existence of a pattern in a map before a mapping is applied.
9490
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009491The "-u" flag forces the unique id of the ACL. This unique id is used with the
9492socket interface to identify ACL and dynamically change its values. Note that a
9493file is always identified by its name even if an id is set.
9494
Willy Tarreau74ca5042013-06-11 23:12:07 +02009495Also, note that the "-i" flag applies to subsequent entries and not to entries
9496loaded from files preceding it. For instance :
9497
9498 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
9499
9500In this example, each line of "exact-ua.lst" will be exactly matched against
9501the "user-agent" header of the request. Then each line of "generic-ua" will be
9502case-insensitively matched. Then the word "test" will be insensitively matched
9503as well.
9504
9505The "-m" flag is used to select a specific pattern matching method on the input
9506sample. All ACL-specific criteria imply a pattern matching method and generally
9507do not need this flag. However, this flag is useful with generic sample fetch
9508methods to describe how they're going to be matched against the patterns. This
9509is required for sample fetches which return data type for which there is no
9510obvious matching method (eg: string or binary). When "-m" is specified and
9511followed by a pattern matching method name, this method is used instead of the
9512default one for the criterion. This makes it possible to match contents in ways
9513that were not initially planned, or with sample fetch methods which return a
9514string. The matching method also affects the way the patterns are parsed.
9515
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009516The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
9517By default, if the parser cannot parse ip address it considers that the parsed
9518string is maybe a domain name and try dns resolution. The flag "-n" disable this
9519resolution. It is useful for detecting malformed ip lists. Note that if the DNS
9520server is not reachable, the haproxy configuration parsing may last many minutes
9521waiting fir the timeout. During this time no error messages are displayed. The
9522flag "-n" disable this behavior. Note also that during the runtime, this
9523function is disabled for the dynamic acl modifications.
9524
Willy Tarreau74ca5042013-06-11 23:12:07 +02009525There are some restrictions however. Not all methods can be used with all
9526sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
9527be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +02009528
9529 - "found" : only check if the requested sample could be found in the stream,
9530 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +02009531 to pass any pattern to avoid confusion. This matching method is
9532 particularly useful to detect presence of certain contents such
9533 as headers, cookies, etc... even if they are empty and without
9534 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009535
9536 - "bool" : check the value as a boolean. It can only be applied to fetches
9537 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009538 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009539
9540 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +02009541 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009542
9543 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +02009544 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009545
9546 - "bin" : match the contents against an hexadecimal string representing a
9547 binary sequence. This may be used with binary or string samples.
9548
9549 - "len" : match the sample's length as an integer. This may be used with
9550 binary or string samples.
9551
Willy Tarreau74ca5042013-06-11 23:12:07 +02009552 - "str" : exact match : match the contents against a string. This may be
9553 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009554
Willy Tarreau74ca5042013-06-11 23:12:07 +02009555 - "sub" : substring match : check that the contents contain at least one of
9556 the provided string patterns. This may be used with binary or
9557 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009558
Willy Tarreau74ca5042013-06-11 23:12:07 +02009559 - "reg" : regex match : match the contents against a list of regular
9560 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009561
Willy Tarreau74ca5042013-06-11 23:12:07 +02009562 - "beg" : prefix match : check that the contents begin like the provided
9563 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009564
Willy Tarreau74ca5042013-06-11 23:12:07 +02009565 - "end" : suffix match : check that the contents end like the provided
9566 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009567
Willy Tarreau74ca5042013-06-11 23:12:07 +02009568 - "dir" : subdir match : check that a slash-delimited portion of the
9569 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009570 This may be used with binary or string samples.
9571
Willy Tarreau74ca5042013-06-11 23:12:07 +02009572 - "dom" : domain match : check that a dot-delimited portion of the contents
9573 exactly match one of the provided string patterns. This may be
9574 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009575
9576For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
9577request, it is possible to do :
9578
9579 acl jsess_present cook(JSESSIONID) -m found
9580
9581In order to apply a regular expression on the 500 first bytes of data in the
9582buffer, one would use the following acl :
9583
9584 acl script_tag payload(0,500) -m reg -i <script>
9585
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009586On systems where the regex library is much slower when using "-i", it is
9587possible to convert the sample to lowercase before matching, like this :
9588
9589 acl script_tag payload(0,500),lower -m reg <script>
9590
Willy Tarreau74ca5042013-06-11 23:12:07 +02009591All ACL-specific criteria imply a default matching method. Most often, these
9592criteria are composed by concatenating the name of the original sample fetch
9593method and the matching method. For example, "hdr_beg" applies the "beg" match
9594to samples retrieved using the "hdr" fetch method. Since all ACL-specific
9595criteria rely on a sample fetch method, it is always possible instead to use
9596the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009597
Willy Tarreau74ca5042013-06-11 23:12:07 +02009598If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009599the matching method is simply applied to the underlying sample fetch method.
9600For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009601
Willy Tarreau74ca5042013-06-11 23:12:07 +02009602 acl short_form hdr_beg(host) www.
9603 acl alternate1 hdr_beg(host) -m beg www.
9604 acl alternate2 hdr_dom(host) -m beg www.
9605 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009606
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009607
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009608The table below summarizes the compatibility matrix between sample or converter
9609types and the pattern types to fetch against. It indicates for each compatible
9610combination the name of the matching method to be used, surrounded with angle
9611brackets ">" and "<" when the method is the default one and will work by
9612default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009613
Willy Tarreau74ca5042013-06-11 23:12:07 +02009614 +-------------------------------------------------+
9615 | Input sample type |
9616 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009617 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009618 +----------------------+---------+---------+---------+---------+---------+
9619 | none (presence only) | found | found | found | found | found |
9620 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009621 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009622 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009623 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009624 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009625 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009626 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009627 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009628 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009629 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009630 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009631 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009632 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009633 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009634 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009635 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009636 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009637 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009638 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009639 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009640 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009641 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009642 +----------------------+---------+---------+---------+---------+---------+
9643 | hex block | | | | bin | bin |
9644 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +02009645
9646
Willy Tarreau74ca5042013-06-11 23:12:07 +020096477.1.1. Matching booleans
9648------------------------
9649
9650In order to match a boolean, no value is needed and all values are ignored.
9651Boolean matching is used by default for all fetch methods of type "boolean".
9652When boolean matching is used, the fetched value is returned as-is, which means
9653that a boolean "true" will always match and a boolean "false" will never match.
9654
9655Boolean matching may also be enforced using "-m bool" on fetch methods which
9656return an integer value. Then, integer value 0 is converted to the boolean
9657"false" and all other values are converted to "true".
9658
Willy Tarreau6a06a402007-07-15 20:15:28 +02009659
Willy Tarreau74ca5042013-06-11 23:12:07 +020096607.1.2. Matching integers
9661------------------------
9662
9663Integer matching applies by default to integer fetch methods. It can also be
9664enforced on boolean fetches using "-m int". In this case, "false" is converted
9665to the integer 0, and "true" is converted to the integer 1.
9666
9667Integer matching also supports integer ranges and operators. Note that integer
9668matching only applies to positive values. A range is a value expressed with a
9669lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009670
9671For instance, "1024:65535" is a valid range to represent a range of
9672unprivileged ports, and "1024:" would also work. "0:1023" is a valid
9673representation of privileged ports, and ":1023" would also work.
9674
Willy Tarreau62644772008-07-16 18:36:06 +02009675As a special case, some ACL functions support decimal numbers which are in fact
9676two integers separated by a dot. This is used with some version checks for
9677instance. All integer properties apply to those decimal numbers, including
9678ranges and operators.
9679
Willy Tarreau6a06a402007-07-15 20:15:28 +02009680For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +01009681operators with ranges does not make much sense and is strongly discouraged.
9682Similarly, it does not make much sense to perform order comparisons with a set
9683of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009684
Willy Tarreau0ba27502007-12-24 16:55:16 +01009685Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009686
9687 eq : true if the tested value equals at least one value
9688 ge : true if the tested value is greater than or equal to at least one value
9689 gt : true if the tested value is greater than at least one value
9690 le : true if the tested value is less than or equal to at least one value
9691 lt : true if the tested value is less than at least one value
9692
Willy Tarreau0ba27502007-12-24 16:55:16 +01009693For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009694
9695 acl negative-length hdr_val(content-length) lt 0
9696
Willy Tarreau62644772008-07-16 18:36:06 +02009697This one matches SSL versions between 3.0 and 3.1 (inclusive) :
9698
9699 acl sslv3 req_ssl_ver 3:3.1
9700
Willy Tarreau6a06a402007-07-15 20:15:28 +02009701
Willy Tarreau74ca5042013-06-11 23:12:07 +020097027.1.3. Matching strings
9703-----------------------
9704
9705String matching applies to string or binary fetch methods, and exists in 6
9706different forms :
9707
9708 - exact match (-m str) : the extracted string must exactly match the
9709 patterns ;
9710
9711 - substring match (-m sub) : the patterns are looked up inside the
9712 extracted string, and the ACL matches if any of them is found inside ;
9713
9714 - prefix match (-m beg) : the patterns are compared with the beginning of
9715 the extracted string, and the ACL matches if any of them matches.
9716
9717 - suffix match (-m end) : the patterns are compared with the end of the
9718 extracted string, and the ACL matches if any of them matches.
9719
9720 - subdir match (-m sub) : the patterns are looked up inside the extracted
9721 string, delimited with slashes ("/"), and the ACL matches if any of them
9722 matches.
9723
9724 - domain match (-m dom) : the patterns are looked up inside the extracted
9725 string, delimited with dots ("."), and the ACL matches if any of them
9726 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009727
9728String matching applies to verbatim strings as they are passed, with the
9729exception of the backslash ("\") which makes it possible to escape some
9730characters such as the space. If the "-i" flag is passed before the first
9731string, then the matching will be performed ignoring the case. In order
9732to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +01009733before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009734
9735
Willy Tarreau74ca5042013-06-11 23:12:07 +020097367.1.4. Matching regular expressions (regexes)
9737---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009738
9739Just like with string matching, regex matching applies to verbatim strings as
9740they are passed, with the exception of the backslash ("\") which makes it
9741possible to escape some characters such as the space. If the "-i" flag is
9742passed before the first regex, then the matching will be performed ignoring
9743the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +01009744the "--" flag before the first string. Same principle applies of course to
9745match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009746
9747
Willy Tarreau74ca5042013-06-11 23:12:07 +020097487.1.5. Matching arbitrary data blocks
9749-------------------------------------
9750
9751It is possible to match some extracted samples against a binary block which may
9752not safely be represented as a string. For this, the patterns must be passed as
9753a series of hexadecimal digits in an even number, when the match method is set
9754to binary. Each sequence of two digits will represent a byte. The hexadecimal
9755digits may be used upper or lower case.
9756
9757Example :
9758 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
9759 acl hello payload(0,6) -m bin 48656c6c6f0a
9760
9761
97627.1.6. Matching IPv4 and IPv6 addresses
9763---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009764
9765IPv4 addresses values can be specified either as plain addresses or with a
9766netmask appended, in which case the IPv4 address matches whenever it is
9767within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009768host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +01009769difficult to read and debug configurations. If hostnames are used, you should
9770at least ensure that they are present in /etc/hosts so that the configuration
9771does not depend on any random DNS match at the moment the configuration is
9772parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009773
Willy Tarreauceb4ac92012-04-28 00:41:46 +02009774IPv6 may be entered in their usual form, with or without a netmask appended.
9775Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
9776trouble with randomly resolved IP addresses, host names are never allowed in
9777IPv6 patterns.
9778
9779HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
9780following situations :
9781 - tested address is IPv4, pattern address is IPv4, the match applies
9782 in IPv4 using the supplied mask if any.
9783 - tested address is IPv6, pattern address is IPv6, the match applies
9784 in IPv6 using the supplied mask if any.
9785 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
9786 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
9787 ::IPV4 or ::ffff:IPV4, otherwise it fails.
9788 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
9789 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
9790 applied in IPv6 using the supplied IPv6 mask.
9791
Willy Tarreau74ca5042013-06-11 23:12:07 +02009792
97937.2. Using ACLs to form conditions
9794----------------------------------
9795
9796Some actions are only performed upon a valid condition. A condition is a
9797combination of ACLs with operators. 3 operators are supported :
9798
9799 - AND (implicit)
9800 - OR (explicit with the "or" keyword or the "||" operator)
9801 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +02009802
Willy Tarreau74ca5042013-06-11 23:12:07 +02009803A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +02009804
Willy Tarreau74ca5042013-06-11 23:12:07 +02009805 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +02009806
Willy Tarreau74ca5042013-06-11 23:12:07 +02009807Such conditions are generally used after an "if" or "unless" statement,
9808indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +02009809
Willy Tarreau74ca5042013-06-11 23:12:07 +02009810For instance, to block HTTP requests to the "*" URL with methods other than
9811"OPTIONS", as well as POST requests without content-length, and GET or HEAD
9812requests with a content-length greater than 0, and finally every request which
9813is not either GET/HEAD/POST/OPTIONS !
9814
9815 acl missing_cl hdr_cnt(Content-length) eq 0
9816 block if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
9817 block if METH_GET HTTP_CONTENT
9818 block unless METH_GET or METH_POST or METH_OPTIONS
9819
9820To select a different backend for requests to static contents on the "www" site
9821and to every request on the "img", "video", "download" and "ftp" hosts :
9822
9823 acl url_static path_beg /static /images /img /css
9824 acl url_static path_end .gif .png .jpg .css .js
9825 acl host_www hdr_beg(host) -i www
9826 acl host_static hdr_beg(host) -i img. video. download. ftp.
9827
9828 # now use backend "static" for all static-only hosts, and for static urls
9829 # of host "www". Use backend "www" for the rest.
9830 use_backend static if host_static or host_www url_static
9831 use_backend www if host_www
9832
9833It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
9834expressions that are built on the fly without needing to be declared. They must
9835be enclosed between braces, with a space before and after each brace (because
9836the braces must be seen as independent words). Example :
9837
9838 The following rule :
9839
9840 acl missing_cl hdr_cnt(Content-length) eq 0
9841 block if METH_POST missing_cl
9842
9843 Can also be written that way :
9844
9845 block if METH_POST { hdr_cnt(Content-length) eq 0 }
9846
9847It is generally not recommended to use this construct because it's a lot easier
9848to leave errors in the configuration when written that way. However, for very
9849simple rules matching only one source IP address for instance, it can make more
9850sense to use them than to declare ACLs with random names. Another example of
9851good use is the following :
9852
9853 With named ACLs :
9854
9855 acl site_dead nbsrv(dynamic) lt 2
9856 acl site_dead nbsrv(static) lt 2
9857 monitor fail if site_dead
9858
9859 With anonymous ACLs :
9860
9861 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
9862
9863See section 4.2 for detailed help on the "block" and "use_backend" keywords.
9864
9865
98667.3. Fetching samples
9867---------------------
9868
9869Historically, sample fetch methods were only used to retrieve data to match
9870against patterns using ACLs. With the arrival of stick-tables, a new class of
9871sample fetch methods was created, most often sharing the same syntax as their
9872ACL counterpart. These sample fetch methods are also known as "fetches". As
9873of now, ACLs and fetches have converged. All ACL fetch methods have been made
9874available as fetch methods, and ACLs may use any sample fetch method as well.
9875
9876This section details all available sample fetch methods and their output type.
9877Some sample fetch methods have deprecated aliases that are used to maintain
9878compatibility with existing configurations. They are then explicitly marked as
9879deprecated and should not be used in new setups.
9880
9881The ACL derivatives are also indicated when available, with their respective
9882matching methods. These ones all have a well defined default pattern matching
9883method, so it is never necessary (though allowed) to pass the "-m" option to
9884indicate how the sample will be matched using ACLs.
9885
9886As indicated in the sample type versus matching compatibility matrix above,
9887when using a generic sample fetch method in an ACL, the "-m" option is
9888mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
9889the same keyword exists as an ACL keyword and as a standard fetch method, the
9890ACL engine will automatically pick the ACL-only one by default.
9891
9892Some of these keywords support one or multiple mandatory arguments, and one or
9893multiple optional arguments. These arguments are strongly typed and are checked
9894when the configuration is parsed so that there is no risk of running with an
9895incorrect argument (eg: an unresolved backend name). Fetch function arguments
9896are passed between parenthesis and are delimited by commas. When an argument
9897is optional, it will be indicated below between square brackets ('[ ]'). When
9898all arguments are optional, the parenthesis may be omitted.
9899
9900Thus, the syntax of a standard sample fetch method is one of the following :
9901 - name
9902 - name(arg1)
9903 - name(arg1,arg2)
9904
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009905
99067.3.1. Converters
9907-----------------
9908
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009909Sample fetch methods may be combined with transformations to be applied on top
9910of the fetched sample (also called "converters"). These combinations form what
9911is called "sample expressions" and the result is a "sample". Initially this
9912was only supported by "stick on" and "stick store-request" directives but this
9913has now be extended to all places where samples may be used (acls, log-format,
9914unique-id-format, add-header, ...).
9915
9916These transformations are enumerated as a series of specific keywords after the
9917sample fetch method. These keywords may equally be appended immediately after
9918the fetch keyword's argument, delimited by a comma. These keywords can also
9919support some arguments (eg: a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009920
Willy Tarreau74ca5042013-06-11 23:12:07 +02009921The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +01009922
Emeric Brun53d1a982014-04-30 18:21:37 +02009923base64
9924 Converts a binary input sample to a base64 string. It is used to log or
9925 transfer binary content in a way that can be reliably transferred (eg:
9926 an SSL ID can be copied in a header).
9927
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009928hex
9929 Converts a binary input sample to an hex string containing two hex digits per
9930 input byte. It is used to log or transfer hex dumps of some binary input data
9931 in a way that can be reliably transferred (eg: an SSL ID can be copied in a
9932 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +01009933
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009934http_date([<offset>])
9935 Converts an integer supposed to contain a date since epoch to a string
9936 representing this date in a format suitable for use in HTTP header fields. If
9937 an offset value is specified, then it is a number of seconds that is added to
9938 the date before the conversion is operated. This is particularly useful to
9939 emit Date header fields, Expires values in responses when combined with a
9940 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009941
Willy Tarreau644d9ef2014-07-10 16:29:08 +02009942ipmask(<mask>)
9943 Apply a mask to an IPv4 address, and use the result for lookups and storage.
9944 This can be used to make all hosts within a certain mask to share the same
9945 table entries and as such use the same server. The mask can be passed in
9946 dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24).
9947
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009948language(<value>[,<default>])
9949 Returns the value with the highest q-factor from a list as extracted from the
9950 "accept-language" header using "req.fhdr". Values with no q-factor have a
9951 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
9952 belong to the list of semi-colon delimited <values> will be considered. The
9953 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
9954 given list and a default value is provided, it is returned. Note that language
9955 names may have a variant after a dash ('-'). If this variant is present in the
9956 list, it will be matched, but if it is not, only the base language is checked.
9957 The match is case-sensitive, and the output string is always one of those
9958 provided in arguments. The ordering of arguments is meaningless, only the
9959 ordering of the values in the request counts, as the first value among
9960 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +02009961
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009962 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +02009963
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009964 # this configuration switches to the backend matching a
9965 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +02009966
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009967 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
9968 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
9969 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
9970 use_backend spanish if es
9971 use_backend french if fr
9972 use_backend english if en
9973 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +02009974
Willy Tarreau644d9ef2014-07-10 16:29:08 +02009975lower
9976 Convert a string sample to lower case. This can only be placed after a string
9977 sample fetch function or after a transformation keyword returning a string
9978 type. The result is of type string.
9979
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009980map(<map_file>[,<default_value>])
9981map_<match_type>(<map_file>[,<default_value>])
9982map_<match_type>_<output_type>(<map_file>[,<default_value>])
9983 Search the input value from <map_file> using the <match_type> matching method,
9984 and return the associated value converted to the type <output_type>. If the
9985 input value cannot be found in the <map_file>, the converter returns the
9986 <default_value>. If the <default_value> is not set, the converter fails and
9987 acts as if no input value could be fetched. If the <match_type> is not set, it
9988 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
9989 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
9990 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +01009991
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009992 It is important to avoid overlapping between the keys : IP addresses and
9993 strings are stored in trees, so the first of the finest match will be used.
9994 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +01009995
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009996 The following array contains the list of all map functions avalaible sorted by
9997 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +01009998
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009999 input type | match method | output type str | output type int | output type ip
10000 -----------+--------------+-----------------+-----------------+---------------
10001 str | str | map_str | map_str_int | map_str_ip
10002 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020010003 str | beg | map_beg | map_beg_int | map_end_ip
10004 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010005 str | sub | map_sub | map_sub_int | map_sub_ip
10006 -----------+--------------+-----------------+-----------------+---------------
10007 str | dir | map_dir | map_dir_int | map_dir_ip
10008 -----------+--------------+-----------------+-----------------+---------------
10009 str | dom | map_dom | map_dom_int | map_dom_ip
10010 -----------+--------------+-----------------+-----------------+---------------
10011 str | end | map_end | map_end_int | map_end_ip
10012 -----------+--------------+-----------------+-----------------+---------------
10013 str | reg | map_reg | map_reg_int | map_reg_ip
10014 -----------+--------------+-----------------+-----------------+---------------
10015 int | int | map_int | map_int_int | map_int_ip
10016 -----------+--------------+-----------------+-----------------+---------------
10017 ip | ip | map_ip | map_ip_int | map_ip_ip
10018 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010019
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010020 The file contains one key + value per line. Lines which start with '#' are
10021 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
10022 is then the first "word" (series of non-space/tabs characters), and the value
10023 is what follows this series of space/tab till the end of the line excluding
10024 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010025
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010026 Example :
10027
10028 # this is a comment and is ignored
10029 2.22.246.0/23 United Kingdom \n
10030 <-><-----------><--><------------><---->
10031 | | | | `- trailing spaces ignored
10032 | | | `---------- value
10033 | | `-------------------- middle spaces ignored
10034 | `---------------------------- key
10035 `------------------------------------ leading spaces ignored
10036
Willy Tarreau644d9ef2014-07-10 16:29:08 +020010037upper
10038 Convert a string sample to upper case. This can only be placed after a string
10039 sample fetch function or after a transformation keyword returning a string
10040 type. The result is of type string.
10041
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010042
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200100437.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020010044--------------------------------------------
10045
10046A first set of sample fetch methods applies to internal information which does
10047not even relate to any client information. These ones are sometimes used with
10048"monitor-fail" directives to report an internal status to external watchers.
10049The sample fetch methods described in this section are usable anywhere.
10050
10051always_false : boolean
10052 Always returns the boolean "false" value. It may be used with ACLs as a
10053 temporary replacement for another one when adjusting configurations.
10054
10055always_true : boolean
10056 Always returns the boolean "true" value. It may be used with ACLs as a
10057 temporary replacement for another one when adjusting configurations.
10058
10059avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010060 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020010061 divided by the number of active servers. The current backend is used if no
10062 backend is specified. This is very similar to "queue" except that the size of
10063 the farm is considered, in order to give a more accurate measurement of the
10064 time it may take for a new connection to be processed. The main usage is with
10065 ACL to return a sorry page to new users when it becomes certain they will get
10066 a degraded service, or to pass to the backend servers in a header so that
10067 they decide to work in degraded mode or to disable some functions to speed up
10068 the processing a bit. Note that in the event there would not be any active
10069 server anymore, twice the number of queued connections would be considered as
10070 the measured value. This is a fair estimate, as we expect one server to get
10071 back soon anyway, but we still prefer to send new traffic to another backend
10072 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
10073 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010010074
Willy Tarreau74ca5042013-06-11 23:12:07 +020010075be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020010076 Applies to the number of currently established connections on the backend,
10077 possibly including the connection being evaluated. If no backend name is
10078 specified, the current one is used. But it is also possible to check another
10079 backend. It can be used to use a specific farm when the nominal one is full.
