blob: 516319902a81ac2188d3c74fb104061f46224e7c [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 Tarreau15480d72014-06-19 21:10:58 +02005 version 1.6
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau9229f122014-06-19 21:01:06 +02007 2014/06/19
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
Simon Horman98637e52014-06-20 12:30:16 +0900451 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200452 - gid
453 - group
454 - log
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100455 - log-send-hostname
Willy Tarreau6a06a402007-07-15 20:15:28 +0200456 - nbproc
457 - pidfile
458 - uid
459 - ulimit-n
460 - user
Willy Tarreaufbee7132007-10-18 13:53:22 +0200461 - stats
Emeric Brun850efd52014-01-29 12:24:34 +0100462 - ssl-server-verify
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +0200463 - node
464 - description
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100465 - unix-bind
Willy Tarreaud72758d2010-01-12 10:42:19 +0100466
Willy Tarreau6a06a402007-07-15 20:15:28 +0200467 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200468 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200469 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200470 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100471 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100472 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100473 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200474 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200475 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200476 - maxsslrate
Willy Tarreau6a06a402007-07-15 20:15:28 +0200477 - noepoll
478 - nokqueue
479 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100480 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300481 - nogetaddrinfo
Willy Tarreaufe255b72007-10-14 23:09:26 +0200482 - spread-checks
Willy Tarreau27a674e2009-08-17 07:23:33 +0200483 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200484 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100485 - tune.comp.maxlevel
Willy Tarreau193b8c62012-11-22 00:17:38 +0100486 - tune.http.cookielen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200487 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100488 - tune.idletimer
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100489 - tune.maxaccept
490 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200491 - tune.maxrewrite
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200492 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100493 - tune.rcvbuf.client
494 - tune.rcvbuf.server
495 - tune.sndbuf.client
496 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100497 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100498 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200499 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100500 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200501 - tune.ssl.default-dh-param
William Lallemanda509e4c2012-11-07 16:54:34 +0100502 - tune.zlib.memlevel
503 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100504
Willy Tarreau6a06a402007-07-15 20:15:28 +0200505 * Debugging
506 - debug
507 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200508
509
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005103.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200511------------------------------------
512
Emeric Brunc8e8d122012-10-02 18:42:10 +0200513ca-base <dir>
514 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200515 relative path is used with "ca-file" or "crl-file" directives. Absolute
516 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200517
Willy Tarreau6a06a402007-07-15 20:15:28 +0200518chroot <jail dir>
519 Changes current directory to <jail dir> and performs a chroot() there before
520 dropping privileges. This increases the security level in case an unknown
521 vulnerability would be exploited, since it would make it very hard for the
522 attacker to exploit the system. This only works when the process is started
523 with superuser privileges. It is important to ensure that <jail_dir> is both
524 empty and unwritable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100525
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100526cpu-map <"all"|"odd"|"even"|process_num> <cpu-set>...
527 On Linux 2.6 and above, it is possible to bind a process to a specific CPU
528 set. This means that the process will never run on other CPUs. The "cpu-map"
529 directive specifies CPU sets for process sets. The first argument is the
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100530 process number to bind. This process must have a number between 1 and 32 or
531 64, depending on the machine's word size, and any process IDs above nbproc
532 are ignored. It is possible to specify all processes at once using "all",
533 only odd numbers using "odd" or even numbers using "even", just like with the
534 "bind-process" directive. The second and forthcoming arguments are CPU sets.
535 Each CPU set is either a unique number between 0 and 31 or 63 or a range with
536 two such numbers delimited by a dash ('-'). Multiple CPU numbers or ranges
537 may be specified, and the processes will be allowed to bind to all of them.
538 Obviously, multiple "cpu-map" directives may be specified. Each "cpu-map"
539 directive will replace the previous ones when they overlap.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100540
Emeric Brunc8e8d122012-10-02 18:42:10 +0200541crt-base <dir>
542 Assigns a default directory to fetch SSL certificates from when a relative
543 path is used with "crtfile" directives. Absolute locations specified after
544 "crtfile" prevail and ignore "crt-base".
545
Willy Tarreau6a06a402007-07-15 20:15:28 +0200546daemon
547 Makes the process fork into background. This is the recommended mode of
548 operation. It is equivalent to the command line "-D" argument. It can be
549 disabled by the command line "-db" argument.
550
Simon Horman98637e52014-06-20 12:30:16 +0900551external-check
552 Allows the use of an external agent to perform health checks.
553 This is disabled by default as a security precaution.
554 See "option external-check".
555
Willy Tarreau6a06a402007-07-15 20:15:28 +0200556gid <number>
557 Changes the process' group ID to <number>. It is recommended that the group
558 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
559 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100560 Note that if haproxy is started from a user having supplementary groups, it
561 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200562 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100563
Willy Tarreau6a06a402007-07-15 20:15:28 +0200564group <group name>
565 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
566 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100567
Willy Tarreau18324f52014-06-27 18:10:07 +0200568log <address> [len <length>] <facility> [max level [min level]]
Willy Tarreau6a06a402007-07-15 20:15:28 +0200569 Adds a global syslog server. Up to two global servers can be defined. They
570 will receive logs for startups and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100571 configured with "log global".
572
573 <address> can be one of:
574
Willy Tarreau2769aa02007-12-27 18:26:09 +0100575 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100576 no port is specified, 514 is used by default (the standard syslog
577 port).
578
David du Colombier24bb5f52011-03-17 10:40:23 +0100579 - An IPv6 address followed by a colon and optionally a UDP port. If
580 no port is specified, 514 is used by default (the standard syslog
581 port).
582
Robert Tsai81ae1952007-12-05 10:47:29 +0100583 - A filesystem path to a UNIX domain socket, keeping in mind
584 considerations for chroot (be sure the path is accessible inside
585 the chroot) and uid/gid (be sure the path is appropriately
586 writeable).
587
Willy Tarreaudad36a32013-03-11 01:20:04 +0100588 Any part of the address string may reference any number of environment
589 variables by preceding their name with a dollar sign ('$') and
590 optionally enclosing them with braces ('{}'), similarly to what is done
591 in Bourne shell.
592
Willy Tarreau18324f52014-06-27 18:10:07 +0200593 <length> is an optional maximum line length. Log lines larger than this value
594 will be truncated before being sent. The reason is that syslog
595 servers act differently on log line length. All servers support the
596 default value of 1024, but some servers simply drop larger lines
597 while others do log them. If a server supports long lines, it may
598 make sense to set this value here in order to avoid truncating long
599 lines. Similarly, if a server drops long lines, it is preferable to
600 truncate them before sending them. Accepted values are 80 to 65535
601 inclusive. The default value of 1024 is generally fine for all
602 standard usages. Some specific cases of long captures or
603 JSON-formated logs may require larger values.
604
Robert Tsai81ae1952007-12-05 10:47:29 +0100605 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200606
607 kern user mail daemon auth syslog lpr news
608 uucp cron auth2 ftp ntp audit alert cron2
609 local0 local1 local2 local3 local4 local5 local6 local7
610
611 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200612 all messages are sent. If a maximum level is specified, only messages with a
613 severity at least as important as this level will be sent. An optional minimum
614 level can be specified. If it is set, logs emitted with a more severe level
615 than this one will be capped to this level. This is used to avoid sending
616 "emerg" messages on all terminals on some default syslog configurations.
617 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200618
Cyril Bontédc4d9032012-04-08 21:57:39 +0200619 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200620
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100621log-send-hostname [<string>]
622 Sets the hostname field in the syslog header. If optional "string" parameter
623 is set the header is set to the string contents, otherwise uses the hostname
624 of the system. Generally used if one is not relaying logs through an
625 intermediate syslog server or for simply customizing the hostname printed in
626 the logs.
627
Kevinm48936af2010-12-22 16:08:21 +0000628log-tag <string>
629 Sets the tag field in the syslog header to this string. It defaults to the
630 program name as launched from the command line, which usually is "haproxy".
631 Sometimes it can be useful to differentiate between multiple processes
632 running on the same host.
633
Willy Tarreau6a06a402007-07-15 20:15:28 +0200634nbproc <number>
635 Creates <number> processes when going daemon. This requires the "daemon"
636 mode. By default, only one process is created, which is the recommended mode
637 of operation. For systems limited to small sets of file descriptors per
638 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
639 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
640
641pidfile <pidfile>
642 Writes pids of all daemons into file <pidfile>. This option is equivalent to
643 the "-p" command line argument. The file must be accessible to the user
644 starting the process. See also "daemon".
645
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100646stats bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200647 Limits the stats socket to a certain set of processes numbers. By default the
648 stats socket is bound to all processes, causing a warning to be emitted when
649 nbproc is greater than 1 because there is no way to select the target process
650 when connecting. However, by using this setting, it becomes possible to pin
651 the stats socket to a specific set of processes, typically the first one. The
652 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100653 the number of processes used. The maximum process ID depends on the machine's
Willy Tarreauae302532014-05-07 19:22:24 +0200654 word size (32 or 64). A better option consists in using the "process" setting
655 of the "stats socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +0200656
Willy Tarreau610f04b2014-02-13 11:36:41 +0100657ssl-default-bind-ciphers <ciphers>
658 This setting is only available when support for OpenSSL was built in. It sets
659 the default string describing the list of cipher algorithms ("cipher suite")
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300660 that are negotiated during the SSL/TLS handshake for all "bind" lines which
Willy Tarreau610f04b2014-02-13 11:36:41 +0100661 do not explicitly define theirs. The format of the string is defined in
662 "man 1 ciphers" from OpenSSL man pages, and can be for instance a string such
663 as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). Please check the
664 "bind" keyword for more information.
665
666ssl-default-server-ciphers <ciphers>
667 This setting is only available when support for OpenSSL was built in. It
668 sets the default string describing the list of cipher algorithms that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300669 negotiated during the SSL/TLS handshake with the server, for all "server"
Willy Tarreau610f04b2014-02-13 11:36:41 +0100670 lines which do not explicitly define theirs. The format of the string is
671 defined in "man 1 ciphers". Please check the "server" keyword for more
672 information.
673
Emeric Brun850efd52014-01-29 12:24:34 +0100674ssl-server-verify [none|required]
675 The default behavior for SSL verify on servers side. If specified to 'none',
676 servers certificates are not verified. The default is 'required' except if
677 forced using cmdline option '-dV'.
678
Willy Tarreauabb175f2012-09-24 12:43:26 +0200679stats socket [<address:port>|<path>] [param*]
680 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
681 Connections to this socket will return various statistics outputs and even
682 allow some commands to be issued to change some runtime settings. Please
683 consult section 9.2 "Unix Socket commands" for more details.
Willy Tarreau6162db22009-10-10 17:13:00 +0200684
Willy Tarreauabb175f2012-09-24 12:43:26 +0200685 All parameters supported by "bind" lines are supported, for instance to
686 restrict access to some users or their access rights. Please consult
687 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200688
689stats timeout <timeout, in milliseconds>
690 The default timeout on the stats socket is set to 10 seconds. It is possible
691 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +0100692 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200693
694stats maxconn <connections>
695 By default, the stats socket is limited to 10 concurrent connections. It is
696 possible to change this value with "stats maxconn".
697
Willy Tarreau6a06a402007-07-15 20:15:28 +0200698uid <number>
699 Changes the process' user ID to <number>. It is recommended that the user ID
700 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
701 be started with superuser privileges in order to be able to switch to another
702 one. See also "gid" and "user".
703
704ulimit-n <number>
705 Sets the maximum number of per-process file-descriptors to <number>. By
706 default, it is automatically computed, so it is recommended not to use this
707 option.
708
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100709unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
710 [ group <group> ] [ gid <gid> ]
711
712 Fixes common settings to UNIX listening sockets declared in "bind" statements.
713 This is mainly used to simplify declaration of those UNIX sockets and reduce
714 the risk of errors, since those settings are most commonly required but are
715 also process-specific. The <prefix> setting can be used to force all socket
716 path to be relative to that directory. This might be needed to access another
717 component's chroot. Note that those paths are resolved before haproxy chroots
718 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
719 all have the same meaning as their homonyms used by the "bind" statement. If
720 both are specified, the "bind" statement has priority, meaning that the
721 "unix-bind" settings may be seen as process-wide default settings.
722
Willy Tarreau6a06a402007-07-15 20:15:28 +0200723user <user name>
724 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
725 See also "uid" and "group".
726
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +0200727node <name>
728 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
729
730 This statement is useful in HA configurations where two or more processes or
731 servers share the same IP address. By setting a different node-name on all
732 nodes, it becomes easy to immediately spot what server is handling the
733 traffic.
734
735description <text>
736 Add a text that describes the instance.
737
738 Please note that it is required to escape certain characters (# for example)
739 and this text is inserted into a html page so you should avoid using
740 "<" and ">" characters.
741
Willy Tarreau6a06a402007-07-15 20:15:28 +0200742
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007433.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200744-----------------------
745
Willy Tarreau1746eec2014-04-25 10:46:47 +0200746max-spread-checks <delay in milliseconds>
747 By default, haproxy tries to spread the start of health checks across the
748 smallest health check interval of all the servers in a farm. The principle is
749 to avoid hammering services running on the same server. But when using large
750 check intervals (10 seconds or more), the last servers in the farm take some
751 time before starting to be tested, which can be a problem. This parameter is
752 used to enforce an upper bound on delay between the first and the last check,
753 even if the servers' check intervals are larger. When servers run with
754 shorter intervals, their intervals will be respected though.
755
Willy Tarreau6a06a402007-07-15 20:15:28 +0200756maxconn <number>
757 Sets the maximum per-process number of concurrent connections to <number>. It
758 is equivalent to the command-line argument "-n". Proxies will stop accepting
759 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +0200760 automatically adjusted according to this value. See also "ulimit-n". Note:
761 the "select" poller cannot reliably use more than 1024 file descriptors on
762 some platforms. If your platform only supports select and reports "select
763 FAILED" on startup, you need to reduce maxconn until it works (slightly
764 below 500 in general).
Willy Tarreau6a06a402007-07-15 20:15:28 +0200765
Willy Tarreau81c25d02011-09-07 15:17:21 +0200766maxconnrate <number>
767 Sets the maximum per-process number of connections per second to <number>.
768 Proxies will stop accepting connections when this limit is reached. It can be
769 used to limit the global capacity regardless of each frontend capacity. It is
770 important to note that this can only be used as a service protection measure,
771 as there will not necessarily be a fair share between frontends when the
772 limit is reached, so it's a good idea to also limit each frontend to some
773 value close to its expected share. Also, lowering tune.maxaccept can improve
774 fairness.
775
William Lallemandd85f9172012-11-09 17:05:39 +0100776maxcomprate <number>
777 Sets the maximum per-process input compression rate to <number> kilobytes
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300778 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +0100779 level will be decreased during the session. If the maximum is reached at the
780 beginning of a session, the session will not compress at all. If the maximum
781 is not reached, the compression level will be increased up to
782 tune.comp.maxlevel. A value of zero means there is no limit, this is the
783 default value.
784
William Lallemand072a2bf2012-11-20 17:01:01 +0100785maxcompcpuusage <number>
786 Sets the maximum CPU usage HAProxy can reach before stopping the compression
787 for new requests or decreasing the compression level of current requests.
788 It works like 'maxcomprate' but measures CPU usage instead of incoming data
789 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
790 case of multiple processes (nbproc > 1), each process manages its individual
791 usage. A value of 100 disable the limit. The default value is 100. Setting
792 a lower value will prevent the compression work from slowing the whole
793 process down and from introducing high latencies.
794
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100795maxpipes <number>
796 Sets the maximum per-process number of pipes to <number>. Currently, pipes
797 are only used by kernel-based tcp splicing. Since a pipe contains two file
798 descriptors, the "ulimit-n" value will be increased accordingly. The default
799 value is maxconn/4, which seems to be more than enough for most heavy usages.
800 The splice code dynamically allocates and releases pipes, and can fall back
801 to standard copy, so setting this value too low may only impact performance.
802
Willy Tarreau93e7c002013-10-07 18:51:07 +0200803maxsessrate <number>
804 Sets the maximum per-process number of sessions per second to <number>.
805 Proxies will stop accepting connections when this limit is reached. It can be
806 used to limit the global capacity regardless of each frontend capacity. It is
807 important to note that this can only be used as a service protection measure,
808 as there will not necessarily be a fair share between frontends when the
809 limit is reached, so it's a good idea to also limit each frontend to some
810 value close to its expected share. Also, lowering tune.maxaccept can improve
811 fairness.
812
Willy Tarreau403edff2012-09-06 11:58:37 +0200813maxsslconn <number>
814 Sets the maximum per-process number of concurrent SSL connections to
815 <number>. By default there is no SSL-specific limit, which means that the
816 global maxconn setting will apply to all connections. Setting this limit
817 avoids having openssl use too much memory and crash when malloc returns NULL
818 (since it unfortunately does not reliably check for such conditions). Note
819 that the limit applies both to incoming and outgoing connections, so one
820 connection which is deciphered then ciphered accounts for 2 SSL connections.
821
Willy Tarreaue43d5322013-10-07 20:01:52 +0200822maxsslrate <number>
823 Sets the maximum per-process number of SSL sessions per second to <number>.
824 SSL listeners will stop accepting connections when this limit is reached. It
825 can be used to limit the global SSL CPU usage regardless of each frontend
826 capacity. It is important to note that this can only be used as a service
827 protection measure, as there will not necessarily be a fair share between
828 frontends when the limit is reached, so it's a good idea to also limit each
829 frontend to some value close to its expected share. It is also important to
830 note that the sessions are accounted before they enter the SSL stack and not
831 after, which also protects the stack against bad handshakes. Also, lowering
832 tune.maxaccept can improve fairness.
833
William Lallemand9d5f5482012-11-07 16:12:57 +0100834maxzlibmem <number>
835 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
836 When the maximum amount is reached, future sessions will not compress as long
837 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +0100838 The default value is 0. The value is available in bytes on the UNIX socket
839 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
840 "ZlibMemUsage" in bytes.
841
Willy Tarreau6a06a402007-07-15 20:15:28 +0200842noepoll
843 Disables the use of the "epoll" event polling system on Linux. It is
844 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +0100845 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +0200846
847nokqueue
848 Disables the use of the "kqueue" event polling system on BSD. It is
849 equivalent to the command-line argument "-dk". The next polling system
850 used will generally be "poll". See also "nopoll".
851
852nopoll
853 Disables the use of the "poll" event polling system. It is equivalent to the
854 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +0100855 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +0100856 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +0200857
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100858nosplice
859 Disables the use of kernel tcp splicing between sockets on Linux. It is
860 equivalent to the command line argument "-dS". Data will then be copied
861 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100862 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100863 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
864 be used. This option makes it easier to globally disable kernel splicing in
865 case of doubt. See also "option splice-auto", "option splice-request" and
866 "option splice-response".
867
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300868nogetaddrinfo
869 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
870 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
871
Willy Tarreaufe255b72007-10-14 23:09:26 +0200872spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +0900873 Sometimes it is desirable to avoid sending agent and health checks to
874 servers at exact intervals, for instance when many logical servers are
875 located on the same physical server. With the help of this parameter, it
876 becomes possible to add some randomness in the check interval between 0
877 and +/- 50%. A value between 2 and 5 seems to show good results. The
878 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +0200879
Willy Tarreau27a674e2009-08-17 07:23:33 +0200880tune.bufsize <number>
881 Sets the buffer size to this size (in bytes). Lower values allow more
882 sessions to coexist in the same amount of RAM, and higher values allow some
883 applications with very large cookies to work. The default value is 16384 and
884 can be changed at build time. It is strongly recommended not to change this
885 from the default value, as very low values will break some services such as
886 statistics, and values larger than default size will increase memory usage,
887 possibly causing the system to run out of memory. At least the global maxconn
888 parameter should be decreased by the same factor as this one is increased.
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +0400889 If HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
890 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
891 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +0200892
Willy Tarreau43961d52010-10-04 20:39:20 +0200893tune.chksize <number>
894 Sets the check buffer size to this size (in bytes). Higher values may help
895 find string or regex patterns in very large pages, though doing so may imply
896 more memory and CPU usage. The default value is 16384 and can be changed at
897 build time. It is not recommended to change this value, but to use better
898 checks whenever possible.
899
William Lallemandf3747832012-11-09 12:33:10 +0100900tune.comp.maxlevel <number>
901 Sets the maximum compression level. The compression level affects CPU
902 usage during compression. This value affects CPU usage during compression.
903 Each session using compression initializes the compression algorithm with
904 this value. The default value is 1.
905
Willy Tarreau193b8c62012-11-22 00:17:38 +0100906tune.http.cookielen <number>
907 Sets the maximum length of captured cookies. This is the maximum value that
908 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
909 will automatically be truncated to this one. It is important not to set too
910 high a value because all cookie captures still allocate this size whatever
911 their configured value (they share a same pool). This value is per request
912 per response, so the memory allocated is twice this value per connection.
913 When not specified, the limit is set to 63 characters. It is recommended not
914 to change this value.
915
Willy Tarreauac1932d2011-10-24 19:14:41 +0200916tune.http.maxhdr <number>
917 Sets the maximum number of headers in a request. When a request comes with a
918 number of headers greater than this value (including the first line), it is
919 rejected with a "400 Bad Request" status code. Similarly, too large responses
920 are blocked with "502 Bad Gateway". The default value is 101, which is enough
921 for all usages, considering that the widely deployed Apache server uses the
922 same limit. It can be useful to push this limit further to temporarily allow
923 a buggy application to work by the time it gets fixed. Keep in mind that each
924 new header consumes 32bits of memory for each session, so don't push this
925 limit too high.
926
Willy Tarreau7e312732014-02-12 16:35:14 +0100927tune.idletimer <timeout>
928 Sets the duration after which haproxy will consider that an empty buffer is
929 probably associated with an idle stream. This is used to optimally adjust
930 some packet sizes while forwarding large and small data alternatively. The
931 decision to use splice() or to send large buffers in SSL is modulated by this
932 parameter. The value is in milliseconds between 0 and 65535. A value of zero
933 means that haproxy will not try to detect idle streams. The default is 1000,
934 which seems to correctly detect end user pauses (eg: read a page before
935 clicking). There should be not reason for changing this value. Please check
936 tune.ssl.maxrecord below.
937
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100938tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +0100939 Sets the maximum number of consecutive connections a process may accept in a
940 row before switching to other work. In single process mode, higher numbers
941 give better performance at high connection rates. However in multi-process
942 modes, keeping a bit of fairness between processes generally is better to
943 increase performance. This value applies individually to each listener, so
944 that the number of processes a listener is bound to is taken into account.
945 This value defaults to 64. In multi-process mode, it is divided by twice
946 the number of processes the listener is bound to. Setting this value to -1
947 completely disables the limitation. It should normally not be needed to tweak
948 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100949
950tune.maxpollevents <number>
951 Sets the maximum amount of events that can be processed at once in a call to
952 the polling system. The default value is adapted to the operating system. It
953 has been noticed that reducing it below 200 tends to slightly decrease
954 latency at the expense of network bandwidth, and increasing it above 200
955 tends to trade latency for slightly increased bandwidth.
956
Willy Tarreau27a674e2009-08-17 07:23:33 +0200957tune.maxrewrite <number>
958 Sets the reserved buffer space to this size in bytes. The reserved space is
959 used for header rewriting or appending. The first reads on sockets will never
960 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
961 bufsize, though that does not make much sense since there are rarely large
962 numbers of headers to add. Setting it too high prevents processing of large
963 requests or responses. Setting it too low prevents addition of new headers
964 to already large requests or to POST requests. It is generally wise to set it
965 to about 1024. It is automatically readjusted to half of bufsize if it is
966 larger than that. This means you don't have to worry about it when changing
967 bufsize.
968
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200969tune.pipesize <number>
970 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
971 are the default size for the system. But sometimes when using TCP splicing,
972 it can improve performance to increase pipe sizes, especially if it is
973 suspected that pipes are not filled and that many calls to splice() are
974 performed. This has an impact on the kernel's memory footprint, so this must
975 not be changed if impacts are not understood.
976
Willy Tarreaue803de22010-01-21 17:43:04 +0100977tune.rcvbuf.client <number>
978tune.rcvbuf.server <number>
979 Forces the kernel socket receive buffer size on the client or the server side
980 to the specified value in bytes. This value applies to all TCP/HTTP frontends
981 and backends. It should normally never be set, and the default size (0) lets
982 the kernel autotune this value depending on the amount of available memory.
983 However it can sometimes help to set it to very low values (eg: 4096) in
984 order to save kernel memory by preventing it from buffering too large amounts
985 of received data. Lower values will significantly increase CPU usage though.
986
987tune.sndbuf.client <number>
988tune.sndbuf.server <number>
989 Forces the kernel socket send buffer size on the client or the server side to
990 the specified value in bytes. This value applies to all TCP/HTTP frontends
991 and backends. It should normally never be set, and the default size (0) lets
992 the kernel autotune this value depending on the amount of available memory.
993 However it can sometimes help to set it to very low values (eg: 4096) in
994 order to save kernel memory by preventing it from buffering too large amounts
995 of received data. Lower values will significantly increase CPU usage though.
996 Another use case is to prevent write timeouts with extremely slow clients due
997 to the kernel waiting for a large part of the buffer to be read before
998 notifying haproxy again.
999
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001000tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001001 Sets the size of the global SSL session cache, in a number of blocks. A block
1002 is large enough to contain an encoded session without peer certificate.