10080 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020010081
Willy Tarreau74ca5042013-06-11 23:12:07 +020010082be_sess_rate([<backend>]) : integer
10083 Returns an integer value corresponding to the sessions creation rate on the
10084 backend, in number of new sessions per second. This is used with ACLs to
10085 switch to an alternate backend when an expensive or fragile one reaches too
10086 high a session rate, or to limit abuse of service (eg. prevent sucking of an
10087 online dictionary). It can also be useful to add this element to logs using a
10088 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010089
10090 Example :
10091 # Redirect to an error page if the dictionary is requested too often
10092 backend dynamic
10093 mode http
10094 acl being_scanned be_sess_rate gt 100
10095 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010010096
Willy Tarreau74ca5042013-06-11 23:12:07 +020010097connslots([<backend>]) : integer
10098 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010099 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020010100 connections on all servers and the maximum queue size. This is probably only
10101 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050010102
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010103 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020010104 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010105 usage; see "use_backend" keyword) can be redirected to a different backend.
10106
Willy Tarreau55165fe2009-05-10 12:02:55 +020010107 'connslots' = number of available server connection slots, + number of
10108 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010109
Willy Tarreaua36af912009-10-10 12:02:45 +020010110 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020010111 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020010112 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020010113 you want to be able to differentiate between different backends, and their
10114 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020010115 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020010116 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010117
Willy Tarreau55165fe2009-05-10 12:02:55 +020010118 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
10119 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020010120 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020010121 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010122
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010123date([<offset>]) : integer
10124 Returns the current date as the epoch (number of seconds since 01/01/1970).
10125 If an offset value is specified, then it is a number of seconds that is added
10126 to the current date before returning the value. This is particularly useful
10127 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020010128 It is useful combined with the http_date converter.
10129
10130 Example :
10131
10132 # set an expires header to now+1 hour in every response
10133 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010134
Willy Tarreau595ec542013-06-12 21:34:28 +020010135env(<name>) : string
10136 Returns a string containing the value of environment variable <name>. As a
10137 reminder, environment variables are per-process and are sampled when the
10138 process starts. This can be useful to pass some information to a next hop
10139 server, or with ACLs to take specific action when the process is started a
10140 certain way.
10141
10142 Examples :
10143 # Pass the Via header to next hop with the local hostname in it
10144 http-request add-header Via 1.1\ %[env(HOSTNAME)]
10145
10146 # reject cookie-less requests when the STOP environment variable is set
10147 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
10148
Willy Tarreau74ca5042013-06-11 23:12:07 +020010149fe_conn([<frontend>]) : integer
10150 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010010151 possibly including the connection being evaluated. If no frontend name is
10152 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010153 frontend. It can be used to return a sorry page before hard-blocking, or to
10154 use a specific backend to drain new requests when the farm is considered
10155 full. This is mostly used with ACLs but can also be used to pass some
10156 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
10157 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020010158
Willy Tarreau74ca5042013-06-11 23:12:07 +020010159fe_sess_rate([<frontend>]) : integer
10160 Returns an integer value corresponding to the sessions creation rate on the
10161 frontend, in number of new sessions per second. This is used with ACLs to
10162 limit the incoming session rate to an acceptable range in order to prevent
10163 abuse of service at the earliest moment, for example when combined with other
10164 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
10165 down below the limit. It can also be useful to add this element to logs using
10166 a log-format directive. See also the "rate-limit sessions" directive for use
10167 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010168
10169 Example :
10170 # This frontend limits incoming mails to 10/s with a max of 100
10171 # concurrent connections. We accept any connection below 10/s, and
10172 # force excess clients to wait for 100 ms. Since clients are limited to
10173 # 100 max, there cannot be more than 10 incoming mails per second.
10174 frontend mail
10175 bind :25
10176 mode tcp
10177 maxconn 100
10178 acl too_fast fe_sess_rate ge 10
10179 tcp-request inspect-delay 100ms
10180 tcp-request content accept if ! too_fast
10181 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010010182
Willy Tarreau74ca5042013-06-11 23:12:07 +020010183nbsrv([<backend>]) : integer
10184 Returns an integer value corresponding to the number of usable servers of
10185 either the current backend or the named backend. This is mostly used with
10186 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010010187 switch to an alternate backend when the number of servers is too low to
10188 to handle some load. It is useful to report a failure when combined with
10189 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010190
Willy Tarreau74ca5042013-06-11 23:12:07 +020010191queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010192 Returns the total number of queued connections of the designated backend,
10193 including all the connections in server queues. If no backend name is
10194 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010195 one. This is useful with ACLs or to pass statistics to backend servers. This
10196 can be used to take actions when queuing goes above a known level, generally
10197 indicating a surge of traffic or a massive slowdown on the servers. One
10198 possible action could be to reject new users but still accept old ones. See
10199 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
10200
Willy Tarreau84310e22014-02-14 11:59:04 +010010201rand([<range>]) : integer
10202 Returns a random integer value within a range of <range> possible values,
10203 starting at zero. If the range is not specified, it defaults to 2^32, which
10204 gives numbers between 0 and 4294967295. It can be useful to pass some values
10205 needed to take some routing decisions for example, or just for debugging
10206 purposes. This random must not be used for security purposes.
10207
Willy Tarreau74ca5042013-06-11 23:12:07 +020010208srv_conn([<backend>/]<server>) : integer
10209 Returns an integer value corresponding to the number of currently established
10210 connections on the designated server, possibly including the connection being
10211 evaluated. If <backend> is omitted, then the server is looked up in the
10212 current backend. It can be used to use a specific farm when one server is
10213 full, or to inform the server about our view of the number of active
10214 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
10215 methods.
10216
10217srv_is_up([<backend>/]<server>) : boolean
10218 Returns true when the designated server is UP, and false when it is either
10219 DOWN or in maintenance mode. If <backend> is omitted, then the server is
10220 looked up in the current backend. It is mainly used to take action based on
10221 an external status reported via a health check (eg: a geographical site's
10222 availability). Another possible use which is more of a hack consists in
10223 using dummy servers as boolean variables that can be enabled or disabled from
10224 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
10225
10226srv_sess_rate([<backend>/]<server>) : integer
10227 Returns an integer corresponding to the sessions creation rate on the
10228 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010229 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020010230 used with ACLs but can make sense with logs too. This is used to switch to an
10231 alternate backend when an expensive or fragile one reaches too high a session
10232 rate, or to limit abuse of service (eg. prevent latent requests from
10233 overloading servers).
10234
10235 Example :
10236 # Redirect to a separate back
10237 acl srv1_full srv_sess_rate(be1/srv1) gt 50
10238 acl srv2_full srv_sess_rate(be1/srv2) gt 50
10239 use_backend be2 if srv1_full or srv2_full
10240
10241table_avl([<table>]) : integer
10242 Returns the total number of available entries in the current proxy's
10243 stick-table or in the designated stick-table. See also table_cnt.
10244
10245table_cnt([<table>]) : integer
10246 Returns the total number of entries currently in use in the current proxy's
10247 stick-table or in the designated stick-table. See also src_conn_cnt and
10248 table_avl for other entry counting methods.
10249
10250
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200102517.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020010252----------------------------------
10253
10254The layer 4 usually describes just the transport layer which in haproxy is
10255closest to the connection, where no content is yet made available. The fetch
10256methods described here are usable as low as the "tcp-request connection" rule
10257sets unless they require some future information. Those generally include
10258TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010259the incoming connection. For retrieving a value from a sticky counters, the
10260counter number can be explicitly set as 0, 1, or 2 using the pre-defined
10261"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010262argument when using the "sc_" prefix. An optional table may be specified with
10263the "sc*" form, in which case the currently tracked key will be looked up into
10264this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010265
10266be_id : integer
10267 Returns an integer containing the current backend's id. It can be used in
10268 frontends with responses to check which backend processed the request.
10269
10270dst : ip
10271 This is the destination IPv4 address of the connection on the client side,
10272 which is the address the client connected to. It can be useful when running
10273 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
10274 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
10275 RFC 4291.
10276
10277dst_conn : integer
10278 Returns an integer value corresponding to the number of currently established
10279 connections on the same socket including the one being evaluated. It is
10280 normally used with ACLs but can as well be used to pass the information to
10281 servers in an HTTP header or in logs. It can be used to either return a sorry
10282 page before hard-blocking, or to use a specific backend to drain new requests
10283 when the socket is considered saturated. This offers the ability to assign
10284 different limits to different listening ports or addresses. See also the
10285 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010286
Willy Tarreau74ca5042013-06-11 23:12:07 +020010287dst_port : integer
10288 Returns an integer value corresponding to the destination TCP port of the
10289 connection on the client side, which is the port the client connected to.
10290 This might be used when running in transparent mode, when assigning dynamic
10291 ports to some clients for a whole application session, to stick all users to
10292 a same server, or to pass the destination port information to a server using
10293 an HTTP header.
10294
10295fe_id : integer
10296 Returns an integer containing the current frontend's id. It can be used in
10297 backends to check from which backend it was called, or to stick all users
10298 coming via a same frontend to the same server.
10299
Cyril Bonté62ba8702014-04-22 23:52:25 +020010300sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010301sc0_bytes_in_rate([<table>]) : integer
10302sc1_bytes_in_rate([<table>]) : integer
10303sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010304 Returns the average client-to-server bytes rate from the currently tracked
10305 counters, measured in amount of bytes over the period configured in the
10306 table. See also src_bytes_in_rate.
10307
Cyril Bonté62ba8702014-04-22 23:52:25 +020010308sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010309sc0_bytes_out_rate([<table>]) : integer
10310sc1_bytes_out_rate([<table>]) : integer
10311sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010312 Returns the average server-to-client bytes rate from the currently tracked
10313 counters, measured in amount of bytes over the period configured in the
10314 table. See also src_bytes_out_rate.
10315
Cyril Bonté62ba8702014-04-22 23:52:25 +020010316sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010317sc0_clr_gpc0([<table>]) : integer
10318sc1_clr_gpc0([<table>]) : integer
10319sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020010320 Clears the first General Purpose Counter associated to the currently tracked
10321 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010010322 stored value is zero, so first invocation will always return zero. This is
10323 typically used as a second ACL in an expression in order to mark a connection
10324 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010325
10326 # block if 5 consecutive requests continue to come faster than 10 sess
10327 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010328 acl abuse sc0_http_req_rate gt 10
10329 acl kill sc0_inc_gpc0 gt 5
10330 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010331 tcp-request connection accept if !abuse save
10332 tcp-request connection reject if abuse kill
10333
Cyril Bonté62ba8702014-04-22 23:52:25 +020010334sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010335sc0_conn_cnt([<table>]) : integer
10336sc1_conn_cnt([<table>]) : integer
10337sc2_conn_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010338 Returns the cumulated number of incoming connections from currently tracked
10339 counters. See also src_conn_cnt.
10340
Cyril Bonté62ba8702014-04-22 23:52:25 +020010341sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010342sc0_conn_cur([<table>]) : integer
10343sc1_conn_cur([<table>]) : integer
10344sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010345 Returns the current amount of concurrent connections tracking the same
10346 tracked counters. This number is automatically incremented when tracking
10347 begins and decremented when tracking stops. See also src_conn_cur.
10348
Cyril Bonté62ba8702014-04-22 23:52:25 +020010349sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010350sc0_conn_rate([<table>]) : integer
10351sc1_conn_rate([<table>]) : integer
10352sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010353 Returns the average connection rate from the currently tracked counters,
10354 measured in amount of connections over the period configured in the table.
10355 See also src_conn_rate.
10356
Cyril Bonté62ba8702014-04-22 23:52:25 +020010357sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010358sc0_get_gpc0([<table>]) : integer
10359sc1_get_gpc0([<table>]) : integer
10360sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010361 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010362 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010363
Cyril Bonté62ba8702014-04-22 23:52:25 +020010364sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010365sc0_gpc0_rate([<table>]) : integer
10366sc1_gpc0_rate([<table>]) : integer
10367sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010368 Returns the average increment rate of the first General Purpose Counter
10369 associated to the currently tracked counters. It reports the frequency
10370 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010371 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10372 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10373 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020010374
Cyril Bonté62ba8702014-04-22 23:52:25 +020010375sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010376sc0_http_err_cnt([<table>]) : integer
10377sc1_http_err_cnt([<table>]) : integer
10378sc2_http_err_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010379 Returns the cumulated number of HTTP errors from the currently tracked
10380 counters. This includes the both request errors and 4xx error responses.
10381 See also src_http_err_cnt.
10382
Cyril Bonté62ba8702014-04-22 23:52:25 +020010383sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010384sc0_http_err_rate([<table>]) : integer
10385sc1_http_err_rate([<table>]) : integer
10386sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010387 Returns the average rate of HTTP errors from the currently tracked counters,
10388 measured in amount of errors over the period configured in the table. This
10389 includes the both request errors and 4xx error responses. See also
10390 src_http_err_rate.
10391
Cyril Bonté62ba8702014-04-22 23:52:25 +020010392sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010393sc0_http_req_cnt([<table>]) : integer
10394sc1_http_req_cnt([<table>]) : integer
10395sc2_http_req_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010396 Returns the cumulated number of HTTP requests from the currently tracked
10397 counters. This includes every started request, valid or not. See also
10398 src_http_req_cnt.
10399
Cyril Bonté62ba8702014-04-22 23:52:25 +020010400sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010401sc0_http_req_rate([<table>]) : integer
10402sc1_http_req_rate([<table>]) : integer
10403sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010404 Returns the average rate of HTTP requests from the currently tracked
10405 counters, measured in amount of requests over the period configured in
10406 the table. This includes every started request, valid or not. See also
10407 src_http_req_rate.
10408
Cyril Bonté62ba8702014-04-22 23:52:25 +020010409sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010410sc0_inc_gpc0([<table>]) : integer
10411sc1_inc_gpc0([<table>]) : integer
10412sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010413 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010010414 tracked counters, and returns its new value. Before the first invocation,
10415 the stored value is zero, so first invocation will increase it to 1 and will
10416 return 1. This is typically used as a second ACL in an expression in order
10417 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020010418
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010419 acl abuse sc0_http_req_rate gt 10
10420 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010421 tcp-request connection reject if abuse kill
10422
Cyril Bonté62ba8702014-04-22 23:52:25 +020010423sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010424sc0_kbytes_in([<table>]) : integer
10425sc1_kbytes_in([<table>]) : integer
10426sc2_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010427 Returns the total amount of client-to-server data from the currently tracked
10428 counters, measured in kilobytes. The test is currently performed on 32-bit
10429 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020010430
Cyril Bonté62ba8702014-04-22 23:52:25 +020010431sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010432sc0_kbytes_out([<table>]) : integer
10433sc1_kbytes_out([<table>]) : integer
10434sc2_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010435 Returns the total amount of server-to-client data from the currently tracked
10436 counters, measured in kilobytes. The test is currently performed on 32-bit
10437 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020010438
Cyril Bonté62ba8702014-04-22 23:52:25 +020010439sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010440sc0_sess_cnt([<table>]) : integer
10441sc1_sess_cnt([<table>]) : integer
10442sc2_sess_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010443 Returns the cumulated number of incoming connections that were transformed
10444 into sessions, which means that they were accepted by a "tcp-request
10445 connection" rule, from the currently tracked counters. A backend may count
10446 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010447 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020010448 with the client. See also src_sess_cnt.
10449
Cyril Bonté62ba8702014-04-22 23:52:25 +020010450sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010451sc0_sess_rate([<table>]) : integer
10452sc1_sess_rate([<table>]) : integer
10453sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010454 Returns the average session rate from the currently tracked counters,
10455 measured in amount of sessions over the period configured in the table. A
10456 session is a connection that got past the early "tcp-request connection"
10457 rules. A backend may count more sessions than connections because each
10458 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010459 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020010460
Cyril Bonté62ba8702014-04-22 23:52:25 +020010461sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020010462sc0_tracked([<table>]) : boolean
10463sc1_tracked([<table>]) : boolean
10464sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020010465 Returns true if the designated session counter is currently being tracked by
10466 the current session. This can be useful when deciding whether or not we want
10467 to set some values in a header passed to the server.
10468
Cyril Bonté62ba8702014-04-22 23:52:25 +020010469sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010470sc0_trackers([<table>]) : integer
10471sc1_trackers([<table>]) : integer
10472sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010010473 Returns the current amount of concurrent connections tracking the same
10474 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010475 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010010476 that it does not rely on any stored information but on the table's reference
10477 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020010478 may sometimes be more suited for layer7 tracking. It can be used to tell a
10479 server how many concurrent connections there are from a given address for
10480 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010010481
Willy Tarreau74ca5042013-06-11 23:12:07 +020010482so_id : integer
10483 Returns an integer containing the current listening socket's id. It is useful
10484 in frontends involving many "bind" lines, or to stick all users coming via a
10485 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010486
Willy Tarreau74ca5042013-06-11 23:12:07 +020010487src : ip
10488 This is the source IPv4 address of the client of the session. It is of type
10489 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
10490 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
10491 TCP-level source address which is used, and not the address of a client
10492 behind a proxy. However if the "accept-proxy" bind directive is used, it can
10493 be the address of a client behind another PROXY-protocol compatible component
10494 for all rule sets except "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010495
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010496 Example:
10497 # add an HTTP header in requests with the originating address' country
10498 http-request set-header X-Country %[src,map_ip(geoip.lst)]
10499
Willy Tarreau74ca5042013-06-11 23:12:07 +020010500src_bytes_in_rate([<table>]) : integer
10501 Returns the average bytes rate from the incoming connection's source address
10502 in the current proxy's stick-table or in the designated stick-table, measured
10503 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010504 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010505
Willy Tarreau74ca5042013-06-11 23:12:07 +020010506src_bytes_out_rate([<table>]) : integer
10507 Returns the average bytes rate to the incoming connection's source address in
10508 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010509 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010510 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010511
Willy Tarreau74ca5042013-06-11 23:12:07 +020010512src_clr_gpc0([<table>]) : integer
10513 Clears the first General Purpose Counter associated to the incoming
10514 connection's source address in the current proxy's stick-table or in the
10515 designated stick-table, and returns its previous value. If the address is not
10516 found, an entry is created and 0 is returned. This is typically used as a
10517 second ACL in an expression in order to mark a connection when a first ACL
10518 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010519
10520 # block if 5 consecutive requests continue to come faster than 10 sess
10521 # per second, and reset the counter as soon as the traffic slows down.
10522 acl abuse src_http_req_rate gt 10
10523 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010010524 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010525 tcp-request connection accept if !abuse save
10526 tcp-request connection reject if abuse kill
10527
Willy Tarreau74ca5042013-06-11 23:12:07 +020010528src_conn_cnt([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010529 Returns the cumulated number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020010530 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010531 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010532 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010533
Willy Tarreau74ca5042013-06-11 23:12:07 +020010534src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010535 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010536 current incoming connection's source address in the current proxy's
10537 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010538 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010539
Willy Tarreau74ca5042013-06-11 23:12:07 +020010540src_conn_rate([<table>]) : integer
10541 Returns the average connection rate from the incoming connection's source
10542 address in the current proxy's stick-table or in the designated stick-table,
10543 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010544 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010545
Willy Tarreau74ca5042013-06-11 23:12:07 +020010546src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010547 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010548 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010549 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010550 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010551
Willy Tarreau74ca5042013-06-11 23:12:07 +020010552src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010553 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020010554 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010555 stick-table or in the designated stick-table. It reports the frequency
10556 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010557 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10558 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10559 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010560
Willy Tarreau74ca5042013-06-11 23:12:07 +020010561src_http_err_cnt([<table>]) : integer
10562 Returns the cumulated number of HTTP errors from the incoming connection's
10563 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020010564 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010565 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020010566 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010567
Willy Tarreau74ca5042013-06-11 23:12:07 +020010568src_http_err_rate([<table>]) : integer
10569 Returns the average rate of HTTP errors from the incoming connection's source
10570 address in the current proxy's stick-table or in the designated stick-table,
10571 measured in amount of errors over the period configured in the table. This
10572 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010573 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010574
Willy Tarreau74ca5042013-06-11 23:12:07 +020010575src_http_req_cnt([<table>]) : integer
10576 Returns the cumulated number of HTTP requests from the incoming connection's
10577 source address in the current proxy's stick-table or in the designated stick-
10578 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010579 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010580
Willy Tarreau74ca5042013-06-11 23:12:07 +020010581src_http_req_rate([<table>]) : integer
10582 Returns the average rate of HTTP requests from the incoming connection's
10583 source address in the current proxy's stick-table or in the designated stick-
10584 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020010585 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010586 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010587
Willy Tarreau74ca5042013-06-11 23:12:07 +020010588src_inc_gpc0([<table>]) : integer
10589 Increments the first General Purpose Counter associated to the incoming
10590 connection's source address in the current proxy's stick-table or in the
10591 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010592 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010593 This is typically used as a second ACL in an expression in order to mark a
10594 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020010595
10596 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010010597 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010598 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020010599
Willy Tarreau74ca5042013-06-11 23:12:07 +020010600src_kbytes_in([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010601 Returns the total amount of data received from the incoming connection's
10602 source address in the current proxy's stick-table or in the designated
10603 stick-table, measured in kilobytes. If the address is not found, zero is
10604 returned. The test is currently performed on 32-bit integers, which limits
10605 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010606
Willy Tarreau74ca5042013-06-11 23:12:07 +020010607src_kbytes_out([<table>]) : integer
Willy Tarreau911780c2014-07-10 15:29:24 +020010608 Returns the total amount of data sent to the incoming connection's source
10609 address in the current proxy's stick-table or in the designated stick-table,
10610 measured in kilobytes. If the address is not found, zero is returned. The
10611 test is currently performed on 32-bit integers, which limits values to 4
10612 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010613
Willy Tarreau74ca5042013-06-11 23:12:07 +020010614src_port : integer
10615 Returns an integer value corresponding to the TCP source port of the
10616 connection on the client side, which is the port the client connected from.
10617 Usage of this function is very limited as modern protocols do not care much
10618 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010619
Willy Tarreau74ca5042013-06-11 23:12:07 +020010620src_sess_cnt([<table>]) : integer
10621 Returns the cumulated number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020010622 connection's source IPv4 address in the current proxy's stick-table or in the
10623 designated stick-table, that were transformed into sessions, which means that
10624 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010625 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010626
Willy Tarreau74ca5042013-06-11 23:12:07 +020010627src_sess_rate([<table>]) : integer
10628 Returns the average session rate from the incoming connection's source
10629 address in the current proxy's stick-table or in the designated stick-table,
10630 measured in amount of sessions over the period configured in the table. A
10631 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010632 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010633
Willy Tarreau74ca5042013-06-11 23:12:07 +020010634src_updt_conn_cnt([<table>]) : integer
10635 Creates or updates the entry associated to the incoming connection's source
10636 address in the current proxy's stick-table or in the designated stick-table.
10637 This table must be configured to store the "conn_cnt" data type, otherwise
10638 the match will be ignored. The current count is incremented by one, and the
10639 expiration timer refreshed. The updated count is returned, so this match
10640 can't return zero. This was used to reject service abusers based on their
10641 source address. Note: it is recommended to use the more complete "track-sc*"
10642 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010643
10644 Example :
10645 # This frontend limits incoming SSH connections to 3 per 10 second for
10646 # each source address, and rejects excess connections until a 10 second
10647 # silence is observed. At most 20 addresses are tracked.
10648 listen ssh
10649 bind :22
10650 mode tcp
10651 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020010652 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020010653 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010654 server local 127.0.0.1:22
10655
Willy Tarreau74ca5042013-06-11 23:12:07 +020010656srv_id : integer
10657 Returns an integer containing the server's id when processing the response.
10658 While it's almost only used with ACLs, it may be used for logging or
10659 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020010660
Hervé COMMOWICK35ed8012010-12-15 14:04:51 +010010661
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200106627.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020010663----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020010664
Willy Tarreau74ca5042013-06-11 23:12:07 +020010665The layer 5 usually describes just the session layer which in haproxy is
10666closest to the session once all the connection handshakes are finished, but
10667when no content is yet made available. The fetch methods described here are
10668usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010669future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020010670
Emeric Brun645ae792014-04-30 14:21:06 +020010671ssl_bc : boolean
10672 Returns true when the back connection was made via an SSL/TLS transport
10673 layer and is locally deciphered. This means the outgoing connection was made
10674 other a server with the "ssl" option.
10675
10676ssl_bc_alg_keysize : integer
10677 Returns the symmetric cipher key size supported in bits when the outgoing
10678 connection was made over an SSL/TLS transport layer.
10679
10680ssl_bc_cipher : string
10681 Returns the name of the used cipher when the outgoing connection was made
10682 over an SSL/TLS transport layer.
10683
10684ssl_bc_protocol : string
10685 Returns the name of the used protocol when the outgoing connection was made
10686 over an SSL/TLS transport layer.
10687
Emeric Brunb73a9b02014-04-30 18:49:19 +020010688ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020010689 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010690 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10691 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020010692
10693ssl_bc_session_id : binary
10694 Returns the SSL ID of the back connection when the outgoing connection was
10695 made over an SSL/TLS transport layer. It is useful to log if we want to know
10696 if session was reused or not.
10697
10698ssl_bc_use_keysize : integer
10699 Returns the symmetric cipher key size used in bits when the outgoing
10700 connection was made over an SSL/TLS transport layer.
10701
Willy Tarreau74ca5042013-06-11 23:12:07 +020010702ssl_c_ca_err : integer
10703 When the incoming connection was made over an SSL/TLS transport layer,
10704 returns the ID of the first error detected during verification of the client
10705 certificate at depth > 0, or 0 if no error was encountered during this
10706 verification process. Please refer to your SSL library's documentation to
10707 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020010708
Willy Tarreau74ca5042013-06-11 23:12:07 +020010709ssl_c_ca_err_depth : integer
10710 When the incoming connection was made over an SSL/TLS transport layer,
10711 returns the depth in the CA chain of the first error detected during the
10712 verification of the client certificate. If no error is encountered, 0 is
10713 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010714
Emeric Brunb3cc4252014-10-29 19:03:26 +010010715ssl_c_der : binary
10716 Returns the DER formatted certificate presented by the client when the
10717 incoming connection was made over an SSL/TLS transport layer. When used for
10718 an ACL, the value(s) to match against can be passed in hexadecimal form.
10719
Willy Tarreau74ca5042013-06-11 23:12:07 +020010720ssl_c_err : integer
10721 When the incoming connection was made over an SSL/TLS transport layer,
10722 returns the ID of the first error detected during verification at depth 0, or
10723 0 if no error was encountered during this verification process. Please refer
10724 to your SSL library's documentation to find the exhaustive list of error
10725 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020010726
Willy Tarreau74ca5042013-06-11 23:12:07 +020010727ssl_c_i_dn([<entry>[,<occ>]]) : string
10728 When the incoming connection was made over an SSL/TLS transport layer,
10729 returns the full distinguished name of the issuer of the certificate
10730 presented by the client when no <entry> is specified, or the value of the
10731 first given entry found from the beginning of the DN. If a positive/negative
10732 occurrence number is specified as the optional second argument, it returns
10733 the value of the nth given entry value from the beginning/end of the DN.
10734 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
10735 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020010736
Willy Tarreau74ca5042013-06-11 23:12:07 +020010737ssl_c_key_alg : string
10738 Returns the name of the algorithm used to generate the key of the certificate
10739 presented by the client when the incoming connection was made over an SSL/TLS
10740 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020010741
Willy Tarreau74ca5042013-06-11 23:12:07 +020010742ssl_c_notafter : string
10743 Returns the end date presented by the client as a formatted string
10744 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10745 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020010746
Willy Tarreau74ca5042013-06-11 23:12:07 +020010747ssl_c_notbefore : string
10748 Returns the start date presented by the client as a formatted string
10749 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10750 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010751
Willy Tarreau74ca5042013-06-11 23:12:07 +020010752ssl_c_s_dn([<entry>[,<occ>]]) : string
10753 When the incoming connection was made over an SSL/TLS transport layer,
10754 returns the full distinguished name of the subject of the certificate
10755 presented by the client when no <entry> is specified, or the value of the
10756 first given entry found from the beginning of the DN. If a positive/negative
10757 occurrence number is specified as the optional second argument, it returns
10758 the value of the nth given entry value from the beginning/end of the DN.
10759 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
10760 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010010761
Willy Tarreau74ca5042013-06-11 23:12:07 +020010762ssl_c_serial : binary
10763 Returns the serial of the certificate presented by the client when the
10764 incoming connection was made over an SSL/TLS transport layer. When used for
10765 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010766
Willy Tarreau74ca5042013-06-11 23:12:07 +020010767ssl_c_sha1 : binary
10768 Returns the SHA-1 fingerprint of the certificate presented by the client when
10769 the incoming connection was made over an SSL/TLS transport layer. This can be
10770 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau9fe4cb62014-07-02 19:01:22 +020010771 Note that the output is binary, so if you want to pass that signature to the
10772 server, you need to encode it in hex or base64, such as in the example below:
10773
10774 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020010775
Willy Tarreau74ca5042013-06-11 23:12:07 +020010776ssl_c_sig_alg : string
10777 Returns the name of the algorithm used to sign the certificate presented by
10778 the client when the incoming connection was made over an SSL/TLS transport
10779 layer.
Emeric Brun87855892012-10-17 17:39:35 +020010780
Willy Tarreau74ca5042013-06-11 23:12:07 +020010781ssl_c_used : boolean
10782 Returns true if current SSL session uses a client certificate even if current
10783 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020010784
Willy Tarreau74ca5042013-06-11 23:12:07 +020010785ssl_c_verify : integer
10786 Returns the verify result error ID when the incoming connection was made over
10787 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
10788 refer to your SSL library's documentation for an exhaustive list of error
10789 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020010790
Willy Tarreau74ca5042013-06-11 23:12:07 +020010791ssl_c_version : integer
10792 Returns the version of the certificate presented by the client when the
10793 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020010794
Emeric Brunb3cc4252014-10-29 19:03:26 +010010795ssl_f_der : binary
10796 Returns the DER formatted certificate presented by the frontend 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_f_i_dn([<entry>[,<occ>]]) : string
10801 When the incoming connection was made over an SSL/TLS transport layer,
10802 returns the full distinguished name of the issuer of the certificate
10803 presented by the frontend when no <entry> is specified, or the value of the
10804 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020010805 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020010806 the value of the nth given entry value from the beginning/end of the DN.
10807 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
10808 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020010809
Willy Tarreau74ca5042013-06-11 23:12:07 +020010810ssl_f_key_alg : string
10811 Returns the name of the algorithm used to generate the key of the certificate
10812 presented by the frontend when the incoming connection was made over an
10813 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010814
Willy Tarreau74ca5042013-06-11 23:12:07 +020010815ssl_f_notafter : string
10816 Returns the end date presented by the frontend as a formatted string
10817 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10818 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010819
Willy Tarreau74ca5042013-06-11 23:12:07 +020010820ssl_f_notbefore : string
10821 Returns the start date presented by the frontend as a formatted string
10822 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
10823 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020010824
Willy Tarreau74ca5042013-06-11 23:12:07 +020010825ssl_f_s_dn([<entry>[,<occ>]]) : string
10826 When the incoming connection was made over an SSL/TLS transport layer,
10827 returns the full distinguished name of the subject of the certificate
10828 presented by the frontend when no <entry> is specified, or the value of the
10829 first given entry found from the beginning of the DN. If a positive/negative
10830 occurrence number is specified as the optional second argument, it returns
10831 the value of the nth given entry value from the beginning/end of the DN.
10832 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
10833 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020010834
Willy Tarreau74ca5042013-06-11 23:12:07 +020010835ssl_f_serial : binary
10836 Returns the serial of the certificate presented by the frontend when the
10837 incoming connection was made over an SSL/TLS transport layer. When used for
10838 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020010839
Emeric Brun55f4fa82014-04-30 17:11:25 +020010840ssl_f_sha1 : binary
10841 Returns the SHA-1 fingerprint of the certificate presented by the frontend
10842 when the incoming connection was made over an SSL/TLS transport layer. This
10843 can be used to know which certificate was chosen using SNI.
10844
Willy Tarreau74ca5042013-06-11 23:12:07 +020010845ssl_f_sig_alg : string
10846 Returns the name of the algorithm used to sign the certificate presented by
10847 the frontend when the incoming connection was made over an SSL/TLS transport
10848 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020010849
Willy Tarreau74ca5042013-06-11 23:12:07 +020010850ssl_f_version : integer
10851 Returns the version of the certificate presented by the frontend when the
10852 incoming connection was made over an SSL/TLS transport layer.
10853
10854ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010855 Returns true when the front connection was made via an SSL/TLS transport
10856 layer and is locally deciphered. This means it has matched a socket declared
10857 with a "bind" line having the "ssl" option.
10858
Willy Tarreau74ca5042013-06-11 23:12:07 +020010859 Example :
10860 # This passes "X-Proto: https" to servers when client connects over SSL
10861 listen http-https
10862 bind :80
10863 bind :443 ssl crt /etc/haproxy.pem
10864 http-request add-header X-Proto https if { ssl_fc }
10865
10866ssl_fc_alg_keysize : integer
10867 Returns the symmetric cipher key size supported in bits when the incoming
10868 connection was made over an SSL/TLS transport layer.
10869
10870ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010871 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020010872 incoming connection made via a TLS transport layer and locally deciphered by
10873 haproxy. The result is a string containing the protocol name advertised by
10874 the client. The SSL library must have been built with support for TLS
10875 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
10876 not advertised unless the "alpn" keyword on the "bind" line specifies a
10877 protocol list. Also, nothing forces the client to pick a protocol from this
10878 list, any other one may be requested. The TLS ALPN extension is meant to
10879 replace the TLS NPN extension. See also "ssl_fc_npn".
10880
Willy Tarreau74ca5042013-06-11 23:12:07 +020010881ssl_fc_cipher : string
10882 Returns the name of the used cipher when the incoming connection was made
10883 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020010884
Willy Tarreau74ca5042013-06-11 23:12:07 +020010885ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020010886 Returns true if a client certificate is present in an incoming connection over
10887 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010010888 Note: on SSL session resumption with Session ID or TLS ticket, client
10889 certificate is not present in the current connection but may be retrieved
10890 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
10891 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020010892
Willy Tarreau74ca5042013-06-11 23:12:07 +020010893ssl_fc_has_sni : boolean
10894 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020010895 in an incoming connection was made over an SSL/TLS transport layer. Returns
10896 true when the incoming connection presents a TLS SNI field. This requires
10897 that the SSL library is build with support for TLS extensions enabled (check
10898 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020010899
Willy Tarreau74ca5042013-06-11 23:12:07 +020010900ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010901 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020010902 made via a TLS transport layer and locally deciphered by haproxy. The result
10903 is a string containing the protocol name advertised by the client. The SSL
10904 library must have been built with support for TLS extensions enabled (check
10905 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
10906 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
10907 forces the client to pick a protocol from this list, any other one may be
10908 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020010909
Willy Tarreau74ca5042013-06-11 23:12:07 +020010910ssl_fc_protocol : string
10911 Returns the name of the used protocol when the incoming connection was made
10912 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020010913
Emeric Brunb73a9b02014-04-30 18:49:19 +020010914ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040010915 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010916 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10917 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040010918
Willy Tarreau74ca5042013-06-11 23:12:07 +020010919ssl_fc_session_id : binary
10920 Returns the SSL ID of the front connection when the incoming connection was
10921 made over an SSL/TLS transport layer. It is useful to stick a given client to
10922 a server. It is important to note that some browsers refresh their session ID
10923 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020010924
Willy Tarreau74ca5042013-06-11 23:12:07 +020010925ssl_fc_sni : string
10926 This extracts the Server Name Indication TLS extension (SNI) field from an
10927 incoming connection made via an SSL/TLS transport layer and locally
10928 deciphered by haproxy. The result (when present) typically is a string
10929 matching the HTTPS host name (253 chars or less). The SSL library must have
10930 been built with support for TLS extensions enabled (check haproxy -vv).
10931
10932 This fetch is different from "req_ssl_sni" above in that it applies to the
10933 connection being deciphered by haproxy and not to SSL contents being blindly
10934 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020010935 requires that the SSL library is build with support for TLS extensions
10936 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020010937
Willy Tarreau74ca5042013-06-11 23:12:07 +020010938 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020010939 ssl_fc_sni_end : suffix match
10940 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020010941
Willy Tarreau74ca5042013-06-11 23:12:07 +020010942ssl_fc_use_keysize : integer
10943 Returns the symmetric cipher key size used in bits when the incoming
10944 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020010945
Willy Tarreaub6fb4202008-07-20 11:18:28 +020010946
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200109477.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020010948------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020010949
Willy Tarreau74ca5042013-06-11 23:12:07 +020010950Fetching samples from buffer contents is a bit different from the previous
10951sample fetches above because the sampled data are ephemeral. These data can
10952only be used when they're available and will be lost when they're forwarded.
10953For this reason, samples fetched from buffer contents during a request cannot
10954be used in a response for example. Even while the data are being fetched, they
10955can change. Sometimes it is necessary to set some delays or combine multiple
10956sample fetch methods to ensure that the expected data are complete and usable,
10957for example through TCP request content inspection. Please see the "tcp-request
10958content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020010959
Willy Tarreau74ca5042013-06-11 23:12:07 +020010960payload(<offset>,<length>) : binary (deprecated)
10961 This is an alias for "req.payload" when used in the context of a request (eg:
10962 "stick on", "stick match"), and for "res.payload" when used in the context of
10963 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010964
Willy Tarreau74ca5042013-06-11 23:12:07 +020010965payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
10966 This is an alias for "req.payload_lv" when used in the context of a request
10967 (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the
10968 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010969
Willy Tarreau74ca5042013-06-11 23:12:07 +020010970req.len : integer
10971req_len : integer (deprecated)
10972 Returns an integer value corresponding to the number of bytes present in the
10973 request buffer. This is mostly used in ACL. It is important to understand
10974 that this test does not return false as long as the buffer is changing. This
10975 means that a check with equality to zero will almost always immediately match
10976 at the beginning of the session, while a test for more data will wait for
10977 that data to come in and return false only when haproxy is certain that no
10978 more data will come in. This test was designed to be used with TCP request
10979 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020010980
Willy Tarreau74ca5042013-06-11 23:12:07 +020010981req.payload(<offset>,<length>) : binary
10982 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020010983 in the request buffer. As a special case, if the <length> argument is zero,
10984 the the whole buffer from <offset> to the end is extracted. This can be used
10985 with ACLs in order to check for the presence of some content in a buffer at
10986 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020010987
Willy Tarreau74ca5042013-06-11 23:12:07 +020010988 ACL alternatives :
10989 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020010990
Willy Tarreau74ca5042013-06-11 23:12:07 +020010991req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
10992 This extracts a binary block whose size is specified at <offset1> for <length>
10993 bytes, and which starts at <offset2> if specified or just after the length in
10994 the request buffer. The <offset2> parameter also supports relative offsets if
10995 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020010996
Willy Tarreau74ca5042013-06-11 23:12:07 +020010997 ACL alternatives :
10998 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020010999
Willy Tarreau74ca5042013-06-11 23:12:07 +020011000 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011001
Willy Tarreau74ca5042013-06-11 23:12:07 +020011002req.proto_http : boolean
11003req_proto_http : boolean (deprecated)
11004 Returns true when data in the request buffer look like HTTP and correctly
11005 parses as such. It is the same parser as the common HTTP request parser which
11006 is used so there should be no surprises. The test does not match until the
11007 request is complete, failed or timed out. This test may be used to report the
11008 protocol in TCP logs, but the biggest use is to block TCP request analysis
11009 until a complete HTTP request is present in the buffer, for example to track
11010 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011011
Willy Tarreau74ca5042013-06-11 23:12:07 +020011012 Example:
11013 # track request counts per "base" (concatenation of Host+URL)
11014 tcp-request inspect-delay 10s
11015 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011016 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011017
Willy Tarreau74ca5042013-06-11 23:12:07 +020011018req.rdp_cookie([<name>]) : string
11019rdp_cookie([<name>]) : string (deprecated)
11020 When the request buffer looks like the RDP protocol, extracts the RDP cookie
11021 <name>, or any cookie if unspecified. The parser only checks for the first
11022 cookie, as illustrated in the RDP protocol specification. The cookie name is
11023 case insensitive. Generally the "MSTS" cookie name will be used, as it can
11024 contain the user name of the client connecting to the server if properly
11025 configured on the client. The "MSTSHASH" cookie is often used as well for
11026 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011027
Willy Tarreau74ca5042013-06-11 23:12:07 +020011028 This differs from "balance rdp-cookie" in that any balancing algorithm may be
11029 used and thus the distribution of clients to backend servers is not linked to
11030 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
11031 such as "balance roundrobin" or "balance leastconn" will lead to a more even
11032 distribution of clients to backend servers than the hash used by "balance
11033 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011034
Willy Tarreau74ca5042013-06-11 23:12:07 +020011035 ACL derivatives :
11036 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011037
Willy Tarreau74ca5042013-06-11 23:12:07 +020011038 Example :
11039 listen tse-farm
11040 bind 0.0.0.0:3389
11041 # wait up to 5s for an RDP cookie in the request
11042 tcp-request inspect-delay 5s
11043 tcp-request content accept if RDP_COOKIE
11044 # apply RDP cookie persistence
11045 persist rdp-cookie
11046 # Persist based on the mstshash cookie
11047 # This is only useful makes sense if
11048 # balance rdp-cookie is not used
11049 stick-table type string size 204800
11050 stick on req.rdp_cookie(mstshash)
11051 server srv1 1.1.1.1:3389
11052 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011053
Willy Tarreau74ca5042013-06-11 23:12:07 +020011054 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
11055 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011056
Willy Tarreau74ca5042013-06-11 23:12:07 +020011057req.rdp_cookie_cnt([name]) : integer
11058rdp_cookie_cnt([name]) : integer (deprecated)
11059 Tries to parse the request buffer as RDP protocol, then returns an integer
11060 corresponding to the number of RDP cookies found. If an optional cookie name
11061 is passed, only cookies matching this name are considered. This is mostly
11062 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011063
Willy Tarreau74ca5042013-06-11 23:12:07 +020011064 ACL derivatives :
11065 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011066
Willy Tarreau74ca5042013-06-11 23:12:07 +020011067req.ssl_hello_type : integer
11068req_ssl_hello_type : integer (deprecated)
11069 Returns an integer value containing the type of the SSL hello message found
11070 in the request buffer if the buffer contains data that parse as a complete
11071 SSL (v3 or superior) client hello message. Note that this only applies to raw
11072 contents found in the request buffer and not to contents deciphered via an
11073 SSL data layer, so this will not work with "bind" lines having the "ssl"
11074 option. This is mostly used in ACL to detect presence of an SSL hello message
11075 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011076
Willy Tarreau74ca5042013-06-11 23:12:07 +020011077req.ssl_sni : string
11078req_ssl_sni : string (deprecated)
11079 Returns a string containing the value of the Server Name TLS extension sent
11080 by a client in a TLS stream passing through the request buffer if the buffer
11081 contains data that parse as a complete SSL (v3 or superior) client hello
11082 message. Note that this only applies to raw contents found in the request
11083 buffer and not to contents deciphered via an SSL data layer, so this will not
11084 work with "bind" lines having the "ssl" option. SNI normally contains the
11085 name of the host the client tries to connect to (for recent browsers). SNI is
11086 useful for allowing or denying access to certain hosts when SSL/TLS is used
11087 by the client. This test was designed to be used with TCP request content
11088 inspection. If content switching is needed, it is recommended to first wait
11089 for a complete client hello (type 1), like in the example below. See also
11090 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011091
Willy Tarreau74ca5042013-06-11 23:12:07 +020011092 ACL derivatives :
11093 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011094
Willy Tarreau74ca5042013-06-11 23:12:07 +020011095 Examples :
11096 # Wait for a client hello for at most 5 seconds
11097 tcp-request inspect-delay 5s
11098 tcp-request content accept if { req_ssl_hello_type 1 }
11099 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
11100 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011101
Willy Tarreau74ca5042013-06-11 23:12:07 +020011102res.ssl_hello_type : integer
11103rep_ssl_hello_type : integer (deprecated)
11104 Returns an integer value containing the type of the SSL hello message found
11105 in the response buffer if the buffer contains data that parses as a complete
11106 SSL (v3 or superior) hello message. Note that this only applies to raw
11107 contents found in the response buffer and not to contents deciphered via an
11108 SSL data layer, so this will not work with "server" lines having the "ssl"
11109 option. This is mostly used in ACL to detect presence of an SSL hello message
11110 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau51539362012-05-08 12:46:28 +020011111
Willy Tarreau74ca5042013-06-11 23:12:07 +020011112req.ssl_ver : integer
11113req_ssl_ver : integer (deprecated)
11114 Returns an integer value containing the version of the SSL/TLS protocol of a
11115 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
11116 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
11117 composed of the major version multiplied by 65536, added to the minor
11118 version. Note that this only applies to raw contents found in the request
11119 buffer and not to contents deciphered via an SSL data layer, so this will not
11120 work with "bind" lines having the "ssl" option. The ACL version of the test
11121 matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This
11122 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011123
Willy Tarreau74ca5042013-06-11 23:12:07 +020011124 ACL derivatives :
11125 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011126
Willy Tarreau47e8eba2013-09-11 23:28:46 +020011127res.len : integer
11128 Returns an integer value corresponding to the number of bytes present in the
11129 response buffer. This is mostly used in ACL. It is important to understand
11130 that this test does not return false as long as the buffer is changing. This
11131 means that a check with equality to zero will almost always immediately match
11132 at the beginning of the session, while a test for more data will wait for
11133 that data to come in and return false only when haproxy is certain that no
11134 more data will come in. This test was designed to be used with TCP response
11135 content inspection.
11136
Willy Tarreau74ca5042013-06-11 23:12:07 +020011137res.payload(<offset>,<length>) : binary
11138 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011139 in the response buffer. As a special case, if the <length> argument is zero,
11140 the the whole buffer from <offset> to the end is extracted. This can be used
11141 with ACLs in order to check for the presence of some content in a buffer at
11142 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011143
Willy Tarreau74ca5042013-06-11 23:12:07 +020011144res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11145 This extracts a binary block whose size is specified at <offset1> for <length>
11146 bytes, and which starts at <offset2> if specified or just after the length in
11147 the response buffer. The <offset2> parameter also supports relative offsets
11148 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011149
Willy Tarreau74ca5042013-06-11 23:12:07 +020011150 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011151
Willy Tarreau74ca5042013-06-11 23:12:07 +020011152wait_end : boolean
11153 This fetch either returns true when the inspection period is over, or does
11154 not fetch. It is only used in ACLs, in conjunction with content analysis to
11155 avoid returning a wrong verdict early. It may also be used to delay some
11156 actions, such as a delayed reject for some special addresses. Since it either
11157 stops the rules evaluation or immediately returns true, it is recommended to
11158 use this acl as the last one in a rule. Please note that the default ACL
11159 "WAIT_END" is always usable without prior declaration. This test was designed
11160 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011161
Willy Tarreau74ca5042013-06-11 23:12:07 +020011162 Examples :
11163 # delay every incoming request by 2 seconds
11164 tcp-request inspect-delay 2s
11165 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010011166
Willy Tarreau74ca5042013-06-11 23:12:07 +020011167 # don't immediately tell bad guys they are rejected
11168 tcp-request inspect-delay 10s
11169 acl goodguys src 10.0.0.0/24
11170 acl badguys src 10.0.1.0/24
11171 tcp-request content accept if goodguys
11172 tcp-request content reject if badguys WAIT_END
11173 tcp-request content reject
11174
11175
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200111767.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011177--------------------------------------
11178
11179It is possible to fetch samples from HTTP contents, requests and responses.
11180This application layer is also called layer 7. It is only possible to fetch the
11181data in this section when a full HTTP request or response has been parsed from
11182its respective request or response buffer. This is always the case with all
11183HTTP specific rules and for sections running with "mode http". When using TCP
11184content inspection, it may be necessary to support an inspection delay in order
11185to let the request or response come in first. These fetches may require a bit
11186more CPU resources than the layer 4 ones, but not much since the request and
11187response are indexed.
11188
11189base : string
11190 This returns the concatenation of the first Host header and the path part of
11191 the request, which starts at the first slash and ends before the question
11192 mark. It can be useful in virtual hosted environments to detect URL abuses as
11193 well as to improve shared caches efficiency. Using this with a limited size
11194 stick table also allows one to collect statistics about most commonly
11195 requested objects by host/path. With ACLs it can allow simple content
11196 switching rules involving the host and the path at the same time, such as
11197 "www.example.com/favicon.ico". See also "path" and "uri".