1003 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001004 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001005 200 bytes of memory. The default value may be forced at build time, otherwise
1006 defaults to 20000. When the cache is full, the most idle entries are purged
1007 and reassigned. Higher values reduce the occurrence of such a purge, hence
1008 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1009 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001010 and are shared between all processes if "nbproc" is greater than 1. Setting
1011 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001012
Emeric Brun8dc60392014-05-09 13:52:00 +02001013tune.ssl.force-private-cache
1014 This boolean disables SSL session cache sharing between all processes. It
1015 should normally not be used since it will force many renegotiations due to
1016 clients hitting a random process. But it may be required on some operating
1017 systems where none of the SSL cache synchronization method may be used. In
1018 this case, adding a first layer of hash-based load balancing before the SSL
1019 layer might limit the impact of the lack of session sharing.
1020
Emeric Brun4f65bff2012-11-16 15:11:00 +01001021tune.ssl.lifetime <timeout>
1022 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001023 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001024 does not guarantee that sessions will last that long, because if the cache is
1025 full, the longest idle sessions will be purged despite their configured
1026 lifetime. The real usefulness of this setting is to prevent sessions from
1027 being used for too long.
1028
Willy Tarreaubfd59462013-02-21 07:46:09 +01001029tune.ssl.maxrecord <number>
1030 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1031 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1032 data only once it has received a full record. With large records, it means
1033 that clients might have to download up to 16kB of data before starting to
1034 process them. Limiting the value can improve page load times on browsers
1035 located over high latency or low bandwidth networks. It is suggested to find
1036 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1037 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1038 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1039 2859 gave good results during tests. Use "strace -e trace=write" to find the
Willy Tarreau7e312732014-02-12 16:35:14 +01001040 best value. Haproxy will automatically switch to this setting after an idle
1041 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001042
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001043tune.ssl.default-dh-param <number>
1044 Sets the maximum size of the Diffie-Hellman parameters used for generating
1045 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1046 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1047 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1048 this maximum value. Default value if 1024. Only 1024 or higher values are
1049 allowed. Higher values will increase the CPU load, and values greater than
1050 1024 bits are not supported by Java 7 and earlier clients. This value is not
1051 used if static Diffie-Hellman parameters are supplied via the certificate file.
1052
William Lallemanda509e4c2012-11-07 16:54:34 +01001053tune.zlib.memlevel <number>
1054 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001055 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001056 state. A value of 1 uses minimum memory but is slow and reduces compression
1057 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
1058 between 1 and 9. The default value is 8.
1059
1060tune.zlib.windowsize <number>
1061 Sets the window size (the size of the history buffer) as a parameter of the
1062 zlib initialization for each session. Larger values of this parameter result
1063 in better compression at the expense of memory usage. Can be a value between
1064 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001065
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010663.3. Debugging
1067--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001068
1069debug
1070 Enables debug mode which dumps to stdout all exchanges, and disables forking
1071 into background. It is the equivalent of the command-line argument "-d". It
1072 should never be used in a production configuration since it may prevent full
1073 system startup.
1074
1075quiet
1076 Do not display any message during startup. It is equivalent to the command-
1077 line argument "-q".
1078
Emeric Brunf099e792010-09-27 12:05:28 +02001079
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010010803.4. Userlists
1081--------------
1082It is possible to control access to frontend/backend/listen sections or to
1083http stats by allowing only authenticated and authorized users. To do this,
1084it is required to create at least one userlist and to define users.
1085
1086userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001087 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001088 used to store authentication & authorization data for independent customers.
1089
1090group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001091 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001092 attach users to this group by using a comma separated list of names
1093 proceeded by "users" keyword.
1094
Cyril Bontéf0c60612010-02-06 14:44:47 +01001095user <username> [password|insecure-password <password>]
1096 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001097 Adds user <username> to the current userlist. Both secure (encrypted) and
1098 insecure (unencrypted) passwords can be used. Encrypted passwords are
Cyril Bonté78caf842010-03-10 22:41:43 +01001099 evaluated using the crypt(3) function so depending of the system's
1100 capabilities, different algorithms are supported. For example modern Glibc
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001101 based Linux system supports MD5, SHA-256, SHA-512 and of course classic,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001102 DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001103
1104
1105 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001106 userlist L1
1107 group G1 users tiger,scott
1108 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001109
Cyril Bontéf0c60612010-02-06 14:44:47 +01001110 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1111 user scott insecure-password elgato
1112 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001113
Cyril Bontéf0c60612010-02-06 14:44:47 +01001114 userlist L2
1115 group G1
1116 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001117
Cyril Bontéf0c60612010-02-06 14:44:47 +01001118 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1119 user scott insecure-password elgato groups G1,G2
1120 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001121
1122 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001123
Emeric Brunf099e792010-09-27 12:05:28 +02001124
11253.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001126----------
Emeric Brunf099e792010-09-27 12:05:28 +02001127It is possible to synchronize server entries in stick tables between several
1128haproxy instances over TCP connections in a multi-master fashion. Each instance
1129pushes its local updates and insertions to remote peers. Server IDs are used to
1130identify servers remotely, so it is important that configurations look similar
1131or at least that the same IDs are forced on each server on all participants.
1132Interrupted exchanges are automatically detected and recovered from the last
1133known point. In addition, during a soft restart, the old process connects to
1134the new one using such a TCP connection to push all its entries before the new
1135process tries to connect to other peers. That ensures very fast replication
1136during a reload, it typically takes a fraction of a second even for large
1137tables.
1138
1139peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001140 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001141 which is referenced by one or more stick-tables.
1142
1143peer <peername> <ip>:<port>
1144 Defines a peer inside a peers section.
1145 If <peername> is set to the local peer name (by default hostname, or forced
1146 using "-L" command line option), haproxy will listen for incoming remote peer
1147 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1148 to join the remote peer, and <peername> is used at the protocol level to
1149 identify and validate the remote peer on the server side.
1150
1151 During a soft restart, local peer <ip>:<port> is used by the old instance to
1152 connect the new one and initiate a complete replication (teaching process).
1153
1154 It is strongly recommended to have the exact same peers declaration on all
1155 peers and to only rely on the "-L" command line argument to change the local
1156 peer name. This makes it easier to maintain coherent configuration files
1157 across all peers.
1158
Willy Tarreaudad36a32013-03-11 01:20:04 +01001159 Any part of the address string may reference any number of environment
1160 variables by preceding their name with a dollar sign ('$') and optionally
1161 enclosing them with braces ('{}'), similarly to what is done in Bourne shell.
1162
Cyril Bontédc4d9032012-04-08 21:57:39 +02001163 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001164 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001165 peer haproxy1 192.168.0.1:1024
1166 peer haproxy2 192.168.0.2:1024
1167 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001168
1169 backend mybackend
1170 mode tcp
1171 balance roundrobin
1172 stick-table type ip size 20k peers mypeers
1173 stick on src
1174
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001175 server srv1 192.168.0.30:80
1176 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001177
1178
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011794. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001180----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001181
Willy Tarreau6a06a402007-07-15 20:15:28 +02001182Proxy configuration can be located in a set of sections :
1183 - defaults <name>
1184 - frontend <name>
1185 - backend <name>
1186 - listen <name>
1187
1188A "defaults" section sets default parameters for all other sections following
1189its declaration. Those default parameters are reset by the next "defaults"
1190section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001191section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001192
1193A "frontend" section describes a set of listening sockets accepting client
1194connections.
1195
1196A "backend" section describes a set of servers to which the proxy will connect
1197to forward incoming connections.
1198
1199A "listen" section defines a complete proxy with its frontend and backend
1200parts combined in one section. It is generally useful for TCP-only traffic.
1201
Willy Tarreau0ba27502007-12-24 16:55:16 +01001202All proxy names must be formed from upper and lower case letters, digits,
1203'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1204case-sensitive, which means that "www" and "WWW" are two different proxies.
1205
1206Historically, all proxy names could overlap, it just caused troubles in the
1207logs. Since the introduction of content switching, it is mandatory that two
1208proxies with overlapping capabilities (frontend/backend) have different names.
1209However, it is still permitted that a frontend and a backend share the same
1210name, as this configuration seems to be commonly encountered.
1211
1212Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1213and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001214bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001215protocol, and can interact with it by allowing, blocking, switching, adding,
1216modifying, or removing arbitrary contents in requests or responses, based on
1217arbitrary criteria.
1218
Willy Tarreau70dffda2014-01-30 03:07:23 +01001219In HTTP mode, the processing applied to requests and responses flowing over
1220a connection depends in the combination of the frontend's HTTP options and
1221the backend's. HAProxy supports 5 connection modes :
1222
1223 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1224 requests and responses are processed, and connections remain open but idle
1225 between responses and new requests.
1226
1227 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1228 1.0 to 1.5-dev21 : only the first request and response are processed, and
1229 everything else is forwarded with no analysis at all. This mode should not
1230 be used as it creates lots of trouble with logging and HTTP processing.
1231
1232 - PCL: passive close ("option httpclose") : exactly the same as tunnel mode,
1233 but with "Connection: close" appended in both directions to try to make
1234 both ends close after the first request/response exchange.
1235
1236 - SCL: server close ("option http-server-close") : the server-facing
1237 connection is closed after the end of the response is received, but the
1238 client-facing connection remains open.
1239
1240 - FCL: forced close ("option forceclose") : the connection is actively closed
1241 after the end of the response.
1242
1243The effective mode that will be applied to a connection passing through a
1244frontend and a backend can be determined by both proxy modes according to the
1245following matrix, but in short, the modes are symmetric, keep-alive is the
1246weakest option and force close is the strongest.
1247
1248 Backend mode
1249
1250 | KAL | TUN | PCL | SCL | FCL
1251 ----+-----+-----+-----+-----+----
1252 KAL | KAL | TUN | PCL | SCL | FCL
1253 ----+-----+-----+-----+-----+----
1254 TUN | TUN | TUN | PCL | SCL | FCL
1255 Frontend ----+-----+-----+-----+-----+----
1256 mode PCL | PCL | PCL | PCL | FCL | FCL
1257 ----+-----+-----+-----+-----+----
1258 SCL | SCL | SCL | FCL | SCL | FCL
1259 ----+-----+-----+-----+-----+----
1260 FCL | FCL | FCL | FCL | FCL | FCL
1261
Willy Tarreau0ba27502007-12-24 16:55:16 +01001262
Willy Tarreau70dffda2014-01-30 03:07:23 +01001263
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012644.1. Proxy keywords matrix
1265--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001266
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001267The following list of keywords is supported. Most of them may only be used in a
1268limited set of section types. Some of them are marked as "deprecated" because
1269they are inherited from an old syntax which may be confusing or functionally
1270limited, and there are new recommended keywords to replace them. Keywords
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001271marked with "(*)" can be optionally inverted using the "no" prefix, eg. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001272option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02001273and must be disabled for a specific instance. Such options may also be prefixed
1274with "default" in order to restore default settings regardless of what has been
1275specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001276
Willy Tarreau6a06a402007-07-15 20:15:28 +02001277
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001278 keyword defaults frontend listen backend
1279------------------------------------+----------+----------+---------+---------
1280acl - X X X
1281appsession - - X X
1282backlog X X X -
1283balance X - X X
1284bind - X X -
1285bind-process X X X X
1286block - X X X
1287capture cookie - X X -
1288capture request header - X X -
1289capture response header - X X -
1290clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02001291compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001292contimeout (deprecated) X - X X
1293cookie X - X X
1294default-server X - X X
1295default_backend X X X -
1296description - X X X
1297disabled X X X X
1298dispatch - - X X
1299enabled X X X X
1300errorfile X X X X
1301errorloc X X X X
1302errorloc302 X X X X
1303-- keyword -------------------------- defaults - frontend - listen -- backend -
1304errorloc303 X X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001305force-persist - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001306fullconn X - X X
1307grace X X X X
1308hash-type X - X X
1309http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01001310http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02001311http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001312http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02001313http-response - X X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02001314http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001315id - X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001316ignore-persist - X X X
William Lallemand0f99e342011-10-12 17:50:54 +02001317log (*) X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02001318max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001319maxconn X X X -
1320mode X X X X
1321monitor fail - X X -
1322monitor-net X X X -
1323monitor-uri X X X -
1324option abortonclose (*) X - X X
1325option accept-invalid-http-request (*) X X X -
1326option accept-invalid-http-response (*) X - X X
1327option allbackups (*) X - X X
1328option checkcache (*) X - X X
1329option clitcpka (*) X X X -
1330option contstats (*) X X X -
1331option dontlog-normal (*) X X X -
1332option dontlognull (*) X X X -
1333option forceclose (*) X X X X
1334-- keyword -------------------------- defaults - frontend - listen -- backend -
1335option forwardfor X X X X
Willy Tarreau16bfb022010-01-16 19:48:41 +01001336option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02001337option http-no-delay (*) X X X X
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02001338option http-pretend-keepalive (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001339option http-server-close (*) X X X X
Willy Tarreau02bce8b2014-01-30 00:15:28 +01001340option http-tunnel (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001341option http-use-proxy-header (*) X X X -
1342option httpchk X - X X
1343option httpclose (*) X X X X
1344option httplog X X X X
1345option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001346option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02001347option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09001348option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001349option log-health-checks (*) X - X X
1350option log-separate-errors (*) X X X -
1351option logasap (*) X X X -
1352option mysql-check X - X X
Rauf Kuliyev38b41562011-01-04 15:14:13 +01001353option pgsql-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001354option nolinger (*) X X X X
1355option originalto X X X X
1356option persist (*) X - X X
1357option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02001358option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001359option smtpchk X - X X
1360option socket-stats (*) X X X -
1361option splice-auto (*) X X X X
1362option splice-request (*) X X X X
1363option splice-response (*) X X X X
1364option srvtcpka (*) X - X X
1365option ssl-hello-chk X - X X
1366-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01001367option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001368option tcp-smart-accept (*) X X X -
1369option tcp-smart-connect (*) X - X X
1370option tcpka X X X X
1371option tcplog X X X X
1372option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09001373external-check command X - X X
1374external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001375persist rdp-cookie X - X X
1376rate-limit sessions X X X -
1377redirect - X X X
1378redisp (deprecated) X - X X
1379redispatch (deprecated) X - X X
1380reqadd - X X X
1381reqallow - X X X
1382reqdel - X X X
1383reqdeny - X X X
1384reqiallow - X X X
1385reqidel - X X X
1386reqideny - X X X
1387reqipass - X X X
1388reqirep - X X X
1389reqisetbe - X X X
1390reqitarpit - X X X
1391reqpass - X X X
1392reqrep - X X X
1393-- keyword -------------------------- defaults - frontend - listen -- backend -
1394reqsetbe - X X X
1395reqtarpit - X X X
1396retries X - X X
1397rspadd - X X X
1398rspdel - X X X
1399rspdeny - X X X
1400rspidel - X X X
1401rspideny - X X X
1402rspirep - X X X
1403rsprep - X X X
1404server - - X X
1405source X - X X
1406srvtimeout (deprecated) X - X X
Cyril Bonté66c327d2010-10-12 00:14:37 +02001407stats admin - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001408stats auth X - X X
1409stats enable X - X X
1410stats hide-version X - X X
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02001411stats http-request - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001412stats realm X - X X
1413stats refresh X - X X
1414stats scope X - X X
1415stats show-desc X - X X
1416stats show-legends X - X X
1417stats show-node X - X X
1418stats uri X - X X
1419-- keyword -------------------------- defaults - frontend - listen -- backend -
1420stick match - - X X
1421stick on - - X X
1422stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02001423stick store-response - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001424stick-table - - X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02001425tcp-check connect - - X X
1426tcp-check expect - - X X
1427tcp-check send - - X X
1428tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02001429tcp-request connection - X X -
1430tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02001431tcp-request inspect-delay - X X X
Emeric Brun0a3b67f2010-09-24 15:34:53 +02001432tcp-response content - - X X
1433tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001434timeout check X - X X
1435timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02001436timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001437timeout clitimeout (deprecated) X X X -
1438timeout connect X - X X
1439timeout contimeout (deprecated) X - X X
1440timeout http-keep-alive X X X X
1441timeout http-request X X X X
1442timeout queue X - X X
1443timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02001444timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001445timeout srvtimeout (deprecated) X - X X
1446timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02001447timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001448transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01001449unique-id-format X X X -
1450unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001451use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02001452use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001453------------------------------------+----------+----------+---------+---------
1454 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02001455
Willy Tarreau0ba27502007-12-24 16:55:16 +01001456
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014574.2. Alphabetically sorted keywords reference
1458---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001459
1460This section provides a description of each keyword and its usage.
1461
1462
1463acl <aclname> <criterion> [flags] [operator] <value> ...
1464 Declare or complete an access list.
1465 May be used in sections : defaults | frontend | listen | backend
1466 no | yes | yes | yes
1467 Example:
1468 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1469 acl invalid_src src_port 0:1023
1470 acl local_dst hdr(host) -i localhost
1471
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001472 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001473
1474
Cyril Bontéb21570a2009-11-29 20:04:48 +01001475appsession <cookie> len <length> timeout <holdtime>
1476 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001477 Define session stickiness on an existing application cookie.
1478 May be used in sections : defaults | frontend | listen | backend
1479 no | no | yes | yes
1480 Arguments :
1481 <cookie> this is the name of the cookie used by the application and which
1482 HAProxy will have to learn for each new session.
1483
Cyril Bontéb21570a2009-11-29 20:04:48 +01001484 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01001485 checked in each cookie value.
1486
1487 <holdtime> this is the time after which the cookie will be removed from
1488 memory if unused. If no unit is specified, this time is in
1489 milliseconds.
1490
Cyril Bontébf47aeb2009-10-15 00:15:40 +02001491 request-learn
1492 If this option is specified, then haproxy will be able to learn
1493 the cookie found in the request in case the server does not
1494 specify any in response. This is typically what happens with
1495 PHPSESSID cookies, or when haproxy's session expires before
1496 the application's session and the correct server is selected.
1497 It is recommended to specify this option to improve reliability.
1498
Cyril Bontéb21570a2009-11-29 20:04:48 +01001499 prefix When this option is specified, haproxy will match on the cookie
1500 prefix (or URL parameter prefix). The appsession value is the
1501 data following this prefix.
1502
1503 Example :
1504 appsession ASPSESSIONID len 64 timeout 3h prefix
1505
1506 This will match the cookie ASPSESSIONIDXXXX=XXXXX,
1507 the appsession value will be XXXX=XXXXX.
1508
1509 mode This option allows to change the URL parser mode.
1510 2 modes are currently supported :
1511 - path-parameters :
1512 The parser looks for the appsession in the path parameters
1513 part (each parameter is separated by a semi-colon), which is
1514 convenient for JSESSIONID for example.
1515 This is the default mode if the option is not set.
1516 - query-string :
1517 In this mode, the parser will look for the appsession in the
1518 query string.
1519
Willy Tarreau0ba27502007-12-24 16:55:16 +01001520 When an application cookie is defined in a backend, HAProxy will check when
1521 the server sets such a cookie, and will store its value in a table, and
1522 associate it with the server's identifier. Up to <length> characters from
1523 the value will be retained. On each connection, haproxy will look for this
Cyril Bontéb21570a2009-11-29 20:04:48 +01001524 cookie both in the "Cookie:" headers, and as a URL parameter (depending on
1525 the mode used). If a known value is found, the client will be directed to the
1526 server associated with this value. Otherwise, the load balancing algorithm is
Willy Tarreau0ba27502007-12-24 16:55:16 +01001527 applied. Cookies are automatically removed from memory when they have been
1528 unused for a duration longer than <holdtime>.
1529
1530 The definition of an application cookie is limited to one per backend.
1531
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001532 Note : Consider not using this feature in multi-process mode (nbproc > 1)
1533 unless you know what you do : memory is not shared between the
1534 processes, which can result in random behaviours.
1535
Willy Tarreau0ba27502007-12-24 16:55:16 +01001536 Example :
1537 appsession JSESSIONID len 52 timeout 3h
1538
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01001539 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
1540 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001541
1542
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001543backlog <conns>
1544 Give hints to the system about the approximate listen backlog desired size
1545 May be used in sections : defaults | frontend | listen | backend
1546 yes | yes | yes | no
1547 Arguments :
1548 <conns> is the number of pending connections. Depending on the operating
1549 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02001550 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01001551
1552 In order to protect against SYN flood attacks, one solution is to increase
1553 the system's SYN backlog size. Depending on the system, sometimes it is just
1554 tunable via a system parameter, sometimes it is not adjustable at all, and
1555 sometimes the system relies on hints given by the application at the time of
1556 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
1557 to the listen() syscall. On systems which can make use of this value, it can
1558 sometimes be useful to be able to specify a different value, hence this
1559 backlog parameter.
1560
1561 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
1562 used as a hint and the system accepts up to the smallest greater power of
1563 two, and never more than some limits (usually 32768).
1564
1565 See also : "maxconn" and the target operating system's tuning guide.
1566
1567
Willy Tarreau0ba27502007-12-24 16:55:16 +01001568balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02001569balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001570 Define the load balancing algorithm to be used in a backend.
1571 May be used in sections : defaults | frontend | listen | backend
1572 yes | no | yes | yes
1573 Arguments :
1574 <algorithm> is the algorithm used to select a server when doing load
1575 balancing. This only applies when no persistence information
1576 is available, or when a connection is redispatched to another
1577 server. <algorithm> may be one of the following :
1578
1579 roundrobin Each server is used in turns, according to their weights.
1580 This is the smoothest and fairest algorithm when the server's
1581 processing time remains equally distributed. This algorithm
1582 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02001583 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08001584 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02001585 large farms, when a server becomes up after having been down
1586 for a very short time, it may sometimes take a few hundreds
1587 requests for it to be re-integrated into the farm and start
1588 receiving traffic. This is normal, though very rare. It is
1589 indicated here in case you would have the chance to observe
1590 it, so that you don't worry.
1591
1592 static-rr Each server is used in turns, according to their weights.
1593 This algorithm is as similar to roundrobin except that it is
1594 static, which means that changing a server's weight on the
1595 fly will have no effect. On the other hand, it has no design
1596 limitation on the number of servers, and when a server goes
1597 up, it is always immediately reintroduced into the farm, once
1598 the full map is recomputed. It also uses slightly less CPU to
1599 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01001600
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01001601 leastconn The server with the lowest number of connections receives the
1602 connection. Round-robin is performed within groups of servers
1603 of the same load to ensure that all servers will be used. Use
1604 of this algorithm is recommended where very long sessions are
1605 expected, such as LDAP, SQL, TSE, etc... but is not very well
1606 suited for protocols using short sessions such as HTTP. This
1607 algorithm is dynamic, which means that server weights may be
1608 adjusted on the fly for slow starts for instance.
1609
Willy Tarreauf09c6602012-02-13 17:12:08 +01001610 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001611 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01001612 identifier to the highest (see server parameter "id"), which
1613 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02001614 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01001615 not make sense to use this algorithm without setting maxconn.
1616 The purpose of this algorithm is to always use the smallest
1617 number of servers so that extra servers can be powered off
1618 during non-intensive hours. This algorithm ignores the server
1619 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02001620 or IMAP than HTTP, though it can be useful there too. In
1621 order to use this algorithm efficiently, it is recommended
1622 that a cloud controller regularly checks server usage to turn
1623 them off when unused, and regularly checks backend queue to
1624 turn new servers on when the queue inflates. Alternatively,
1625 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01001626
Willy Tarreau0ba27502007-12-24 16:55:16 +01001627 source The source IP address is hashed and divided by the total
1628 weight of the running servers to designate which server will
1629 receive the request. This ensures that the same client IP
1630 address will always reach the same server as long as no
1631 server goes down or up. If the hash result changes due to the
1632 number of running servers changing, many clients will be
1633 directed to a different server. This algorithm is generally
1634 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001635 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01001636 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001637 static by default, which means that changing a server's
1638 weight on the fly will have no effect, but this can be
1639 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001640
Oskar Stolc8dc41842012-05-19 10:19:54 +01001641 uri This algorithm hashes either the left part of the URI (before
1642 the question mark) or the whole URI (if the "whole" parameter
1643 is present) and divides the hash value by the total weight of
1644 the running servers. The result designates which server will
1645 receive the request. This ensures that the same URI will
1646 always be directed to the same server as long as no server
1647 goes up or down. This is used with proxy caches and
1648 anti-virus proxies in order to maximize the cache hit rate.
1649 Note that this algorithm may only be used in an HTTP backend.
1650 This algorithm is static by default, which means that
1651 changing a server's weight on the fly will have no effect,
1652 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001653
Oskar Stolc8dc41842012-05-19 10:19:54 +01001654 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001655 "depth", both followed by a positive integer number. These
1656 options may be helpful when it is needed to balance servers
1657 based on the beginning of the URI only. The "len" parameter
1658 indicates that the algorithm should only consider that many
1659 characters at the beginning of the URI to compute the hash.
1660 Note that having "len" set to 1 rarely makes sense since most
1661 URIs start with a leading "/".
1662
1663 The "depth" parameter indicates the maximum directory depth
1664 to be used to compute the hash. One level is counted for each
1665 slash in the request. If both parameters are specified, the
1666 evaluation stops when either is reached.
1667
Willy Tarreau0ba27502007-12-24 16:55:16 +01001668 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001669 the query string of each HTTP GET request.
1670
1671 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02001672 request entity will be searched for the parameter argument,
1673 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02001674 ('?') in the URL. The message body will only start to be
1675 analyzed once either the advertised amount of data has been
1676 received or the request buffer is full. In the unlikely event
1677 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02001678 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02001679 be randomly balanced if at all. This keyword used to support
1680 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001681
1682 If the parameter is found followed by an equal sign ('=') and
1683 a value, then the value is hashed and divided by the total
1684 weight of the running servers. The result designates which
1685 server will receive the request.
1686
1687 This is used to track user identifiers in requests and ensure
1688 that a same user ID will always be sent to the same server as
1689 long as no server goes up or down. If no value is found or if
1690 the parameter is not found, then a round robin algorithm is
1691 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001692 backend. This algorithm is static by default, which means
1693 that changing a server's weight on the fly will have no
1694 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001695
Cyril Bontédc4d9032012-04-08 21:57:39 +02001696 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
1697 request. Just as with the equivalent ACL 'hdr()' function,
1698 the header name in parenthesis is not case sensitive. If the
1699 header is absent or if it does not contain any value, the
1700 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01001701
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001702 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01001703 reducing the hash algorithm to the main domain part with some
1704 specific headers such as 'Host'. For instance, in the Host
1705 value "haproxy.1wt.eu", only "1wt" will be considered.
1706
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001707 This algorithm is static by default, which means that
1708 changing a server's weight on the fly will have no effect,
1709 but this can be changed using "hash-type".
1710
Emeric Brun736aa232009-06-30 17:56:00 +02001711 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02001712 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02001713 The RDP cookie <name> (or "mstshash" if omitted) will be
1714 looked up and hashed for each incoming TCP request. Just as
1715 with the equivalent ACL 'req_rdp_cookie()' function, the name
1716 is not case-sensitive. This mechanism is useful as a degraded
1717 persistence mode, as it makes it possible to always send the
1718 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001719 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02001720 used instead.
1721
1722 Note that for this to work, the frontend must ensure that an
1723 RDP cookie is already present in the request buffer. For this
1724 you must use 'tcp-request content accept' rule combined with
1725 a 'req_rdp_cookie_cnt' ACL.
1726
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001727 This algorithm is static by default, which means that
1728 changing a server's weight on the fly will have no effect,
1729 but this can be changed using "hash-type".
1730
Cyril Bontédc4d9032012-04-08 21:57:39 +02001731 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09001732
Willy Tarreau0ba27502007-12-24 16:55:16 +01001733 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02001734 algorithms. Right now, only "url_param" and "uri" support an
1735 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001736
Willy Tarreau3cd9af22009-03-15 14:06:41 +01001737 The load balancing algorithm of a backend is set to roundrobin when no other
1738 algorithm, mode nor option have been set. The algorithm may only be set once
1739 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001740
1741 Examples :
1742 balance roundrobin
1743 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001744 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01001745 balance hdr(User-Agent)
1746 balance hdr(host)
1747 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001748
1749 Note: the following caveats and limitations on using the "check_post"
1750 extension with "url_param" must be considered :
1751
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001752 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001753 to determine if the parameters will be found in the body or entity which
1754 may contain binary data. Therefore another method may be required to
1755 restrict consideration of POST requests that have no URL parameters in
1756 the body. (see acl reqideny http_end)
1757
1758 - using a <max_wait> value larger than the request buffer size does not
1759 make sense and is useless. The buffer size is set at build time, and
1760 defaults to 16 kB.
1761
1762 - Content-Encoding is not supported, the parameter search will probably
1763 fail; and load balancing will fall back to Round Robin.
1764
1765 - Expect: 100-continue is not supported, load balancing will fall back to
1766 Round Robin.
1767
1768 - Transfer-Encoding (RFC2616 3.6.1) is only supported in the first chunk.
1769 If the entire parameter value is not present in the first chunk, the
1770 selection of server is undefined (actually, defined by how little
1771 actually appeared in the first chunk).
1772
1773 - This feature does not support generation of a 100, 411 or 501 response.
1774
1775 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001776 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02001777 white space or control characters are found, indicating the end of what
1778 might be a URL parameter list. This is probably not a concern with SGML
1779 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001780
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02001781 See also : "dispatch", "cookie", "appsession", "transparent", "hash-type" and
1782 "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001783
1784
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001785bind [<address>]:<port_range> [, ...] [param*]
1786bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01001787 Define one or several listening addresses and/or ports in a frontend.
1788 May be used in sections : defaults | frontend | listen | backend
1789 no | yes | yes | no
1790 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001791 <address> is optional and can be a host name, an IPv4 address, an IPv6
1792 address, or '*'. It designates the address the frontend will
1793 listen on. If unset, all IPv4 addresses of the system will be
1794 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01001795 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01001796 Optionally, an address family prefix may be used before the
1797 address to force the family regardless of the address format,
1798 which can be useful to specify a path to a unix socket with
1799 no slash ('/'). Currently supported prefixes are :
1800 - 'ipv4@' -> address is always IPv4
1801 - 'ipv6@' -> address is always IPv6
1802 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02001803 - 'abns@' -> address is in abstract namespace (Linux only).
1804 Note: since abstract sockets are not "rebindable", they
1805 do not cope well with multi-process mode during
1806 soft-restart, so it is better to avoid them if
1807 nbproc is greater than 1. The effect is that if the
1808 new process fails to start, only one of the old ones
1809 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01001810 - 'fd@<n>' -> use file descriptor <n> inherited from the
1811 parent. The fd must be bound and may or may not already
1812 be listening.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001813 Any part of the address string may reference any number of
1814 environment variables by preceding their name with a dollar
1815 sign ('$') and optionally enclosing them with braces ('{}'),
1816 similarly to what is done in Bourne shell.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01001817
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001818 <port_range> is either a unique TCP port, or a port range for which the
1819 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001820 above. The port is mandatory for TCP listeners. Note that in
1821 the case of an IPv6 address, the port is always the number
1822 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001823 - a numerical port (ex: '80')
1824 - a dash-delimited ports range explicitly stating the lower
1825 and upper bounds (ex: '2000-2100') which are included in
1826 the range.
1827
1828 Particular care must be taken against port ranges, because
1829 every <address:port> couple consumes one socket (= a file
1830 descriptor), so it's easy to consume lots of descriptors
1831 with a simple range, and to run out of sockets. Also, each
1832 <address:port> couple must be used only once among all
1833 instances running on a same system. Please note that binding
1834 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001835 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01001836 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001837
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001838 <path> is a UNIX socket path beginning with a slash ('/'). This is
1839 alternative to the TCP listening port. Haproxy will then
1840 receive UNIX connections on the socket located at this place.
1841 The path must begin with a slash and by default is absolute.
1842 It can be relative to the prefix defined by "unix-bind" in
1843 the global section. Note that the total length of the prefix
1844 followed by the socket path cannot exceed some system limits
1845 for UNIX sockets, which commonly are set to 107 characters.
1846
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001847 <param*> is a list of parameters common to all sockets declared on the
1848 same line. These numerous parameters depend on OS and build
1849 options and have a complete section dedicated to them. Please
1850 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001851
Willy Tarreau0ba27502007-12-24 16:55:16 +01001852 It is possible to specify a list of address:port combinations delimited by
1853 commas. The frontend will then listen on all of these addresses. There is no
1854 fixed limit to the number of addresses and ports which can be listened on in
1855 a frontend, as well as there is no limit to the number of "bind" statements
1856 in a frontend.
1857
1858 Example :
1859 listen http_proxy
1860 bind :80,:443
1861 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001862 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01001863
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001864 listen http_https_proxy
1865 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02001866 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02001867
Willy Tarreau24709282013-03-10 21:32:12 +01001868 listen http_https_proxy_explicit
1869 bind ipv6@:80
1870 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
1871 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
1872
Willy Tarreaudad36a32013-03-11 01:20:04 +01001873 listen external_bind_app1
1874 bind fd@${FD_APP1}
1875
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001876 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02001877 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001878
1879
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001880bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001881 Limit visibility of an instance to a certain set of processes numbers.
1882 May be used in sections : defaults | frontend | listen | backend
1883 yes | yes | yes | yes
1884 Arguments :
1885 all All process will see this instance. This is the default. It
1886 may be used to override a default value.
1887
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001888 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001889 option may be combined with other numbers.
1890
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001891 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001892 option may be combined with other numbers. Do not use it
1893 with less than 2 processes otherwise some instances might be
1894 missing from all processes.
1895
Willy Tarreau110ecc12012-11-15 17:50:01 +01001896 number The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001897 whose values must all be between 1 and 32 or 64 depending on
Willy Tarreau102df612014-05-07 23:56:38 +02001898 the machine's word size. If a proxy is bound to process
1899 numbers greater than the configured global.nbproc, it will
1900 either be forced to process #1 if a single process was
1901 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001902
1903 This keyword limits binding of certain instances to certain processes. This
1904 is useful in order not to have too many processes listening to the same
1905 ports. For instance, on a dual-core machine, it might make sense to set
1906 'nbproc 2' in the global section, then distributes the listeners among 'odd'
1907 and 'even' instances.
1908
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001909 At the moment, it is not possible to reference more than 32 or 64 processes
1910 using this keyword, but this should be more than enough for most setups.
1911 Please note that 'all' really means all processes regardless of the machine's
1912 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001913
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001914 Each "bind" line may further be limited to a subset of the proxy's processes,
1915 please consult the "process" bind keyword in section 5.1.
1916
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001917 If some backends are referenced by frontends bound to other processes, the
1918 backend automatically inherits the frontend's processes.
1919
1920 Example :
1921 listen app_ip1
1922 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001923 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001924
1925 listen app_ip2
1926 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001927 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001928
1929 listen management
1930 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02001931 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001932
Willy Tarreau110ecc12012-11-15 17:50:01 +01001933 listen management
1934 bind 10.0.0.4:80
1935 bind-process 1-4
1936
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02001937 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01001938
1939
Willy Tarreau0ba27502007-12-24 16:55:16 +01001940block { if | unless } <condition>
1941 Block a layer 7 request if/unless a condition is matched
1942 May be used in sections : defaults | frontend | listen | backend
1943 no | yes | yes | yes
1944
1945 The HTTP request will be blocked very early in the layer 7 processing
1946 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001947 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02001948 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01001949 conditions are met or not met. There is no fixed limit to the number of
1950 "block" statements per instance.
1951
1952 Example:
1953 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
1954 acl invalid_src src_port 0:1023
1955 acl local_dst hdr(host) -i localhost
1956 block if invalid_src || local_dst
1957
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001958 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001959
1960
1961capture cookie <name> len <length>
1962 Capture and log a cookie in the request and in the response.
1963 May be used in sections : defaults | frontend | listen | backend
1964 no | yes | yes | no
1965 Arguments :
1966 <name> is the beginning of the name of the cookie to capture. In order
1967 to match the exact name, simply suffix the name with an equal
1968 sign ('='). The full name will appear in the logs, which is
1969 useful with application servers which adjust both the cookie name
1970 and value (eg: ASPSESSIONXXXXX).
1971
1972 <length> is the maximum number of characters to report in the logs, which
1973 include the cookie name, the equal sign and the value, all in the
1974 standard "name=value" form. The string will be truncated on the
1975 right if it exceeds <length>.
1976
1977 Only the first cookie is captured. Both the "cookie" request headers and the
1978 "set-cookie" response headers are monitored. This is particularly useful to
1979 check for application bugs causing session crossing or stealing between
1980 users, because generally the user's cookies can only change on a login page.
1981
1982 When the cookie was not presented by the client, the associated log column
1983 will report "-". When a request does not cause a cookie to be assigned by the
1984 server, a "-" is reported in the response column.
1985
1986 The capture is performed in the frontend only because it is necessary that
1987 the log format does not change for a given frontend depending on the
1988 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01001989 "capture cookie" statement in a frontend. The maximum capture length is set
1990 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
1991 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001992
1993 Example:
1994 capture cookie ASPSESSION len 32
1995
1996 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001997 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001998
1999
2000capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002001 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002002 May be used in sections : defaults | frontend | listen | backend
2003 no | yes | yes | no
2004 Arguments :
2005 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002006 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002007 appear in the requests, with the first letter of each word in
2008 upper case. The header name will not appear in the logs, only the
2009 value is reported, but the position in the logs is respected.
2010
2011 <length> is the maximum number of characters to extract from the value and
2012 report in the logs. The string will be truncated on the right if
2013 it exceeds <length>.
2014
Willy Tarreau4460d032012-11-21 23:37:37 +01002015 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002016 value will be added to the logs between braces ('{}'). If multiple headers
2017 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002018 in the same order they were declared in the configuration. Non-existent
2019 headers will be logged just as an empty string. Common uses for request
2020 header captures include the "Host" field in virtual hosting environments, the
2021 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002022 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002023 environments to find where the request came from.
2024
2025 Note that when capturing headers such as "User-agent", some spaces may be
2026 logged, making the log analysis more difficult. Thus be careful about what
2027 you log if you know your log parser is not smart enough to rely on the
2028 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002029
Willy Tarreau0900abb2012-11-22 00:21:46 +01002030 There is no limit to the number of captured request headers nor to their
2031 length, though it is wise to keep them low to limit memory usage per session.
2032 In order to keep log format consistent for a same frontend, header captures
2033 can only be declared in a frontend. It is not possible to specify a capture
2034 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002035
2036 Example:
2037 capture request header Host len 15
2038 capture request header X-Forwarded-For len 15
2039 capture request header Referrer len 15
2040
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002041 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002042 about logging.
2043
2044
2045capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002046 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002047 May be used in sections : defaults | frontend | listen | backend
2048 no | yes | yes | no
2049 Arguments :
2050 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002051 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002052 appear in the response, with the first letter of each word in
2053 upper case. The header name will not appear in the logs, only the
2054 value is reported, but the position in the logs is respected.
2055
2056 <length> is the maximum number of characters to extract from the value and
2057 report in the logs. The string will be truncated on the right if
2058 it exceeds <length>.
2059
Willy Tarreau4460d032012-11-21 23:37:37 +01002060 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002061 result will be added to the logs between braces ('{}') after the captured
2062 request headers. If multiple headers are captured, they will be delimited by
2063 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002064 the configuration. Non-existent headers will be logged just as an empty
2065 string. Common uses for response header captures include the "Content-length"
2066 header which indicates how many bytes are expected to be returned, the
2067 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002068
Willy Tarreau0900abb2012-11-22 00:21:46 +01002069 There is no limit to the number of captured response headers nor to their
2070 length, though it is wise to keep them low to limit memory usage per session.
2071 In order to keep log format consistent for a same frontend, header captures
2072 can only be declared in a frontend. It is not possible to specify a capture
2073 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002074
2075 Example:
2076 capture response header Content-length len 9
2077 capture response header Location len 15
2078
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002079 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002080 about logging.
2081
2082
Cyril Bontéf0c60612010-02-06 14:44:47 +01002083clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002084 Set the maximum inactivity time on the client side.
2085 May be used in sections : defaults | frontend | listen | backend
2086 yes | yes | yes | no
2087 Arguments :
2088 <timeout> is the timeout value is specified in milliseconds by default, but
2089 can be in any other unit if the number is suffixed by the unit,
2090 as explained at the top of this document.
2091
2092 The inactivity timeout applies when the client is expected to acknowledge or
2093 send data. In HTTP mode, this timeout is particularly important to consider
2094 during the first phase, when the client sends the request, and during the
2095 response while it is reading data sent by the server. The value is specified
2096 in milliseconds by default, but can be in any other unit if the number is
2097 suffixed by the unit, as specified at the top of this document. In TCP mode
2098 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2099 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002100 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002101 losses by specifying timeouts that are slightly above multiples of 3 seconds
2102 (eg: 4 or 5 seconds).
2103
2104 This parameter is specific to frontends, but can be specified once for all in
2105 "defaults" sections. This is in fact one of the easiest solutions not to
2106 forget about it. An unspecified timeout results in an infinite timeout, which
2107 is not recommended. Such a usage is accepted and works but reports a warning
2108 during startup because it may results in accumulation of expired sessions in
2109 the system if the system's timeouts are not configured either.
2110
2111 This parameter is provided for compatibility but is currently deprecated.
2112 Please use "timeout client" instead.
2113
Willy Tarreau036fae02008-01-06 13:24:40 +01002114 See also : "timeout client", "timeout http-request", "timeout server", and
2115 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002116
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002117compression algo <algorithm> ...
2118compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002119compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002120 Enable HTTP compression.
2121 May be used in sections : defaults | frontend | listen | backend
2122 yes | yes | yes | yes
2123 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002124 algo is followed by the list of supported compression algorithms.
2125 type is followed by the list of MIME types that will be compressed.
2126 offload makes haproxy work as a compression offloader only (see notes).
2127
2128 The currently supported algorithms are :
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002129 identity this is mostly for debugging, and it was useful for developing
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002130 the compression feature. Identity does not apply any change on
2131 data.
2132
2133 gzip applies gzip compression. This setting is only available when
2134 support for zlib was built in.
2135
2136 deflate same as gzip, but with deflate algorithm and zlib format.
2137 Note that this algorithm has ambiguous support on many browsers
2138 and no support at all from recent ones. It is strongly
2139 recommended not to use it for anything else than experimentation.
2140 This setting is only available when support for zlib was built
2141 in.
2142
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002143 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002144 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002145 If backend servers support HTTP compression, these directives
2146 will be no-op: haproxy will see the compressed response and will not
2147 compress again. If backend servers do not support HTTP compression and
2148 there is Accept-Encoding header in request, haproxy will compress the
2149 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002150
2151 The "offload" setting makes haproxy remove the Accept-Encoding header to
2152 prevent backend servers from compressing responses. It is strongly
2153 recommended not to do this because this means that all the compression work
2154 will be done on the single point where haproxy is located. However in some
2155 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002156 with broken HTTP compression implementation which can't be turned off.
2157 In that case haproxy can be used to prevent that gateway from emitting
2158 invalid payloads. In this case, simply removing the header in the
2159 configuration does not work because it applies before the header is parsed,
2160 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02002161 then be used for such scenarios. Note: for now, the "offload" setting is
2162 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002163
William Lallemand05097442012-11-20 12:14:28 +01002164 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002165 * the request does not advertise a supported compression algorithm in the
2166 "Accept-Encoding" header
2167 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002168 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002169 * response header "Transfer-Encoding" contains "chunked" (Temporary
2170 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002171 * response contain neither a "Content-Length" header nor a
2172 "Transfer-Encoding" whose last value is "chunked"
2173 * response contains a "Content-Type" header whose first value starts with
2174 "multipart"
2175 * the response contains the "no-transform" value in the "Cache-control"
2176 header
2177 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2178 and later
2179 * The response contains a "Content-Encoding" header, indicating that the
2180 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002181
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002182 Note: The compression does not rewrite Etag headers, and does not emit the
2183 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002184
William Lallemand82fe75c2012-10-23 10:25:10 +02002185 Examples :
2186 compression algo gzip
2187 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002188
Cyril Bontéf0c60612010-02-06 14:44:47 +01002189contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002190 Set the maximum time to wait for a connection attempt to a server to succeed.
2191 May be used in sections : defaults | frontend | listen | backend
2192 yes | no | yes | yes
2193 Arguments :
2194 <timeout> is the timeout value is specified in milliseconds by default, but
2195 can be in any other unit if the number is suffixed by the unit,
2196 as explained at the top of this document.
2197
2198 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002199 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002200 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01002201 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
2202 connect timeout also presets the queue timeout to the same value if this one
2203 has not been specified. Historically, the contimeout was also used to set the
2204 tarpit timeout in a listen section, which is not possible in a pure frontend.
2205
2206 This parameter is specific to backends, but can be specified once for all in
2207 "defaults" sections. This is in fact one of the easiest solutions not to
2208 forget about it. An unspecified timeout results in an infinite timeout, which
2209 is not recommended. Such a usage is accepted and works but reports a warning
2210 during startup because it may results in accumulation of failed sessions in
2211 the system if the system's timeouts are not configured either.
2212
2213 This parameter is provided for backwards compatibility but is currently
2214 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2215 instead.
2216
2217 See also : "timeout connect", "timeout queue", "timeout tarpit",
2218 "timeout server", "contimeout".
2219
2220
Willy Tarreau55165fe2009-05-10 12:02:55 +02002221cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002222 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2223 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002224 Enable cookie-based persistence in a backend.
2225 May be used in sections : defaults | frontend | listen | backend
2226 yes | no | yes | yes
2227 Arguments :
2228 <name> is the name of the cookie which will be monitored, modified or
2229 inserted in order to bring persistence. This cookie is sent to
2230 the client via a "Set-Cookie" header in the response, and is
2231 brought back by the client in a "Cookie" header in all requests.
2232 Special care should be taken to choose a name which does not
2233 conflict with any likely application cookie. Also, if the same
2234 backends are subject to be used by the same clients (eg:
2235 HTTP/HTTPS), care should be taken to use different cookie names
2236 between all backends if persistence between them is not desired.
2237
2238 rewrite This keyword indicates that the cookie will be provided by the
2239 server and that haproxy will have to modify its value to set the
2240 server's identifier in it. This mode is handy when the management
2241 of complex combinations of "Set-cookie" and "Cache-control"
2242 headers is left to the application. The application can then
2243 decide whether or not it is appropriate to emit a persistence
2244 cookie. Since all responses should be monitored, this mode only
2245 works in HTTP close mode. Unless the application behaviour is
2246 very complex and/or broken, it is advised not to start with this
2247 mode for new deployments. This keyword is incompatible with
2248 "insert" and "prefix".
2249
2250 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02002251 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002252
Willy Tarreaua79094d2010-08-31 22:54:15 +02002253 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002254 server. When used without the "preserve" option, if the server
2255 emits a cookie with the same name, it will be remove before
2256 processing. For this reason, this mode can be used to upgrade
2257 existing configurations running in the "rewrite" mode. The cookie
2258 will only be a session cookie and will not be stored on the
2259 client's disk. By default, unless the "indirect" option is added,
2260 the server will see the cookies emitted by the client. Due to
2261 caching effects, it is generally wise to add the "nocache" or
2262 "postonly" keywords (see below). The "insert" keyword is not
2263 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002264
2265 prefix This keyword indicates that instead of relying on a dedicated
2266 cookie for the persistence, an existing one will be completed.
2267 This may be needed in some specific environments where the client
2268 does not support more than one single cookie and the application
2269 already needs it. In this case, whenever the server sets a cookie
2270 named <name>, it will be prefixed with the server's identifier
2271 and a delimiter. The prefix will be removed from all client
2272 requests so that the server still finds the cookie it emitted.
2273 Since all requests and responses are subject to being modified,
2274 this mode requires the HTTP close mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02002275 not compatible with "rewrite" and "insert". Note: it is highly
2276 recommended not to use "indirect" with "prefix", otherwise server
2277 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002278
Willy Tarreaua79094d2010-08-31 22:54:15 +02002279 indirect When this option is specified, no cookie will be emitted to a
2280 client which already has a valid one for the server which has
2281 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002282 it will be removed, unless the "preserve" option is also set. In
2283 "insert" mode, this will additionally remove cookies from the
2284 requests transmitted to the server, making the persistence
2285 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02002286 Note: it is highly recommended not to use "indirect" with
2287 "prefix", otherwise server cookie updates would not be sent to
2288 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002289
2290 nocache This option is recommended in conjunction with the insert mode
2291 when there is a cache between the client and HAProxy, as it
2292 ensures that a cacheable response will be tagged non-cacheable if
2293 a cookie needs to be inserted. This is important because if all
2294 persistence cookies are added on a cacheable home page for
2295 instance, then all customers will then fetch the page from an
2296 outer cache and will all share the same persistence cookie,
2297 leading to one server receiving much more traffic than others.
2298 See also the "insert" and "postonly" options.
2299
2300 postonly This option ensures that cookie insertion will only be performed
2301 on responses to POST requests. It is an alternative to the
2302 "nocache" option, because POST responses are not cacheable, so
2303 this ensures that the persistence cookie will never get cached.
2304 Since most sites do not need any sort of persistence before the
2305 first POST which generally is a login request, this is a very
2306 efficient method to optimize caching without risking to find a
2307 persistence cookie in the cache.
2308 See also the "insert" and "nocache" options.
2309
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002310 preserve This option may only be used with "insert" and/or "indirect". It
2311 allows the server to emit the persistence cookie itself. In this
2312 case, if a cookie is found in the response, haproxy will leave it
2313 untouched. This is useful in order to end persistence after a
2314 logout request for instance. For this, the server just has to
2315 emit a cookie with an invalid value (eg: empty) or with a date in
2316 the past. By combining this mechanism with the "disable-on-404"
2317 check option, it is possible to perform a completely graceful
2318 shutdown because users will definitely leave the server after
2319 they logout.
2320
Willy Tarreau4992dd22012-05-31 21:02:17 +02002321 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
2322 when a cookie is inserted. This attribute is used so that a
2323 user agent doesn't share the cookie with non-HTTP components.
2324 Please check RFC6265 for more information on this attribute.
2325
2326 secure This option tells haproxy to add a "Secure" cookie attribute when
2327 a cookie is inserted. This attribute is used so that a user agent
2328 never emits this cookie over non-secure channels, which means
2329 that a cookie learned with this flag will be presented only over
2330 SSL/TLS connections. Please check RFC6265 for more information on
2331 this attribute.
2332
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002333 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002334 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01002335 name. If the domain begins with a dot, the browser is allowed to
2336 use it for any host ending with that name. It is also possible to
2337 specify several domain names by invoking this option multiple
2338 times. Some browsers might have small limits on the number of
2339 domains, so be careful when doing that. For the record, sending
2340 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002341
Willy Tarreau996a92c2010-10-13 19:30:47 +02002342 maxidle This option allows inserted cookies to be ignored after some idle
2343 time. It only works with insert-mode cookies. When a cookie is
2344 sent to the client, the date this cookie was emitted is sent too.
2345 Upon further presentations of this cookie, if the date is older
2346 than the delay indicated by the parameter (in seconds), it will
2347 be ignored. Otherwise, it will be refreshed if needed when the
2348 response is sent to the client. This is particularly useful to
2349 prevent users who never close their browsers from remaining for
2350 too long on the same server (eg: after a farm size change). When
2351 this option is set and a cookie has no date, it is always
2352 accepted, but gets refreshed in the response. This maintains the
2353 ability for admins to access their sites. Cookies that have a
2354 date in the future further than 24 hours are ignored. Doing so
2355 lets admins fix timezone issues without risking kicking users off
2356 the site.