11198
11199 ACL derivatives :
11200 base : exact string match
11201 base_beg : prefix match
11202 base_dir : subdir match
11203 base_dom : domain match
11204 base_end : suffix match
11205 base_len : length match
11206 base_reg : regex match
11207 base_sub : substring match
11208
11209base32 : integer
11210 This returns a 32-bit hash of the value returned by the "base" fetch method
11211 above. This is useful to track per-URL activity on high traffic sites without
11212 having to store all URLs. Instead a shorter hash is stored, saving a lot of
11213 memory. The output type is an unsigned integer.
11214
11215base32+src : binary
11216 This returns the concatenation of the base32 fetch above and the src fetch
11217 below. The resulting type is of type binary, with a size of 8 or 20 bytes
11218 depending on the source address family. This can be used to track per-IP,
11219 per-URL counters.
11220
William Lallemand65ad6e12014-01-31 15:08:02 +010011221capture.req.hdr(<idx>) : string
11222 This extracts the content of the header captured by the "capture request
11223 header", idx is the position of the capture keyword in the configuration.
11224 The first entry is an index of 0. See also: "capture request header".
11225
11226capture.req.method : string
11227 This extracts the METHOD of an HTTP request. It can be used in both request
11228 and response. Unlike "method", it can be used in both request and response
11229 because it's allocated.
11230
11231capture.req.uri : string
11232 This extracts the request's URI, which starts at the first slash and ends
11233 before the first space in the request (without the host part). Unlike "path"
11234 and "url", it can be used in both request and response because it's
11235 allocated.
11236
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011237capture.req.ver : string
11238 This extracts the request's HTTP version and returns either "HTTP/1.0" or
11239 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
11240 logs because it relies on a persistent flag.
11241
William Lallemand65ad6e12014-01-31 15:08:02 +010011242capture.res.hdr(<idx>) : string
11243 This extracts the content of the header captured by the "capture response
11244 header", idx is the position of the capture keyword in the configuration.
11245 The first entry is an index of 0.
11246 See also: "capture response header"
11247
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011248capture.res.ver : string
11249 This extracts the response's HTTP version and returns either "HTTP/1.0" or
11250 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
11251 persistent flag.
11252
Willy Tarreau74ca5042013-06-11 23:12:07 +020011253req.cook([<name>]) : string
11254cook([<name>]) : string (deprecated)
11255 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11256 header line from the request, and returns its value as string. If no name is
11257 specified, the first cookie value is returned. When used with ACLs, all
11258 matching cookies are evaluated. Spaces around the name and the value are
11259 ignored as requested by the Cookie header specification (RFC6265). The cookie
11260 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
11261 well return an empty value if it is present. Use the "found" match to detect
11262 presence. Use the res.cook() variant for response cookies sent by the server.
11263
11264 ACL derivatives :
11265 cook([<name>]) : exact string match
11266 cook_beg([<name>]) : prefix match
11267 cook_dir([<name>]) : subdir match
11268 cook_dom([<name>]) : domain match
11269 cook_end([<name>]) : suffix match
11270 cook_len([<name>]) : length match
11271 cook_reg([<name>]) : regex match
11272 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011273
Willy Tarreau74ca5042013-06-11 23:12:07 +020011274req.cook_cnt([<name>]) : integer
11275cook_cnt([<name>]) : integer (deprecated)
11276 Returns an integer value representing the number of occurrences of the cookie
11277 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011278
Willy Tarreau74ca5042013-06-11 23:12:07 +020011279req.cook_val([<name>]) : integer
11280cook_val([<name>]) : integer (deprecated)
11281 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11282 header line from the request, and converts its value to an integer which is
11283 returned. If no name is specified, the first cookie value is returned. When
11284 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020011285
Willy Tarreau74ca5042013-06-11 23:12:07 +020011286cookie([<name>]) : string (deprecated)
11287 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11288 header line from the request, or a "Set-Cookie" header from the response, and
11289 returns its value as a string. A typical use is to get multiple clients
11290 sharing a same profile use the same server. This can be similar to what
11291 "appsession" does with the "request-learn" statement, but with support for
11292 multi-peer synchronization and state keeping across restarts. If no name is
11293 specified, the first cookie value is returned. This fetch should not be used
11294 anymore and should be replaced by req.cook() or res.cook() instead as it
11295 ambiguously uses the direction based on the context where it is used.
11296 See also : "appsession".
Willy Tarreaud63335a2010-02-26 12:56:52 +010011297
Willy Tarreau74ca5042013-06-11 23:12:07 +020011298hdr([<name>[,<occ>]]) : string
11299 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
11300 used on responses. Please refer to these respective fetches for more details.
11301 In case of doubt about the fetch direction, please use the explicit ones.
11302 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011303 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011304
Willy Tarreau74ca5042013-06-11 23:12:07 +020011305req.fhdr(<name>[,<occ>]) : string
11306 This extracts the last occurrence of header <name> in an HTTP request. When
11307 used from an ACL, all occurrences are iterated over until a match is found.
11308 Optionally, a specific occurrence might be specified as a position number.
11309 Positive values indicate a position from the first occurrence, with 1 being
11310 the first one. Negative values indicate positions relative to the last one,
11311 with -1 being the last one. It differs from req.hdr() in that any commas
11312 present in the value are returned and are not used as delimiters. This is
11313 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011314
Willy Tarreau74ca5042013-06-11 23:12:07 +020011315req.fhdr_cnt([<name>]) : integer
11316 Returns an integer value representing the number of occurrences of request
11317 header field name <name>, or the total number of header fields if <name> is
11318 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
11319 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011320
Willy Tarreau74ca5042013-06-11 23:12:07 +020011321req.hdr([<name>[,<occ>]]) : string
11322 This extracts the last occurrence of header <name> in an HTTP request. When
11323 used from an ACL, all occurrences are iterated over until a match is found.
11324 Optionally, a specific occurrence might be specified as a position number.
11325 Positive values indicate a position from the first occurrence, with 1 being
11326 the first one. Negative values indicate positions relative to the last one,
11327 with -1 being the last one. A typical use is with the X-Forwarded-For header
11328 once converted to IP, associated with an IP stick-table. The function
11329 considers any comma as a delimiter for distinct values. If full-line headers
11330 are desired instead, use req.fhdr(). Please carefully check RFC2616 to know
11331 how certain headers are supposed to be parsed. Also, some of them are case
11332 insensitive (eg: Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010011333
Willy Tarreau74ca5042013-06-11 23:12:07 +020011334 ACL derivatives :
11335 hdr([<name>[,<occ>]]) : exact string match
11336 hdr_beg([<name>[,<occ>]]) : prefix match
11337 hdr_dir([<name>[,<occ>]]) : subdir match
11338 hdr_dom([<name>[,<occ>]]) : domain match
11339 hdr_end([<name>[,<occ>]]) : suffix match
11340 hdr_len([<name>[,<occ>]]) : length match
11341 hdr_reg([<name>[,<occ>]]) : regex match
11342 hdr_sub([<name>[,<occ>]]) : substring match
11343
11344req.hdr_cnt([<name>]) : integer
11345hdr_cnt([<header>]) : integer (deprecated)
11346 Returns an integer value representing the number of occurrences of request
11347 header field name <name>, or the total number of header field values if
11348 <name> is not specified. It is important to remember that one header line may
11349 count as several headers if it has several values. The function considers any
11350 comma as a delimiter for distinct values. If full-line headers are desired
11351 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
11352 detect presence, absence or abuse of a specific header, as well as to block
11353 request smuggling attacks by rejecting requests which contain more than one
11354 of certain headers. See "req.hdr" for more information on header matching.
11355
11356req.hdr_ip([<name>[,<occ>]]) : ip
11357hdr_ip([<name>[,<occ>]]) : ip (deprecated)
11358 This extracts the last occurrence of header <name> in an HTTP request,
11359 converts it to an IPv4 or IPv6 address and returns this address. When used
11360 with ACLs, all occurrences are checked, and if <name> is omitted, every value
11361 of every header is checked. Optionally, a specific occurrence might be
11362 specified as a position number. Positive values indicate a position from the
11363 first occurrence, with 1 being the first one. Negative values indicate
11364 positions relative to the last one, with -1 being the last one. A typical use
11365 is with the X-Forwarded-For and X-Client-IP headers.
11366
11367req.hdr_val([<name>[,<occ>]]) : integer
11368hdr_val([<name>[,<occ>]]) : integer (deprecated)
11369 This extracts the last occurrence of header <name> in an HTTP request, and
11370 converts it to an integer value. When used with ACLs, all occurrences are
11371 checked, and if <name> is omitted, every value of every header is checked.
11372 Optionally, a specific occurrence might be specified as a position number.
11373 Positive values indicate a position from the first occurrence, with 1 being
11374 the first one. Negative values indicate positions relative to the last one,
11375 with -1 being the last one. A typical use is with the X-Forwarded-For header.
11376
11377http_auth(<userlist>) : boolean
11378 Returns a boolean indicating whether the authentication data received from
11379 the client match a username & password stored in the specified userlist. This
11380 fetch function is not really useful outside of ACLs. Currently only http
11381 basic auth is supported.
11382
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011383http_auth_group(<userlist>) : string
11384 Returns a string corresponding to the user name found in the authentication
11385 data received from the client if both the user name and password are valid
11386 according to the specified userlist. The main purpose is to use it in ACLs
11387 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011388 This fetch function is not really useful outside of ACLs. Currently only http
11389 basic auth is supported.
11390
11391 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011392 http_auth_group(<userlist>) : group ...
11393 Returns true when the user extracted from the request and whose password is
11394 valid according to the specified userlist belongs to at least one of the
11395 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011396
11397http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020011398 Returns true when the request being processed is the first one of the
11399 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020011400 from some requests when a request is not the first one, or to help grouping
11401 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020011402
Willy Tarreau74ca5042013-06-11 23:12:07 +020011403method : integer + string
11404 Returns an integer value corresponding to the method in the HTTP request. For
11405 example, "GET" equals 1 (check sources to establish the matching). Value 9
11406 means "other method" and may be converted to a string extracted from the
11407 stream. This should not be used directly as a sample, this is only meant to
11408 be used from ACLs, which transparently convert methods from patterns to these
11409 integer + string values. Some predefined ACL already check for most common
11410 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011411
Willy Tarreau74ca5042013-06-11 23:12:07 +020011412 ACL derivatives :
11413 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011414
Willy Tarreau74ca5042013-06-11 23:12:07 +020011415 Example :
11416 # only accept GET and HEAD requests
11417 acl valid_method method GET HEAD
11418 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020011419
Willy Tarreau74ca5042013-06-11 23:12:07 +020011420path : string
11421 This extracts the request's URL path, which starts at the first slash and
11422 ends before the question mark (without the host part). A typical use is with
11423 prefetch-capable caches, and with portals which need to aggregate multiple
11424 information from databases and keep them in caches. Note that with outgoing
11425 caches, it would be wiser to use "url" instead. With ACLs, it's typically
11426 used to match exact file names (eg: "/login.php"), or directory parts using
11427 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011428
Willy Tarreau74ca5042013-06-11 23:12:07 +020011429 ACL derivatives :
11430 path : exact string match
11431 path_beg : prefix match
11432 path_dir : subdir match
11433 path_dom : domain match
11434 path_end : suffix match
11435 path_len : length match
11436 path_reg : regex match
11437 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011438
Willy Tarreau74ca5042013-06-11 23:12:07 +020011439req.ver : string
11440req_ver : string (deprecated)
11441 Returns the version string from the HTTP request, for example "1.1". This can
11442 be useful for logs, but is mostly there for ACL. Some predefined ACL already
11443 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011444
Willy Tarreau74ca5042013-06-11 23:12:07 +020011445 ACL derivatives :
11446 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020011447
Willy Tarreau74ca5042013-06-11 23:12:07 +020011448res.comp : boolean
11449 Returns the boolean "true" value if the response has been compressed by
11450 HAProxy, otherwise returns boolean "false". This may be used to add
11451 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011452
Willy Tarreau74ca5042013-06-11 23:12:07 +020011453res.comp_algo : string
11454 Returns a string containing the name of the algorithm used if the response
11455 was compressed by HAProxy, for example : "deflate". This may be used to add
11456 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011457
Willy Tarreau74ca5042013-06-11 23:12:07 +020011458res.cook([<name>]) : string
11459scook([<name>]) : string (deprecated)
11460 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11461 header line from the response, and returns its value as string. If no name is
11462 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011463
Willy Tarreau74ca5042013-06-11 23:12:07 +020011464 ACL derivatives :
11465 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011466
Willy Tarreau74ca5042013-06-11 23:12:07 +020011467res.cook_cnt([<name>]) : integer
11468scook_cnt([<name>]) : integer (deprecated)
11469 Returns an integer value representing the number of occurrences of the cookie
11470 <name> in the response, or all cookies if <name> is not specified. This is
11471 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011472
Willy Tarreau74ca5042013-06-11 23:12:07 +020011473res.cook_val([<name>]) : integer
11474scook_val([<name>]) : integer (deprecated)
11475 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11476 header line from the response, and converts its value to an integer which is
11477 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011478
Willy Tarreau74ca5042013-06-11 23:12:07 +020011479res.fhdr([<name>[,<occ>]]) : string
11480 This extracts the last occurrence of header <name> in an HTTP response, or of
11481 the last header if no <name> is specified. Optionally, a specific occurrence
11482 might be specified as a position number. Positive values indicate a position
11483 from the first occurrence, with 1 being the first one. Negative values
11484 indicate positions relative to the last one, with -1 being the last one. It
11485 differs from res.hdr() in that any commas present in the value are returned
11486 and are not used as delimiters. If this is not desired, the res.hdr() fetch
11487 should be used instead. This is sometimes useful with headers such as Date or
11488 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011489
Willy Tarreau74ca5042013-06-11 23:12:07 +020011490res.fhdr_cnt([<name>]) : integer
11491 Returns an integer value representing the number of occurrences of response
11492 header field name <name>, or the total number of header fields if <name> is
11493 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
11494 the number of full line headers and does not stop on commas. If this is not
11495 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011496
Willy Tarreau74ca5042013-06-11 23:12:07 +020011497res.hdr([<name>[,<occ>]]) : string
11498shdr([<name>[,<occ>]]) : string (deprecated)
11499 This extracts the last occurrence of header <name> in an HTTP response, or of
11500 the last header if no <name> is specified. Optionally, a specific occurrence
11501 might be specified as a position number. Positive values indicate a position
11502 from the first occurrence, with 1 being the first one. Negative values
11503 indicate positions relative to the last one, with -1 being the last one. This
11504 can be useful to learn some data into a stick-table. The function considers
11505 any comma as a delimiter for distinct values. If this is not desired, the
11506 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011507
Willy Tarreau74ca5042013-06-11 23:12:07 +020011508 ACL derivatives :
11509 shdr([<name>[,<occ>]]) : exact string match
11510 shdr_beg([<name>[,<occ>]]) : prefix match
11511 shdr_dir([<name>[,<occ>]]) : subdir match
11512 shdr_dom([<name>[,<occ>]]) : domain match
11513 shdr_end([<name>[,<occ>]]) : suffix match
11514 shdr_len([<name>[,<occ>]]) : length match
11515 shdr_reg([<name>[,<occ>]]) : regex match
11516 shdr_sub([<name>[,<occ>]]) : substring match
11517
11518res.hdr_cnt([<name>]) : integer
11519shdr_cnt([<name>]) : integer (deprecated)
11520 Returns an integer value representing the number of occurrences of response
11521 header field name <name>, or the total number of header fields if <name> is
11522 not specified. The function considers any comma as a delimiter for distinct
11523 values. If this is not desired, the res.fhdr_cnt() fetch should be used
11524 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011525
Willy Tarreau74ca5042013-06-11 23:12:07 +020011526res.hdr_ip([<name>[,<occ>]]) : ip
11527shdr_ip([<name>[,<occ>]]) : ip (deprecated)
11528 This extracts the last occurrence of header <name> in an HTTP response,
11529 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
11530 specific occurrence might be specified as a position number. Positive values
11531 indicate a position from the first occurrence, with 1 being the first one.
11532 Negative values indicate positions relative to the last one, with -1 being
11533 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011534
Willy Tarreau74ca5042013-06-11 23:12:07 +020011535res.hdr_val([<name>[,<occ>]]) : integer
11536shdr_val([<name>[,<occ>]]) : integer (deprecated)
11537 This extracts the last occurrence of header <name> in an HTTP response, and
11538 converts it to an integer value. Optionally, a specific occurrence might be
11539 specified as a position number. Positive values indicate a position from the
11540 first occurrence, with 1 being the first one. Negative values indicate
11541 positions relative to the last one, with -1 being the last one. This can be
11542 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011543
Willy Tarreau74ca5042013-06-11 23:12:07 +020011544res.ver : string
11545resp_ver : string (deprecated)
11546 Returns the version string from the HTTP response, for example "1.1". This
11547 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020011548
Willy Tarreau74ca5042013-06-11 23:12:07 +020011549 ACL derivatives :
11550 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011551
Willy Tarreau74ca5042013-06-11 23:12:07 +020011552set-cookie([<name>]) : string (deprecated)
11553 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11554 header line from the response and uses the corresponding value to match. This
11555 can be comparable to what "appsession" does with default options, but with
11556 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011557
Willy Tarreau74ca5042013-06-11 23:12:07 +020011558 This fetch function is deprecated and has been superseded by the "res.cook"
11559 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011560
Willy Tarreau74ca5042013-06-11 23:12:07 +020011561 See also : "appsession"
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011562
Willy Tarreau74ca5042013-06-11 23:12:07 +020011563status : integer
11564 Returns an integer containing the HTTP status code in the HTTP response, for
11565 example, 302. It is mostly used within ACLs and integer ranges, for example,
11566 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011567
Willy Tarreau74ca5042013-06-11 23:12:07 +020011568url : string
11569 This extracts the request's URL as presented in the request. A typical use is
11570 with prefetch-capable caches, and with portals which need to aggregate
11571 multiple information from databases and keep them in caches. With ACLs, using
11572 "path" is preferred over using "url", because clients may send a full URL as
11573 is normally done with proxies. The only real use is to match "*" which does
11574 not match in "path", and for which there is already a predefined ACL. See
11575 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011576
Willy Tarreau74ca5042013-06-11 23:12:07 +020011577 ACL derivatives :
11578 url : exact string match
11579 url_beg : prefix match
11580 url_dir : subdir match
11581 url_dom : domain match
11582 url_end : suffix match
11583 url_len : length match
11584 url_reg : regex match
11585 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011586
Willy Tarreau74ca5042013-06-11 23:12:07 +020011587url_ip : ip
11588 This extracts the IP address from the request's URL when the host part is
11589 presented as an IP address. Its use is very limited. For instance, a
11590 monitoring system might use this field as an alternative for the source IP in
11591 order to test what path a given source address would follow, or to force an
11592 entry in a table for a given source address. With ACLs it can be used to
11593 restrict access to certain systems through a proxy, for example when combined
11594 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011595
Willy Tarreau74ca5042013-06-11 23:12:07 +020011596url_port : integer
11597 This extracts the port part from the request's URL. Note that if the port is
11598 not specified in the request, port 80 is assumed. With ACLs it can be used to
11599 restrict access to certain systems through a proxy, for example when combined
11600 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011601
Willy Tarreau74ca5042013-06-11 23:12:07 +020011602urlp(<name>[,<delim>]) : string
11603url_param(<name>[,<delim>]) : string
11604 This extracts the first occurrence of the parameter <name> in the query
11605 string, which begins after either '?' or <delim>, and which ends before '&',
11606 ';' or <delim>. The parameter name is case-sensitive. The result is a string
11607 corresponding to the value of the parameter <name> as presented in the
11608 request (no URL decoding is performed). This can be used for session
11609 stickiness based on a client ID, to extract an application cookie passed as a
11610 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
11611 this fetch do not iterate over multiple parameters and stop at the first one
11612 as well.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011613
Willy Tarreau74ca5042013-06-11 23:12:07 +020011614 ACL derivatives :
11615 urlp(<name>[,<delim>]) : exact string match
11616 urlp_beg(<name>[,<delim>]) : prefix match
11617 urlp_dir(<name>[,<delim>]) : subdir match
11618 urlp_dom(<name>[,<delim>]) : domain match
11619 urlp_end(<name>[,<delim>]) : suffix match
11620 urlp_len(<name>[,<delim>]) : length match
11621 urlp_reg(<name>[,<delim>]) : regex match
11622 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011623
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011624
Willy Tarreau74ca5042013-06-11 23:12:07 +020011625 Example :
11626 # match http://example.com/foo?PHPSESSIONID=some_id
11627 stick on urlp(PHPSESSIONID)
11628 # match http://example.com/foo;JSESSIONID=some_id
11629 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011630
Willy Tarreau74ca5042013-06-11 23:12:07 +020011631urlp_val(<name>[,<delim>]) : integer
11632 See "urlp" above. This one extracts the URL parameter <name> in the request
11633 and converts it to an integer value. This can be used for session stickiness
11634 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020011635
Willy Tarreau198a7442008-01-17 12:05:32 +010011636
Willy Tarreau74ca5042013-06-11 23:12:07 +0200116377.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011638---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011639
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011640Some predefined ACLs are hard-coded so that they do not have to be declared in
11641every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020011642order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010011643
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011644ACL name Equivalent to Usage
11645---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011646FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020011647HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011648HTTP_1.0 req_ver 1.0 match HTTP version 1.0
11649HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010011650HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
11651HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
11652HTTP_URL_SLASH url_beg / match URL beginning with "/"
11653HTTP_URL_STAR url * match URL equal to "*"
11654LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011655METH_CONNECT method CONNECT match HTTP CONNECT method
11656METH_GET method GET HEAD match HTTP GET or HEAD method
11657METH_HEAD method HEAD match HTTP HEAD method
11658METH_OPTIONS method OPTIONS match HTTP OPTIONS method
11659METH_POST method POST match HTTP POST method
11660METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020011661RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011662REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010011663TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011664WAIT_END wait_end wait for end of content analysis
11665---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011666
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011667
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200116688. Logging
11669----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010011670
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011671One of HAProxy's strong points certainly lies is its precise logs. It probably
11672provides the finest level of information available for such a product, which is
11673very important for troubleshooting complex environments. Standard information
11674provided in logs include client ports, TCP/HTTP state timers, precise session
11675state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011676to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011677headers.
11678
11679In order to improve administrators reactivity, it offers a great transparency
11680about encountered problems, both internal and external, and it is possible to
11681send logs to different sources at the same time with different level filters :
11682
11683 - global process-level logs (system errors, start/stop, etc..)
11684 - per-instance system and internal errors (lack of resource, bugs, ...)
11685 - per-instance external troubles (servers up/down, max connections)
11686 - per-instance activity (client connections), either at the establishment or
11687 at the termination.
11688
11689The ability to distribute different levels of logs to different log servers
11690allow several production teams to interact and to fix their problems as soon
11691as possible. For example, the system team might monitor system-wide errors,
11692while the application team might be monitoring the up/down for their servers in
11693real time, and the security team might analyze the activity logs with one hour
11694delay.
11695
11696
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200116978.1. Log levels
11698---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011699
Simon Hormandf791f52011-05-29 15:01:10 +090011700TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011701source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090011702HTTP request, HTTP return code, number of bytes transmitted, conditions
11703in which the session ended, and even exchanged cookies values. For example
11704track a particular user's problems. All messages may be sent to up to two
11705syslog servers. Check the "log" keyword in section 4.2 for more information
11706about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011707
11708
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117098.2. Log formats
11710----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011711
William Lallemand48940402012-01-30 16:47:22 +010011712HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090011713and will be detailed in the following sections. A few of them may vary
11714slightly with the configuration, due to indicators specific to certain
11715options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011716
11717 - the default format, which is very basic and very rarely used. It only
11718 provides very basic information about the incoming connection at the moment
11719 it is accepted : source IP:port, destination IP:port, and frontend-name.
11720 This mode will eventually disappear so it will not be described to great
11721 extents.
11722
11723 - the TCP format, which is more advanced. This format is enabled when "option
11724 tcplog" is set on the frontend. HAProxy will then usually wait for the
11725 connection to terminate before logging. This format provides much richer
11726 information, such as timers, connection counts, queue size, etc... This
11727 format is recommended for pure TCP proxies.
11728
11729 - the HTTP format, which is the most advanced for HTTP proxying. This format
11730 is enabled when "option httplog" is set on the frontend. It provides the
11731 same information as the TCP format with some HTTP-specific fields such as
11732 the request, the status code, and captures of headers and cookies. This
11733 format is recommended for HTTP proxies.