2357
2358 maxlife This option allows inserted cookies to be ignored after some life
2359 time, whether they're in use or not. It only works with insert
2360 mode cookies. When a cookie is first sent to the client, the date
2361 this cookie was emitted is sent too. Upon further presentations
2362 of this cookie, if the date is older than the delay indicated by
2363 the parameter (in seconds), it will be ignored. If the cookie in
2364 the request has no date, it is accepted and a date will be set.
2365 Cookies that have a date in the future further than 24 hours are
2366 ignored. Doing so lets admins fix timezone issues without risking
2367 kicking users off the site. Contrary to maxidle, this value is
2368 not refreshed, only the first visit date counts. Both maxidle and
2369 maxlife may be used at the time. This is particularly useful to
2370 prevent users who never close their browsers from remaining for
2371 too long on the same server (eg: after a farm size change). This
2372 is stronger than the maxidle method in that it forces a
2373 redispatch after some absolute delay.
2374
Willy Tarreau0ba27502007-12-24 16:55:16 +01002375 There can be only one persistence cookie per HTTP backend, and it can be
2376 declared in a defaults section. The value of the cookie will be the value
2377 indicated after the "cookie" keyword in a "server" statement. If no cookie
2378 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002379
Willy Tarreau0ba27502007-12-24 16:55:16 +01002380 Examples :
2381 cookie JSESSIONID prefix
2382 cookie SRV insert indirect nocache
2383 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02002384 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01002385
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002386 See also : "appsession", "balance source", "capture cookie", "server"
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002387 and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002388
Willy Tarreau983e01e2010-01-11 18:42:06 +01002389
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002390default-server [param*]
2391 Change default options for a server in a backend
2392 May be used in sections : defaults | frontend | listen | backend
2393 yes | no | yes | yes
2394 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01002395 <param*> is a list of parameters for this server. The "default-server"
2396 keyword accepts an important number of options and has a complete
2397 section dedicated to it. Please refer to section 5 for more
2398 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002399
Willy Tarreau983e01e2010-01-11 18:42:06 +01002400 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01002401 default-server inter 1000 weight 13
2402
2403 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01002404
Willy Tarreau983e01e2010-01-11 18:42:06 +01002405
Willy Tarreau0ba27502007-12-24 16:55:16 +01002406default_backend <backend>
2407 Specify the backend to use when no "use_backend" rule has been matched.
2408 May be used in sections : defaults | frontend | listen | backend
2409 yes | yes | yes | no
2410 Arguments :
2411 <backend> is the name of the backend to use.
2412
2413 When doing content-switching between frontend and backends using the
2414 "use_backend" keyword, it is often useful to indicate which backend will be
2415 used when no rule has matched. It generally is the dynamic backend which
2416 will catch all undetermined requests.
2417
Willy Tarreau0ba27502007-12-24 16:55:16 +01002418 Example :
2419
2420 use_backend dynamic if url_dyn
2421 use_backend static if url_css url_img extension_img
2422 default_backend dynamic
2423
Willy Tarreau2769aa02007-12-27 18:26:09 +01002424 See also : "use_backend", "reqsetbe", "reqisetbe"
2425
Willy Tarreau0ba27502007-12-24 16:55:16 +01002426
Baptiste Assmann27f51342013-10-09 06:51:49 +02002427description <string>
2428 Describe a listen, frontend or backend.
2429 May be used in sections : defaults | frontend | listen | backend
2430 no | yes | yes | yes
2431 Arguments : string
2432
2433 Allows to add a sentence to describe the related object in the HAProxy HTML
2434 stats page. The description will be printed on the right of the object name
2435 it describes.
2436 No need to backslash spaces in the <string> arguments.
2437
2438
Willy Tarreau0ba27502007-12-24 16:55:16 +01002439disabled
2440 Disable a proxy, frontend or backend.
2441 May be used in sections : defaults | frontend | listen | backend
2442 yes | yes | yes | yes
2443 Arguments : none
2444
2445 The "disabled" keyword is used to disable an instance, mainly in order to
2446 liberate a listening port or to temporarily disable a service. The instance
2447 will still be created and its configuration will be checked, but it will be
2448 created in the "stopped" state and will appear as such in the statistics. It
2449 will not receive any traffic nor will it send any health-checks or logs. It
2450 is possible to disable many instances at once by adding the "disabled"
2451 keyword in a "defaults" section.
2452
2453 See also : "enabled"
2454
2455
Willy Tarreau5ce94572010-06-07 14:35:41 +02002456dispatch <address>:<port>
2457 Set a default server address
2458 May be used in sections : defaults | frontend | listen | backend
2459 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02002460 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02002461
2462 <address> is the IPv4 address of the default server. Alternatively, a
2463 resolvable hostname is supported, but this name will be resolved
2464 during start-up.
2465
2466 <ports> is a mandatory port specification. All connections will be sent
2467 to this port, and it is not permitted to use port offsets as is
2468 possible with normal servers.
2469
Willy Tarreau787aed52011-04-15 06:45:37 +02002470 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02002471 server can take the connection. In the past it was used to forward non
2472 persistent connections to an auxiliary load balancer. Due to its simple
2473 syntax, it has also been used for simple TCP relays. It is recommended not to
2474 use it for more clarity, and to use the "server" directive instead.
2475
2476 See also : "server"
2477
2478
Willy Tarreau0ba27502007-12-24 16:55:16 +01002479enabled
2480 Enable a proxy, frontend or backend.
2481 May be used in sections : defaults | frontend | listen | backend
2482 yes | yes | yes | yes
2483 Arguments : none
2484
2485 The "enabled" keyword is used to explicitly enable an instance, when the
2486 defaults has been set to "disabled". This is very rarely used.
2487
2488 See also : "disabled"
2489
2490
2491errorfile <code> <file>
2492 Return a file contents instead of errors generated by HAProxy
2493 May be used in sections : defaults | frontend | listen | backend
2494 yes | yes | yes | yes
2495 Arguments :
2496 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002497 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002498
2499 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002500 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01002501 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01002502 error pages, and to use absolute paths, since files are read
2503 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002504
2505 It is important to understand that this keyword is not meant to rewrite
2506 errors returned by the server, but errors detected and returned by HAProxy.
2507 This is why the list of supported errors is limited to a small set.
2508
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002509 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2510
Willy Tarreau0ba27502007-12-24 16:55:16 +01002511 The files are returned verbatim on the TCP socket. This allows any trick such
2512 as redirections to another URL or site, as well as tricks to clean cookies,
2513 force enable or disable caching, etc... The package provides default error
2514 files returning the same contents as default errors.
2515
Willy Tarreau59140a22009-02-22 12:02:30 +01002516 The files should not exceed the configured buffer size (BUFSIZE), which
2517 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
2518 not to put any reference to local contents (eg: images) in order to avoid
2519 loops between the client and HAProxy when all servers are down, causing an
2520 error to be returned instead of an image. For better HTTP compliance, it is
2521 recommended that all header lines end with CR-LF and not LF alone.
2522
Willy Tarreau0ba27502007-12-24 16:55:16 +01002523 The files are read at the same time as the configuration and kept in memory.
2524 For this reason, the errors continue to be returned even when the process is
2525 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01002526 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002527 403 status code and interrogating a blocked URL.
2528
2529 See also : "errorloc", "errorloc302", "errorloc303"
2530
Willy Tarreau59140a22009-02-22 12:02:30 +01002531 Example :
2532 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau2705a612014-05-23 17:38:34 +02002533 errorfile 408 /dev/null # workaround Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01002534 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
2535 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
2536
Willy Tarreau2769aa02007-12-27 18:26:09 +01002537
2538errorloc <code> <url>
2539errorloc302 <code> <url>
2540 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2541 May be used in sections : defaults | frontend | listen | backend
2542 yes | yes | yes | yes
2543 Arguments :
2544 <code> is the HTTP status code. Currently, HAProxy is capable of
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002545 generating codes 200, 400, 403, 408, 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002546
2547 <url> it is the exact contents of the "Location" header. It may contain
2548 either a relative URI to an error page hosted on the same site,
2549 or an absolute URI designating an error page on another site.
2550 Special care should be given to relative URIs to avoid redirect
2551 loops if the URI itself may generate the same error (eg: 500).
2552
2553 It is important to understand that this keyword is not meant to rewrite
2554 errors returned by the server, but errors detected and returned by HAProxy.
2555 This is why the list of supported errors is limited to a small set.
2556
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002557 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2558
Willy Tarreau2769aa02007-12-27 18:26:09 +01002559 Note that both keyword return the HTTP 302 status code, which tells the
2560 client to fetch the designated URL using the same HTTP method. This can be
2561 quite problematic in case of non-GET methods such as POST, because the URL
2562 sent to the client might not be allowed for something other than GET. To
2563 workaround this problem, please use "errorloc303" which send the HTTP 303
2564 status code, indicating to the client that the URL must be fetched with a GET
2565 request.
2566
2567 See also : "errorfile", "errorloc303"
2568
2569
2570errorloc303 <code> <url>
2571 Return an HTTP redirection to a URL instead of errors generated by HAProxy
2572 May be used in sections : defaults | frontend | listen | backend
2573 yes | yes | yes | yes
2574 Arguments :
2575 <code> is the HTTP status code. Currently, HAProxy is capable of
2576 generating codes 400, 403, 408, 500, 502, 503, and 504.
2577
2578 <url> it is the exact contents of the "Location" header. It may contain
2579 either a relative URI to an error page hosted on the same site,
2580 or an absolute URI designating an error page on another site.
2581 Special care should be given to relative URIs to avoid redirect
2582 loops if the URI itself may generate the same error (eg: 500).
2583
2584 It is important to understand that this keyword is not meant to rewrite
2585 errors returned by the server, but errors detected and returned by HAProxy.
2586 This is why the list of supported errors is limited to a small set.
2587
Willy Tarreauae94d4d2011-05-11 16:28:49 +02002588 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
2589
Willy Tarreau2769aa02007-12-27 18:26:09 +01002590 Note that both keyword return the HTTP 303 status code, which tells the
2591 client to fetch the designated URL using the same HTTP GET method. This
2592 solves the usual problems associated with "errorloc" and the 302 code. It is
2593 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002594 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002595
2596 See also : "errorfile", "errorloc", "errorloc302"
2597
2598
Willy Tarreau4de91492010-01-22 19:10:05 +01002599force-persist { if | unless } <condition>
2600 Declare a condition to force persistence on down servers
2601 May be used in sections: defaults | frontend | listen | backend
2602 no | yes | yes | yes
2603
2604 By default, requests are not dispatched to down servers. It is possible to
2605 force this using "option persist", but it is unconditional and redispatches
2606 to a valid server if "option redispatch" is set. That leaves with very little
2607 possibilities to force some requests to reach a server which is artificially
2608 marked down for maintenance operations.
2609
2610 The "force-persist" statement allows one to declare various ACL-based
2611 conditions which, when met, will cause a request to ignore the down status of
2612 a server and still try to connect to it. That makes it possible to start a
2613 server, still replying an error to the health checks, and run a specially
2614 configured browser to test the service. Among the handy methods, one could
2615 use a specific source IP address, or a specific cookie. The cookie also has
2616 the advantage that it can easily be added/removed on the browser from a test
2617 page. Once the service is validated, it is then possible to open the service
2618 to the world by returning a valid response to health checks.
2619
2620 The forced persistence is enabled when an "if" condition is met, or unless an
2621 "unless" condition is met. The final redispatch is always disabled when this
2622 is used.
2623
Cyril Bonté0d4bf012010-04-25 23:21:46 +02002624 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02002625 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01002626
2627
Willy Tarreau2769aa02007-12-27 18:26:09 +01002628fullconn <conns>
2629 Specify at what backend load the servers will reach their maxconn
2630 May be used in sections : defaults | frontend | listen | backend
2631 yes | no | yes | yes
2632 Arguments :
2633 <conns> is the number of connections on the backend which will make the
2634 servers use the maximal number of connections.
2635
Willy Tarreau198a7442008-01-17 12:05:32 +01002636 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01002637 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01002638 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01002639 load. The server will then always accept at least <minconn> connections,
2640 never more than <maxconn>, and the limit will be on the ramp between both
2641 values when the backend has less than <conns> concurrent connections. This
2642 makes it possible to limit the load on the servers during normal loads, but
2643 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002644 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01002645
Willy Tarreaufbb78422011-06-05 15:38:35 +02002646 Since it's hard to get this value right, haproxy automatically sets it to
2647 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01002648 backend (based on "use_backend" and "default_backend" rules). That way it's
2649 safe to leave it unset. However, "use_backend" involving dynamic names are
2650 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02002651
Willy Tarreau2769aa02007-12-27 18:26:09 +01002652 Example :
2653 # The servers will accept between 100 and 1000 concurrent connections each
2654 # and the maximum of 1000 will be reached when the backend reaches 10000
2655 # connections.
2656 backend dynamic
2657 fullconn 10000
2658 server srv1 dyn1:80 minconn 100 maxconn 1000
2659 server srv2 dyn2:80 minconn 100 maxconn 1000
2660
2661 See also : "maxconn", "server"
2662
2663
2664grace <time>
2665 Maintain a proxy operational for some time after a soft stop
2666 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01002667 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01002668 Arguments :
2669 <time> is the time (by default in milliseconds) for which the instance
2670 will remain operational with the frontend sockets still listening
2671 when a soft-stop is received via the SIGUSR1 signal.
2672
2673 This may be used to ensure that the services disappear in a certain order.
2674 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002675 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01002676 needed by the equipment to detect the failure.
2677
2678 Note that currently, there is very little benefit in using this parameter,
2679 and it may in fact complicate the soft-reconfiguration process more than
2680 simplify it.
2681
Willy Tarreau0ba27502007-12-24 16:55:16 +01002682
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002683hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002684 Specify a method to use for mapping hashes to servers
2685 May be used in sections : defaults | frontend | listen | backend
2686 yes | no | yes | yes
2687 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04002688 <method> is the method used to select a server from the hash computed by
2689 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002690
Bhaskar98634f02013-10-29 23:30:51 -04002691 map-based the hash table is a static array containing all alive servers.
2692 The hashes will be very smooth, will consider weights, but
2693 will be static in that weight changes while a server is up
2694 will be ignored. This means that there will be no slow start.
2695 Also, since a server is selected by its position in the array,
2696 most mappings are changed when the server count changes. This
2697 means that when a server goes up or down, or when a server is
2698 added to a farm, most connections will be redistributed to
2699 different servers. This can be inconvenient with caches for
2700 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01002701
Bhaskar98634f02013-10-29 23:30:51 -04002702 consistent the hash table is a tree filled with many occurrences of each
2703 server. The hash key is looked up in the tree and the closest
2704 server is chosen. This hash is dynamic, it supports changing
2705 weights while the servers are up, so it is compatible with the
2706 slow start feature. It has the advantage that when a server
2707 goes up or down, only its associations are moved. When a
2708 server is added to the farm, only a few part of the mappings
2709 are redistributed, making it an ideal method for caches.
2710 However, due to its principle, the distribution will never be
2711 very smooth and it may sometimes be necessary to adjust a
2712 server's weight or its ID to get a more balanced distribution.
2713 In order to get the same distribution on multiple load
2714 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002715 same IDs. Note: consistent hash uses sdbm and avalanche if no
2716 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04002717
2718 <function> is the hash function to be used :
2719
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002720 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04002721 reimplementation of ndbm) database library. It was found to do
2722 well in scrambling bits, causing better distribution of the keys
2723 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002724 function with good distribution, unless the total server weight
2725 is a multiple of 64, in which case applying the avalanche
2726 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04002727
2728 djb2 this function was first proposed by Dan Bernstein many years ago
2729 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002730 function provides a better distribution than sdbm. It generally
2731 works well with text-based inputs though it can perform extremely
2732 poorly with numeric-only input or when the total server weight is
2733 a multiple of 33, unless the avalanche modifier is also used.
2734
Willy Tarreaua0f42712013-11-14 14:30:35 +01002735 wt6 this function was designed for haproxy while testing other
2736 functions in the past. It is not as smooth as the other ones, but
2737 is much less sensible to the input data set or to the number of
2738 servers. It can make sense as an alternative to sdbm+avalanche or
2739 djb2+avalanche for consistent hashing or when hashing on numeric
2740 data such as a source IP address or a visitor identifier in a URL
2741 parameter.
2742
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05002743 <modifier> indicates an optional method applied after hashing the key :
2744
2745 avalanche This directive indicates that the result from the hash
2746 function above should not be used in its raw form but that
2747 a 4-byte full avalanche hash must be applied first. The
2748 purpose of this step is to mix the resulting bits from the
2749 previous hash in order to avoid any undesired effect when
2750 the input contains some limited values or when the number of
2751 servers is a multiple of one of the hash's components (64
2752 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
2753 result less predictable, but it's also not as smooth as when
2754 using the original function. Some testing might be needed
2755 with some workloads. This hash is one of the many proposed
2756 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002757
Bhaskar98634f02013-10-29 23:30:51 -04002758 The default hash type is "map-based" and is recommended for most usages. The
2759 default function is "sdbm", the selection of a function should be based on
2760 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002761
2762 See also : "balance", "server"
2763
2764
Willy Tarreau0ba27502007-12-24 16:55:16 +01002765http-check disable-on-404
2766 Enable a maintenance mode upon HTTP/404 response to health-checks
2767 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01002768 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01002769 Arguments : none
2770
2771 When this option is set, a server which returns an HTTP code 404 will be
2772 excluded from further load-balancing, but will still receive persistent
2773 connections. This provides a very convenient method for Web administrators
2774 to perform a graceful shutdown of their servers. It is also important to note
2775 that a server which is detected as failed while it was in this mode will not
2776 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
2777 will immediately be reinserted into the farm. The status on the stats page
2778 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01002779 option only works in conjunction with the "httpchk" option. If this option
2780 is used with "http-check expect", then it has precedence over it so that 404
2781 responses will still be considered as soft-stop.
2782
2783 See also : "option httpchk", "http-check expect"
2784
2785
2786http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002787 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01002788 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02002789 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01002790 Arguments :
2791 <match> is a keyword indicating how to look for a specific pattern in the
2792 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002793 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01002794 exclamation mark ("!") to negate the match. Spaces are allowed
2795 between the exclamation mark and the keyword. See below for more
2796 details on the supported keywords.
2797
2798 <pattern> is the pattern to look for. It may be a string or a regular
2799 expression. If the pattern contains spaces, they must be escaped
2800 with the usual backslash ('\').
2801
2802 By default, "option httpchk" considers that response statuses 2xx and 3xx
2803 are valid, and that others are invalid. When "http-check expect" is used,
2804 it defines what is considered valid or invalid. Only one "http-check"
2805 statement is supported in a backend. If a server fails to respond or times
2806 out, the check obviously fails. The available matches are :
2807
2808 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002809 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002810 response's status code is exactly this string. If the
2811 "status" keyword is prefixed with "!", then the response
2812 will be considered invalid if the status code matches.
2813
2814 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002815 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002816 response's status code matches the expression. If the
2817 "rstatus" keyword is prefixed with "!", then the response
2818 will be considered invalid if the status code matches.
2819 This is mostly used to check for multiple codes.
2820
2821 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002822 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002823 response's body contains this exact string. If the
2824 "string" keyword is prefixed with "!", then the response
2825 will be considered invalid if the body contains this
2826 string. This can be used to look for a mandatory word at
2827 the end of a dynamic page, or to detect a failure when a
2828 specific error appears on the check page (eg: a stack
2829 trace).
2830
2831 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04002832 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01002833 response's body matches this expression. If the "rstring"
2834 keyword is prefixed with "!", then the response will be
2835 considered invalid if the body matches the expression.
2836 This can be used to look for a mandatory word at the end
2837 of a dynamic page, or to detect a failure when a specific
2838 error appears on the check page (eg: a stack trace).
2839
2840 It is important to note that the responses will be limited to a certain size
2841 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
2842 Thus, too large responses may not contain the mandatory pattern when using
2843 "string" or "rstring". If a large response is absolutely required, it is
2844 possible to change the default max size by setting the global variable.
2845 However, it is worth keeping in mind that parsing very large responses can
2846 waste some CPU cycles, especially when regular expressions are used, and that
2847 it is always better to focus the checks on smaller resources.
2848
2849 Last, if "http-check expect" is combined with "http-check disable-on-404",
2850 then this last one has precedence when the server responds with 404.
2851
2852 Examples :
2853 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002854 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01002855
2856 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002857 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01002858
2859 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002860 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01002861
2862 # check that we have a correct hexadecimal tag before /html
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01002863 http-check expect rstring <!--tag:[0-9a-f]*</html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01002864
Willy Tarreaubd741542010-03-16 18:46:54 +01002865 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01002866
2867
Willy Tarreauef781042010-01-27 11:53:01 +01002868http-check send-state
2869 Enable emission of a state header with HTTP health checks
2870 May be used in sections : defaults | frontend | listen | backend
2871 yes | no | yes | yes
2872 Arguments : none
2873
2874 When this option is set, haproxy will systematically send a special header
2875 "X-Haproxy-Server-State" with a list of parameters indicating to each server
2876 how they are seen by haproxy. This can be used for instance when a server is
2877 manipulated without access to haproxy and the operator needs to know whether
2878 haproxy still sees it up or not, or if the server is the last one in a farm.
2879
2880 The header is composed of fields delimited by semi-colons, the first of which
2881 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
2882 checks on the total number before transition, just as appears in the stats
2883 interface. Next headers are in the form "<variable>=<value>", indicating in
2884 no specific order some values available in the stats interface :
2885 - a variable "name", containing the name of the backend followed by a slash
2886 ("/") then the name of the server. This can be used when a server is
2887 checked in multiple backends.
2888
2889 - a variable "node" containing the name of the haproxy node, as set in the
2890 global "node" variable, otherwise the system's hostname if unspecified.
2891
2892 - a variable "weight" indicating the weight of the server, a slash ("/")
2893 and the total weight of the farm (just counting usable servers). This
2894 helps to know if other servers are available to handle the load when this
2895 one fails.
2896
2897 - a variable "scur" indicating the current number of concurrent connections
2898 on the server, followed by a slash ("/") then the total number of
2899 connections on all servers of the same backend.
2900
2901 - a variable "qcur" indicating the current number of requests in the
2902 server's queue.
2903
2904 Example of a header received by the application server :
2905 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
2906 scur=13/22; qcur=0
2907
2908 See also : "option httpchk", "http-check disable-on-404"
2909
Willy Tarreauccbcc372012-12-27 12:37:57 +01002910http-request { allow | deny | tarpit | auth [realm <realm>] | redirect <rule> |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02002911 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002912 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06002913 replace-header <name> <match-regex> <replace-fmt> |
2914 replace-value <name> <match-regex> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02002915 set-tos <tos> | set-mark <mark> |
2916 add-acl(<file name>) <key fmt> |
2917 del-acl(<file name>) <key fmt> |
2918 del-map(<file name>) <key fmt> |
2919 set-map(<file name>) <key fmt> <value fmt>
2920 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01002921 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002922 Access control for Layer 7 requests
2923
2924 May be used in sections: defaults | frontend | listen | backend
2925 no | yes | yes | yes
2926
Willy Tarreau20b0de52012-12-24 15:45:22 +01002927 The http-request statement defines a set of rules which apply to layer 7
2928 processing. The rules are evaluated in their declaration order when they are
2929 met in a frontend, listen or backend section. Any rule may optionally be
2930 followed by an ACL-based condition, in which case it will only be evaluated
2931 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002932
Willy Tarreau20b0de52012-12-24 15:45:22 +01002933 The first keyword is the rule's action. Currently supported actions include :
2934 - "allow" : this stops the evaluation of the rules and lets the request
2935 pass the check. No further "http-request" rules are evaluated.
2936
2937 - "deny" : this stops the evaluation of the rules and immediately rejects
2938 the request and emits an HTTP 403 error. No further "http-request" rules
2939 are evaluated.
2940
Willy Tarreauccbcc372012-12-27 12:37:57 +01002941 - "tarpit" : this stops the evaluation of the rules and immediately blocks
2942 the request without responding for a delay specified by "timeout tarpit"
2943 or "timeout connect" if the former is not set. After that delay, if the
2944 client is still connected, an HTTP error 500 is returned so that the
2945 client does not suspect it has been tarpitted. Logs will report the flags
2946 "PT". The goal of the tarpit rule is to slow down robots during an attack
2947 when they're limited on the number of concurrent requests. It can be very
2948 efficient against very dumb robots, and will significantly reduce the
2949 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002950 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreauccbcc372012-12-27 12:37:57 +01002951 front firewall to support insane number of concurrent connections.
2952
Willy Tarreau20b0de52012-12-24 15:45:22 +01002953 - "auth" : this stops the evaluation of the rules and immediately responds
2954 with an HTTP 401 or 407 error code to invite the user to present a valid
2955 user name and password. No further "http-request" rules are evaluated. An
2956 optional "realm" parameter is supported, it sets the authentication realm
2957 that is returned with the response (typically the application's name).
2958
Willy Tarreau81499eb2012-12-27 12:19:02 +01002959 - "redirect" : this performs an HTTP redirection based on a redirect rule.
2960 This is exactly the same as the "redirect" statement except that it
2961 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01002962 "http-request" rules and that these rules use the "log-format" strings.
2963 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01002964
Willy Tarreau20b0de52012-12-24 15:45:22 +01002965 - "add-header" appends an HTTP header field whose name is specified in
2966 <name> and whose value is defined by <fmt> which follows the log-format
2967 rules (see Custom Log Format in section 8.2.4). This is particularly
2968 useful to pass connection-specific information to the server (eg: the
2969 client's SSL certificate), or to combine several headers into one. This
2970 rule is not final, so it is possible to add other similar rules. Note
2971 that header addition is performed immediately, so one rule might reuse
2972 the resulting header from a previous rule.
2973
2974 - "set-header" does the same as "add-header" except that the header name
2975 is first removed if it existed. This is useful when passing security
2976 information to the server, where the header must not be manipulated by
2977 external users.
2978
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02002979 - "del-header" removes all HTTP header fields whose name is specified in
2980 <name>.
2981
Sasha Pachev218f0642014-06-16 12:05:59 -06002982 - "replace-header" matches the regular expression in all occurrences of
2983 header field <name> according to <match-regex>, and replaces them with
2984 the <replace-fmt> argument. Format characters are allowed in replace-fmt
2985 and work like in <fmt> arguments in "add-header". The match is only
2986 case-sensitive. It is important to understand that this action only
2987 considers whole header lines, regardless of the number of values they
2988 may contain. This usage is suited to headers naturally containing commas
2989 in their value, such as If-Modified-Since and so on.
2990
2991 Example:
2992
2993 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
2994
2995 applied to:
2996
2997 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
2998
2999 outputs:
3000
3001 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3002
3003 assuming the backend IP is 192.168.1.20
3004
3005 - "replace-value" works like "replace-header" except that it matches the
3006 regex against every comma-delimited value of the header field <name>
3007 instead of the entire header. This is suited for all headers which are
3008 allowed to carry more than one value. An example could be the Accept
3009 header.
3010
3011 Example:
3012
3013 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
3014
3015 applied to:
3016
3017 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
3018
3019 outputs:
3020
3021 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
3022
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003023 - "set-nice" sets the "nice" factor of the current request being processed.
3024 It only has effect against the other requests being processed at the same
3025 time. The default value is 0, unless altered by the "nice" setting on the
3026 "bind" line. The accepted range is -1024..1024. The higher the value, the
3027 nicest the request will be. Lower values will make the request more
3028 important than other ones. This can be useful to improve the speed of
3029 some requests, or lower the priority of non-important requests. Using
3030 this setting without prior experimentation can cause some major slowdown.
3031
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003032 - "set-log-level" is used to change the log level of the current request
3033 when a certain condition is met. Valid levels are the 8 syslog levels
3034 (see the "log" keyword) plus the special level "silent" which disables
3035 logging for this request. This rule is not final so the last matching
3036 rule wins. This rule can be useful to disable health checks coming from
3037 another equipment.
3038
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003039 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3040 the client to the value passed in <tos> on platforms which support this.
3041 This value represents the whole 8 bits of the IP TOS field, and can be
3042 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3043 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3044 bits are always 0. This can be used to adjust some routing behaviour on
3045 border routers based on some information from the request. See RFC 2474,
3046 2597, 3260 and 4594 for more information.
3047
Willy Tarreau51347ed2013-06-11 19:34:13 +02003048 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3049 client to the value passed in <mark> on platforms which support it. This
3050 value is an unsigned 32 bit value which can be matched by netfilter and
3051 by the routing table. It can be expressed both in decimal or hexadecimal
3052 format (prefixed by "0x"). This can be useful to force certain packets to
3053 take a different route (for example a cheaper network path for bulk
3054 downloads). This works on Linux kernels 2.6.32 and above and requires
3055 admin privileges.
3056
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003057 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3058 from a file (even a dummy empty file). The file name of the ACL to be
3059 updated is passed between parentheses. It takes one argument: <key fmt>,
3060 which follows log-format rules, to collect content of the new entry. It
3061 performs a lookup in the ACL before insertion, to avoid duplicated (or
3062 more) values. This lookup is done by a linear search and can be expensive
3063 with large lists! It is the equivalent of the "add acl" command from the
3064 stats socket, but can be triggered by an HTTP request.
3065
3066 - "del-acl" is used to delete an entry from 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 entry to delete.
3070 It is the equivalent of the "del acl" command from the stats socket, but
3071 can be triggered by an HTTP request.
3072
3073 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3074 from a file (even a dummy empty file). The file name of the MAP to be
3075 updated is passed between parentheses. It takes one argument: <key fmt>,
3076 which follows log-format rules, to collect content of the entry to delete.
3077 It takes one argument: "file name" It is the equivalent of the "del map"
3078 command from the stats socket, but can be triggered by an HTTP request.
3079
3080 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3081 from a file (even a dummy empty file). The file name of the MAP to be
3082 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3083 which follows log-format rules, used to collect MAP key, and <value fmt>,
3084 which follows log-format rules, used to collect content for the new entry.
3085 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3086 more) values. This lookup is done by a linear search and can be expensive
3087 with large lists! It is the equivalent of the "set map" command from the
3088 stats socket, but can be triggered by an HTTP request.
3089
Willy Tarreau09448f72014-06-25 18:12:15 +02003090 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
3091 enables tracking of sticky counters from current request. These rules
3092 do not stop evaluation and do not change default action. Three sets of
3093 counters may be simultaneously tracked by the same connection. The first
3094 "track-sc0" rule executed enables tracking of the counters of the
3095 specified table as the first set. The first "track-sc1" rule executed
3096 enables tracking of the counters of the specified table as the second
3097 set. The first "track-sc2" rule executed enables tracking of the
3098 counters of the specified table as the third set. It is a recommended
3099 practice to use the first set of counters for the per-frontend counters
3100 and the second set for the per-backend ones. But this is just a
3101 guideline, all may be used everywhere.
3102
3103 These actions take one or two arguments :
3104 <key> is mandatory, and is a sample expression rule as described
3105 in section 7.3. It describes what elements of the incoming
3106 request or connection will be analysed, extracted, combined,
3107 and used to select which table entry to update the counters.
3108
3109 <table> is an optional table to be used instead of the default one,
3110 which is the stick-table declared in the current proxy. All
3111 the counters for the matches and updates for the key will
3112 then be performed in that table until the session ends.
3113
3114 Once a "track-sc*" rule is executed, the key is looked up in the table
3115 and if it is not found, an entry is allocated for it. Then a pointer to
3116 that entry is kept during all the session's life, and this entry's
3117 counters are updated as often as possible, every time the session's
3118 counters are updated, and also systematically when the session ends.
3119 Counters are only updated for events that happen after the tracking has
3120 been started. As an exception, connection counters and request counters
3121 are systematically updated so that they reflect useful information.
3122
3123 If the entry tracks concurrent connection counters, one connection is
3124 counted for as long as the entry is tracked, and the entry will not
3125 expire during that time. Tracking counters also provides a performance
3126 advantage over just checking the keys, because only one table lookup is
3127 performed for all ACL checks that make use of it.
3128
Willy Tarreau20b0de52012-12-24 15:45:22 +01003129 There is no limit to the number of http-request statements per instance.
3130
3131 It is important to know that http-request rules are processed very early in
3132 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
3133 rules. That way, headers added by "add-header"/"set-header" are visible by
3134 almost all further ACL rules.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003135
3136 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01003137 acl nagios src 192.168.129.3
3138 acl local_net src 192.168.0.0/16
3139 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003140
Cyril Bonté78caf842010-03-10 22:41:43 +01003141 http-request allow if nagios
3142 http-request allow if local_net auth_ok
3143 http-request auth realm Gimme if local_net auth_ok
3144 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003145
Cyril Bonté78caf842010-03-10 22:41:43 +01003146 Example:
3147 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01003148 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003149
Willy Tarreau20b0de52012-12-24 15:45:22 +01003150 Example:
3151 http-request set-header X-Haproxy-Current-Date %T
3152 http-request set-header X-SSL %[ssl_fc]
3153 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id]
3154 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
3155 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
3156 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
3157 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
3158 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
3159 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
3160
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003161 Example:
3162 acl key req.hdr(X-Add-Acl-Key) -m found
3163 acl add path /addacl
3164 acl del path /delacl
3165
3166 acl myhost hdr(Host) -f myhost.lst
3167
3168 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
3169 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
3170
3171 Example:
3172 acl value req.hdr(X-Value) -m found
3173 acl setmap path /setmap
3174 acl delmap path /delmap
3175
3176 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3177
3178 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
3179 http-request del-map(map.lst) %[src] if delmap
3180
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02003181 See also : "stats http-request", section 3.4 about userlists and section 7
3182 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01003183
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003184http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003185 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003186 replace-header <name> <regex-match> <replace-fmt> |
3187 replace-value <name> <regex-match> <replace-fmt> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003188 set-log-level <level> | set-mark <mark> | set-tos <tos> |
3189 add-acl(<file name>) <key fmt> |
3190 del-acl(<file name>) <key fmt> |
3191 del-map(<file name>) <key fmt> |
3192 set-map(<file name>) <key fmt> <value fmt>
3193 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02003194 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003195 Access control for Layer 7 responses
3196
3197 May be used in sections: defaults | frontend | listen | backend
3198 no | yes | yes | yes
3199
3200 The http-response statement defines a set of rules which apply to layer 7
3201 processing. The rules are evaluated in their declaration order when they are
3202 met in a frontend, listen or backend section. Any rule may optionally be
3203 followed by an ACL-based condition, in which case it will only be evaluated
3204 if the condition is true. Since these rules apply on responses, the backend
3205 rules are applied first, followed by the frontend's rules.
3206
3207 The first keyword is the rule's action. Currently supported actions include :
3208 - "allow" : this stops the evaluation of the rules and lets the response
3209 pass the check. No further "http-response" rules are evaluated for the
3210 current section.
3211
3212 - "deny" : this stops the evaluation of the rules and immediately rejects
3213 the response and emits an HTTP 502 error. No further "http-response"
3214 rules are evaluated.
3215
3216 - "add-header" appends an HTTP header field whose name is specified in
3217 <name> and whose value is defined by <fmt> which follows the log-format
3218 rules (see Custom Log Format in section 8.2.4). This may be used to send
3219 a cookie to a client for example, or to pass some internal information.
3220 This rule is not final, so it is possible to add other similar rules.
3221 Note that header addition is performed immediately, so one rule might
3222 reuse the resulting header from a previous rule.
3223
3224 - "set-header" does the same as "add-header" except that the header name
3225 is first removed if it existed. This is useful when passing security
3226 information to the server, where the header must not be manipulated by
3227 external users.
3228
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003229 - "del-header" removes all HTTP header fields whose name is specified in
3230 <name>.
3231
Sasha Pachev218f0642014-06-16 12:05:59 -06003232 - "replace-header" matches the regular expression in all occurrences of
3233 header field <name> according to <match-regex>, and replaces them with
3234 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3235 and work like in <fmt> arguments in "add-header". The match is only
3236 case-sensitive. It is important to understand that this action only
3237 considers whole header lines, regardless of the number of values they
3238 may contain. This usage is suited to headers naturally containing commas
3239 in their value, such as Set-Cookie, Expires and so on.
3240
3241 Example:
3242
3243 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
3244
3245 applied to:
3246
3247 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
3248
3249 outputs:
3250
3251 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
3252
3253 assuming the backend IP is 192.168.1.20.
3254
3255 - "replace-value" works like "replace-header" except that it matches the
3256 regex against every comma-delimited value of the header field <name>
3257 instead of the entire header. This is suited for all headers which are
3258 allowed to carry more than one value. An example could be the Accept
3259 header.
3260
3261 Example:
3262
3263 http-response replace-value Cache-control ^public$ private
3264
3265 applied to:
3266
3267 Cache-Control: max-age=3600, public
3268
3269 outputs:
3270
3271 Cache-Control: max-age=3600, private
3272
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003273 - "set-nice" sets the "nice" factor of the current request being processed.
3274 It only has effect against the other requests being processed at the same
3275 time. The default value is 0, unless altered by the "nice" setting on the
3276 "bind" line. The accepted range is -1024..1024. The higher the value, the
3277 nicest the request will be. Lower values will make the request more
3278 important than other ones. This can be useful to improve the speed of
3279 some requests, or lower the priority of non-important requests. Using
3280 this setting without prior experimentation can cause some major slowdown.
3281
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003282 - "set-log-level" is used to change the log level of the current request
3283 when a certain condition is met. Valid levels are the 8 syslog levels
3284 (see the "log" keyword) plus the special level "silent" which disables
3285 logging for this request. This rule is not final so the last matching
3286 rule wins. This rule can be useful to disable health checks coming from
3287 another equipment.
3288
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003289 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3290 the client to the value passed in <tos> on platforms which support this.
3291 This value represents the whole 8 bits of the IP TOS field, and can be
3292 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3293 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3294 bits are always 0. This can be used to adjust some routing behaviour on
3295 border routers based on some information from the request. See RFC 2474,
3296 2597, 3260 and 4594 for more information.
3297
Willy Tarreau51347ed2013-06-11 19:34:13 +02003298 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3299 client to the value passed in <mark> on platforms which support it. This
3300 value is an unsigned 32 bit value which can be matched by netfilter and
3301 by the routing table. It can be expressed both in decimal or hexadecimal
3302 format (prefixed by "0x"). This can be useful to force certain packets to
3303 take a different route (for example a cheaper network path for bulk
3304 downloads). This works on Linux kernels 2.6.32 and above and requires
3305 admin privileges.
3306
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003307 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3308 from a file (even a dummy empty file). The file name of the ACL to be
3309 updated is passed between parentheses. It takes one argument: <key fmt>,
3310 which follows log-format rules, to collect content of the new entry. It
3311 performs a lookup in the ACL before insertion, to avoid duplicated (or
3312 more) values. This lookup is done by a linear search and can be expensive
3313 with large lists! It is the equivalent of the "add acl" command from the
3314 stats socket, but can be triggered by an HTTP response.
3315
3316 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3317 from a file (even a dummy empty file). The file name of the ACL to be
3318 updated is passed between parentheses. It takes one argument: <key fmt>,
3319 which follows log-format rules, to collect content of the entry to delete.
3320 It is the equivalent of the "del acl" command from the stats socket, but
3321 can be triggered by an HTTP response.
3322
3323 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3324 from a file (even a dummy empty file). The file name of the MAP to be
3325 updated is passed between parentheses. It takes one argument: <key fmt>,
3326 which follows log-format rules, to collect content of the entry to delete.
3327 It takes one argument: "file name" It is the equivalent of the "del map"
3328 command from the stats socket, but can be triggered by an HTTP response.
3329
3330 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3331 from a file (even a dummy empty file). The file name of the MAP to be
3332 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3333 which follows log-format rules, used to collect MAP key, and <value fmt>,
3334 which follows log-format rules, used to collect content for the new entry.
3335 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3336 more) values. This lookup is done by a linear search and can be expensive
3337 with large lists! It is the equivalent of the "set map" command from the
3338 stats socket, but can be triggered by an HTTP response.
3339
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003340 There is no limit to the number of http-response statements per instance.
3341
Godbach09250262013-07-02 01:19:15 +08003342 It is important to know that http-response rules are processed very early in
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003343 the HTTP processing, before "reqdel" or "reqrep" rules. That way, headers
3344 added by "add-header"/"set-header" are visible by almost all further ACL
3345 rules.
3346
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003347 Example:
3348 acl key_acl res.hdr(X-Acl-Key) -m found
3349
3350 acl myhost hdr(Host) -f myhost.lst
3351
3352 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3353 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
3354
3355 Example:
3356 acl value res.hdr(X-Value) -m found
3357
3358 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
3359
3360 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
3361 http-response del-map(map.lst) %[src] if ! value
3362
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003363 See also : "http-request", section 3.4 about userlists and section 7 about
3364 ACL usage.
3365
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02003366
Mark Lamourinec2247f02012-01-04 13:02:01 -05003367http-send-name-header [<header>]
3368 Add the server name to a request. Use the header string given by <header>
3369
3370 May be used in sections: defaults | frontend | listen | backend
3371 yes | no | yes | yes
3372
3373 Arguments :
3374
3375 <header> The header string to use to send the server name
3376
3377 The "http-send-name-header" statement causes the name of the target
3378 server to be added to the headers of an HTTP request. The name
3379 is added with the header string proved.
3380
3381 See also : "server"
3382
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003383id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02003384 Set a persistent ID to a proxy.
3385 May be used in sections : defaults | frontend | listen | backend
3386 no | yes | yes | yes
3387 Arguments : none
3388
3389 Set a persistent ID for the proxy. This ID must be unique and positive.
3390 An unused ID will automatically be assigned if unset. The first assigned
3391 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01003392
3393
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003394ignore-persist { if | unless } <condition>
3395 Declare a condition to ignore persistence
3396 May be used in sections: defaults | frontend | listen | backend
3397 no | yes | yes | yes
3398
3399 By default, when cookie persistence is enabled, every requests containing
3400 the cookie are unconditionally persistent (assuming the target server is up
3401 and running).
3402
3403 The "ignore-persist" statement allows one to declare various ACL-based
3404 conditions which, when met, will cause a request to ignore persistence.
3405 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003406 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003407 persistence for a specific User-Agent (for example, some web crawler bots).
3408
3409 Combined with "appsession", it can also help reduce HAProxy memory usage, as
3410 the appsession table won't grow if persistence is ignored.
3411
3412 The persistence is ignored when an "if" condition is met, or unless an
3413 "unless" condition is met.
3414
3415 See also : "force-persist", "cookie", and section 7 about ACL usage.
3416
3417
Willy Tarreau2769aa02007-12-27 18:26:09 +01003418log global
Willy Tarreau18324f52014-06-27 18:10:07 +02003419log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02003420no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01003421 Enable per-instance logging of events and traffic.
3422 May be used in sections : defaults | frontend | listen | backend
3423 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02003424
3425 Prefix :
3426 no should be used when the logger list must be flushed. For example,
3427 if you don't want to inherit from the default logger list. This
3428 prefix does not allow arguments.
3429
Willy Tarreau2769aa02007-12-27 18:26:09 +01003430 Arguments :
3431 global should be used when the instance's logging parameters are the
3432 same as the global ones. This is the most common usage. "global"
3433 replaces <address>, <facility> and <level> with those of the log
3434 entries found in the "global" section. Only one "log global"
3435 statement may be used per instance, and this form takes no other
3436 parameter.
3437
3438 <address> indicates where to send the logs. It takes the same format as
3439 for the "global" section's logs, and can be one of :
3440
3441 - An IPv4 address optionally followed by a colon (':') and a UDP
3442 port. If no port is specified, 514 is used by default (the
3443 standard syslog port).
3444
David du Colombier24bb5f52011-03-17 10:40:23 +01003445 - An IPv6 address followed by a colon (':') and optionally a UDP
3446 port. If no port is specified, 514 is used by default (the
3447 standard syslog port).
3448
Willy Tarreau2769aa02007-12-27 18:26:09 +01003449 - A filesystem path to a UNIX domain socket, keeping in mind
3450 considerations for chroot (be sure the path is accessible
3451 inside the chroot) and uid/gid (be sure the path is
3452 appropriately writeable).
3453
Willy Tarreaudad36a32013-03-11 01:20:04 +01003454 Any part of the address string may reference any number of
3455 environment variables by preceding their name with a dollar
3456 sign ('$') and optionally enclosing them with braces ('{}'),
3457 similarly to what is done in Bourne shell.
3458
Willy Tarreau18324f52014-06-27 18:10:07 +02003459 <length> is an optional maximum line length. Log lines larger than this
3460 value will be truncated before being sent. The reason is that
3461 syslog servers act differently on log line length. All servers
3462 support the default value of 1024, but some servers simply drop
3463 larger lines while others do log them. If a server supports long
3464 lines, it may make sense to set this value here in order to avoid
3465 truncating long lines. Similarly, if a server drops long lines,
3466 it is preferable to truncate them before sending them. Accepted
3467 values are 80 to 65535 inclusive. The default value of 1024 is
3468 generally fine for all standard usages. Some specific cases of
3469 long captures or JSON-formated logs may require larger values.
3470
Willy Tarreau2769aa02007-12-27 18:26:09 +01003471 <facility> must be one of the 24 standard syslog facilities :
3472
3473 kern user mail daemon auth syslog lpr news
3474 uucp cron auth2 ftp ntp audit alert cron2
3475 local0 local1 local2 local3 local4 local5 local6 local7
3476
3477 <level> is optional and can be specified to filter outgoing messages. By
3478 default, all messages are sent. If a level is specified, only
3479 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003480 will be sent. An optional minimum level can be specified. If it
3481 is set, logs emitted with a more severe level than this one will
3482 be capped to this level. This is used to avoid sending "emerg"
3483 messages on all terminals on some default syslog configurations.
3484 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003485
3486 emerg alert crit err warning notice info debug
3487
William Lallemand0f99e342011-10-12 17:50:54 +02003488 It is important to keep in mind that it is the frontend which decides what to
3489 log from a connection, and that in case of content switching, the log entries
3490 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003491
3492 However, backend log declaration define how and where servers status changes
3493 will be logged. Level "notice" will be used to indicate a server going up,
3494 "warning" will be used for termination signals and definitive service
3495 termination, and "alert" will be used for when a server goes down.
3496
3497 Note : According to RFC3164, messages are truncated to 1024 bytes before
3498 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003499
3500 Example :
3501 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02003502 log 127.0.0.1:514 local0 notice # only send important events
3503 log 127.0.0.1:514 local0 notice notice # same but limit output level
Willy Tarreaudad36a32013-03-11 01:20:04 +01003504 log ${LOCAL_SYSLOG}:514 local0 notice # send to local server
3505
Willy Tarreau2769aa02007-12-27 18:26:09 +01003506
William Lallemand48940402012-01-30 16:47:22 +01003507log-format <string>
3508 Allows you to custom a log line.
3509
3510 See also : Custom Log Format (8.2.4)
3511
Willy Tarreau2769aa02007-12-27 18:26:09 +01003512
Willy Tarreauc35362a2014-04-25 13:58:37 +02003513max-keep-alive-queue <value>
3514 Set the maximum server queue size for maintaining keep-alive connections
3515 May be used in sections: defaults | frontend | listen | backend
3516 yes | no | yes | yes
3517
3518 HTTP keep-alive tries to reuse the same server connection whenever possible,
3519 but sometimes it can be counter-productive, for example if a server has a lot
3520 of connections while other ones are idle. This is especially true for static
3521 servers.
3522
3523 The purpose of this setting is to set a threshold on the number of queued
3524 connections at which haproxy stops trying to reuse the same server and prefers
3525 to find another one. The default value, -1, means there is no limit. A value
3526 of zero means that keep-alive requests will never be queued. For very close
3527 servers which can be reached with a low latency and which are not sensible to
3528 breaking keep-alive, a low value is recommended (eg: local static server can
3529 use a value of 10 or less). For remote servers suffering from a high latency,
3530 higher values might be needed to cover for the latency and/or the cost of
3531 picking a different server.
3532
3533 Note that this has no impact on responses which are maintained to the same
3534 server consecutively to a 401 response. They will still go to the same server
3535 even if they have to be queued.
3536
3537 See also : "option http-server-close", "option prefer-last-server", server
3538 "maxconn" and cookie persistence.
3539
3540
Willy Tarreau2769aa02007-12-27 18:26:09 +01003541maxconn <conns>
3542 Fix the maximum number of concurrent connections on a frontend
3543 May be used in sections : defaults | frontend | listen | backend
3544 yes | yes | yes | no
3545 Arguments :
3546 <conns> is the maximum number of concurrent connections the frontend will
3547 accept to serve. Excess connections will be queued by the system
3548 in the socket's listen queue and will be served once a connection
3549 closes.
3550
3551 If the system supports it, it can be useful on big sites to raise this limit
3552 very high so that haproxy manages connection queues, instead of leaving the
3553 clients with unanswered connection attempts. This value should not exceed the
3554 global maxconn. Also, keep in mind that a connection contains two buffers
3555 of 8kB each, as well as some other data resulting in about 17 kB of RAM being
3556 consumed per established connection. That means that a medium system equipped
3557 with 1GB of RAM can withstand around 40000-50000 concurrent connections if
3558 properly tuned.
3559
3560 Also, when <conns> is set to large values, it is possible that the servers
3561 are not sized to accept such loads, and for this reason it is generally wise
3562 to assign them some reasonable connection limits.
3563
Vincent Bernat6341be52012-06-27 17:18:30 +02003564 By default, this value is set to 2000.
3565
Willy Tarreau2769aa02007-12-27 18:26:09 +01003566 See also : "server", global section's "maxconn", "fullconn"
3567
3568
3569mode { tcp|http|health }
3570 Set the running mode or protocol of the instance
3571 May be used in sections : defaults | frontend | listen | backend
3572 yes | yes | yes | yes
3573 Arguments :
3574 tcp The instance will work in pure TCP mode. A full-duplex connection
3575 will be established between clients and servers, and no layer 7
3576 examination will be performed. This is the default mode. It
3577 should be used for SSL, SSH, SMTP, ...
3578
3579 http The instance will work in HTTP mode. The client request will be
3580 analyzed in depth before connecting to any server. Any request
3581 which is not RFC-compliant will be rejected. Layer 7 filtering,
3582 processing and switching will be possible. This is the mode which
3583 brings HAProxy most of its value.
3584
3585 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02003586 to incoming connections and close the connection. Alternatively,
3587 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
3588 instead. Nothing will be logged in either case. This mode is used
3589 to reply to external components health checks. This mode is
3590 deprecated and should not be used anymore as it is possible to do
3591 the same and even better by combining TCP or HTTP modes with the
3592 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003593
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003594 When doing content switching, it is mandatory that the frontend and the
3595 backend are in the same mode (generally HTTP), otherwise the configuration
3596 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003597
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003598 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01003599 defaults http_instances
3600 mode http
3601
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003602 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003603
Willy Tarreau0ba27502007-12-24 16:55:16 +01003604
Cyril Bontéf0c60612010-02-06 14:44:47 +01003605monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01003606 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003607 May be used in sections : defaults | frontend | listen | backend
3608 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01003609 Arguments :
3610 if <cond> the monitor request will fail if the condition is satisfied,
3611 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003612 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01003613 are met, for instance a low number of servers both in a
3614 backend and its backup.