11734
Emeric Brun3a058f32009-06-30 18:26:00 +020011735 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
11736 fields arranged in the same order as the CLF format. In this mode, all
11737 timers, captures, flags, etc... appear one per field after the end of the
11738 common fields, in the same order they appear in the standard HTTP format.
11739
William Lallemand48940402012-01-30 16:47:22 +010011740 - the custom log format, allows you to make your own log line.
11741
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011742Next sections will go deeper into details for each of these formats. Format
11743specification will be performed on a "field" basis. Unless stated otherwise, a
11744field is a portion of text delimited by any number of spaces. Since syslog
11745servers are susceptible of inserting fields at the beginning of a line, it is
11746always assumed that the first field is the one containing the process name and
11747identifier.
11748
11749Note : Since log lines may be quite long, the log examples in sections below
11750 might be broken into multiple lines. The example log lines will be
11751 prefixed with 3 closing angle brackets ('>>>') and each time a log is
11752 broken into multiple lines, each non-final line will end with a
11753 backslash ('\') and the next line will start indented by two characters.
11754
11755
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117568.2.1. Default log format
11757-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011758
11759This format is used when no specific option is set. The log is emitted as soon
11760as the connection is accepted. One should note that this currently is the only
11761format which logs the request's destination IP and ports.
11762
11763 Example :
11764 listen www
11765 mode http
11766 log global
11767 server srv1 127.0.0.1:8000
11768
11769 >>> Feb 6 12:12:09 localhost \
11770 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
11771 (www/HTTP)
11772
11773 Field Format Extract from the example above
11774 1 process_name '[' pid ']:' haproxy[14385]:
11775 2 'Connect from' Connect from
11776 3 source_ip ':' source_port 10.0.1.2:33312
11777 4 'to' to
11778 5 destination_ip ':' destination_port 10.0.3.31:8012
11779 6 '(' frontend_name '/' mode ')' (www/HTTP)
11780
11781Detailed fields description :
11782 - "source_ip" is the IP address of the client which initiated the connection.
11783 - "source_port" is the TCP port of the client which initiated the connection.
11784 - "destination_ip" is the IP address the client connected to.
11785 - "destination_port" is the TCP port the client connected to.
11786 - "frontend_name" is the name of the frontend (or listener) which received
11787 and processed the connection.
11788 - "mode is the mode the frontend is operating (TCP or HTTP).
11789
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011790In case of a UNIX socket, the source and destination addresses are marked as
11791"unix:" and the ports reflect the internal ID of the socket which accepted the
11792connection (the same ID as reported in the stats).
11793
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011794It is advised not to use this deprecated format for newer installations as it
11795will eventually disappear.
11796
11797
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117988.2.2. TCP log format
11799---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011800
11801The TCP format is used when "option tcplog" is specified in the frontend, and
11802is the recommended format for pure TCP proxies. It provides a lot of precious
11803information for troubleshooting. Since this format includes timers and byte
11804counts, the log is normally emitted at the end of the session. It can be
11805emitted earlier if "option logasap" is specified, which makes sense in most
11806environments with long sessions such as remote terminals. Sessions which match
11807the "monitor" rules are never logged. It is also possible not to emit logs for
11808sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020011809specifying "option dontlognull" in the frontend. Successful connections will
11810not be logged if "option dontlog-normal" is specified in the frontend. A few
11811fields may slightly vary depending on some configuration options, those are
11812marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011813
11814 Example :
11815 frontend fnt
11816 mode tcp
11817 option tcplog
11818 log global
11819 default_backend bck
11820
11821 backend bck
11822 server srv1 127.0.0.1:8000
11823
11824 >>> Feb 6 12:12:56 localhost \
11825 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
11826 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
11827
11828 Field Format Extract from the example above
11829 1 process_name '[' pid ']:' haproxy[14387]:
11830 2 client_ip ':' client_port 10.0.1.2:33313
11831 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
11832 4 frontend_name fnt
11833 5 backend_name '/' server_name bck/srv1
11834 6 Tw '/' Tc '/' Tt* 0/0/5007
11835 7 bytes_read* 212
11836 8 termination_state --
11837 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
11838 10 srv_queue '/' backend_queue 0/0
11839
11840Detailed fields description :
11841 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011842 connection to haproxy. If the connection was accepted on a UNIX socket
11843 instead, the IP address would be replaced with the word "unix". Note that
11844 when the connection is accepted on a socket configured with "accept-proxy"
11845 and the PROXY protocol is correctly used, then the logs will reflect the
11846 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011847
11848 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010011849 If the connection was accepted on a UNIX socket instead, the port would be
11850 replaced with the ID of the accepting socket, which is also reported in the
11851 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011852
11853 - "accept_date" is the exact date when the connection was received by haproxy
11854 (which might be very slightly different from the date observed on the
11855 network if there was some queuing in the system's backlog). This is usually
11856 the same date which may appear in any upstream firewall's log.
11857
11858 - "frontend_name" is the name of the frontend (or listener) which received
11859 and processed the connection.
11860
11861 - "backend_name" is the name of the backend (or listener) which was selected
11862 to manage the connection to the server. This will be the same as the
11863 frontend if no switching rule has been applied, which is common for TCP
11864 applications.
11865
11866 - "server_name" is the name of the last server to which the connection was
11867 sent, which might differ from the first one if there were connection errors
11868 and a redispatch occurred. Note that this server belongs to the backend
11869 which processed the request. If the connection was aborted before reaching
11870 a server, "<NOSRV>" is indicated instead of a server name.
11871
11872 - "Tw" is the total time in milliseconds spent waiting in the various queues.
11873 It can be "-1" if the connection was aborted before reaching the queue.
11874 See "Timers" below for more details.
11875
11876 - "Tc" is the total time in milliseconds spent waiting for the connection to
11877 establish to the final server, including retries. It can be "-1" if the
11878 connection was aborted before a connection could be established. See
11879 "Timers" below for more details.
11880
11881 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011882 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011883 "option logasap" was specified, then the time counting stops at the moment
11884 the log is emitted. In this case, a '+' sign is prepended before the value,
11885 indicating that the final one will be larger. See "Timers" below for more
11886 details.
11887
11888 - "bytes_read" is the total number of bytes transmitted from the server to
11889 the client when the log is emitted. If "option logasap" is specified, the
11890 this value will be prefixed with a '+' sign indicating that the final one
11891 may be larger. Please note that this value is a 64-bit counter, so log
11892 analysis tools must be able to handle it without overflowing.
11893
11894 - "termination_state" is the condition the session was in when the session
11895 ended. This indicates the session state, which side caused the end of
11896 session to happen, and for what reason (timeout, error, ...). The normal
11897 flags should be "--", indicating the session was closed by either end with
11898 no data remaining in buffers. See below "Session state at disconnection"
11899 for more details.
11900
11901 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011902 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011903 limits have been reached. For instance, if actconn is close to 512 when
11904 multiple connection errors occur, chances are high that the system limits
11905 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011906 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011907
11908 - "feconn" is the total number of concurrent connections on the frontend when
11909 the session was logged. It is useful to estimate the amount of resource
11910 required to sustain high loads, and to detect when the frontend's "maxconn"
11911 has been reached. Most often when this value increases by huge jumps, it is
11912 because there is congestion on the backend servers, but sometimes it can be
11913 caused by a denial of service attack.
11914
11915 - "beconn" is the total number of concurrent connections handled by the
11916 backend when the session was logged. It includes the total number of
11917 concurrent connections active on servers as well as the number of
11918 connections pending in queues. It is useful to estimate the amount of
11919 additional servers needed to support high loads for a given application.
11920 Most often when this value increases by huge jumps, it is because there is
11921 congestion on the backend servers, but sometimes it can be caused by a
11922 denial of service attack.
11923
11924 - "srv_conn" is the total number of concurrent connections still active on
11925 the server when the session was logged. It can never exceed the server's
11926 configured "maxconn" parameter. If this value is very often close or equal
11927 to the server's "maxconn", it means that traffic regulation is involved a
11928 lot, meaning that either the server's maxconn value is too low, or that
11929 there aren't enough servers to process the load with an optimal response
11930 time. When only one of the server's "srv_conn" is high, it usually means
11931 that this server has some trouble causing the connections to take longer to
11932 be processed than on other servers.
11933
11934 - "retries" is the number of connection retries experienced by this session
11935 when trying to connect to the server. It must normally be zero, unless a
11936 server is being stopped at the same moment the connection was attempted.
11937 Frequent retries generally indicate either a network problem between
11938 haproxy and the server, or a misconfigured system backlog on the server
11939 preventing new connections from being queued. This field may optionally be
11940 prefixed with a '+' sign, indicating that the session has experienced a
11941 redispatch after the maximal retry count has been reached on the initial
11942 server. In this case, the server name appearing in the log is the one the
11943 connection was redispatched to, and not the first one, though both may
11944 sometimes be the same in case of hashing for instance. So as a general rule
11945 of thumb, when a '+' is present in front of the retry count, this count
11946 should not be attributed to the logged server.
11947
11948 - "srv_queue" is the total number of requests which were processed before
11949 this one in the server queue. It is zero when the request has not gone
11950 through the server queue. It makes it possible to estimate the approximate
11951 server's response time by dividing the time spent in queue by the number of
11952 requests in the queue. It is worth noting that if a session experiences a
11953 redispatch and passes through two server queues, their positions will be
11954 cumulated. A request should not pass through both the server queue and the
11955 backend queue unless a redispatch occurs.
11956
11957 - "backend_queue" is the total number of requests which were processed before
11958 this one in the backend's global queue. It is zero when the request has not
11959 gone through the global queue. It makes it possible to estimate the average
11960 queue length, which easily translates into a number of missing servers when
11961 divided by a server's "maxconn" parameter. It is worth noting that if a
11962 session experiences a redispatch, it may pass twice in the backend's queue,
11963 and then both positions will be cumulated. A request should not pass
11964 through both the server queue and the backend queue unless a redispatch
11965 occurs.
11966
11967
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200119688.2.3. HTTP log format
11969----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011970
11971The HTTP format is the most complete and the best suited for HTTP proxies. It
11972is enabled by when "option httplog" is specified in the frontend. It provides
11973the same level of information as the TCP format with additional features which
11974are specific to the HTTP protocol. Just like the TCP format, the log is usually
11975emitted at the end of the session, unless "option logasap" is specified, which
11976generally only makes sense for download sites. A session which matches the
11977"monitor" rules will never logged. It is also possible not to log sessions for
11978which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020011979frontend. Successful connections will not be logged if "option dontlog-normal"
11980is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011981
11982Most fields are shared with the TCP log, some being different. A few fields may
11983slightly vary depending on some configuration options. Those ones are marked
11984with a star ('*') after the field name below.
11985
11986 Example :
11987 frontend http-in
11988 mode http
11989 option httplog
11990 log global
11991 default_backend bck
11992
11993 backend static
11994 server srv1 127.0.0.1:8000
11995
11996 >>> Feb 6 12:14:14 localhost \
11997 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
11998 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 +010011999 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012000
12001 Field Format Extract from the example above
12002 1 process_name '[' pid ']:' haproxy[14389]:
12003 2 client_ip ':' client_port 10.0.1.2:33317
12004 3 '[' accept_date ']' [06/Feb/2009:12:14:14.655]
12005 4 frontend_name http-in
12006 5 backend_name '/' server_name static/srv1
12007 6 Tq '/' Tw '/' Tc '/' Tr '/' Tt* 10/0/30/69/109
12008 7 status_code 200
12009 8 bytes_read* 2750
12010 9 captured_request_cookie -
12011 10 captured_response_cookie -
12012 11 termination_state ----
12013 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
12014 13 srv_queue '/' backend_queue 0/0
12015 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
12016 15 '{' captured_response_headers* '}' {}
12017 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010012018
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012019
12020Detailed fields description :
12021 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012022 connection to haproxy. If the connection was accepted on a UNIX socket
12023 instead, the IP address would be replaced with the word "unix". Note that
12024 when the connection is accepted on a socket configured with "accept-proxy"
12025 and the PROXY protocol is correctly used, then the logs will reflect the
12026 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012027
12028 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012029 If the connection was accepted on a UNIX socket instead, the port would be
12030 replaced with the ID of the accepting socket, which is also reported in the
12031 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012032
12033 - "accept_date" is the exact date when the TCP connection was received by
12034 haproxy (which might be very slightly different from the date observed on
12035 the network if there was some queuing in the system's backlog). This is
12036 usually the same date which may appear in any upstream firewall's log. This
12037 does not depend on the fact that the client has sent the request or not.
12038
12039 - "frontend_name" is the name of the frontend (or listener) which received
12040 and processed the connection.
12041
12042 - "backend_name" is the name of the backend (or listener) which was selected
12043 to manage the connection to the server. This will be the same as the
12044 frontend if no switching rule has been applied.
12045
12046 - "server_name" is the name of the last server to which the connection was
12047 sent, which might differ from the first one if there were connection errors
12048 and a redispatch occurred. Note that this server belongs to the backend
12049 which processed the request. If the request was aborted before reaching a
12050 server, "<NOSRV>" is indicated instead of a server name. If the request was
12051 intercepted by the stats subsystem, "<STATS>" is indicated instead.
12052
12053 - "Tq" is the total time in milliseconds spent waiting for the client to send
12054 a full HTTP request, not counting data. It can be "-1" if the connection
12055 was aborted before a complete request could be received. It should always
12056 be very small because a request generally fits in one single packet. Large
12057 times here generally indicate network trouble between the client and
12058 haproxy. See "Timers" below for more details.
12059
12060 - "Tw" is the total time in milliseconds spent waiting in the various queues.
12061 It can be "-1" if the connection was aborted before reaching the queue.
12062 See "Timers" below for more details.
12063
12064 - "Tc" is the total time in milliseconds spent waiting for the connection to
12065 establish to the final server, including retries. It can be "-1" if the
12066 request was aborted before a connection could be established. See "Timers"
12067 below for more details.
12068
12069 - "Tr" is the total time in milliseconds spent waiting for the server to send
12070 a full HTTP response, not counting data. It can be "-1" if the request was
12071 aborted before a complete response could be received. It generally matches
12072 the server's processing time for the request, though it may be altered by
12073 the amount of data sent by the client to the server. Large times here on
12074 "GET" requests generally indicate an overloaded server. See "Timers" below
12075 for more details.
12076
12077 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012078 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012079 "option logasap" was specified, then the time counting stops at the moment
12080 the log is emitted. In this case, a '+' sign is prepended before the value,
12081 indicating that the final one will be larger. See "Timers" below for more
12082 details.
12083
12084 - "status_code" is the HTTP status code returned to the client. This status
12085 is generally set by the server, but it might also be set by haproxy when
12086 the server cannot be reached or when its response is blocked by haproxy.
12087
12088 - "bytes_read" is the total number of bytes transmitted to the client when
12089 the log is emitted. This does include HTTP headers. If "option logasap" is
12090 specified, the this value will be prefixed with a '+' sign indicating that
12091 the final one may be larger. Please note that this value is a 64-bit
12092 counter, so log analysis tools must be able to handle it without
12093 overflowing.
12094
12095 - "captured_request_cookie" is an optional "name=value" entry indicating that
12096 the client had this cookie in the request. The cookie name and its maximum
12097 length are defined by the "capture cookie" statement in the frontend
12098 configuration. The field is a single dash ('-') when the option is not
12099 set. Only one cookie may be captured, it is generally used to track session
12100 ID exchanges between a client and a server to detect session crossing
12101 between clients due to application bugs. For more details, please consult
12102 the section "Capturing HTTP headers and cookies" below.
12103
12104 - "captured_response_cookie" is an optional "name=value" entry indicating
12105 that the server has returned a cookie with its response. The cookie name
12106 and its maximum length are defined by the "capture cookie" statement in the
12107 frontend configuration. The field is a single dash ('-') when the option is
12108 not set. Only one cookie may be captured, it is generally used to track
12109 session ID exchanges between a client and a server to detect session
12110 crossing between clients due to application bugs. For more details, please
12111 consult the section "Capturing HTTP headers and cookies" below.
12112
12113 - "termination_state" is the condition the session was in when the session
12114 ended. This indicates the session state, which side caused the end of
12115 session to happen, for what reason (timeout, error, ...), just like in TCP
12116 logs, and information about persistence operations on cookies in the last
12117 two characters. The normal flags should begin with "--", indicating the
12118 session was closed by either end with no data remaining in buffers. See
12119 below "Session state at disconnection" for more details.
12120
12121 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012122 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012123 limits have been reached. For instance, if actconn is close to 512 or 1024
12124 when multiple connection errors occur, chances are high that the system
12125 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012126 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012127 system.
12128
12129 - "feconn" is the total number of concurrent connections on the frontend when
12130 the session was logged. It is useful to estimate the amount of resource
12131 required to sustain high loads, and to detect when the frontend's "maxconn"
12132 has been reached. Most often when this value increases by huge jumps, it is
12133 because there is congestion on the backend servers, but sometimes it can be
12134 caused by a denial of service attack.
12135
12136 - "beconn" is the total number of concurrent connections handled by the
12137 backend when the session was logged. It includes the total number of
12138 concurrent connections active on servers as well as the number of
12139 connections pending in queues. It is useful to estimate the amount of
12140 additional servers needed to support high loads for a given application.
12141 Most often when this value increases by huge jumps, it is because there is
12142 congestion on the backend servers, but sometimes it can be caused by a
12143 denial of service attack.
12144
12145 - "srv_conn" is the total number of concurrent connections still active on
12146 the server when the session was logged. It can never exceed the server's
12147 configured "maxconn" parameter. If this value is very often close or equal
12148 to the server's "maxconn", it means that traffic regulation is involved a
12149 lot, meaning that either the server's maxconn value is too low, or that
12150 there aren't enough servers to process the load with an optimal response
12151 time. When only one of the server's "srv_conn" is high, it usually means
12152 that this server has some trouble causing the requests to take longer to be
12153 processed than on other servers.
12154
12155 - "retries" is the number of connection retries experienced by this session
12156 when trying to connect to the server. It must normally be zero, unless a
12157 server is being stopped at the same moment the connection was attempted.
12158 Frequent retries generally indicate either a network problem between
12159 haproxy and the server, or a misconfigured system backlog on the server
12160 preventing new connections from being queued. This field may optionally be
12161 prefixed with a '+' sign, indicating that the session has experienced a
12162 redispatch after the maximal retry count has been reached on the initial
12163 server. In this case, the server name appearing in the log is the one the
12164 connection was redispatched to, and not the first one, though both may
12165 sometimes be the same in case of hashing for instance. So as a general rule
12166 of thumb, when a '+' is present in front of the retry count, this count
12167 should not be attributed to the logged server.
12168
12169 - "srv_queue" is the total number of requests which were processed before
12170 this one in the server queue. It is zero when the request has not gone
12171 through the server queue. It makes it possible to estimate the approximate
12172 server's response time by dividing the time spent in queue by the number of
12173 requests in the queue. It is worth noting that if a session experiences a
12174 redispatch and passes through two server queues, their positions will be
12175 cumulated. A request should not pass through both the server queue and the
12176 backend queue unless a redispatch occurs.
12177
12178 - "backend_queue" is the total number of requests which were processed before
12179 this one in the backend's global queue. It is zero when the request has not
12180 gone through the global queue. It makes it possible to estimate the average
12181 queue length, which easily translates into a number of missing servers when
12182 divided by a server's "maxconn" parameter. It is worth noting that if a
12183 session experiences a redispatch, it may pass twice in the backend's queue,
12184 and then both positions will be cumulated. A request should not pass
12185 through both the server queue and the backend queue unless a redispatch
12186 occurs.
12187
12188 - "captured_request_headers" is a list of headers captured in the request due
12189 to the presence of the "capture request header" statement in the frontend.
12190 Multiple headers can be captured, they will be delimited by a vertical bar
12191 ('|'). When no capture is enabled, the braces do not appear, causing a
12192 shift of remaining fields. It is important to note that this field may
12193 contain spaces, and that using it requires a smarter log parser than when
12194 it's not used. Please consult the section "Capturing HTTP headers and
12195 cookies" below for more details.
12196
12197 - "captured_response_headers" is a list of headers captured in the response
12198 due to the presence of the "capture response header" statement in the
12199 frontend. Multiple headers can be captured, they will be delimited by a
12200 vertical bar ('|'). When no capture is enabled, the braces do not appear,
12201 causing a shift of remaining fields. It is important to note that this
12202 field may contain spaces, and that using it requires a smarter log parser
12203 than when it's not used. Please consult the section "Capturing HTTP headers
12204 and cookies" below for more details.
12205
12206 - "http_request" is the complete HTTP request line, including the method,
12207 request and HTTP version string. Non-printable characters are encoded (see
12208 below the section "Non-printable characters"). This is always the last
12209 field, and it is always delimited by quotes and is the only one which can
12210 contain quotes. If new fields are added to the log format, they will be
12211 added before this field. This field might be truncated if the request is
12212 huge and does not fit in the standard syslog buffer (1024 characters). This
12213 is the reason why this field must always remain the last one.
12214
12215
Cyril Bontédc4d9032012-04-08 21:57:39 +0200122168.2.4. Custom log format
12217------------------------
William Lallemand48940402012-01-30 16:47:22 +010012218
Willy Tarreau2beef582012-12-20 17:22:52 +010012219The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010012220mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010012221
12222HAproxy understands some log format variables. % precedes log format variables.
12223Variables can take arguments using braces ('{}'), and multiple arguments are
12224separated by commas within the braces. Flags may be added or removed by
12225prefixing them with a '+' or '-' sign.
12226
12227Special variable "%o" may be used to propagate its flags to all other
12228variables on the same format string. This is particularly handy with quoted
12229string formats ("Q").
12230
Willy Tarreauc8368452012-12-21 00:09:23 +010012231If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020012232as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010012233less common information such as the client's SSL certificate's DN, or to log
12234the key that would be used to store an entry into a stick table.
12235
William Lallemand48940402012-01-30 16:47:22 +010012236Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012237In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010012238in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010012239
12240Flags are :
12241 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012242 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
William Lallemand48940402012-01-30 16:47:22 +010012243
12244 Example:
12245
12246 log-format %T\ %t\ Some\ Text
12247 log-format %{+Q}o\ %t\ %s\ %{-Q}r
12248
12249At the moment, the default HTTP format is defined this way :
12250
Willy Tarreau2beef582012-12-20 17:22:52 +010012251 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tq/%Tw/%Tc/%Tr/%Tt\ %ST\ %B\ %CC\ \
12252 %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r
William Lallemand48940402012-01-30 16:47:22 +010012253
William Lallemandbddd4fd2012-02-27 11:23:10 +010012254the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010012255
Willy Tarreau2beef582012-12-20 17:22:52 +010012256 log-format %{+Q}o\ %{-Q}ci\ -\ -\ [%T]\ %r\ %ST\ %B\ \"\"\ \"\"\ %cp\ \
Willy Tarreau773d65f2012-10-12 14:56:11 +020012257 %ms\ %ft\ %b\ %s\ \%Tq\ %Tw\ %Tc\ %Tr\ %Tt\ %tsc\ %ac\ %fc\ \
Willy Tarreau2beef582012-12-20 17:22:52 +010012258 %bc\ %sc\ %rc\ %sq\ %bq\ %CC\ %CS\ \%hrl\ %hsl
William Lallemand48940402012-01-30 16:47:22 +010012259
William Lallemandbddd4fd2012-02-27 11:23:10 +010012260and the default TCP format is defined this way :
12261
Willy Tarreau2beef582012-12-20 17:22:52 +010012262 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ \
William Lallemandbddd4fd2012-02-27 11:23:10 +010012263 %ac/%fc/%bc/%sc/%rc\ %sq/%bq
12264
William Lallemand48940402012-01-30 16:47:22 +010012265Please refer to the table below for currently defined variables :
12266
William Lallemandbddd4fd2012-02-27 11:23:10 +010012267 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012268 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012269 +---+------+-----------------------------------------------+-------------+
12270 | | %o | special variable, apply flags on all next var | |
12271 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010012272 | | %B | bytes_read (from server to client) | numeric |
12273 | H | %CC | captured_request_cookie | string |
12274 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020012275 | | %H | hostname | string |
William Lallemanda73203e2012-03-12 12:48:57 +010012276 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020012277 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012278 | | %T | gmt_date_time | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012279 | | %Tc | Tc | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080012280 | | %Tl | local_date_time | date |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012281 | H | %Tq | Tq | numeric |
12282 | H | %Tr | Tr | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012283 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012284 | | %Tt | Tt | numeric |
12285 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012286 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012287 | | %ac | actconn | numeric |
12288 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012289 | | %bc | beconn (backend concurrent connections) | numeric |
12290 | | %bi | backend_source_ip (connecting address) | IP |
12291 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012292 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012293 | | %ci | client_ip (accepted address) | IP |
12294 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012295 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012296 | | %fc | feconn (frontend concurrent connections) | numeric |
12297 | | %fi | frontend_ip (accepting address) | IP |
12298 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020012299 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020012300 | | %hr | captured_request_headers default style | string |
12301 | | %hrl | captured_request_headers CLF style | string list |
12302 | | %hs | captured_response_headers default style | string |
12303 | | %hsl | captured_response_headers CLF style | string list |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012304 | | %ms | accept date milliseconds | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012305 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012306 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012307 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010012308 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012309 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012310 | | %sc | srv_conn (server concurrent connections) | numeric |
12311 | | %si | server_IP (target address) | IP |
12312 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012313 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012314 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
12315 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012316 | | %t | date_time (with millisecond resolution) | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012317 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012318 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012319 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010012320
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012321 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010012322
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010012323
123248.2.5. Error log format
12325-----------------------
12326
12327When an incoming connection fails due to an SSL handshake or an invalid PROXY
12328protocol header, haproxy will log the event using a shorter, fixed line format.