3615
3616 unless <cond> the monitor request will succeed only if the condition is
3617 satisfied, and will fail otherwise. Such a condition may be
3618 based on a test on the presence of a minimum number of active
3619 servers in a list of backends.
3620
3621 This statement adds a condition which can force the response to a monitor
3622 request to report a failure. By default, when an external component queries
3623 the URI dedicated to monitoring, a 200 response is returned. When one of the
3624 conditions above is met, haproxy will return 503 instead of 200. This is
3625 very useful to report a site failure to an external component which may base
3626 routing advertisements between multiple sites on the availability reported by
3627 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003628 criterion. Note that "monitor fail" only works in HTTP mode. Both status
3629 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003630
3631 Example:
3632 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01003633 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01003634 acl site_dead nbsrv(dynamic) lt 2
3635 acl site_dead nbsrv(static) lt 2
3636 monitor-uri /site_alive
3637 monitor fail if site_dead
3638
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003639 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003640
3641
3642monitor-net <source>
3643 Declare a source network which is limited to monitor requests
3644 May be used in sections : defaults | frontend | listen | backend
3645 yes | yes | yes | no
3646 Arguments :
3647 <source> is the source IPv4 address or network which will only be able to
3648 get monitor responses to any request. It can be either an IPv4
3649 address, a host name, or an address followed by a slash ('/')
3650 followed by a mask.
3651
3652 In TCP mode, any connection coming from a source matching <source> will cause
3653 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003654 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01003655 forwarding the connection to a remote server.
3656
3657 In HTTP mode, a connection coming from a source matching <source> will be
3658 accepted, the following response will be sent without waiting for a request,
3659 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
3660 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02003661 running without forwarding the request to a backend server. Note that this
3662 response is sent in raw format, without any transformation. This is important
3663 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003664
Willy Tarreau82569f92012-09-27 23:48:56 +02003665 Monitor requests are processed very early, just after tcp-request connection
3666 ACLs which are the only ones able to block them. These connections are short
3667 lived and never wait for any data from the client. They cannot be logged, and
3668 it is the intended purpose. They are only used to report HAProxy's health to
3669 an upper component, nothing more. Please note that "monitor fail" rules do
3670 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01003671
Willy Tarreau95cd2832010-03-04 23:36:33 +01003672 Last, please note that only one "monitor-net" statement can be specified in
3673 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003674
Willy Tarreau2769aa02007-12-27 18:26:09 +01003675 Example :
3676 # addresses .252 and .253 are just probing us.
3677 frontend www
3678 monitor-net 192.168.0.252/31
3679
3680 See also : "monitor fail", "monitor-uri"
3681
3682
3683monitor-uri <uri>
3684 Intercept a URI used by external components' monitor requests
3685 May be used in sections : defaults | frontend | listen | backend
3686 yes | yes | yes | no
3687 Arguments :
3688 <uri> is the exact URI which we want to intercept to return HAProxy's
3689 health status instead of forwarding the request.
3690
3691 When an HTTP request referencing <uri> will be received on a frontend,
3692 HAProxy will not forward it nor log it, but instead will return either
3693 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
3694 conditions defined with "monitor fail". This is normally enough for any
3695 front-end HTTP probe to detect that the service is UP and running without
3696 forwarding the request to a backend server. Note that the HTTP method, the
3697 version and all headers are ignored, but the request must at least be valid
3698 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
3699
3700 Monitor requests are processed very early. It is not possible to block nor
3701 divert them using ACLs. They cannot be logged either, and it is the intended
3702 purpose. They are only used to report HAProxy's health to an upper component,
3703 nothing more. However, it is possible to add any number of conditions using
3704 "monitor fail" and ACLs so that the result can be adjusted to whatever check
3705 can be imagined (most often the number of available servers in a backend).
3706
3707 Example :
3708 # Use /haproxy_test to report haproxy's status
3709 frontend www
3710 mode http
3711 monitor-uri /haproxy_test
3712
3713 See also : "monitor fail", "monitor-net"
3714
Willy Tarreau0ba27502007-12-24 16:55:16 +01003715
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003716option abortonclose
3717no option abortonclose
3718 Enable or disable early dropping of aborted requests pending in queues.
3719 May be used in sections : defaults | frontend | listen | backend
3720 yes | no | yes | yes
3721 Arguments : none
3722
3723 In presence of very high loads, the servers will take some time to respond.
3724 The per-instance connection queue will inflate, and the response time will
3725 increase respective to the size of the queue times the average per-session
3726 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01003727 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003728 the queue, and slowing down other users, and the servers as well, because the
3729 request will eventually be served, then aborted at the first error
3730 encountered while delivering the response.
3731
3732 As there is no way to distinguish between a full STOP and a simple output
3733 close on the client side, HTTP agents should be conservative and consider
3734 that the client might only have closed its output channel while waiting for
3735 the response. However, this introduces risks of congestion when lots of users
3736 do the same, and is completely useless nowadays because probably no client at
3737 all will close the session while waiting for the response. Some HTTP agents
Willy Tarreaud72758d2010-01-12 10:42:19 +01003738 support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003739 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01003740 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003741 of being the single component to break rare but valid traffic is extremely
3742 low, which adds to the temptation to be able to abort a session early while
3743 still not served and not pollute the servers.
3744
3745 In HAProxy, the user can choose the desired behaviour using the option
3746 "abortonclose". By default (without the option) the behaviour is HTTP
3747 compliant and aborted requests will be served. But when the option is
3748 specified, a session with an incoming channel closed will be aborted while
3749 it is still possible, either pending in the queue for a connection slot, or
3750 during the connection establishment if the server has not yet acknowledged
3751 the connection request. This considerably reduces the queue size and the load
3752 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01003753 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003754
3755 If this option has been enabled in a "defaults" section, it can be disabled
3756 in a specific instance by prepending the "no" keyword before it.
3757
3758 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
3759
3760
Willy Tarreau4076a152009-04-02 15:18:36 +02003761option accept-invalid-http-request
3762no option accept-invalid-http-request
3763 Enable or disable relaxing of HTTP request parsing
3764 May be used in sections : defaults | frontend | listen | backend
3765 yes | yes | yes | no
3766 Arguments : none
3767
3768 By default, HAProxy complies with RFC2616 in terms of message parsing. This
3769 means that invalid characters in header names are not permitted and cause an
3770 error to be returned to the client. This is the desired behaviour as such
3771 forbidden characters are essentially used to build attacks exploiting server
3772 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3773 server will emit invalid header names for whatever reason (configuration,
3774 implementation) and the issue will not be immediately fixed. In such a case,
3775 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01003776 even if that does not make sense, by specifying this option. Similarly, the
3777 list of characters allowed to appear in a URI is well defined by RFC3986, and
3778 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
3779 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
3780 not allowed at all. Haproxy always blocks a number of them (0..32, 127). The
3781 remaining ones are blocked by default unless this option is enabled.
Willy Tarreau4076a152009-04-02 15:18:36 +02003782
3783 This option should never be enabled by default as it hides application bugs
3784 and open security breaches. It should only be deployed after a problem has
3785 been confirmed.
3786
3787 When this option is enabled, erroneous header names will still be accepted in
3788 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01003789 analysis using the "show errors" request on the UNIX stats socket. Similarly,
3790 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02003791 also helps confirming that the issue has been solved.
3792
3793 If this option has been enabled in a "defaults" section, it can be disabled
3794 in a specific instance by prepending the "no" keyword before it.
3795
3796 See also : "option accept-invalid-http-response" and "show errors" on the
3797 stats socket.
3798
3799
3800option accept-invalid-http-response
3801no option accept-invalid-http-response
3802 Enable or disable relaxing of HTTP response parsing
3803 May be used in sections : defaults | frontend | listen | backend
3804 yes | no | yes | yes
3805 Arguments : none
3806
3807 By default, HAProxy complies with RFC2616 in terms of message parsing. This
3808 means that invalid characters in header names are not permitted and cause an
3809 error to be returned to the client. This is the desired behaviour as such
3810 forbidden characters are essentially used to build attacks exploiting server
3811 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
3812 server will emit invalid header names for whatever reason (configuration,
3813 implementation) and the issue will not be immediately fixed. In such a case,
3814 it is possible to relax HAProxy's header name parser to accept any character
3815 even if that does not make sense, by specifying this option.
3816
3817 This option should never be enabled by default as it hides application bugs
3818 and open security breaches. It should only be deployed after a problem has
3819 been confirmed.
3820
3821 When this option is enabled, erroneous header names will still be accepted in
3822 responses, but the complete response will be captured in order to permit
3823 later analysis using the "show errors" request on the UNIX stats socket.
3824 Doing this also helps confirming that the issue has been solved.
3825
3826 If this option has been enabled in a "defaults" section, it can be disabled
3827 in a specific instance by prepending the "no" keyword before it.
3828
3829 See also : "option accept-invalid-http-request" and "show errors" on the
3830 stats socket.
3831
3832
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003833option allbackups
3834no option allbackups
3835 Use either all backup servers at a time or only the first one
3836 May be used in sections : defaults | frontend | listen | backend
3837 yes | no | yes | yes
3838 Arguments : none
3839
3840 By default, the first operational backup server gets all traffic when normal
3841 servers are all down. Sometimes, it may be preferred to use multiple backups
3842 at once, because one will not be enough. When "option allbackups" is enabled,
3843 the load balancing will be performed among all backup servers when all normal
3844 ones are unavailable. The same load balancing algorithm will be used and the
3845 servers' weights will be respected. Thus, there will not be any priority
3846 order between the backup servers anymore.
3847
3848 This option is mostly used with static server farms dedicated to return a
3849 "sorry" page when an application is completely offline.
3850
3851 If this option has been enabled in a "defaults" section, it can be disabled
3852 in a specific instance by prepending the "no" keyword before it.
3853
3854
3855option checkcache
3856no option checkcache
Godbach7056a352013-12-11 20:01:07 +08003857 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003858 May be used in sections : defaults | frontend | listen | backend
3859 yes | no | yes | yes
3860 Arguments : none
3861
3862 Some high-level frameworks set application cookies everywhere and do not
3863 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003864 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003865 high risk of session crossing or stealing between users traversing the same
3866 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003867 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003868
3869 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003870 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01003871 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003872 response to check if there's a risk of caching a cookie on a client-side
3873 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01003874 to the client are :
3875 - all those without "Set-Cookie" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003876 - all those with a return code other than 200, 203, 206, 300, 301, 410,
Willy Tarreau198a7442008-01-17 12:05:32 +01003877 provided that the server has not set a "Cache-control: public" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003878 - all those that come from a POST request, provided that the server has not
3879 set a 'Cache-Control: public' header ;
3880 - those with a 'Pragma: no-cache' header
3881 - those with a 'Cache-control: private' header
3882 - those with a 'Cache-control: no-store' header
3883 - those with a 'Cache-control: max-age=0' header
3884 - those with a 'Cache-control: s-maxage=0' header
3885 - those with a 'Cache-control: no-cache' header
3886 - those with a 'Cache-control: no-cache="set-cookie"' header
3887 - those with a 'Cache-control: no-cache="set-cookie,' header
3888 (allowing other fields after set-cookie)
3889
3890 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01003891 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003892 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003893 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003894 that admins are informed that there's something to be fixed.
3895
3896 Due to the high impact on the application, the application should be tested
3897 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003898 good practice to always activate it during tests, even if it is not used in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003899 production, as it will report potentially dangerous application behaviours.
3900
3901 If this option has been enabled in a "defaults" section, it can be disabled
3902 in a specific instance by prepending the "no" keyword before it.
3903
3904
3905option clitcpka
3906no option clitcpka
3907 Enable or disable the sending of TCP keepalive packets on the client side
3908 May be used in sections : defaults | frontend | listen | backend
3909 yes | yes | yes | no
3910 Arguments : none
3911
3912 When there is a firewall or any session-aware component between a client and
3913 a server, and when the protocol involves very long sessions with long idle
3914 periods (eg: remote desktops), there is a risk that one of the intermediate
3915 components decides to expire a session which has remained idle for too long.
3916
3917 Enabling socket-level TCP keep-alives makes the system regularly send packets
3918 to the other end of the connection, leaving it active. The delay between
3919 keep-alive probes is controlled by the system only and depends both on the
3920 operating system and its tuning parameters.
3921
3922 It is important to understand that keep-alive packets are neither emitted nor
3923 received at the application level. It is only the network stacks which sees
3924 them. For this reason, even if one side of the proxy already uses keep-alives
3925 to maintain its connection alive, those keep-alive packets will not be
3926 forwarded to the other side of the proxy.
3927
3928 Please note that this has nothing to do with HTTP keep-alive.
3929
3930 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
3931 client side of a connection, which should help when session expirations are
3932 noticed between HAProxy and a client.
3933
3934 If this option has been enabled in a "defaults" section, it can be disabled
3935 in a specific instance by prepending the "no" keyword before it.
3936
3937 See also : "option srvtcpka", "option tcpka"
3938
3939
Willy Tarreau0ba27502007-12-24 16:55:16 +01003940option contstats
3941 Enable continuous traffic statistics updates
3942 May be used in sections : defaults | frontend | listen | backend
3943 yes | yes | yes | no
3944 Arguments : none
3945
3946 By default, counters used for statistics calculation are incremented
3947 only when a session finishes. It works quite well when serving small
3948 objects, but with big ones (for example large images or archives) or
3949 with A/V streaming, a graph generated from haproxy counters looks like
3950 a hedgehog. With this option enabled counters get incremented continuously,
3951 during a whole session. Recounting touches a hotpath directly so
3952 it is not enabled by default, as it has small performance impact (~0.5%).
3953
3954
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003955option dontlog-normal
3956no option dontlog-normal
3957 Enable or disable logging of normal, successful connections
3958 May be used in sections : defaults | frontend | listen | backend
3959 yes | yes | yes | no
3960 Arguments : none
3961
3962 There are large sites dealing with several thousand connections per second
3963 and for which logging is a major pain. Some of them are even forced to turn
3964 logs off and cannot debug production issues. Setting this option ensures that
3965 normal connections, those which experience no error, no timeout, no retry nor
3966 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
3967 mode, the response status code is checked and return codes 5xx will still be
3968 logged.
3969
3970 It is strongly discouraged to use this option as most of the time, the key to
3971 complex issues is in the normal logs which will not be logged here. If you
3972 need to separate logs, see the "log-separate-errors" option instead.
3973
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003974 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02003975 logging.
3976
3977
Willy Tarreaubf1f8162007-12-28 17:42:56 +01003978option dontlognull
3979no option dontlognull
3980 Enable or disable logging of null connections
3981 May be used in sections : defaults | frontend | listen | backend
3982 yes | yes | yes | no
3983 Arguments : none
3984
3985 In certain environments, there are components which will regularly connect to
3986 various systems to ensure that they are still alive. It can be the case from
3987 another load balancer as well as from monitoring systems. By default, even a
3988 simple port probe or scan will produce a log. If those connections pollute
3989 the logs too much, it is possible to enable option "dontlognull" to indicate
3990 that a connection on which no data has been transferred will not be logged,
3991 which typically corresponds to those probes.
3992
3993 It is generally recommended not to use this option in uncontrolled
3994 environments (eg: internet), otherwise scans and other malicious activities
3995 would not be logged.
3996
3997 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 Tarreauc57f0e22009-05-10 13:12:33 +02004000 See also : "log", "monitor-net", "monitor-uri" and section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004001
4002
4003option forceclose
4004no option forceclose
4005 Enable or disable active connection closing after response is transferred.
4006 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01004007 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004008 Arguments : none
4009
4010 Some HTTP servers do not necessarily close the connections when they receive
4011 the "Connection: close" set by "option httpclose", and if the client does not
4012 close either, then the connection remains open till the timeout expires. This
4013 causes high number of simultaneous connections on the servers and shows high
4014 global session times in the logs.
4015
4016 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01004017 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01004018 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004019
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004020 This option may also be combined with "option http-pretend-keepalive", which
4021 will disable sending of the "Connection: close" header, but will still cause
4022 the connection to be closed once the whole response is received.
4023
Cyril Bonté653dcd62014-02-20 00:13:15 +01004024 This option disables and replaces any previous "option httpclose", "option
4025 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004026
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004027 If this option has been enabled in a "defaults" section, it can be disabled
4028 in a specific instance by prepending the "no" keyword before it.
4029
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004030 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01004031
4032
Willy Tarreau87cf5142011-08-19 22:57:24 +02004033option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004034 Enable insertion of the X-Forwarded-For header to requests sent to servers
4035 May be used in sections : defaults | frontend | listen | backend
4036 yes | yes | yes | yes
4037 Arguments :
4038 <network> is an optional argument used to disable this option for sources
4039 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02004040 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01004041 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004042
4043 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
4044 their client address. This is sometimes annoying when the client's IP address
4045 is expected in server logs. To solve this problem, the well-known HTTP header
4046 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
4047 This header contains a value representing the client's IP address. Since this
4048 header is always appended at the end of the existing header list, the server
4049 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02004050 the server's manual to find how to enable use of this standard header. Note
4051 that only the last occurrence of the header must be used, since it is really
4052 possible that the client has already brought one.
4053
Willy Tarreaud72758d2010-01-12 10:42:19 +01004054 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02004055 the default "X-Forwarded-For". This can be useful where you might already
Willy Tarreaud72758d2010-01-12 10:42:19 +01004056 have a "X-Forwarded-For" header from a different application (eg: stunnel),
4057 and you need preserve it. Also if your backend server doesn't use the
Ross Westaf72a1d2008-08-03 10:51:45 +02004058 "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers
4059 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01004060
4061 Sometimes, a same HAProxy instance may be shared between a direct client
4062 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4063 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4064 header for a known source address or network by adding the "except" keyword
4065 followed by the network address. In this case, any source IP matching the
4066 network will not cause an addition of this header. Most common uses are with
4067 private networks or 127.0.0.1.
4068
Willy Tarreau87cf5142011-08-19 22:57:24 +02004069 Alternatively, the keyword "if-none" states that the header will only be
4070 added if it is not present. This should only be used in perfectly trusted
4071 environment, as this might cause a security issue if headers reaching haproxy
4072 are under the control of the end-user.
4073
Willy Tarreauc27debf2008-01-06 08:57:02 +01004074 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02004075 least one of them uses it, the header will be added. Note that the backend's
4076 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02004077 both are defined. In the case of the "if-none" argument, if at least one of
4078 the frontend or the backend does not specify it, it wants the addition to be
4079 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004080
Ross Westaf72a1d2008-08-03 10:51:45 +02004081 Examples :
Willy Tarreauc27debf2008-01-06 08:57:02 +01004082 # Public HTTP address also used by stunnel on the same machine
4083 frontend www
4084 mode http
4085 option forwardfor except 127.0.0.1 # stunnel already adds the header
4086
Ross Westaf72a1d2008-08-03 10:51:45 +02004087 # Those servers want the IP Address in X-Client
4088 backend www
4089 mode http
4090 option forwardfor header X-Client
4091
Willy Tarreau87cf5142011-08-19 22:57:24 +02004092 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01004093 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01004094
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004095
Willy Tarreau16bfb022010-01-16 19:48:41 +01004096option http-keep-alive
4097no option http-keep-alive
4098 Enable or disable HTTP keep-alive from client to server
4099 May be used in sections : defaults | frontend | listen | backend
4100 yes | yes | yes | yes
4101 Arguments : none
4102
Willy Tarreau70dffda2014-01-30 03:07:23 +01004103 By default HAProxy operates in keep-alive mode with regards to persistent
4104 connections: for each connection it processes each request and response, and
4105 leaves the connection idle on both sides between the end of a response and the
4106 start of a new request. This mode may be changed by several options such as
4107 "option http-server-close", "option forceclose", "option httpclose" or
4108 "option http-tunnel". This option allows to set back the keep-alive mode,
4109 which can be useful when another mode was used in a defaults section.
4110
4111 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
4112 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01004113 network) and the fastest session reuse on the server side at the expense
4114 of maintaining idle connections to the servers. In general, it is possible
4115 with this option to achieve approximately twice the request rate that the
4116 "http-server-close" option achieves on small objects. There are mainly two
4117 situations where this option may be useful :
4118
4119 - when the server is non-HTTP compliant and authenticates the connection
4120 instead of requests (eg: NTLM authentication)
4121
4122 - when the cost of establishing the connection to the server is significant
4123 compared to the cost of retrieving the associated object from the server.
4124
4125 This last case can happen when the server is a fast static server of cache.
4126 In this case, the server will need to be properly tuned to support high enough
4127 connection counts because connections will last until the client sends another
4128 request.
4129
4130 If the client request has to go to another backend or another server due to
4131 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01004132 immediately be closed and a new one re-opened. Option "prefer-last-server" is
4133 available to try optimize server selection so that if the server currently
4134 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01004135
4136 In general it is preferred to use "option http-server-close" with application
4137 servers, and some static servers might benefit from "option http-keep-alive".
4138
4139 At the moment, logs will not indicate whether requests came from the same
4140 session or not. The accept date reported in the logs corresponds to the end
4141 of the previous request, and the request time corresponds to the time spent
4142 waiting for a new request. The keep-alive request time is still bound to the
4143 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4144 not set.
4145
Cyril Bonté653dcd62014-02-20 00:13:15 +01004146 This option disables and replaces any previous "option httpclose", "option
4147 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01004148 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01004149 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004150
4151 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01004152 "option prefer-last-server", "option http-pretend-keepalive",
4153 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01004154
4155
Willy Tarreau96e31212011-05-30 18:10:30 +02004156option http-no-delay
4157no option http-no-delay
4158 Instruct the system to favor low interactive delays over performance in HTTP
4159 May be used in sections : defaults | frontend | listen | backend
4160 yes | yes | yes | yes
4161 Arguments : none
4162
4163 In HTTP, each payload is unidirectional and has no notion of interactivity.
4164 Any agent is expected to queue data somewhat for a reasonably low delay.
4165 There are some very rare server-to-server applications that abuse the HTTP
4166 protocol and expect the payload phase to be highly interactive, with many
4167 interleaved data chunks in both directions within a single request. This is
4168 absolutely not supported by the HTTP specification and will not work across
4169 most proxies or servers. When such applications attempt to do this through
4170 haproxy, it works but they will experience high delays due to the network
4171 optimizations which favor performance by instructing the system to wait for
4172 enough data to be available in order to only send full packets. Typical
4173 delays are around 200 ms per round trip. Note that this only happens with
4174 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
4175 affected.
4176
4177 When "option http-no-delay" is present in either the frontend or the backend
4178 used by a connection, all such optimizations will be disabled in order to
4179 make the exchanges as fast as possible. Of course this offers no guarantee on
4180 the functionality, as it may break at any other place. But if it works via
4181 HAProxy, it will work as fast as possible. This option should never be used
4182 by default, and should never be used at all unless such a buggy application
4183 is discovered. The impact of using this option is an increase of bandwidth
4184 usage and CPU usage, which may significantly lower performance in high
4185 latency environments.
4186
4187
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004188option http-pretend-keepalive
4189no option http-pretend-keepalive
4190 Define whether haproxy will announce keepalive to the server or not
4191 May be used in sections : defaults | frontend | listen | backend
4192 yes | yes | yes | yes
4193 Arguments : none
4194
4195 When running with "option http-server-close" or "option forceclose", haproxy
4196 adds a "Connection: close" header to the request forwarded to the server.
4197 Unfortunately, when some servers see this header, they automatically refrain
4198 from using the chunked encoding for responses of unknown length, while this
4199 is totally unrelated. The immediate effect is that this prevents haproxy from
4200 maintaining the client connection alive. A second effect is that a client or
4201 a cache could receive an incomplete response without being aware of it, and
4202 consider the response complete.
4203
4204 By setting "option http-pretend-keepalive", haproxy will make the server
4205 believe it will keep the connection alive. The server will then not fall back
4206 to the abnormal undesired above. When haproxy gets the whole response, it
4207 will close the connection with the server just as it would do with the
4208 "forceclose" option. That way the client gets a normal response and the
4209 connection is correctly closed on the server side.
4210
4211 It is recommended not to enable this option by default, because most servers
4212 will more efficiently close the connection themselves after the last packet,
4213 and release its buffers slightly earlier. Also, the added packet on the
4214 network could slightly reduce the overall peak performance. However it is
4215 worth noting that when this option is enabled, haproxy will have slightly
4216 less work to do. So if haproxy is the bottleneck on the whole architecture,
4217 enabling this option might save a few CPU cycles.
4218
4219 This option may be set both in a frontend and in a backend. It is enabled if
4220 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004221 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02004222 keepalive to be announced to the server and close to be announced to the
4223 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004224
4225 If this option has been enabled in a "defaults" section, it can be disabled
4226 in a specific instance by prepending the "no" keyword before it.
4227
Willy Tarreau16bfb022010-01-16 19:48:41 +01004228 See also : "option forceclose", "option http-server-close", and
4229 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02004230
Willy Tarreauc27debf2008-01-06 08:57:02 +01004231
Willy Tarreaub608feb2010-01-02 22:47:18 +01004232option http-server-close
4233no option http-server-close
4234 Enable or disable HTTP connection closing on the server side
4235 May be used in sections : defaults | frontend | listen | backend
4236 yes | yes | yes | yes
4237 Arguments : none
4238
Willy Tarreau70dffda2014-01-30 03:07:23 +01004239 By default HAProxy operates in keep-alive mode with regards to persistent
4240 connections: for each connection it processes each request and response, and
4241 leaves the connection idle on both sides between the end of a response and
4242 the start of a new request. This mode may be changed by several options such
4243 as "option http-server-close", "option forceclose", "option httpclose" or
4244 "option http-tunnel". Setting "option http-server-close" enables HTTP
4245 connection-close mode on the server side while keeping the ability to support
4246 HTTP keep-alive and pipelining on the client side. This provides the lowest
4247 latency on the client side (slow network) and the fastest session reuse on
4248 the server side to save server resources, similarly to "option forceclose".
4249 It also permits non-keepalive capable servers to be served in keep-alive mode
4250 to the clients if they conform to the requirements of RFC2616. Please note
4251 that some servers do not always conform to those requirements when they see
4252 "Connection: close" in the request. The effect will be that keep-alive will
4253 never be used. A workaround consists in enabling "option
4254 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004255
4256 At the moment, logs will not indicate whether requests came from the same
4257 session or not. The accept date reported in the logs corresponds to the end
4258 of the previous request, and the request time corresponds to the time spent
4259 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01004260 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
4261 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004262
4263 This option may be set both in a frontend and in a backend. It is enabled if
4264 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004265 It disables and replaces any previous "option httpclose", "option forceclose",
4266 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01004267 ("Proxies") to see how this option combines with others when frontend and
4268 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01004269
4270 If this option has been enabled in a "defaults" section, it can be disabled
4271 in a specific instance by prepending the "no" keyword before it.
4272
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004273 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01004274 "option httpclose", "option http-keep-alive", and
4275 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01004276
4277
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004278option http-tunnel
4279no option http-tunnel
4280 Disable or enable HTTP connection processing after first transaction
4281 May be used in sections : defaults | frontend | listen | backend
4282 yes | yes | yes | yes
4283 Arguments : none
4284
Willy Tarreau70dffda2014-01-30 03:07:23 +01004285 By default HAProxy operates in keep-alive mode with regards to persistent
4286 connections: for each connection it processes each request and response, and
4287 leaves the connection idle on both sides between the end of a response and
4288 the start of a new request. This mode may be changed by several options such
4289 as "option http-server-close", "option forceclose", "option httpclose" or
4290 "option http-tunnel".
4291
4292 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004293 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01004294 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
4295 is normally not needed anymore unless in very specific cases such as when
4296 using an in-house protocol that looks like HTTP but is not compatible, or
4297 just to log one request per client in order to reduce log size. Note that
4298 everything which works at the HTTP level, including header parsing/addition,
4299 cookie processing or content switching will only work for the first request
4300 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01004301
4302 If this option has been enabled in a "defaults" section, it can be disabled
4303 in a specific instance by prepending the "no" keyword before it.
4304
4305 See also : "option forceclose", "option http-server-close",
4306 "option httpclose", "option http-keep-alive", and
4307 "1.1. The HTTP transaction model".
4308
4309
Willy Tarreau88d349d2010-01-25 12:15:43 +01004310option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01004311no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01004312 Make use of non-standard Proxy-Connection header instead of Connection
4313 May be used in sections : defaults | frontend | listen | backend
4314 yes | yes | yes | no
4315 Arguments : none
4316
4317 While RFC2616 explicitly states that HTTP/1.1 agents must use the
4318 Connection header to indicate their wish of persistent or non-persistent
4319 connections, both browsers and proxies ignore this header for proxied
4320 connections and make use of the undocumented, non-standard Proxy-Connection
4321 header instead. The issue begins when trying to put a load balancer between
4322 browsers and such proxies, because there will be a difference between what
4323 haproxy understands and what the client and the proxy agree on.
4324
4325 By setting this option in a frontend, haproxy can automatically switch to use
4326 that non-standard header if it sees proxied requests. A proxied request is
4327 defined here as one where the URI begins with neither a '/' nor a '*'. The
4328 choice of header only affects requests passing through proxies making use of
4329 one of the "httpclose", "forceclose" and "http-server-close" options. Note
4330 that this option can only be specified in a frontend and will affect the
4331 request along its whole life.
4332
Willy Tarreau844a7e72010-01-31 21:46:18 +01004333 Also, when this option is set, a request which requires authentication will
4334 automatically switch to use proxy authentication headers if it is itself a
4335 proxied request. That makes it possible to check or enforce authentication in
4336 front of an existing proxy.
4337
Willy Tarreau88d349d2010-01-25 12:15:43 +01004338 This option should normally never be used, except in front of a proxy.
4339
4340 See also : "option httpclose", "option forceclose" and "option
4341 http-server-close".
4342
4343
Willy Tarreaud63335a2010-02-26 12:56:52 +01004344option httpchk
4345option httpchk <uri>
4346option httpchk <method> <uri>
4347option httpchk <method> <uri> <version>
4348 Enable HTTP protocol to check on the servers health
4349 May be used in sections : defaults | frontend | listen | backend
4350 yes | no | yes | yes
4351 Arguments :
4352 <method> is the optional HTTP method used with the requests. When not set,
4353 the "OPTIONS" method is used, as it generally requires low server
4354 processing and is easy to filter out from the logs. Any method
4355 may be used, though it is not recommended to invent non-standard
4356 ones.
4357
4358 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
4359 which is accessible by default on almost any server, but may be
4360 changed to any other URI. Query strings are permitted.
4361
4362 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
4363 but some servers might behave incorrectly in HTTP 1.0, so turning
4364 it to HTTP/1.1 may sometimes help. Note that the Host field is
4365 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
4366 after "\r\n" following the version string.
4367
4368 By default, server health checks only consist in trying to establish a TCP
4369 connection. When "option httpchk" is specified, a complete HTTP request is
4370 sent once the TCP connection is established, and responses 2xx and 3xx are
4371 considered valid, while all other ones indicate a server failure, including
4372 the lack of any response.
4373
4374 The port and interval are specified in the server configuration.
4375
4376 This option does not necessarily require an HTTP backend, it also works with
4377 plain TCP backends. This is particularly useful to check simple scripts bound
4378 to some dedicated ports using the inetd daemon.
4379
4380 Examples :
4381 # Relay HTTPS traffic to Apache instance and check service availability
4382 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
4383 backend https_relay
4384 mode tcp
4385 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
4386 server apache1 192.168.1.1:443 check port 80
4387
Simon Hormanafc47ee2013-11-25 10:46:35 +09004388 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
4389 "option pgsql-check", "http-check" and the "check", "port" and
4390 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01004391
4392
Willy Tarreauc27debf2008-01-06 08:57:02 +01004393option httpclose
4394no option httpclose
4395 Enable or disable passive HTTP connection closing
4396 May be used in sections : defaults | frontend | listen | backend
4397 yes | yes | yes | yes
4398 Arguments : none
4399
Willy Tarreau70dffda2014-01-30 03:07:23 +01004400 By default HAProxy operates in keep-alive mode with regards to persistent
4401 connections: for each connection it processes each request and response, and
4402 leaves the connection idle on both sides between the end of a response and
4403 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01004404 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01004405 "option http-tunnel".
4406
4407 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
4408 if a "Connection: close" header is already set in each direction, and will
4409 add one if missing. Each end should react to this by actively closing the TCP
4410 connection after each transfer, thus resulting in a switch to the HTTP close
4411 mode. Any "Connection" header different from "close" will also be removed.
4412 Note that this option is deprecated since what it does is very cheap but not
4413 reliable. Using "option http-server-close" or "option forceclose" is strongly
4414 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004415
4416 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004417 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01004418 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
4419 it is possible to use the "option forceclose" which actively closes the
4420 request connection once the server responds. Option "forceclose" also
4421 releases the server connection earlier because it does not have to wait for
4422 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004423
4424 This option may be set both in a frontend and in a backend. It is enabled if
4425 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01004426 It disables and replaces any previous "option http-server-close",
4427 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01004428 check section 4 ("Proxies") to see how this option combines with others when
4429 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004430
4431 If this option has been enabled in a "defaults" section, it can be disabled
4432 in a specific instance by prepending the "no" keyword before it.
4433
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02004434 See also : "option forceclose", "option http-server-close" and
4435 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01004436
4437
Emeric Brun3a058f32009-06-30 18:26:00 +02004438option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01004439 Enable logging of HTTP request, session state and timers
4440 May be used in sections : defaults | frontend | listen | backend
4441 yes | yes | yes | yes
Emeric Brun3a058f32009-06-30 18:26:00 +02004442 Arguments :
4443 clf if the "clf" argument is added, then the output format will be
4444 the CLF format instead of HAProxy's default HTTP format. You can
4445 use this when you need to feed HAProxy's logs through a specific
4446 log analyser which only support the CLF format and which is not
4447 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004448
4449 By default, the log output format is very poor, as it only contains the
4450 source and destination addresses, and the instance name. By specifying
4451 "option httplog", each log line turns into a much richer format including,
4452 but not limited to, the HTTP request, the connection timers, the session
4453 status, the connections numbers, the captured headers and cookies, the
4454 frontend, backend and server name, and of course the source address and
4455 ports.
4456
4457 This option may be set either in the frontend or the backend.
4458
Emeric Brun3a058f32009-06-30 18:26:00 +02004459 If this option has been enabled in a "defaults" section, it can be disabled
4460 in a specific instance by prepending the "no" keyword before it. Specifying
4461 only "option httplog" will automatically clear the 'clf' mode if it was set
4462 by default.
4463
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004464 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01004465
Willy Tarreau55165fe2009-05-10 12:02:55 +02004466
4467option http_proxy
4468no option http_proxy
4469 Enable or disable plain HTTP proxy mode
4470 May be used in sections : defaults | frontend | listen | backend
4471 yes | yes | yes | yes
4472 Arguments : none
4473
4474 It sometimes happens that people need a pure HTTP proxy which understands
4475 basic proxy requests without caching nor any fancy feature. In this case,
4476 it may be worth setting up an HAProxy instance with the "option http_proxy"
4477 set. In this mode, no server is declared, and the connection is forwarded to
4478 the IP address and port found in the URL after the "http://" scheme.
4479
4480 No host address resolution is performed, so this only works when pure IP
4481 addresses are passed. Since this option's usage perimeter is rather limited,
4482 it will probably be used only by experts who know they need exactly it. Last,
4483 if the clients are susceptible of sending keep-alive requests, it will be
Cyril Bonté2409e682010-12-14 22:47:51 +01004484 needed to add "option httpclose" to ensure that all requests will correctly
Willy Tarreau55165fe2009-05-10 12:02:55 +02004485 be analyzed.
4486
4487 If this option has been enabled in a "defaults" section, it can be disabled
4488 in a specific instance by prepending the "no" keyword before it.
4489
4490 Example :
4491 # this backend understands HTTP proxy requests and forwards them directly.
4492 backend direct_forward
4493 option httpclose
4494 option http_proxy
4495
4496 See also : "option httpclose"
4497
Willy Tarreau211ad242009-10-03 21:45:07 +02004498
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004499option independent-streams
4500no option independent-streams
4501 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004502 May be used in sections : defaults | frontend | listen | backend
4503 yes | yes | yes | yes
4504 Arguments : none
4505
4506 By default, when data is sent over a socket, both the write timeout and the
4507 read timeout for that socket are refreshed, because we consider that there is
4508 activity on that socket, and we have no other means of guessing if we should
4509 receive data or not.
4510
4511 While this default behaviour is desirable for almost all applications, there
4512 exists a situation where it is desirable to disable it, and only refresh the
4513 read timeout if there are incoming data. This happens on sessions with large
4514 timeouts and low amounts of exchanged data such as telnet session. If the
4515 server suddenly disappears, the output data accumulates in the system's
4516 socket buffers, both timeouts are correctly refreshed, and there is no way
4517 to know the server does not receive them, so we don't timeout. However, when
4518 the underlying protocol always echoes sent data, it would be enough by itself
4519 to detect the issue using the read timeout. Note that this problem does not
4520 happen with more verbose protocols because data won't accumulate long in the
4521 socket buffers.
4522
4523 When this option is set on the frontend, it will disable read timeout updates
4524 on data sent to the client. There probably is little use of this case. When
4525 the option is set on the backend, it will disable read timeout updates on
4526 data sent to the server. Doing so will typically break large HTTP posts from
4527 slow lines, so use it with caution.
4528
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004529 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004530 with a spelling mistake. This spelling is still supported but
4531 deprecated.
4532
Willy Tarreauce887fd2012-05-12 12:50:00 +02004533 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02004534
4535
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02004536option ldap-check
4537 Use LDAPv3 health checks for server testing
4538 May be used in sections : defaults | frontend | listen | backend
4539 yes | no | yes | yes
4540 Arguments : none
4541
4542 It is possible to test that the server correctly talks LDAPv3 instead of just
4543 testing that it accepts the TCP connection. When this option is set, an
4544 LDAPv3 anonymous simple bind message is sent to the server, and the response
4545 is analyzed to find an LDAPv3 bind response message.
4546
4547 The server is considered valid only when the LDAP response contains success
4548 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
4549
4550 Logging of bind requests is server dependent see your documentation how to
4551 configure it.
4552
4553 Example :
4554 option ldap-check
4555
4556 See also : "option httpchk"
4557
4558
Simon Horman98637e52014-06-20 12:30:16 +09004559option external-check
4560 Use external processes for server health checks
4561 May be used in sections : defaults | frontend | listen | backend
4562 yes | no | yes | yes
4563
4564 It is possible to test the health of a server using an external command.
4565 This is achieved by running the executable set using "external-check
4566 command".
4567
4568 Requires the "external-check" global to be set.
4569
4570 See also : "external-check", "external-check command", "external-check path"
4571
4572
Willy Tarreau211ad242009-10-03 21:45:07 +02004573option log-health-checks
4574no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02004575 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02004576 May be used in sections : defaults | frontend | listen | backend
4577 yes | no | yes | yes
4578 Arguments : none
4579
Willy Tarreaubef1b322014-05-13 21:01:39 +02004580 By default, failed health check are logged if server is UP and successful
4581 health checks are logged if server is DOWN, so the amount of additional
4582 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02004583
Willy Tarreaubef1b322014-05-13 21:01:39 +02004584 When this option is enabled, any change of the health check status or to
4585 the server's health will be logged, so that it becomes possible to know
4586 that a server was failing occasional checks before crashing, or exactly when
4587 it failed to respond a valid HTTP status, then when the port started to
4588 reject connections, then when the server stopped responding at all.
4589
4590 Note that status changes not caused by health checks (eg: enable/disable on
4591 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02004592
Willy Tarreaubef1b322014-05-13 21:01:39 +02004593 See also: "option httpchk", "option ldap-check", "option mysql-check",
4594 "option pgsql-check", "option redis-check", "option smtpchk",
4595 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02004596
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004597
4598option log-separate-errors
4599no option log-separate-errors
4600 Change log level for non-completely successful connections
4601 May be used in sections : defaults | frontend | listen | backend
4602 yes | yes | yes | no
4603 Arguments : none
4604
4605 Sometimes looking for errors in logs is not easy. This option makes haproxy
4606 raise the level of logs containing potentially interesting information such
4607 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
4608 level changes from "info" to "err". This makes it possible to log them
4609 separately to a different file with most syslog daemons. Be careful not to
4610 remove them from the original file, otherwise you would lose ordering which
4611 provides very important information.
4612
4613 Using this option, large sites dealing with several thousand connections per
4614 second may log normal traffic to a rotating buffer and only archive smaller
4615 error logs.
4616
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004617 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02004618 logging.
4619
Willy Tarreauc27debf2008-01-06 08:57:02 +01004620
4621option logasap
4622no option logasap
4623 Enable or disable early logging of HTTP requests
4624 May be used in sections : defaults | frontend | listen | backend
4625 yes | yes | yes | no
4626 Arguments : none
4627
4628 By default, HTTP requests are logged upon termination so that the total
4629 transfer time and the number of bytes appear in the logs. When large objects
4630 are being transferred, it may take a while before the request appears in the
4631 logs. Using "option logasap", the request gets logged as soon as the server
4632 sends the complete headers. The only missing information in the logs will be
4633 the total number of bytes which will indicate everything except the amount
4634 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004635 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01004636 "Content-Length" response header so that the logs at least indicate how many
4637 bytes are expected to be transferred.
4638
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004639 Examples :
4640 listen http_proxy 0.0.0.0:80
4641 mode http
4642 option httplog
4643 option logasap
4644 log 192.168.2.200 local3
4645
4646 >>> Feb 6 12:14:14 localhost \
4647 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
4648 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
4649 "GET /image.iso HTTP/1.0"
4650
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004651 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01004652 logging.
4653
4654
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004655option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004656 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004657 May be used in sections : defaults | frontend | listen | backend
4658 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004659 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004660 <username> This is the username which will be used when connecting to MySQL
4661 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02004662 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02004663
4664 If you specify a username, the check consists of sending two MySQL packet,
4665 one Client Authentication packet, and one QUIT packet, to correctly close
4666 MySQL session. We then parse the MySQL Handshake Initialisation packet and/or
4667 Error packet. It is a basic but useful test which does not produce error nor
4668 aborted connect on the server. However, it requires adding an authorization
4669 in the MySQL table, like this :
4670
4671 USE mysql;
4672 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
4673 FLUSH PRIVILEGES;
4674
4675 If you don't specify a username (it is deprecated and not recommended), the
4676 check only consists in parsing the Mysql Handshake Initialisation packet or
4677 Error packet, we don't send anything in this mode. It was reported that it
4678 can generate lockout if check is too frequent and/or if there is not enough
4679 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
4680 value as if a connection is established successfully within fewer than MySQL
4681 "max_connect_errors" attempts after a previous connection was interrupted,
4682 the error count for the host is cleared to zero. If HAProxy's server get
4683 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
4684
4685 Remember that this does not check database presence nor database consistency.
4686 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004687
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02004688 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01004689
4690 Most often, an incoming MySQL server needs to see the client's IP address for
4691 various purposes, including IP privilege matching and connection logging.
4692 When possible, it is often wise to masquerade the client's IP address when
4693 connecting to the server using the "usesrc" argument of the "source" keyword,
4694 which requires the cttproxy feature to be compiled in, and the MySQL server
4695 to route the client via the machine hosting haproxy.
4696
4697 See also: "option httpchk"
4698
4699
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004700option nolinger
4701no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004702 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004703 May be used in sections: defaults | frontend | listen | backend
4704 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004705 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004706
4707 When clients or servers abort connections in a dirty way (eg: they are
4708 physically disconnected), the session timeouts triggers and the session is
4709 closed. But it will remain in FIN_WAIT1 state for some time in the system,
4710 using some resources and possibly limiting the ability to establish newer
4711 connections.
4712
4713 When this happens, it is possible to activate "option nolinger" which forces
4714 the system to immediately remove any socket's pending data on close. Thus,
4715 the session is instantly purged from the system's tables. This usually has
4716 side effects such as increased number of TCP resets due to old retransmits
4717 getting immediately rejected. Some firewalls may sometimes complain about
4718 this too.
4719
4720 For this reason, it is not recommended to use this option when not absolutely
4721 needed. You know that you need it when you have thousands of FIN_WAIT1
4722 sessions on your system (TIME_WAIT ones do not count).
4723
4724 This option may be used both on frontends and backends, depending on the side
4725 where it is required. Use it on the frontend for clients, and on the backend
4726 for servers.
4727
4728 If this option has been enabled in a "defaults" section, it can be disabled
4729 in a specific instance by prepending the "no" keyword before it.
4730
4731
Willy Tarreau55165fe2009-05-10 12:02:55 +02004732option originalto [ except <network> ] [ header <name> ]
4733 Enable insertion of the X-Original-To header to requests sent to servers
4734 May be used in sections : defaults | frontend | listen | backend
4735 yes | yes | yes | yes
4736 Arguments :
4737 <network> is an optional argument used to disable this option for sources
4738 matching <network>
4739 <name> an optional argument to specify a different "X-Original-To"
4740 header name.
4741
4742 Since HAProxy can work in transparent mode, every request from a client can
4743 be redirected to the proxy and HAProxy itself can proxy every request to a
4744 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
4745 be lost. This is annoying when you want access rules based on destination ip
4746 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
4747 added by HAProxy to all requests sent to the server. This header contains a
4748 value representing the original destination IP address. Since this must be
4749 configured to always use the last occurrence of this header only. Note that
4750 only the last occurrence of the header must be used, since it is really
4751 possible that the client has already brought one.
4752
4753 The keyword "header" may be used to supply a different header name to replace
4754 the default "X-Original-To". This can be useful where you might already
4755 have a "X-Original-To" header from a different application, and you need
4756 preserve it. Also if your backend server doesn't use the "X-Original-To"
4757 header and requires different one.
4758
4759 Sometimes, a same HAProxy instance may be shared between a direct client
4760 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
4761 used to decrypt HTTPS traffic). It is possible to disable the addition of the
4762 header for a known source address or network by adding the "except" keyword
4763 followed by the network address. In this case, any source IP matching the
4764 network will not cause an addition of this header. Most common uses are with
4765 private networks or 127.0.0.1.
4766
4767 This option may be specified either in the frontend or in the backend. If at
4768 least one of them uses it, the header will be added. Note that the backend's
4769 setting of the header subargument takes precedence over the frontend's if
4770 both are defined.
4771
Willy Tarreau55165fe2009-05-10 12:02:55 +02004772 Examples :
4773 # Original Destination address
4774 frontend www
4775 mode http
4776 option originalto except 127.0.0.1
4777
4778 # Those servers want the IP Address in X-Client-Dst
4779 backend www
4780 mode http
4781 option originalto header X-Client-Dst
4782
Willy Tarreau87cf5142011-08-19 22:57:24 +02004783 See also : "option httpclose", "option http-server-close",
4784 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02004785
4786
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004787option persist
4788no option persist
4789 Enable or disable forced persistence on down servers
4790 May be used in sections: defaults | frontend | listen | backend
4791 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004792 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004793
4794 When an HTTP request reaches a backend with a cookie which references a dead
4795 server, by default it is redispatched to another server. It is possible to
4796 force the request to be sent to the dead server first using "option persist"
4797 if absolutely needed. A common use case is when servers are under extreme
4798 load and spend their time flapping. In this case, the users would still be
4799 directed to the server they opened the session on, in the hope they would be
4800 correctly served. It is recommended to use "option redispatch" in conjunction
4801 with this option so that in the event it would not be possible to connect to
4802 the server at all (server definitely dead), the client would finally be
4803 redirected to another valid server.
4804
4805 If this option has been enabled in a "defaults" section, it can be disabled
4806 in a specific instance by prepending the "no" keyword before it.
4807
Willy Tarreau4de91492010-01-22 19:10:05 +01004808 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004809
4810
Willy Tarreau0c122822013-12-15 18:49:01 +01004811option pgsql-check [ user <username> ]
4812 Use PostgreSQL health checks for server testing
4813 May be used in sections : defaults | frontend | listen | backend
4814 yes | no | yes | yes
4815 Arguments :
4816 <username> This is the username which will be used when connecting to
4817 PostgreSQL server.
4818
4819 The check sends a PostgreSQL StartupMessage and waits for either
4820 Authentication request or ErrorResponse message. It is a basic but useful
4821 test which does not produce error nor aborted connect on the server.
4822 This check is identical with the "mysql-check".
4823
4824 See also: "option httpchk"
4825
4826
Willy Tarreau9420b122013-12-15 18:58:25 +01004827option prefer-last-server
4828no option prefer-last-server
4829 Allow multiple load balanced requests to remain on the same server
4830 May be used in sections: defaults | frontend | listen | backend
4831 yes | no | yes | yes
4832 Arguments : none
4833
4834 When the load balancing algorithm in use is not deterministic, and a previous
4835 request was sent to a server to which haproxy still holds a connection, it is
4836 sometimes desirable that subsequent requests on a same session go to the same
4837 server as much as possible. Note that this is different from persistence, as
4838 we only indicate a preference which haproxy tries to apply without any form
4839 of warranty. The real use is for keep-alive connections sent to servers. When
4840 this option is used, haproxy will try to reuse the same connection that is
4841 attached to the server instead of rebalancing to another server, causing a
4842 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01004843 not make much sense to use this in combination with hashing algorithms. Note,
4844 haproxy already automatically tries to stick to a server which sends a 401 or
4845 to a proxy which sends a 407 (authentication required). This is mandatory for
4846 use with the broken NTLM authentication challenge, and significantly helps in
4847 troubleshooting some faulty applications. Option prefer-last-server might be
4848 desirable in these environments as well, to avoid redistributing the traffic
4849 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01004850
4851 If this option has been enabled in a "defaults" section, it can be disabled
4852 in a specific instance by prepending the "no" keyword before it.
4853
4854 See also: "option http-keep-alive"
4855
4856
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004857option redispatch
4858no option redispatch
4859 Enable or disable session redistribution in case of connection failure
4860 May be used in sections: defaults | frontend | listen | backend
4861 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01004862 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004863
4864 In HTTP mode, if a server designated by a cookie is down, clients may
4865 definitely stick to it because they cannot flush the cookie, so they will not
4866 be able to access the service anymore.
4867
4868 Specifying "option redispatch" will allow the proxy to break their
4869 persistence and redistribute them to a working server.
4870
4871 It also allows to retry last connection to another server in case of multiple
4872 connection failures. Of course, it requires having "retries" set to a nonzero
4873 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01004874
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004875 This form is the preferred form, which replaces both the "redispatch" and
4876 "redisp" keywords.
4877
4878 If this option has been enabled in a "defaults" section, it can be disabled
4879 in a specific instance by prepending the "no" keyword before it.