12329By default, logs are emitted at the LOG_INFO level, unless the option
12330"log-separate-errors" is set in the backend, in which case the LOG_ERR level
12331will be used. Connections on which no data are exchanged (eg: probes) are not
12332logged if the "dontlognull" option is set.
12333
12334The format looks like this :
12335
12336 >>> Dec 3 18:27:14 localhost \
12337 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
12338 Connection error during SSL handshake
12339
12340 Field Format Extract from the example above
12341 1 process_name '[' pid ']:' haproxy[6103]:
12342 2 client_ip ':' client_port 127.0.0.1:56059
12343 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
12344 4 frontend_name "/" bind_name ":" frt/f1:
12345 5 message Connection error during SSL handshake
12346
12347These fields just provide minimal information to help debugging connection
12348failures.
12349
12350
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123518.3. Advanced logging options
12352-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012353
12354Some advanced logging options are often looked for but are not easy to find out
12355just by looking at the various options. Here is an entry point for the few
12356options which can enable better logging. Please refer to the keywords reference
12357for more information about their usage.
12358
12359
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123608.3.1. Disabling logging of external tests
12361------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012362
12363It is quite common to have some monitoring tools perform health checks on
12364haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
12365commercial load-balancer, and sometimes it will simply be a more complete
12366monitoring system such as Nagios. When the tests are very frequent, users often
12367ask how to disable logging for those checks. There are three possibilities :
12368
12369 - if connections come from everywhere and are just TCP probes, it is often
12370 desired to simply disable logging of connections without data exchange, by
12371 setting "option dontlognull" in the frontend. It also disables logging of
12372 port scans, which may or may not be desired.
12373
12374 - if the connection come from a known source network, use "monitor-net" to
12375 declare this network as monitoring only. Any host in this network will then
12376 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012377 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012378 such as other load-balancers.
12379
12380 - if the tests are performed on a known URI, use "monitor-uri" to declare
12381 this URI as dedicated to monitoring. Any host sending this request will
12382 only get the result of a health-check, and the request will not be logged.
12383
12384
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123858.3.2. Logging before waiting for the session to terminate
12386----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012387
12388The problem with logging at end of connection is that you have no clue about
12389what is happening during very long sessions, such as remote terminal sessions
12390or large file downloads. This problem can be worked around by specifying
12391"option logasap" in the frontend. Haproxy will then log as soon as possible,
12392just before data transfer begins. This means that in case of TCP, it will still
12393log the connection status to the server, and in case of HTTP, it will log just
12394after processing the server headers. In this case, the number of bytes reported
12395is the number of header bytes sent to the client. In order to avoid confusion
12396with normal logs, the total time field and the number of bytes are prefixed
12397with a '+' sign which means that real numbers are certainly larger.
12398
12399
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124008.3.3. Raising log level upon errors
12401------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012402
12403Sometimes it is more convenient to separate normal traffic from errors logs,
12404for instance in order to ease error monitoring from log files. When the option
12405"log-separate-errors" is used, connections which experience errors, timeouts,
12406retries, redispatches or HTTP status codes 5xx will see their syslog level
12407raised from "info" to "err". This will help a syslog daemon store the log in
12408a separate file. It is very important to keep the errors in the normal traffic
12409file too, so that log ordering is not altered. You should also be careful if
12410you already have configured your syslog daemon to store all logs higher than
12411"notice" in an "admin" file, because the "err" level is higher than "notice".
12412
12413
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124148.3.4. Disabling logging of successful connections
12415--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012416
12417Although this may sound strange at first, some large sites have to deal with
12418multiple thousands of logs per second and are experiencing difficulties keeping
12419them intact for a long time or detecting errors within them. If the option
12420"dontlog-normal" is set on the frontend, all normal connections will not be
12421logged. In this regard, a normal connection is defined as one without any
12422error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
12423and a response with a status 5xx is not considered normal and will be logged
12424too. Of course, doing is is really discouraged as it will remove most of the
12425useful information from the logs. Do this only if you have no other
12426alternative.
12427
12428
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124298.4. Timing events
12430------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012431
12432Timers provide a great help in troubleshooting network problems. All values are
12433reported in milliseconds (ms). These timers should be used in conjunction with
12434the session termination flags. In TCP mode with "option tcplog" set on the
12435frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
12436mode, 5 control points are reported under the form "Tq/Tw/Tc/Tr/Tt" :
12437
12438 - Tq: total time to get the client request (HTTP mode only). It's the time
12439 elapsed between the moment the client connection was accepted and the
12440 moment the proxy received the last HTTP header. The value "-1" indicates
12441 that the end of headers (empty line) has never been seen. This happens when
12442 the client closes prematurely or times out.
12443
12444 - Tw: total time spent in the queues waiting for a connection slot. It
12445 accounts for backend queue as well as the server queues, and depends on the
12446 queue size, and the time needed for the server to complete previous
12447 requests. The value "-1" means that the request was killed before reaching
12448 the queue, which is generally what happens with invalid or denied requests.
12449
12450 - Tc: total time to establish the TCP connection to the server. It's the time
12451 elapsed between the moment the proxy sent the connection request, and the
12452 moment it was acknowledged by the server, or between the TCP SYN packet and
12453 the matching SYN/ACK packet in return. The value "-1" means that the
12454 connection never established.
12455
12456 - Tr: server response time (HTTP mode only). It's the time elapsed between
12457 the moment the TCP connection was established to the server and the moment
12458 the server sent its complete response headers. It purely shows its request
12459 processing time, without the network overhead due to the data transmission.
12460 It is worth noting that when the client has data to send to the server, for
12461 instance during a POST request, the time already runs, and this can distort
12462 apparent response time. For this reason, it's generally wise not to trust
12463 too much this field for POST requests initiated from clients behind an
12464 untrusted network. A value of "-1" here means that the last the response
12465 header (empty line) was never seen, most likely because the server timeout
12466 stroke before the server managed to process the request.
12467
12468 - Tt: total session duration time, between the moment the proxy accepted it
12469 and the moment both ends were closed. The exception is when the "logasap"
12470 option is specified. In this case, it only equals (Tq+Tw+Tc+Tr), and is
12471 prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012472 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012473
12474 Td = Tt - (Tq + Tw + Tc + Tr)
12475
12476 Timers with "-1" values have to be excluded from this equation. In TCP
12477 mode, "Tq" and "Tr" have to be excluded too. Note that "Tt" can never be
12478 negative.
12479
12480These timers provide precious indications on trouble causes. Since the TCP
12481protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
12482that timers close to multiples of 3s are nearly always related to lost packets
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012483due to network problems (wires, negotiation, congestion). Moreover, if "Tt" is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012484close to a timeout value specified in the configuration, it often means that a
12485session has been aborted on timeout.
12486
12487Most common cases :
12488
12489 - If "Tq" is close to 3000, a packet has probably been lost between the
12490 client and the proxy. This is very rare on local networks but might happen
12491 when clients are on far remote networks and send large requests. It may
12492 happen that values larger than usual appear here without any network cause.
12493 Sometimes, during an attack or just after a resource starvation has ended,
12494 haproxy may accept thousands of connections in a few milliseconds. The time
12495 spent accepting these connections will inevitably slightly delay processing
12496 of other connections, and it can happen that request times in the order of
12497 a few tens of milliseconds are measured after a few thousands of new
Patrick Mezard105faca2010-06-12 17:02:46 +020012498 connections have been accepted at once. Setting "option http-server-close"
12499 may display larger request times since "Tq" also measures the time spent
12500 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012501
12502 - If "Tc" is close to 3000, a packet has probably been lost between the
12503 server and the proxy during the server connection phase. This value should
12504 always be very low, such as 1 ms on local networks and less than a few tens
12505 of ms on remote networks.
12506
Willy Tarreau55165fe2009-05-10 12:02:55 +020012507 - If "Tr" is nearly always lower than 3000 except some rare values which seem
12508 to be the average majored by 3000, there are probably some packets lost
12509 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012510
12511 - If "Tt" is large even for small byte counts, it generally is because
12512 neither the client nor the server decides to close the connection, for
12513 instance because both have agreed on a keep-alive connection mode. In order
12514 to solve this issue, it will be needed to specify "option httpclose" on
12515 either the frontend or the backend. If the problem persists, it means that
12516 the server ignores the "close" connection mode and expects the client to
12517 close. Then it will be required to use "option forceclose". Having the
12518 smallest possible 'Tt' is important when connection regulation is used with
12519 the "maxconn" option on the servers, since no new connection will be sent
12520 to the server until another one is released.
12521
12522Other noticeable HTTP log cases ('xx' means any value to be ignored) :
12523
12524 Tq/Tw/Tc/Tr/+Tt The "option logasap" is present on the frontend and the log
12525 was emitted before the data phase. All the timers are valid
12526 except "Tt" which is shorter than reality.
12527
12528 -1/xx/xx/xx/Tt The client was not able to send a complete request in time
12529 or it aborted too early. Check the session termination flags
12530 then "timeout http-request" and "timeout client" settings.
12531
12532 Tq/-1/xx/xx/Tt It was not possible to process the request, maybe because
12533 servers were out of order, because the request was invalid
12534 or forbidden by ACL rules. Check the session termination
12535 flags.
12536
12537 Tq/Tw/-1/xx/Tt The connection could not establish on the server. Either it
12538 actively refused it or it timed out after Tt-(Tq+Tw) ms.
12539 Check the session termination flags, then check the
12540 "timeout connect" setting. Note that the tarpit action might
12541 return similar-looking patterns, with "Tw" equal to the time
12542 the client connection was maintained open.
12543
12544 Tq/Tw/Tc/-1/Tt The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012545 a complete response in time, or it closed its connection
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012546 unexpectedly after Tt-(Tq+Tw+Tc) ms. Check the session
12547 termination flags, then check the "timeout server" setting.
12548
12549
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200125508.5. Session state at disconnection
12551-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012552
12553TCP and HTTP logs provide a session termination indicator in the
12554"termination_state" field, just before the number of active connections. It is
125552-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
12556each of which has a special meaning :
12557
12558 - On the first character, a code reporting the first event which caused the
12559 session to terminate :
12560
12561 C : the TCP session was unexpectedly aborted by the client.
12562
12563 S : the TCP session was unexpectedly aborted by the server, or the
12564 server explicitly refused it.
12565
12566 P : the session was prematurely aborted by the proxy, because of a
12567 connection limit enforcement, because a DENY filter was matched,
12568 because of a security check which detected and blocked a dangerous
12569 error in server response which might have caused information leak
Willy Tarreau570f2212013-06-10 16:42:09 +020012570 (eg: cacheable cookie).
12571
12572 L : the session was locally processed by haproxy and was not passed to
12573 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012574
12575 R : a resource on the proxy has been exhausted (memory, sockets, source
12576 ports, ...). Usually, this appears during the connection phase, and
12577 system logs should contain a copy of the precise error. If this
12578 happens, it must be considered as a very serious anomaly which
12579 should be fixed as soon as possible by any means.
12580
12581 I : an internal error was identified by the proxy during a self-check.
12582 This should NEVER happen, and you are encouraged to report any log
12583 containing this, because this would almost certainly be a bug. It
12584 would be wise to preventively restart the process after such an
12585 event too, in case it would be caused by memory corruption.
12586
Simon Horman752dc4a2011-06-21 14:34:59 +090012587 D : the session was killed by haproxy because the server was detected
12588 as down and was configured to kill all connections when going down.
12589
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012590 U : the session was killed by haproxy on this backup server because an
12591 active server was detected as up and was configured to kill all
12592 backup connections when going up.
12593
Willy Tarreaua2a64e92011-09-07 23:01:56 +020012594 K : the session was actively killed by an admin operating on haproxy.
12595
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012596 c : the client-side timeout expired while waiting for the client to
12597 send or receive data.
12598
12599 s : the server-side timeout expired while waiting for the server to
12600 send or receive data.
12601
12602 - : normal session completion, both the client and the server closed
12603 with nothing left in the buffers.
12604
12605 - on the second character, the TCP or HTTP session state when it was closed :
12606
Willy Tarreauf7b30a92010-12-06 22:59:17 +010012607 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012608 (HTTP mode only). Nothing was sent to any server.
12609
12610 Q : the proxy was waiting in the QUEUE for a connection slot. This can
12611 only happen when servers have a 'maxconn' parameter set. It can
12612 also happen in the global queue after a redispatch consecutive to
12613 a failed attempt to connect to a dying server. If no redispatch is
12614 reported, then no connection attempt was made to any server.
12615
12616 C : the proxy was waiting for the CONNECTION to establish on the
12617 server. The server might at most have noticed a connection attempt.
12618
12619 H : the proxy was waiting for complete, valid response HEADERS from the
12620 server (HTTP only).
12621
12622 D : the session was in the DATA phase.
12623
12624 L : the proxy was still transmitting LAST data to the client while the
12625 server had already finished. This one is very rare as it can only
12626 happen when the client dies while receiving the last packets.
12627
12628 T : the request was tarpitted. It has been held open with the client
12629 during the whole "timeout tarpit" duration or until the client
12630 closed, both of which will be reported in the "Tw" timer.
12631
12632 - : normal session completion after end of data transfer.
12633
12634 - the third character tells whether the persistence cookie was provided by
12635 the client (only in HTTP mode) :
12636
12637 N : the client provided NO cookie. This is usually the case for new
12638 visitors, so counting the number of occurrences of this flag in the
12639 logs generally indicate a valid trend for the site frequentation.
12640
12641 I : the client provided an INVALID cookie matching no known server.
12642 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020012643 cookies between HTTP/HTTPS sites, persistence conditionally
12644 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012645
12646 D : the client provided a cookie designating a server which was DOWN,
12647 so either "option persist" was used and the client was sent to
12648 this server, or it was not set and the client was redispatched to
12649 another server.
12650
Willy Tarreau996a92c2010-10-13 19:30:47 +020012651 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012652 server.
12653
Willy Tarreau996a92c2010-10-13 19:30:47 +020012654 E : the client provided a valid cookie, but with a last date which was
12655 older than what is allowed by the "maxidle" cookie parameter, so
12656 the cookie is consider EXPIRED and is ignored. The request will be
12657 redispatched just as if there was no cookie.
12658
12659 O : the client provided a valid cookie, but with a first date which was
12660 older than what is allowed by the "maxlife" cookie parameter, so
12661 the cookie is consider too OLD and is ignored. The request will be
12662 redispatched just as if there was no cookie.
12663
Willy Tarreauc89ccb62012-04-05 21:18:22 +020012664 U : a cookie was present but was not used to select the server because
12665 some other server selection mechanism was used instead (typically a
12666 "use-server" rule).
12667
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012668 - : does not apply (no cookie set in configuration).
12669
12670 - the last character reports what operations were performed on the persistence
12671 cookie returned by the server (only in HTTP mode) :
12672
12673 N : NO cookie was provided by the server, and none was inserted either.
12674
12675 I : no cookie was provided by the server, and the proxy INSERTED one.
12676 Note that in "cookie insert" mode, if the server provides a cookie,
12677 it will still be overwritten and reported as "I" here.
12678
Willy Tarreau996a92c2010-10-13 19:30:47 +020012679 U : the proxy UPDATED the last date in the cookie that was presented by
12680 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012681 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020012682 date indicated in the cookie. If any other change happens, such as
12683 a redispatch, then the cookie will be marked as inserted instead.
12684
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012685 P : a cookie was PROVIDED by the server and transmitted as-is.
12686
12687 R : the cookie provided by the server was REWRITTEN by the proxy, which
12688 happens in "cookie rewrite" or "cookie prefix" modes.
12689
12690 D : the cookie provided by the server was DELETED by the proxy.
12691
12692 - : does not apply (no cookie set in configuration).
12693
Willy Tarreau996a92c2010-10-13 19:30:47 +020012694The combination of the two first flags gives a lot of information about what
12695was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012696helpful to detect server saturation, network troubles, local system resource
12697starvation, attacks, etc...
12698
12699The most common termination flags combinations are indicated below. They are
12700alphabetically sorted, with the lowercase set just after the upper case for
12701easier finding and understanding.
12702
12703 Flags Reason
12704
12705 -- Normal termination.
12706
12707 CC The client aborted before the connection could be established to the
12708 server. This can happen when haproxy tries to connect to a recently
12709 dead (or unchecked) server, and the client aborts while haproxy is
12710 waiting for the server to respond or for "timeout connect" to expire.
12711
12712 CD The client unexpectedly aborted during data transfer. This can be
12713 caused by a browser crash, by an intermediate equipment between the
12714 client and haproxy which decided to actively break the connection,
12715 by network routing issues between the client and haproxy, or by a
12716 keep-alive session between the server and the client terminated first
12717 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012718
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012719 cD The client did not send nor acknowledge any data for as long as the
12720 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020012721 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012722
12723 CH The client aborted while waiting for the server to start responding.
12724 It might be the server taking too long to respond or the client
12725 clicking the 'Stop' button too fast.
12726
12727 cH The "timeout client" stroke while waiting for client data during a
12728 POST request. This is sometimes caused by too large TCP MSS values
12729 for PPPoE networks which cannot transport full-sized packets. It can
12730 also happen when client timeout is smaller than server timeout and
12731 the server takes too long to respond.
12732
12733 CQ The client aborted while its session was queued, waiting for a server
12734 with enough empty slots to accept it. It might be that either all the
12735 servers were saturated or that the assigned server was taking too
12736 long a time to respond.
12737
12738 CR The client aborted before sending a full HTTP request. Most likely
12739 the request was typed by hand using a telnet client, and aborted
12740 too early. The HTTP status code is likely a 400 here. Sometimes this
12741 might also be caused by an IDS killing the connection between haproxy
12742 and the client.
12743
12744 cR The "timeout http-request" stroke before the client sent a full HTTP
12745 request. This is sometimes caused by too large TCP MSS values on the
12746 client side for PPPoE networks which cannot transport full-sized
12747 packets, or by clients sending requests by hand and not typing fast
12748 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020012749 request. The HTTP status code is likely a 408 here. Note: recently,
12750 some browsers such as Google Chrome started to break the deployed Web
12751 infrastructure by aggressively implementing a new "pre-connect"
12752 feature, consisting in sending connections to sites recently visited
12753 without sending any request on them until the user starts to browse
12754 the site. This mechanism causes massive disruption among resource-
12755 limited servers, and causes a lot of 408 errors in HAProxy logs.
12756 Worse, some people report that sometimes the browser displays the 408
12757 error when the user expects to see the actual content (Mozilla fixed
12758 this bug in 2004, while Chrome users continue to report it in 2014),
12759 so in this case, using "errorfile 408 /dev/null" can be used as a
12760 workaround. More information on the subject is available here :
12761 https://bugzilla.mozilla.org/show_bug.cgi?id=248827
12762 https://code.google.com/p/chromium/issues/detail?id=85229
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012763
12764 CT The client aborted while its session was tarpitted. It is important to
12765 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020012766 wrong tarpit rules have been written. If a lot of them happen, it
12767 might make sense to lower the "timeout tarpit" value to something
12768 closer to the average reported "Tw" timer, in order not to consume
12769 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012770
Willy Tarreau570f2212013-06-10 16:42:09 +020012771 LR The request was intercepted and locally handled by haproxy. Generally
12772 it means that this was a redirect or a stats request.
12773
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012774 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012775 the TCP connection (the proxy received a TCP RST or an ICMP message
12776 in return). Under some circumstances, it can also be the network
12777 stack telling the proxy that the server is unreachable (eg: no route,
12778 or no ARP response on local network). When this happens in HTTP mode,
12779 the status code is likely a 502 or 503 here.
12780
12781 sC The "timeout connect" stroke before a connection to the server could
12782 complete. When this happens in HTTP mode, the status code is likely a
12783 503 or 504 here.
12784
12785 SD The connection to the server died with an error during the data
12786 transfer. This usually means that haproxy has received an RST from
12787 the server or an ICMP message from an intermediate equipment while
12788 exchanging data with the server. This can be caused by a server crash
12789 or by a network issue on an intermediate equipment.
12790
12791 sD The server did not send nor acknowledge any data for as long as the
12792 "timeout server" setting during the data phase. This is often caused
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012793 by too short timeouts on L4 equipments before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012794 load-balancers, ...), as well as keep-alive sessions maintained
12795 between the client and the server expiring first on haproxy.
12796
12797 SH The server aborted before sending its full HTTP response headers, or
12798 it crashed while processing the request. Since a server aborting at
12799 this moment is very rare, it would be wise to inspect its logs to
12800 control whether it crashed and why. The logged request may indicate a
12801 small set of faulty requests, demonstrating bugs in the application.
12802 Sometimes this might also be caused by an IDS killing the connection
12803 between haproxy and the server.
12804
12805 sH The "timeout server" stroke before the server could return its
12806 response headers. This is the most common anomaly, indicating too
12807 long transactions, probably caused by server or database saturation.
12808 The immediate workaround consists in increasing the "timeout server"
12809 setting, but it is important to keep in mind that the user experience
12810 will suffer from these long response times. The only long term
12811 solution is to fix the application.
12812
12813 sQ The session spent too much time in queue and has been expired. See
12814 the "timeout queue" and "timeout connect" settings to find out how to
12815 fix this if it happens too often. If it often happens massively in
12816 short periods, it may indicate general problems on the affected
12817 servers due to I/O or database congestion, or saturation caused by
12818 external attacks.
12819
12820 PC The proxy refused to establish a connection to the server because the
12821 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020012822 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012823 so that it does not happen anymore. This status is very rare and
12824 might happen when the global "ulimit-n" parameter is forced by hand.
12825
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012826 PD The proxy blocked an incorrectly formatted chunked encoded message in
12827 a request or a response, after the server has emitted its headers. In
12828 most cases, this will indicate an invalid message from the server to
Willy Tarreauf3a3e132013-08-31 08:16:26 +020012829 the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647
12830 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012831
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012832 PH The proxy blocked the server's response, because it was invalid,
12833 incomplete, dangerous (cache control), or matched a security filter.
12834 In any case, an HTTP 502 error is sent to the client. One possible
12835 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010012836 containing unauthorized characters. It is also possible but quite
12837 rare, that the proxy blocked a chunked-encoding request from the
12838 client due to an invalid syntax, before the server responded. In this
12839 case, an HTTP 400 error is sent to the client and reported in the
12840 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012841
12842 PR The proxy blocked the client's HTTP request, either because of an
12843 invalid HTTP syntax, in which case it returned an HTTP 400 error to
12844 the client, or because a deny filter matched, in which case it
12845 returned an HTTP 403 error.
12846
12847 PT The proxy blocked the client's request and has tarpitted its
12848 connection before returning it a 500 server error. Nothing was sent
12849 to the server. The connection was maintained open for as long as
12850 reported by the "Tw" timer field.
12851
12852 RC A local resource has been exhausted (memory, sockets, source ports)
12853 preventing the connection to the server from establishing. The error
12854 logs will tell precisely what was missing. This is very rare and can
12855 only be solved by proper system tuning.
12856
Willy Tarreau996a92c2010-10-13 19:30:47 +020012857The combination of the two last flags gives a lot of information about how
12858persistence was handled by the client, the server and by haproxy. This is very
12859important to troubleshoot disconnections, when users complain they have to
12860re-authenticate. The commonly encountered flags are :
12861
12862 -- Persistence cookie is not enabled.
12863
12864 NN No cookie was provided by the client, none was inserted in the
12865 response. For instance, this can be in insert mode with "postonly"
12866 set on a GET request.
12867
12868 II A cookie designating an invalid server was provided by the client,
12869 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012870 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020012871 value can be presented by a client when no other server knows it.
12872
12873 NI No cookie was provided by the client, one was inserted in the
12874 response. This typically happens for first requests from every user
12875 in "insert" mode, which makes it an easy way to count real users.
12876
12877 VN A cookie was provided by the client, none was inserted in the
12878 response. This happens for most responses for which the client has
12879 already got a cookie.
12880
12881 VU A cookie was provided by the client, with a last visit date which is
12882 not completely up-to-date, so an updated cookie was provided in
12883 response. This can also happen if there was no date at all, or if
12884 there was a date but the "maxidle" parameter was not set, so that the
12885 cookie can be switched to unlimited time.
12886
12887 EI A cookie was provided by the client, with a last visit date which is
12888 too old for the "maxidle" parameter, so the cookie was ignored and a
12889 new cookie was inserted in the response.
12890
12891 OI A cookie was provided by the client, with a first visit date which is
12892 too old for the "maxlife" parameter, so the cookie was ignored and a
12893 new cookie was inserted in the response.
12894
12895 DI The server designated by the cookie was down, a new server was
12896 selected and a new cookie was emitted in the response.
12897
12898 VI The server designated by the cookie was not marked dead but could not
12899 be reached. A redispatch happened and selected another one, which was
12900 then advertised in the response.
12901
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012902
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129038.6. Non-printable characters
12904-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012905
12906In order not to cause trouble to log analysis tools or terminals during log
12907consulting, non-printable characters are not sent as-is into log files, but are
12908converted to the two-digits hexadecimal representation of their ASCII code,
12909prefixed by the character '#'. The only characters that can be logged without
12910being escaped are comprised between 32 and 126 (inclusive). Obviously, the
12911escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
12912is the same for the character '"' which becomes "#22", as well as '{', '|' and
12913'}' when logging headers.
12914
12915Note that the space character (' ') is not encoded in headers, which can cause
12916issues for tools relying on space count to locate fields. A typical header
12917containing spaces is "User-Agent".
12918
12919Last, it has been observed that some syslog daemons such as syslog-ng escape
12920the quote ('"') with a backslash ('\'). The reverse operation can safely be
12921performed since no quote may appear anywhere else in the logs.
12922
12923
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129248.7. Capturing HTTP cookies
12925---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012926
12927Cookie capture simplifies the tracking a complete user session. This can be
12928achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012929section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012930cookie will simultaneously be checked in the request ("Cookie:" header) and in
12931the response ("Set-Cookie:" header). The respective values will be reported in
12932the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012933locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012934not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
12935user switches to a new session for example, because the server will reassign it
12936a new cookie. It is also possible to detect if a server unexpectedly sets a
12937wrong cookie to a client, leading to session crossing.
12938
12939 Examples :
12940 # capture the first cookie whose name starts with "ASPSESSION"
12941 capture cookie ASPSESSION len 32
12942
12943 # capture the first cookie whose name is exactly "vgnvisitor"
12944 capture cookie vgnvisitor= len 32
12945
12946
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129478.8. Capturing HTTP headers
12948---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012949
12950Header captures are useful to track unique request identifiers set by an upper
12951proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
12952the response, one can search for information about the response length, how the
12953server asked the cache to behave, or an object location during a redirection.
12954
12955Header captures are performed using the "capture request header" and "capture
12956response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012957section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012958
12959It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012960time. Non-existent headers are logged as empty strings, and if one header
12961appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012962are grouped within braces '{' and '}' in the same order as they were declared,
12963and delimited with a vertical bar '|' without any space. Response headers
12964follow the same representation, but are displayed after a space following the
12965request headers block. These blocks are displayed just before the HTTP request
12966in the logs.
12967
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020012968As a special case, it is possible to specify an HTTP header capture in a TCP
12969frontend. The purpose is to enable logging of headers which will be parsed in
12970an HTTP backend if the request is then switched to this HTTP backend.
12971
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012972 Example :
12973 # This instance chains to the outgoing proxy
12974 listen proxy-out
12975 mode http
12976 option httplog
12977 option logasap
12978 log global
12979 server cache1 192.168.1.1:3128
12980
12981 # log the name of the virtual server
12982 capture request header Host len 20
12983
12984 # log the amount of data uploaded during a POST
12985 capture request header Content-Length len 10
12986
12987 # log the beginning of the referrer
12988 capture request header Referer len 20
12989
12990 # server name (useful for outgoing proxies only)
12991 capture response header Server len 20
12992
12993 # logging the content-length is useful with "option logasap"
12994 capture response header Content-Length len 10
12995
12996 # log the expected cache behaviour on the response
12997 capture response header Cache-Control len 8
12998
12999 # the Via header will report the next proxy's name
13000 capture response header Via len 20
13001
13002 # log the URL location during a redirection
13003 capture response header Location len 20
13004
13005 >>> Aug 9 20:26:09 localhost \
13006 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
13007 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
13008 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
13009 "GET http://fr.adserver.yahoo.com/"
13010
13011 >>> Aug 9 20:30:46 localhost \
13012 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
13013 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
13014 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013015 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013016
13017 >>> Aug 9 20:30:46 localhost \
13018 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
13019 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
13020 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
13021 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013022 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013023
13024
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130258.9. Examples of logs
13026---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013027
13028These are real-world examples of logs accompanied with an explanation. Some of
13029them have been made up by hand. The syslog part has been removed for better
13030reading. Their sole purpose is to explain how to decipher them.
13031
13032 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
13033 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
13034 "HEAD / HTTP/1.0"
13035
13036 => long request (6.5s) entered by hand through 'telnet'. The server replied
13037 in 147 ms, and the session ended normally ('----')
13038
13039 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
13040 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
13041 0/9 "HEAD / HTTP/1.0"
13042
13043 => Idem, but the request was queued in the global queue behind 9 other
13044 requests, and waited there for 1230 ms.
13045
13046 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
13047 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
13048 "GET /image.iso HTTP/1.0"
13049
13050 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013051 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013052 14 ms, 243 bytes of headers were sent to the client, and total time from
13053 accept to first data byte is 30 ms.
13054
13055 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
13056 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
13057 "GET /cgi-bin/bug.cgi? HTTP/1.0"
13058
13059 => the proxy blocked a server response either because of an "rspdeny" or
13060 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020013061 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013062 risked being cached. In this case, the response is replaced with a "502
13063 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
13064 to return the 502 and not the server.
13065
13066 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013067 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 +010013068
13069 => the client never completed its request and aborted itself ("C---") after
13070 8.5s, while the proxy was waiting for the request headers ("-R--").
13071 Nothing was sent to any server.
13072
13073 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
13074 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
13075
13076 => The client never completed its request, which was aborted by the
13077 time-out ("c---") after 50s, while the proxy was waiting for the request
13078 headers ("-R--"). Nothing was sent to any server, but the proxy could
13079 send a 408 return code to the client.
13080
13081 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
13082 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
13083
13084 => This log was produced with "option tcplog". The client timed out after
13085 5 seconds ("c----").
13086
13087 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
13088 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013089 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013090
13091 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013092 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013093 (config says 'retries 3'), and no redispatch (otherwise we would have
13094 seen "/+3"). Status code 503 was returned to the client. There were 115
13095 connections on this server, 202 connections on this proxy, and 205 on
13096 the global process. It is possible that the server refused the
13097 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013098
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013099
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131009. Statistics and monitoring
13101----------------------------
13102
13103It is possible to query HAProxy about its status. The most commonly used
13104mechanism is the HTTP statistics page. This page also exposes an alternative
13105CSV output format for monitoring tools. The same format is provided on the
13106Unix socket.
13107
13108
131099.1. CSV format
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013110---------------
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +010013111
Willy Tarreau7f062c42009-03-05 18:43:00 +010013112The statistics may be consulted either from the unix socket or from the HTTP
Willy Tarreaua3310dc2014-06-16 15:43:21 +020013113page. Both means provide a CSV format whose fields follow. The first line
13114begins with a sharp ('#') and has one word per comma-delimited field which
13115represents the title of the column. All other lines starting at the second one
13116use a classical CSV format using a comma as the delimiter, and the double quote
13117('"') as an optional text delimiter, but only if the enclosed text is ambiguous
13118(if it contains a quote or a comma). The double-quote character ('"') in the
13119text is doubled ('""'), which is the format that most tools recognize. Please
13120do not insert any column before these ones in order not to break tools which
13121use hard-coded column positions.
Willy Tarreau7f062c42009-03-05 18:43:00 +010013122
James Westby75c98ad2014-07-08 10:14:57 -040013123In brackets after each field name are the types which may have a value for
13124that field. The types are L (Listeners), F (Frontends), B (Backends), and
13125S (Servers).
13126
13127 0. pxname [LFBS]: proxy name
13128 1. svname [LFBS]: service name (FRONTEND for frontend, BACKEND for backend,
13129 any name for server/listener)
13130 2. qcur [..BS]: current queued requests. For the backend this reports the
13131 number queued without a server assigned.
13132 3. qmax [..BS]: max value of qcur
13133 4. scur [LFBS]: current sessions
13134 5. smax [LFBS]: max sessions
13135 6. slim [LFBS]: configured session limit
13136 7. stot [LFBS]: cumulative number of connections
13137 8. bin [LFBS]: bytes in
13138 9. bout [LFBS]: bytes out
13139 10. dreq [LFB.]: requests denied because of security concerns.
13140 - For tcp this is because of a matched tcp-request content rule.
13141 - For http this is because of a matched http-request or tarpit rule.
13142 11. dresp [LFBS]: responses denied because of security concerns.
13143 - For http this is because of a matched http-request rule, or
13144 "option checkcache".
13145 12. ereq [LF..]: request errors. Some of the possible causes are:
13146 - early termination from the client, before the request has been sent.
13147 - read error from the client
13148 - client timeout
13149 - client closed connection
13150 - various bad requests from the client.
13151 - request was tarpitted.
13152 13. econ [..BS]: number of requests that encountered an error trying to
13153 connect to a backend server. The backend stat is the sum of the stat
13154 for all servers of that backend, plus any connection errors not
13155 associated with a particular server (such as the backend having no
13156 active servers).
13157 14. eresp [..BS]: response errors. srv_abrt will be counted here also.
13158 Some other errors are:
13159 - write error on the client socket (won't be counted for the server stat)
13160 - failure applying filters to the response.
13161 15. wretr [..BS]: number of times a connection to a server was retried.
13162 16. wredis [..BS]: number of times a request was redispatched to another
13163 server. The server value counts the number of times that server was
13164 switched away from.
13165 17. status [LFBS]: status (UP/DOWN/NOLB/MAINT/MAINT(via)...)
13166 18. weight [..BS]: server weight (server), total weight (backend)
13167 19. act [..BS]: server is active (server), number of active servers (backend)
13168 20. bck [..BS]: server is backup (server), number of backup servers (backend)
13169 21. chkfail [...S]: number of failed checks. (Only counts checks failed when
13170 the server is up.)
13171 22. chkdown [..BS]: number of UP->DOWN transitions. The backend counter counts
13172 transitions to the whole backend being down, rather than the sum of the
13173 counters for each server.
13174 23. lastchg [..BS]: number of seconds since the last UP<->DOWN transition
13175 24. downtime [..BS]: total downtime (in seconds). The value for the backend
13176 is the downtime for the whole backend, not the sum of the server downtime.
13177 25. qlimit [...S]: configured maxqueue for the server, or nothing in the
13178 value is 0 (default, meaning no limit)
13179 26. pid [LFBS]: process id (0 for first instance, 1 for second, ...)
13180 27. iid [LFBS]: unique proxy id
13181 28. sid [L..S]: server id (unique inside a proxy)
13182 29. throttle [...S]: current throttle percentage for the server, when
13183 slowstart is active, or no value if not in slowstart.
13184 30. lbtot [..BS]: total number of times a server was selected, either for new
13185 sessions, or when re-dispatching. The server counter is the number
13186 of times that server was selected.
13187 31. tracked [...S]: id of proxy/server if tracking is enabled.
13188 32. type [LFBS]: (0=frontend, 1=backend, 2=server, 3=socket/listener)
13189 33. rate [.FBS]: number of sessions per second over last elapsed second
13190 34. rate_lim [.F..]: configured limit on new sessions per second
13191 35. rate_max [.FBS]: max number of new sessions per second
13192 36. check_status [...S]: status of last health check, one of:
Cyril Bontéf0c60612010-02-06 14:44:47 +010013193 UNK -> unknown
13194 INI -> initializing
13195 SOCKERR -> socket error
13196 L4OK -> check passed on layer 4, no upper layers testing enabled
13197 L4TMOUT -> layer 1-4 timeout
13198 L4CON -> layer 1-4 connection problem, for example
13199 "Connection refused" (tcp rst) or "No route to host" (icmp)
13200 L6OK -> check passed on layer 6
13201 L6TOUT -> layer 6 (SSL) timeout
13202 L6RSP -> layer 6 invalid response - protocol error
13203 L7OK -> check passed on layer 7
13204 L7OKC -> check conditionally passed on layer 7, for example 404 with
13205 disable-on-404
13206 L7TOUT -> layer 7 (HTTP/SMTP) timeout
13207 L7RSP -> layer 7 invalid response - protocol error
13208 L7STS -> layer 7 response error, for example HTTP 5xx
James Westby75c98ad2014-07-08 10:14:57 -040013209 37. check_code [...S]: layer5-7 code, if available
13210 38. check_duration [...S]: time in ms took to finish last health check
13211 39. hrsp_1xx [.FBS]: http responses with 1xx code
13212 40. hrsp_2xx [.FBS]: http responses with 2xx code
13213 41. hrsp_3xx [.FBS]: http responses with 3xx code
13214 42. hrsp_4xx [.FBS]: http responses with 4xx code
13215 43. hrsp_5xx [.FBS]: http responses with 5xx code
13216 44. hrsp_other [.FBS]: http responses with other codes (protocol error)
13217 45. hanafail [...S]: failed health checks details
13218 46. req_rate [.F..]: HTTP requests per second over last elapsed second
13219 47. req_rate_max [.F..]: max number of HTTP requests per second observed
13220 48. req_tot [.F..]: total number of HTTP requests received
13221 49. cli_abrt [..BS]: number of data transfers aborted by the client
13222 50. srv_abrt [..BS]: number of data transfers aborted by the server
13223 (inc. in eresp)
13224 51. comp_in [.FB.]: number of HTTP response bytes fed to the compressor
13225 52. comp_out [.FB.]: number of HTTP response bytes emitted by the compressor
13226 53. comp_byp [.FB.]: number of bytes that bypassed the HTTP compressor
13227 (CPU/BW limit)
13228 54. comp_rsp [.FB.]: number of HTTP responses that were compressed
13229 55. lastsess [..BS]: number of seconds since last session assigned to
13230 server/backend
13231 56. last_chk [...S]: last health check contents or textual error
13232 57. last_agt [...S]: last agent check contents or textual error
13233 58. qtime [..BS]: the average queue time in ms over the 1024 last requests
13234 59. ctime [..BS]: the average connect time in ms over the 1024 last requests
13235 60. rtime [..BS]: the average response time in ms over the 1024 last requests
13236 (0 for TCP)
13237 61. ttime [..BS]: the average total session time in ms over the 1024 last
13238 requests
Willy Tarreau844e3c52008-01-11 16:28:18 +010013239
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013240
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200132419.2. Unix Socket commands
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013242-------------------------
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013243
Willy Tarreau468f4932013-08-01 16:50:16 +020013244The stats socket is not enabled by default. In order to enable it, it is
13245necessary to add one line in the global section of the haproxy configuration.
13246A second line is recommended to set a larger timeout, always appreciated when
13247issuing commands by hand :
13248
13249 global
13250 stats socket /var/run/haproxy.sock mode 600 level admin
13251 stats timeout 2m
13252
13253It is also possible to add multiple instances of the stats socket by repeating
13254the line, and make them listen to a TCP port instead of a UNIX socket. This is
13255never done by default because this is dangerous, but can be handy in some
13256situations :
13257
13258 global
13259 stats socket /var/run/haproxy.sock mode 600 level admin
13260 stats socket ipv4@192.168.0.1:9999 level admin
13261 stats timeout 2m
13262
13263To access the socket, an external utility such as "socat" is required. Socat is a
13264swiss-army knife to connect anything to anything. We use it to connect terminals
13265to the socket, or a couple of stdin/stdout pipes to it for scripts. The two main
13266syntaxes we'll use are the following :
13267
13268 # socat /var/run/haproxy.sock stdio
13269 # socat /var/run/haproxy.sock readline
13270
13271The first one is used with scripts. It is possible to send the output of a
13272script to haproxy, and pass haproxy's output to another script. That's useful
13273for retrieving counters or attack traces for example.
13274
13275The second one is only useful for issuing commands by hand. It has the benefit
13276that the terminal is handled by the readline library which supports line
13277editing and history, which is very convenient when issuing repeated commands
13278(eg: watch a counter).
13279
13280The socket supports two operation modes :
13281 - interactive
13282 - non-interactive
13283
13284The non-interactive mode is the default when socat connects to the socket. In
13285this mode, a single line may be sent. It is processed as a whole, responses are
13286sent back, and the connection closes after the end of the response. This is the
13287mode that scripts and monitoring tools use. It is possible to send multiple
13288commands in this mode, they need to be delimited by a semi-colon (';'). For
13289example :
13290
13291 # echo "show info;show stat;show table" | socat /var/run/haproxy stdio
13292
13293The interactive mode displays a prompt ('>') and waits for commands to be
13294entered on the line, then processes them, and displays the prompt again to wait
13295for a new command. This mode is entered via the "prompt" command which must be
13296sent on the first line in non-interactive mode. The mode is a flip switch, if
13297"prompt" is sent in interactive mode, it is disabled and the connection closes
13298after processing the last command of the same line.
13299
13300For this reason, when debugging by hand, it's quite common to start with the
13301"prompt" command :
13302
13303 # socat /var/run/haproxy readline
13304 prompt
13305 > show info
13306 ...
13307 >
13308
13309Since multiple commands may be issued at once, haproxy uses the empty line as a
13310delimiter to mark an end of output for each command, and takes care of ensuring
13311that no command can emit an empty line on output. A script can thus easily
13312parse the output even when multiple commands were pipelined on a single line.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013313
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013314It is important to understand that when multiple haproxy processes are started
13315on the same sockets, any process may pick up the request and will output its
13316own stats.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013317
Willy Tarreau468f4932013-08-01 16:50:16 +020013318The list of commands currently supported on the stats socket is provided below.
13319If an unknown command is sent, haproxy displays the usage message which reminds
13320all supported commands. Some commands support a more complex syntax, generally
13321it will explain what part of the command is invalid when this happens.
13322
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013323add acl <acl> <pattern>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013324 Add an entry into the acl <acl>. <acl> is the #<id> or the <file> returned by
13325 "show acl". This command does not verify if the entry already exists. This
13326 command cannot be used if the reference <acl> is a file also used with a map.
13327 In this case, you must use the command "add map" in place of "add acl".
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013328
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013329add map <map> <key> <value>
13330 Add an entry into the map <map> to associate the value <value> to the key
13331 <key>. This command does not verify if the entry already exists. It is
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013332 mainly used to fill a map after a clear operation. Note that if the reference
13333 <map> is a file and is shared with a map, this map will contain also a new
13334 pattern entry.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013335
Willy Tarreaud63335a2010-02-26 12:56:52 +010013336clear counters
13337 Clear the max values of the statistics counters in each proxy (frontend &
13338 backend) and in each server. The cumulated counters are not affected. This
13339 can be used to get clean counters after an incident, without having to
13340 restart nor to clear traffic counters. This command is restricted and can
13341 only be issued on sockets configured for levels "operator" or "admin".
13342
13343clear counters all
13344 Clear all statistics counters in each proxy (frontend & backend) and in each
13345 server. This has the same effect as restarting. This command is restricted
13346 and can only be issued on sockets configured for level "admin".
13347
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013348clear acl <acl>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013349 Remove all entries from the acl <acl>. <acl> is the #<id> or the <file>
13350 returned by "show acl". Note that if the reference <acl> is a file and is
13351 shared with a map, this map will be also cleared.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013352
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013353clear map <map>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013354 Remove all entries from the map <map>. <map> is the #<id> or the <file>
13355 returned by "show map". Note that if the reference <map> is a file and is
13356 shared with a acl, this acl will be also cleared.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013357
Simon Hormanc88b8872011-06-15 15:18:49 +090013358clear table <table> [ data.<type> <operator> <value> ] | [ key <key> ]
13359 Remove entries from the stick-table <table>.
13360
13361 This is typically used to unblock some users complaining they have been
13362 abusively denied access to a service, but this can also be used to clear some
13363 stickiness entries matching a server that is going to be replaced (see "show
13364 table" below for details). Note that sometimes, removal of an entry will be
13365 refused because it is currently tracked by a session. Retrying a few seconds
13366 later after the session ends is usual enough.
13367
13368 In the case where no options arguments are given all entries will be removed.
13369
13370 When the "data." form is used entries matching a filter applied using the
13371 stored data (see "stick-table" in section 4.2) are removed. A stored data
13372 type must be specified in <type>, and this data type must be stored in the
13373 table otherwise an error is reported. The data is compared according to
13374 <operator> with the 64-bit integer <value>. Operators are the same as with
13375 the ACLs :
13376
13377 - eq : match entries whose data is equal to this value
13378 - ne : match entries whose data is not equal to this value
13379 - le : match entries whose data is less than or equal to this value
13380 - ge : match entries whose data is greater than or equal to this value
13381 - lt : match entries whose data is less than this value
13382 - gt : match entries whose data is greater than this value
13383
13384 When the key form is used the entry <key> is removed. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013385 same type as the table, which currently is limited to IPv4, IPv6, integer and
13386 string.
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013387
13388 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013389 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013390 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013391 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13392 bytes_out_rate(60000)=187
13393 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13394 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013395
13396 $ echo "clear table http_proxy key 127.0.0.1" | socat stdio /tmp/sock1
13397
13398 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013399 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau62a36c42010-08-17 15:53:10 +020013400 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13401 bytes_out_rate(60000)=191
Simon Hormanc88b8872011-06-15 15:18:49 +090013402 $ echo "clear table http_proxy data.gpc0 eq 1" | socat stdio /tmp/sock1
13403 $ echo "show table http_proxy" | socat stdio /tmp/sock1
13404 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013405
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013406del acl <acl> [<key>|#<ref>]
13407 Delete all the acl entries from the acl <acl> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013408 <acl> is the #<id> or the <file> returned by "show acl". If the <ref> is used,
13409 this command delete only the listed reference. The reference can be found with
13410 listing the content of the acl. Note that if the reference <acl> is a file and
13411 is shared with a map, the entry will be also deleted in the map.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013412
13413del map <map> [<key>|#<ref>]
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013414 Delete all the map entries from the map <map> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013415 <map> is the #<id> or the <file> returned by "show map". If the <ref> is used,
13416 this command delete only the listed reference. The reference can be found with
13417 listing the content of the map. Note that if the reference <map> is a file and
13418 is shared with a acl, the entry will be also deleted in the map.
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013419
13420disable agent <backend>/<server>
Simon Horman671b6f02013-11-25 10:46:39 +090013421 Mark the auxiliary agent check as temporarily stopped.
13422
13423 In the case where an agent check is being run as a auxiliary check, due
13424 to the agent-check parameter of a server directive, new checks are only
13425 initialised when the agent is in the enabled. Thus, disable agent will
13426 prevent any new agent checks from begin initiated until the agent
13427 re-enabled using enable agent.
13428
13429 When an agent is disabled the processing of an auxiliary agent check that
13430 was initiated while the agent was set as enabled is as follows: All
13431 results that would alter the weight, specifically "drain" or a weight
13432 returned by the agent, are ignored. The processing of agent check is
13433 otherwise unchanged.
13434
13435 The motivation for this feature is to allow the weight changing effects
13436 of the agent checks to be paused to allow the weight of a server to be
13437 configured using set weight without being overridden by the agent.
13438
13439 This command is restricted and can only be issued on sockets configured for
13440 level "admin".
13441
Willy Tarreau532a4502011-09-07 22:37:44 +020013442disable frontend <frontend>
13443 Mark the frontend as temporarily stopped. This corresponds to the mode which
13444 is used during a soft restart : the frontend releases the port but can be
13445 enabled again if needed. This should be used with care as some non-Linux OSes
13446 are unable to enable it back. This is intended to be used in environments
13447 where stopping a proxy is not even imaginable but a misconfigured proxy must
13448 be fixed. That way it's possible to release the port and bind it into another
13449 process to restore operations. The frontend will appear with status "STOP"
13450 on the stats page.
13451
13452 The frontend may be specified either by its name or by its numeric ID,
13453 prefixed with a sharp ('#').
13454
13455 This command is restricted and can only be issued on sockets configured for
13456 level "admin".
13457
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013458disable health <backend>/<server>
13459 Mark the primary health check as temporarily stopped. This will disable
13460 sending of health checks, and the last health check result will be ignored.
13461 The server will be in unchecked state and considered UP unless an auxiliary
13462 agent check forces it down.
13463
13464 This command is restricted and can only be issued on sockets configured for
13465 level "admin".
13466
Willy Tarreaud63335a2010-02-26 12:56:52 +010013467disable server <backend>/<server>
13468 Mark the server DOWN for maintenance. In this mode, no more checks will be
13469 performed on the server until it leaves maintenance.
13470 If the server is tracked by other servers, those servers will be set to DOWN
13471 during the maintenance.
13472
13473 In the statistics page, a server DOWN for maintenance will appear with a
13474 "MAINT" status, its tracking servers with the "MAINT(via)" one.
13475
13476 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013477 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013478
13479 This command is restricted and can only be issued on sockets configured for
13480 level "admin".
13481
Simon Horman671b6f02013-11-25 10:46:39 +090013482enable agent <backend>/<server>
13483 Resume auxiliary agent check that was temporarily stopped.
13484
13485 See "disable agent" for details of the effect of temporarily starting
13486 and stopping an auxiliary agent.
13487
13488 This command is restricted and can only be issued on sockets configured for
13489 level "admin".
13490
Willy Tarreau532a4502011-09-07 22:37:44 +020013491enable frontend <frontend>
13492 Resume a frontend which was temporarily stopped. It is possible that some of
13493 the listening ports won't be able to bind anymore (eg: if another process
13494 took them since the 'disable frontend' operation). If this happens, an error
13495 is displayed. Some operating systems might not be able to resume a frontend
13496 which was disabled.
13497
13498 The frontend may be specified either by its name or by its numeric ID,
13499 prefixed with a sharp ('#').
13500
13501 This command is restricted and can only be issued on sockets configured for
13502 level "admin".
13503
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013504enable health <backend>/<server>
13505 Resume a primary health check that was temporarily stopped. This will enable
13506 sending of health checks again. Please see "disable health" for details.
13507
13508 This command is restricted and can only be issued on sockets configured for
13509 level "admin".
13510
Willy Tarreaud63335a2010-02-26 12:56:52 +010013511enable server <backend>/<server>
13512 If the server was previously marked as DOWN for maintenance, this marks the
13513 server UP and checks are re-enabled.
13514
13515 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013516 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013517
13518 This command is restricted and can only be issued on sockets configured for
13519 level "admin".
13520
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013521get map <map> <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013522get acl <acl> <value>
13523 Lookup the value <value> in the map <map> or in the ACL <acl>. <map> or <acl>
13524 are the #<id> or the <file> returned by "show map" or "show acl". This command
13525 returns all the matching patterns associated with this map. This is useful for
13526 debugging maps and ACLs. The output format is composed by one line par
13527 matching type. Each line is composed by space-delimited series of words.
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013528
13529 The first two words are:
13530
13531 <match method>: The match method applied. It can be "found", "bool",
13532 "int", "ip", "bin", "len", "str", "beg", "sub", "dir",
13533 "dom", "end" or "reg".
13534
13535 <match result>: The result. Can be "match" or "no-match".
13536
13537 The following words are returned only if the pattern matches an entry.
13538
13539 <index type>: "tree" or "list". The internal lookup algorithm.
13540
13541 <case>: "case-insensitive" or "case-sensitive". The
13542 interpretation of the case.
13543
13544 <entry matched>: match="<entry>". Return the matched pattern. It is
13545 useful with regular expressions.
13546
13547 The two last word are used to show the returned value and its type. With the
13548 "acl" case, the pattern doesn't exist.
13549
13550 return=nothing: No return because there are no "map".
13551 return="<value>": The value returned in the string format.
13552 return=cannot-display: The value cannot be converted as string.
13553
13554 type="<type>": The type of the returned sample.
13555
Willy Tarreaud63335a2010-02-26 12:56:52 +010013556get weight <backend>/<server>
13557 Report the current weight and the initial weight of server <server> in
13558 backend <backend> or an error if either doesn't exist. The initial weight is
13559 the one that appears in the configuration file. Both are normally equal
13560 unless the current weight has been changed. Both the backend and the server
13561 may be specified either by their name or by their numeric ID, prefixed with a
Willy Tarreauf5f31922011-08-02 11:32:07 +020013562 sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013563
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013564help
13565 Print the list of known keywords and their basic usage. The same help screen
13566 is also displayed for unknown commands.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013567
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013568prompt
13569 Toggle the prompt at the beginning of the line and enter or leave interactive
13570 mode. In interactive mode, the connection is not closed after a command
13571 completes. Instead, the prompt will appear again, indicating the user that
13572 the interpreter is waiting for a new command. The prompt consists in a right
13573 angle bracket followed by a space "> ". This mode is particularly convenient
13574 when one wants to periodically check information such as stats or errors.
13575 It is also a good idea to enter interactive mode before issuing a "help"
13576 command.
13577
13578quit
13579 Close the connection when in interactive mode.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013580
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013581set map <map> [<key>|#<ref>] <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013582 Modify the value corresponding to each key <key> in a map <map>. <map> is the
13583 #<id> or <file> returned by "show map". If the <ref> is used in place of
13584 <key>, only the entry pointed by <ref> is changed. The new value is <value>.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013585
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013586set maxconn frontend <frontend> <value>
Willy Tarreau3c7a79d2012-09-26 21:07:15 +020013587 Dynamically change the specified frontend's maxconn setting. Any positive
13588 value is allowed including zero, but setting values larger than the global
13589 maxconn does not make much sense. If the limit is increased and connections
13590 were pending, they will immediately be accepted. If it is lowered to a value
13591 below the current number of connections, new connections acceptation will be
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013592 delayed until the threshold is reached. The frontend might be specified by
13593 either its name or its numeric ID prefixed with a sharp ('#').
13594
Willy Tarreau91886b62011-09-07 14:38:31 +020013595set maxconn global <maxconn>
13596 Dynamically change the global maxconn setting within the range defined by the
13597 initial global maxconn setting. If it is increased and connections were
13598 pending, they will immediately be accepted. If it is lowered to a value below
13599 the current number of connections, new connections acceptation will be
13600 delayed until the threshold is reached. A value of zero restores the initial
13601 setting.
13602
Willy Tarreauf5b22872011-09-07 16:13:44 +020013603set rate-limit connections global <value>
13604 Change the process-wide connection rate limit, which is set by the global
13605 'maxconnrate' setting. A value of zero disables the limitation. This limit
13606 applies to all frontends and the change has an immediate effect. The value
13607 is passed in number of connections per second.
13608
William Lallemandd85f9172012-11-09 17:05:39 +010013609set rate-limit http-compression global <value>
13610 Change the maximum input compression rate, which is set by the global
13611 'maxcomprate' setting. A value of zero disables the limitation. The value is
William Lallemand096f5542012-11-19 17:26:05 +010013612 passed in number of kilobytes per second. The value is available in the "show
13613 info" on the line "CompressBpsRateLim" in bytes.
William Lallemandd85f9172012-11-09 17:05:39 +010013614
Willy Tarreau93e7c002013-10-07 18:51:07 +020013615set rate-limit sessions global <value>
13616 Change the process-wide session rate limit, which is set by the global
13617 'maxsessrate' setting. A value of zero disables the limitation. This limit
13618 applies to all frontends and the change has an immediate effect. The value
13619 is passed in number of sessions per second.
13620
Willy Tarreaue43d5322013-10-07 20:01:52 +020013621set rate-limit ssl-sessions global <value>
13622 Change the process-wide SSL session rate limit, which is set by the global
13623 'maxsslrate' setting. A value of zero disables the limitation. This limit
13624 applies to all frontends and the change has an immediate effect. The value
13625 is passed in number of sessions per second sent to the SSL stack. It applies
13626 before the handshake in order to protect the stack against handshake abuses.
13627
Willy Tarreau2a4b70f2014-05-22 18:42:35 +020013628set server <backend>/<server> agent [ up | down ]
13629 Force a server's agent to a new state. This can be useful to immediately
13630 switch a server's state regardless of some slow agent checks for example.
13631 Note that the change is propagated to tracking servers if any.
13632
13633set server <backend>/<server> health [ up | stopping | down ]
13634 Force a server's health to a new state. This can be useful to immediately
13635 switch a server's state regardless of some slow health checks for example.
13636 Note that the change is propagated to tracking servers if any.
13637
13638set server <backend>/<server> state [ ready | drain | maint ]
13639 Force a server's administrative state to a new state. This can be useful to
13640 disable load balancing and/or any traffic to a server. Setting the state to
13641 "ready" puts the server in normal mode, and the command is the equivalent of
13642 the "enable server" command. Setting the state to "maint" disables any traffic
13643 to the server as well as any health checks. This is the equivalent of the
13644 "disable server" command. Setting the mode to "drain" only removes the server
13645 from load balancing but still allows it to be checked and to accept new
13646 persistent connections. Changes are propagated to tracking servers if any.
13647
13648set server <backend>/<server> weight <weight>[%]
13649 Change a server's weight to the value passed in argument. This is the exact
13650 equivalent of the "set weight" command below.
13651
Emeric Brun4147b2e2014-06-16 18:36:30 +020013652set ssl ocsp-response <response>
13653 This command is used to update an OCSP Response for a certificate (see "crt"
13654 on "bind" lines). Same controls are performed as during the initial loading of
13655 the response. The <response> must be passed as a base64 encoded string of the
13656 DER encoded response from the OCSP server.
13657
13658 Example:
13659 openssl ocsp -issuer issuer.pem -cert server.pem \
13660 -host ocsp.issuer.com:80 -respout resp.der
13661 echo "set ssl ocsp-response $(base64 -w 10000 resp.der)" | \
13662 socat stdio /var/run/haproxy.stat
13663
Willy Tarreau47060b62013-08-01 21:11:42 +020013664set table <table> key <key> [data.<data_type> <value>]*
Willy Tarreau654694e2012-06-07 01:03:16 +020013665 Create or update a stick-table entry in the table. If the key is not present,
13666 an entry is inserted. See stick-table in section 4.2 to find all possible
13667 values for <data_type>. The most likely use consists in dynamically entering
13668 entries for source IP addresses, with a flag in gpc0 to dynamically block an
Willy Tarreau47060b62013-08-01 21:11:42 +020013669 IP address or affect its quality of service. It is possible to pass multiple
13670 data_types in a single call.
Willy Tarreau654694e2012-06-07 01:03:16 +020013671
Willy Tarreaud63335a2010-02-26 12:56:52 +010013672set timeout cli <delay>
13673 Change the CLI interface timeout for current connection. This can be useful
13674 during long debugging sessions where the user needs to constantly inspect
13675 some indicators without being disconnected. The delay is passed in seconds.
13676
13677set weight <backend>/<server> <weight>[%]
13678 Change a server's weight to the value passed in argument. If the value ends
13679 with the '%' sign, then the new weight will be relative to the initially
Simon Horman58b5d292013-02-12 10:45:52 +090013680 configured weight. Absolute weights are permitted between 0 and 256.
13681 Relative weights must be positive with the resulting absolute weight is
13682 capped at 256. Servers which are part of a farm running a static
13683 load-balancing algorithm have stricter limitations because the weight
13684 cannot change once set. Thus for these servers, the only accepted values
13685 are 0 and 100% (or 0 and the initial weight). Changes take effect
13686 immediately, though certain LB algorithms require a certain amount of
13687 requests to consider changes. A typical usage of this command is to
13688 disable a server during an update by setting its weight to zero, then to
13689 enable it again after the update by setting it back to 100%. This command
13690 is restricted and can only be issued on sockets configured for level
13691 "admin". Both the backend and the server may be specified either by their
13692 name or by their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013693
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013694show errors [<iid>]
13695 Dump last known request and response errors collected by frontends and
13696 backends. If <iid> is specified, the limit the dump to errors concerning
Willy Tarreau6162db22009-10-10 17:13:00 +020013697 either frontend or backend whose ID is <iid>. This command is restricted
13698 and can only be issued on sockets configured for levels "operator" or
13699 "admin".
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013700
13701 The errors which may be collected are the last request and response errors
13702 caused by protocol violations, often due to invalid characters in header
13703 names. The report precisely indicates what exact character violated the
13704 protocol. Other important information such as the exact date the error was
13705 detected, frontend and backend names, the server name (when known), the
13706 internal session ID and the source address which has initiated the session
13707 are reported too.
13708
13709 All characters are returned, and non-printable characters are encoded. The
13710 most common ones (\t = 9, \n = 10, \r = 13 and \e = 27) are encoded as one
13711 letter following a backslash. The backslash itself is encoded as '\\' to
13712 avoid confusion. Other non-printable characters are encoded '\xNN' where
13713 NN is the two-digits hexadecimal representation of the character's ASCII
13714 code.
13715
13716 Lines are prefixed with the position of their first character, starting at 0
13717 for the beginning of the buffer. At most one input line is printed per line,
13718 and large lines will be broken into multiple consecutive output lines so that
13719 the output never goes beyond 79 characters wide. It is easy to detect if a
13720 line was broken, because it will not end with '\n' and the next line's offset
13721 will be followed by a '+' sign, indicating it is a continuation of previous
13722 line.
13723
13724 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013725 $ echo "show errors" | socat stdio /tmp/sock1
13726 >>> [04/Mar/2009:15:46:56.081] backend http-in (#2) : invalid response
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013727 src 127.0.0.1, session #54, frontend fe-eth0 (#1), server s2 (#1)
13728 response length 213 bytes, error at position 23:
13729
13730 00000 HTTP/1.0 200 OK\r\n
13731 00017 header/bizarre:blah\r\n
13732 00038 Location: blah\r\n
13733 00054 Long-line: this is a very long line which should b
13734 00104+ e broken into multiple lines on the output buffer,
13735 00154+ otherwise it would be too large to print in a ter
13736 00204+ minal\r\n
13737 00211 \r\n
13738
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013739 In the example above, we see that the backend "http-in" which has internal
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013740 ID 2 has blocked an invalid response from its server s2 which has internal
13741 ID 1. The request was on session 54 initiated by source 127.0.0.1 and
13742 received by frontend fe-eth0 whose ID is 1. The total response length was
13743 213 bytes when the error was detected, and the error was at byte 23. This
13744 is the slash ('/') in header name "header/bizarre", which is not a valid
13745 HTTP character for a header name.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013746
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013747show info
13748 Dump info about haproxy status on current process.
13749
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013750show map [<map>]
13751 Dump info about map converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013752 maps is returned. If a <map> is specified, its contents are dumped. <map> is
13753 the #<id> or <file>. The first column is a unique identifier. It can be used
13754 as reference for the operation "del map" and "set map". The second column is
13755 the pattern and the third column is the sample if available. The data returned
13756 are not directly a list of available maps, but are the list of all patterns
13757 composing any map. Many of these patterns can be shared with ACL.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013758
13759show acl [<acl>]
13760 Dump info about acl converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013761 acls is returned. If a <acl> is specified, its contents are dumped. <acl> if
13762 the #<id> or <file>. The dump format is the same than the map even for the
13763 sample value. The data returned are not a list of available ACL, but are the
13764 list of all patterns composing any ACL. Many of these patterns can be shared
13765 with maps.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013766
Willy Tarreau12833bb2014-01-28 16:49:56 +010013767show pools
13768 Dump the status of internal memory pools. This is useful to track memory
13769 usage when suspecting a memory leak for example. It does exactly the same
13770 as the SIGQUIT when running in foreground except that it does not flush
13771 the pools.
13772
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013773show sess
13774 Dump all known sessions. Avoid doing this on slow connections as this can
Willy Tarreau6162db22009-10-10 17:13:00 +020013775 be huge. This command is restricted and can only be issued on sockets
13776 configured for levels "operator" or "admin".
13777
Willy Tarreau66dc20a2010-03-05 17:53:32 +010013778show sess <id>
13779 Display a lot of internal information about the specified session identifier.
13780 This identifier is the first field at the beginning of the lines in the dumps
13781 of "show sess" (it corresponds to the session pointer). Those information are
13782 useless to most users but may be used by haproxy developers to troubleshoot a
13783 complex bug. The output format is intentionally not documented so that it can
Oliviere99d44d2014-09-05 18:49:10 +020013784 freely evolve depending on demands. You may find a description of all fields
13785 returned in src/dumpstats.c
13786
13787 The special id "all" dumps the states of all sessions, which must be avoided
13788 as much as possible as it is highly CPU intensive and can take a lot of time.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013789
13790show stat [<iid> <type> <sid>]
13791 Dump statistics in the CSV format. By passing <id>, <type> and <sid>, it is
13792 possible to dump only selected items :
13793 - <iid> is a proxy ID, -1 to dump everything
13794 - <type> selects the type of dumpable objects : 1 for frontends, 2 for
13795 backends, 4 for servers, -1 for everything. These values can be ORed,
13796 for example:
13797 1 + 2 = 3 -> frontend + backend.
13798 1 + 2 + 4 = 7 -> frontend + backend + server.
13799 - <sid> is a server ID, -1 to dump everything from the selected proxy.
13800
13801 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013802 $ echo "show info;show stat" | socat stdio unix-connect:/tmp/sock1
13803 >>> Name: HAProxy
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013804 Version: 1.4-dev2-49
13805 Release_date: 2009/09/23
13806 Nbproc: 1
13807 Process_num: 1
13808 (...)
13809
13810 # pxname,svname,qcur,qmax,scur,smax,slim,stot,bin,bout,dreq, (...)
13811 stats,FRONTEND,,,0,0,1000,0,0,0,0,0,0,,,,,OPEN,,,,,,,,,1,1,0, (...)
13812 stats,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,0,0,0,,0,250,(...)
13813 (...)
13814 www1,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,1,1,0,,0,250, (...)
13815
13816 $
13817
13818 Here, two commands have been issued at once. That way it's easy to find
13819 which process the stats apply to in multi-process mode. Notice the empty
13820 line after the information output which marks the end of the first block.
13821 A similar empty line appears at the end of the second block (stats) so that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013822 the reader knows the output has not been truncated.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013823
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013824show table
13825 Dump general information on all known stick-tables. Their name is returned
13826 (the name of the proxy which holds them), their type (currently zero, always
13827 IP), their size in maximum possible number of entries, and the number of
13828 entries currently in use.
13829
13830 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013831 $ echo "show table" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013832 >>> # table: front_pub, type: ip, size:204800, used:171454
13833 >>> # table: back_rdp, type: ip, size:204800, used:0
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013834
Simon Horman17bce342011-06-15 15:18:47 +090013835show table <name> [ data.<type> <operator> <value> ] | [ key <key> ]
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013836 Dump contents of stick-table <name>. In this mode, a first line of generic
13837 information about the table is reported as with "show table", then all
13838 entries are dumped. Since this can be quite heavy, it is possible to specify
Simon Horman17bce342011-06-15 15:18:47 +090013839 a filter in order to specify what entries to display.
13840
13841 When the "data." form is used the filter applies to the stored data (see
13842 "stick-table" in section 4.2). A stored data type must be specified
13843 in <type>, and this data type must be stored in the table otherwise an
13844 error is reported. The data is compared according to <operator> with the
13845 64-bit integer <value>. Operators are the same as with the ACLs :
13846
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013847 - eq : match entries whose data is equal to this value
13848 - ne : match entries whose data is not equal to this value
13849 - le : match entries whose data is less than or equal to this value
13850 - ge : match entries whose data is greater than or equal to this value
13851 - lt : match entries whose data is less than this value
13852 - gt : match entries whose data is greater than this value
13853
Simon Hormanc88b8872011-06-15 15:18:49 +090013854
13855 When the key form is used the entry <key> is shown. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013856 same type as the table, which currently is limited to IPv4, IPv6, integer,
13857 and string.
Simon Horman17bce342011-06-15 15:18:47 +090013858
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013859 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013860 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013861 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013862 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13863 bytes_out_rate(60000)=187
13864 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13865 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013866
Willy Tarreau62a36c42010-08-17 15:53:10 +020013867 $ echo "show table http_proxy data.gpc0 gt 0" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013868 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013869 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13870 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013871
Willy Tarreau62a36c42010-08-17 15:53:10 +020013872 $ echo "show table http_proxy data.conn_rate gt 5" | \
13873 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013874 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013875 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13876 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013877
Simon Horman17bce342011-06-15 15:18:47 +090013878 $ echo "show table http_proxy key 127.0.0.2" | \
13879 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090013880 >>> # table: http_proxy, type: ip, size:204800, used:2
Simon Horman17bce342011-06-15 15:18:47 +090013881 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13882 bytes_out_rate(60000)=191
13883
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013884 When the data criterion applies to a dynamic value dependent on time such as
13885 a bytes rate, the value is dynamically computed during the evaluation of the
13886 entry in order to decide whether it has to be dumped or not. This means that
13887 such a filter could match for some time then not match anymore because as
13888 time goes, the average event rate drops.
13889
13890 It is possible to use this to extract lists of IP addresses abusing the
13891 service, in order to monitor them or even blacklist them in a firewall.
13892 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013893 $ echo "show table http_proxy data.gpc0 gt 0" \
13894 | socat stdio /tmp/sock1 \
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013895 | fgrep 'key=' | cut -d' ' -f2 | cut -d= -f2 > abusers-ip.txt
13896 ( or | awk '/key/{ print a[split($2,a,"=")]; }' )
Krzysztof Piotr Oledzki719e7262009-10-04 15:02:46 +020013897
Willy Tarreau532a4502011-09-07 22:37:44 +020013898shutdown frontend <frontend>
13899 Completely delete the specified frontend. All the ports it was bound to will
13900 be released. It will not be possible to enable the frontend anymore after
13901 this operation. This is intended to be used in environments where stopping a
13902 proxy is not even imaginable but a misconfigured proxy must be fixed. That
13903 way it's possible to release the port and bind it into another process to
13904 restore operations. The frontend will not appear at all on the stats page
13905 once it is terminated.
13906
13907 The frontend may be specified either by its name or by its numeric ID,
13908 prefixed with a sharp ('#').
13909
13910 This command is restricted and can only be issued on sockets configured for
13911 level "admin".
13912
Willy Tarreaua295edc2011-09-07 23:21:03 +020013913shutdown session <id>
13914 Immediately terminate the session matching the specified session identifier.
13915 This identifier is the first field at the beginning of the lines in the dumps
13916 of "show sess" (it corresponds to the session pointer). This can be used to
13917 terminate a long-running session without waiting for a timeout or when an
13918 endless transfer is ongoing. Such terminated sessions are reported with a 'K'
13919 flag in the logs.
13920
Cyril Bontéde9789b2014-07-12 18:22:42 +020013921shutdown sessions server <backend>/<server>
Willy Tarreau52b2d222011-09-07 23:48:48 +020013922 Immediately terminate all the sessions attached to the specified server. This
13923 can be used to terminate long-running sessions after a server is put into
13924 maintenance mode, for instance. Such terminated sessions are reported with a
13925 'K' flag in the logs.
13926
Willy Tarreau0ba27502007-12-24 16:55:16 +010013927/*
13928 * Local variables:
13929 * fill-column: 79
13930 * End:
13931 */