4880
Willy Tarreau4de91492010-01-22 19:10:05 +01004881 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004882
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004883
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02004884option redis-check
4885 Use redis health checks for server testing
4886 May be used in sections : defaults | frontend | listen | backend
4887 yes | no | yes | yes
4888 Arguments : none
4889
4890 It is possible to test that the server correctly talks REDIS protocol instead
4891 of just testing that it accepts the TCP connection. When this option is set,
4892 a PING redis command is sent to the server, and the response is analyzed to
4893 find the "+PONG" response message.
4894
4895 Example :
4896 option redis-check
4897
4898 See also : "option httpchk"
4899
4900
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004901option smtpchk
4902option smtpchk <hello> <domain>
4903 Use SMTP health checks for server testing
4904 May be used in sections : defaults | frontend | listen | backend
4905 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01004906 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01004907 <hello> is an optional argument. It is the "hello" command to use. It can
4908 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
4909 values will be turned into the default command ("HELO").
4910
4911 <domain> is the domain name to present to the server. It may only be
4912 specified (and is mandatory) if the hello command has been
4913 specified. By default, "localhost" is used.
4914
4915 When "option smtpchk" is set, the health checks will consist in TCP
4916 connections followed by an SMTP command. By default, this command is
4917 "HELO localhost". The server's return code is analyzed and only return codes
4918 starting with a "2" will be considered as valid. All other responses,
4919 including a lack of response will constitute an error and will indicate a
4920 dead server.
4921
4922 This test is meant to be used with SMTP servers or relays. Depending on the
4923 request, it is possible that some servers do not log each connection attempt,
4924 so you may want to experiment to improve the behaviour. Using telnet on port
4925 25 is often easier than adjusting the configuration.
4926
4927 Most often, an incoming SMTP server needs to see the client's IP address for
4928 various purposes, including spam filtering, anti-spoofing and logging. When
4929 possible, it is often wise to masquerade the client's IP address when
4930 connecting to the server using the "usesrc" argument of the "source" keyword,
4931 which requires the cttproxy feature to be compiled in.
4932
4933 Example :
4934 option smtpchk HELO mydomain.org
4935
4936 See also : "option httpchk", "source"
4937
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01004938
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02004939option socket-stats
4940no option socket-stats
4941
4942 Enable or disable collecting & providing separate statistics for each socket.
4943 May be used in sections : defaults | frontend | listen | backend
4944 yes | yes | yes | no
4945
4946 Arguments : none
4947
4948
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004949option splice-auto
4950no option splice-auto
4951 Enable or disable automatic kernel acceleration on sockets in both directions
4952 May be used in sections : defaults | frontend | listen | backend
4953 yes | yes | yes | yes
4954 Arguments : none
4955
4956 When this option is enabled either on a frontend or on a backend, haproxy
4957 will automatically evaluate the opportunity to use kernel tcp splicing to
4958 forward data between the client and the server, in either direction. Haproxy
4959 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004960 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004961 are not much aggressive in order to limit excessive use of splicing. This
4962 option requires splicing to be enabled at compile time, and may be globally
4963 disabled with the global option "nosplice". Since splice uses pipes, using it
4964 requires that there are enough spare pipes.
4965
4966 Important note: kernel-based TCP splicing is a Linux-specific feature which
4967 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
4968 transfer data between sockets without copying these data to user-space, thus
4969 providing noticeable performance gains and CPU cycles savings. Since many
4970 early implementations are buggy, corrupt data and/or are inefficient, this
4971 feature is not enabled by default, and it should be used with extreme care.
4972 While it is not possible to detect the correctness of an implementation,
4973 2.6.29 is the first version offering a properly working implementation. In
4974 case of doubt, splicing may be globally disabled using the global "nosplice"
4975 keyword.
4976
4977 Example :
4978 option splice-auto
4979
4980 If this option has been enabled in a "defaults" section, it can be disabled
4981 in a specific instance by prepending the "no" keyword before it.
4982
4983 See also : "option splice-request", "option splice-response", and global
4984 options "nosplice" and "maxpipes"
4985
4986
4987option splice-request
4988no option splice-request
4989 Enable or disable automatic kernel acceleration on sockets for requests
4990 May be used in sections : defaults | frontend | listen | backend
4991 yes | yes | yes | yes
4992 Arguments : none
4993
4994 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004995 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01004996 the client to the server. It might still use the recv/send scheme if there
4997 are no spare pipes left. This option requires splicing to be enabled at
4998 compile time, and may be globally disabled with the global option "nosplice".
4999 Since splice uses pipes, using it requires that there are enough spare pipes.
5000
5001 Important note: see "option splice-auto" for usage limitations.
5002
5003 Example :
5004 option splice-request
5005
5006 If this option has been enabled in a "defaults" section, it can be disabled
5007 in a specific instance by prepending the "no" keyword before it.
5008
5009 See also : "option splice-auto", "option splice-response", and global options
5010 "nosplice" and "maxpipes"
5011
5012
5013option splice-response
5014no option splice-response
5015 Enable or disable automatic kernel acceleration on sockets for responses
5016 May be used in sections : defaults | frontend | listen | backend
5017 yes | yes | yes | yes
5018 Arguments : none
5019
5020 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005021 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01005022 the server to the client. It might still use the recv/send scheme if there
5023 are no spare pipes left. This option requires splicing to be enabled at
5024 compile time, and may be globally disabled with the global option "nosplice".
5025 Since splice uses pipes, using it requires that there are enough spare pipes.
5026
5027 Important note: see "option splice-auto" for usage limitations.
5028
5029 Example :
5030 option splice-response
5031
5032 If this option has been enabled in a "defaults" section, it can be disabled
5033 in a specific instance by prepending the "no" keyword before it.
5034
5035 See also : "option splice-auto", "option splice-request", and global options
5036 "nosplice" and "maxpipes"
5037
5038
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005039option srvtcpka
5040no option srvtcpka
5041 Enable or disable the sending of TCP keepalive packets on the server side
5042 May be used in sections : defaults | frontend | listen | backend
5043 yes | no | yes | yes
5044 Arguments : none
5045
5046 When there is a firewall or any session-aware component between a client and
5047 a server, and when the protocol involves very long sessions with long idle
5048 periods (eg: remote desktops), there is a risk that one of the intermediate
5049 components decides to expire a session which has remained idle for too long.
5050
5051 Enabling socket-level TCP keep-alives makes the system regularly send packets
5052 to the other end of the connection, leaving it active. The delay between
5053 keep-alive probes is controlled by the system only and depends both on the
5054 operating system and its tuning parameters.
5055
5056 It is important to understand that keep-alive packets are neither emitted nor
5057 received at the application level. It is only the network stacks which sees
5058 them. For this reason, even if one side of the proxy already uses keep-alives
5059 to maintain its connection alive, those keep-alive packets will not be
5060 forwarded to the other side of the proxy.
5061
5062 Please note that this has nothing to do with HTTP keep-alive.
5063
5064 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
5065 server side of a connection, which should help when session expirations are
5066 noticed between HAProxy and a server.
5067
5068 If this option has been enabled in a "defaults" section, it can be disabled
5069 in a specific instance by prepending the "no" keyword before it.
5070
5071 See also : "option clitcpka", "option tcpka"
5072
5073
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005074option ssl-hello-chk
5075 Use SSLv3 client hello health checks for server testing
5076 May be used in sections : defaults | frontend | listen | backend
5077 yes | no | yes | yes
5078 Arguments : none
5079
5080 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
5081 possible to test that the server correctly talks SSL instead of just testing
5082 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
5083 SSLv3 client hello messages are sent once the connection is established to
5084 the server, and the response is analyzed to find an SSL server hello message.
5085 The server is considered valid only when the response contains this server
5086 hello message.
5087
5088 All servers tested till there correctly reply to SSLv3 client hello messages,
5089 and most servers tested do not even log the requests containing only hello
5090 messages, which is appreciable.
5091
Willy Tarreau763a95b2012-10-04 23:15:39 +02005092 Note that this check works even when SSL support was not built into haproxy
5093 because it forges the SSL message. When SSL support is available, it is best
5094 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005095
Willy Tarreau763a95b2012-10-04 23:15:39 +02005096 See also: "option httpchk", "check-ssl"
5097
Willy Tarreaua453bdd2008-01-08 19:50:52 +01005098
Willy Tarreaued179852013-12-16 01:07:00 +01005099option tcp-check
5100 Perform health checks using tcp-check send/expect sequences
5101 May be used in sections: defaults | frontend | listen | backend
5102 yes | no | yes | yes
5103
5104 This health check method is intended to be combined with "tcp-check" command
5105 lists in order to support send/expect types of health check sequences.
5106
5107 TCP checks currently support 4 modes of operations :
5108 - no "tcp-check" directive : the health check only consists in a connection
5109 attempt, which remains the default mode.
5110
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005111 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005112 used to send a string along with a connection opening. With some
5113 protocols, it helps sending a "QUIT" message for example that prevents
5114 the server from logging a connection error for each health check. The
5115 check result will still be based on the ability to open the connection
5116 only.
5117
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005118 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01005119 The connection is opened and haproxy waits for the server to present some
5120 contents which must validate some rules. The check result will be based
5121 on the matching between the contents and the rules. This is suited for
5122 POP, IMAP, SMTP, FTP, SSH, TELNET.
5123
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005124 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01005125 used to test a hello-type protocol. Haproxy sends a message, the server
5126 responds and its response is analysed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005127 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01005128 suited for protocols which require a binding or a request/response model.
5129 LDAP, MySQL, Redis and SSL are example of such protocols, though they
5130 already all have their dedicated checks with a deeper understanding of
5131 the respective protocols.
5132 In this mode, many questions may be sent and many answers may be
5133 analysed.
5134
5135 Examples :
5136 # perform a POP check (analyse only server's banner)
5137 option tcp-check
5138 tcp-check expect string +OK\ POP3\ ready
5139
5140 # perform an IMAP check (analyse only server's banner)
5141 option tcp-check
5142 tcp-check expect string *\ OK\ IMAP4\ ready
5143
5144 # look for the redis master server after ensuring it speaks well
5145 # redis protocol, then it exits properly.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005146 # (send a command then analyse the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01005147 option tcp-check
5148 tcp-check send PING\r\n
5149 tcp-check expect +PONG
5150 tcp-check send info\ replication\r\n
5151 tcp-check expect string role:master
5152 tcp-check send QUIT\r\n
5153 tcp-check expect string +OK
5154
5155 forge a HTTP request, then analyse the response
5156 (send many headers before analyzing)
5157 option tcp-check
5158 tcp-check send HEAD\ /\ HTTP/1.1\r\n
5159 tcp-check send Host:\ www.mydomain.com\r\n
5160 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
5161 tcp-check send \r\n
5162 tcp-check expect rstring HTTP/1\..\ (2..|3..)
5163
5164
5165 See also : "tcp-check expect", "tcp-check send"
5166
5167
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005168option tcp-smart-accept
5169no option tcp-smart-accept
5170 Enable or disable the saving of one ACK packet during the accept sequence
5171 May be used in sections : defaults | frontend | listen | backend
5172 yes | yes | yes | no
5173 Arguments : none
5174
5175 When an HTTP connection request comes in, the system acknowledges it on
5176 behalf of HAProxy, then the client immediately sends its request, and the
5177 system acknowledges it too while it is notifying HAProxy about the new
5178 connection. HAProxy then reads the request and responds. This means that we
5179 have one TCP ACK sent by the system for nothing, because the request could
5180 very well be acknowledged by HAProxy when it sends its response.
5181
5182 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
5183 sending this useless ACK on platforms which support it (currently at least
5184 Linux). It must not cause any problem, because the system will send it anyway
5185 after 40 ms if the response takes more time than expected to come.
5186
5187 During complex network debugging sessions, it may be desirable to disable
5188 this optimization because delayed ACKs can make troubleshooting more complex
5189 when trying to identify where packets are delayed. It is then possible to
5190 fall back to normal behaviour by specifying "no option tcp-smart-accept".
5191
5192 It is also possible to force it for non-HTTP proxies by simply specifying
5193 "option tcp-smart-accept". For instance, it can make sense with some services
5194 such as SMTP where the server speaks first.
5195
5196 It is recommended to avoid forcing this option in a defaults section. In case
5197 of doubt, consider setting it back to automatic values by prepending the
5198 "default" keyword before it, or disabling it using the "no" keyword.
5199
Willy Tarreaud88edf22009-06-14 15:48:17 +02005200 See also : "option tcp-smart-connect"
5201
5202
5203option tcp-smart-connect
5204no option tcp-smart-connect
5205 Enable or disable the saving of one ACK packet during the connect sequence
5206 May be used in sections : defaults | frontend | listen | backend
5207 yes | no | yes | yes
5208 Arguments : none
5209
5210 On certain systems (at least Linux), HAProxy can ask the kernel not to
5211 immediately send an empty ACK upon a connection request, but to directly
5212 send the buffer request instead. This saves one packet on the network and
5213 thus boosts performance. It can also be useful for some servers, because they
5214 immediately get the request along with the incoming connection.
5215
5216 This feature is enabled when "option tcp-smart-connect" is set in a backend.
5217 It is not enabled by default because it makes network troubleshooting more
5218 complex.
5219
5220 It only makes sense to enable it with protocols where the client speaks first
5221 such as HTTP. In other situations, if there is no data to send in place of
5222 the ACK, a normal ACK is sent.
5223
5224 If this option has been enabled in a "defaults" section, it can be disabled
5225 in a specific instance by prepending the "no" keyword before it.
5226
5227 See also : "option tcp-smart-accept"
5228
Willy Tarreau9ea05a72009-06-14 12:07:01 +02005229
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005230option tcpka
5231 Enable or disable the sending of TCP keepalive packets on both sides
5232 May be used in sections : defaults | frontend | listen | backend
5233 yes | yes | yes | yes
5234 Arguments : none
5235
5236 When there is a firewall or any session-aware component between a client and
5237 a server, and when the protocol involves very long sessions with long idle
5238 periods (eg: remote desktops), there is a risk that one of the intermediate
5239 components decides to expire a session which has remained idle for too long.
5240
5241 Enabling socket-level TCP keep-alives makes the system regularly send packets
5242 to the other end of the connection, leaving it active. The delay between
5243 keep-alive probes is controlled by the system only and depends both on the
5244 operating system and its tuning parameters.
5245
5246 It is important to understand that keep-alive packets are neither emitted nor
5247 received at the application level. It is only the network stacks which sees
5248 them. For this reason, even if one side of the proxy already uses keep-alives
5249 to maintain its connection alive, those keep-alive packets will not be
5250 forwarded to the other side of the proxy.
5251
5252 Please note that this has nothing to do with HTTP keep-alive.
5253
5254 Using option "tcpka" enables the emission of TCP keep-alive probes on both
5255 the client and server sides of a connection. Note that this is meaningful
5256 only in "defaults" or "listen" sections. If this option is used in a
5257 frontend, only the client side will get keep-alives, and if this option is
5258 used in a backend, only the server side will get keep-alives. For this
5259 reason, it is strongly recommended to explicitly use "option clitcpka" and
5260 "option srvtcpka" when the configuration is split between frontends and
5261 backends.
5262
5263 See also : "option clitcpka", "option srvtcpka"
5264
Willy Tarreau844e3c52008-01-11 16:28:18 +01005265
5266option tcplog
5267 Enable advanced logging of TCP connections with session state and timers
5268 May be used in sections : defaults | frontend | listen | backend
5269 yes | yes | yes | yes
5270 Arguments : none
5271
5272 By default, the log output format is very poor, as it only contains the
5273 source and destination addresses, and the instance name. By specifying
5274 "option tcplog", each log line turns into a much richer format including, but
5275 not limited to, the connection timers, the session status, the connections
5276 numbers, the frontend, backend and server name, and of course the source
5277 address and ports. This option is useful for pure TCP proxies in order to
5278 find which of the client or server disconnects or times out. For normal HTTP
5279 proxies, it's better to use "option httplog" which is even more complete.
5280
5281 This option may be set either in the frontend or the backend.
5282
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005283 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005284
5285
Willy Tarreau844e3c52008-01-11 16:28:18 +01005286option transparent
5287no option transparent
5288 Enable client-side transparent proxying
5289 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01005290 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01005291 Arguments : none
5292
5293 This option was introduced in order to provide layer 7 persistence to layer 3
5294 load balancers. The idea is to use the OS's ability to redirect an incoming
5295 connection for a remote address to a local process (here HAProxy), and let
5296 this process know what address was initially requested. When this option is
5297 used, sessions without cookies will be forwarded to the original destination
5298 IP address of the incoming request (which should match that of another
5299 equipment), while requests with cookies will still be forwarded to the
5300 appropriate server.
5301
5302 Note that contrary to a common belief, this option does NOT make HAProxy
5303 present the client's IP to the server when establishing the connection.
5304
Willy Tarreaua1146052011-03-01 09:51:54 +01005305 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005306 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01005307
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005308
Simon Horman98637e52014-06-20 12:30:16 +09005309external-check command <command>
5310 Executable to run when performing an external-check
5311 May be used in sections : defaults | frontend | listen | backend
5312 yes | no | yes | yes
5313
5314 Arguments :
5315 <command> is the external command to run
5316
5317 The PATH environment variable used when executing the
5318 command may be set using "external-check path".
5319
5320 The arguments passed to the to the command are:
5321
5322 proxy_address proxy_port server_address server_port
5323
5324 The proxy_address and proxy_port are derived from the first listener
5325 that is either IPv4, IPv6 or a UNIX socket. It is an error for no such
5326 listeners to exist. In the case of a UNIX socket listener the
5327 proxy_address will be the path of the socket and the proxy_port will
5328 be the string "NOT_USED".
5329
5330 If the command executed and exits with a zero status then the check is
5331 considered to have passed, otherwise the check is considered to have
5332 failed.
5333
5334 Example :
5335 external-check command /bin/true
5336
5337 See also : "external-check", "option external-check", "external-check path"
5338
5339
5340external-check path <path>
5341 The value of the PATH environment variable used when running an external-check
5342 May be used in sections : defaults | frontend | listen | backend
5343 yes | no | yes | yes
5344
5345 Arguments :
5346 <path> is the path used when executing external command to run
5347
5348 The default path is "".
5349
5350 Example :
5351 external-check path "/usr/bin:/bin"
5352
5353 See also : "external-check", "option external-check",
5354 "external-check command"
5355
5356
Emeric Brun647caf12009-06-30 17:57:00 +02005357persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02005358persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02005359 Enable RDP cookie-based persistence
5360 May be used in sections : defaults | frontend | listen | backend
5361 yes | no | yes | yes
5362 Arguments :
5363 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02005364 default cookie name "msts" will be used. There currently is no
5365 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02005366
5367 This statement enables persistence based on an RDP cookie. The RDP cookie
5368 contains all information required to find the server in the list of known
5369 servers. So when this option is set in the backend, the request is analysed
5370 and if an RDP cookie is found, it is decoded. If it matches a known server
5371 which is still UP (or if "option persist" is set), then the connection is
5372 forwarded to this server.
5373
5374 Note that this only makes sense in a TCP backend, but for this to work, the
5375 frontend must have waited long enough to ensure that an RDP cookie is present
5376 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005377 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02005378 a single "listen" section.
5379
Willy Tarreau61e28f22010-05-16 22:31:05 +02005380 Also, it is important to understand that the terminal server will emit this
5381 RDP cookie only if it is configured for "token redirection mode", which means
5382 that the "IP address redirection" option is disabled.
5383
Emeric Brun647caf12009-06-30 17:57:00 +02005384 Example :
5385 listen tse-farm
5386 bind :3389
5387 # wait up to 5s for an RDP cookie in the request
5388 tcp-request inspect-delay 5s
5389 tcp-request content accept if RDP_COOKIE
5390 # apply RDP cookie persistence
5391 persist rdp-cookie
5392 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02005393 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02005394 balance rdp-cookie
5395 server srv1 1.1.1.1:3389
5396 server srv2 1.1.1.2:3389
5397
Simon Hormanab814e02011-06-24 14:50:20 +09005398 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
5399 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02005400
5401
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005402rate-limit sessions <rate>
5403 Set a limit on the number of new sessions accepted per second on a frontend
5404 May be used in sections : defaults | frontend | listen | backend
5405 yes | yes | yes | no
5406 Arguments :
5407 <rate> The <rate> parameter is an integer designating the maximum number
5408 of new sessions per second to accept on the frontend.
5409
5410 When the frontend reaches the specified number of new sessions per second, it
5411 stops accepting new connections until the rate drops below the limit again.
5412 During this time, the pending sessions will be kept in the socket's backlog
5413 (in system buffers) and haproxy will not even be aware that sessions are
5414 pending. When applying very low limit on a highly loaded service, it may make
5415 sense to increase the socket's backlog using the "backlog" keyword.
5416
5417 This feature is particularly efficient at blocking connection-based attacks
5418 or service abuse on fragile servers. Since the session rate is measured every
5419 millisecond, it is extremely accurate. Also, the limit applies immediately,
5420 no delay is needed at all to detect the threshold.
5421
5422 Example : limit the connection rate on SMTP to 10 per second max
5423 listen smtp
5424 mode tcp
5425 bind :25
5426 rate-limit sessions 10
5427 server 127.0.0.1:1025
5428
Willy Tarreaua17c2d92011-07-25 08:16:20 +02005429 Note : when the maximum rate is reached, the frontend's status is not changed
5430 but its sockets appear as "WAITING" in the statistics if the
5431 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01005432
5433 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
5434
5435
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005436redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
5437redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
5438redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005439 Return an HTTP redirection if/unless a condition is matched
5440 May be used in sections : defaults | frontend | listen | backend
5441 no | yes | yes | yes
5442
5443 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01005444 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005445
Willy Tarreau0140f252008-11-19 21:07:09 +01005446 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005447 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005448 the HTTP "Location" header. When used in an "http-request" rule,
5449 <loc> value follows the log-format rules and can include some
5450 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005451
5452 <pfx> With "redirect prefix", the "Location" header is built from the
5453 concatenation of <pfx> and the complete URI path, including the
5454 query string, unless the "drop-query" option is specified (see
5455 below). As a special case, if <pfx> equals exactly "/", then
5456 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005457 redirect to the same URL (for instance, to insert a cookie). When
5458 used in an "http-request" rule, <pfx> value follows the log-format
5459 rules and can include some dynamic values (see Custom Log Format
5460 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005461
5462 <sch> With "redirect scheme", then the "Location" header is built by
5463 concatenating <sch> with "://" then the first occurrence of the
5464 "Host" header, and then the URI path, including the query string
5465 unless the "drop-query" option is specified (see below). If no
5466 path is found or if the path is "*", then "/" is used instead. If
5467 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005468 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005469 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005470 HTTPS. When used in an "http-request" rule, <sch> value follows
5471 the log-format rules and can include some dynamic values (see
5472 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01005473
5474 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01005475 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
5476 with 302 used by default if no code is specified. 301 means
5477 "Moved permanently", and a browser may cache the Location. 302
5478 means "Moved permanently" and means that the browser should not
5479 cache the redirection. 303 is equivalent to 302 except that the
5480 browser will fetch the location with a GET method. 307 is just
5481 like 302 but makes it clear that the same method must be reused.
5482 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01005483
5484 <option> There are several options which can be specified to adjust the
5485 expected behaviour of a redirection :
5486
5487 - "drop-query"
5488 When this keyword is used in a prefix-based redirection, then the
5489 location will be set without any possible query-string, which is useful
5490 for directing users to a non-secure page for instance. It has no effect
5491 with a location-type redirect.
5492
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005493 - "append-slash"
5494 This keyword may be used in conjunction with "drop-query" to redirect
5495 users who use a URL not ending with a '/' to the same one with the '/'.
5496 It can be useful to ensure that search engines will only see one URL.
5497 For this, a return code 301 is preferred.
5498
Willy Tarreau0140f252008-11-19 21:07:09 +01005499 - "set-cookie NAME[=value]"
5500 A "Set-Cookie" header will be added with NAME (and optionally "=value")
5501 to the response. This is sometimes used to indicate that a user has
5502 been seen, for instance to protect against some types of DoS. No other
5503 cookie option is added, so the cookie will be a session cookie. Note
5504 that for a browser, a sole cookie name without an equal sign is
5505 different from a cookie with an equal sign.
5506
5507 - "clear-cookie NAME[=]"
5508 A "Set-Cookie" header will be added with NAME (and optionally "="), but
5509 with the "Max-Age" attribute set to zero. This will tell the browser to
5510 delete this cookie. It is useful for instance on logout pages. It is
5511 important to note that clearing the cookie "NAME" will not remove a
5512 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
5513 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005514
5515 Example: move the login URL only to HTTPS.
5516 acl clear dst_port 80
5517 acl secure dst_port 8080
5518 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01005519 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01005520 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01005521 acl cookie_set hdr_sub(cookie) SEEN=1
5522
5523 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01005524 redirect prefix https://mysite.com if login_page !secure
5525 redirect prefix http://mysite.com drop-query if login_page !uid_given
5526 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01005527 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005528
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01005529 Example: send redirects for request for articles without a '/'.
5530 acl missing_slash path_reg ^/article/[^/]*$
5531 redirect code 301 prefix / drop-query append-slash if missing_slash
5532
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005533 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01005534 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02005535
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01005536 Example: append 'www.' prefix in front of all hosts not having it
5537 http-request redirect code 301 location www.%[hdr(host)]%[req.uri] \
5538 unless { hdr_beg(host) -i www }
5539
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005540 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02005541
5542
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005543redisp (deprecated)
5544redispatch (deprecated)
5545 Enable or disable session redistribution in case of connection failure
5546 May be used in sections: defaults | frontend | listen | backend
5547 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005548 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005549
5550 In HTTP mode, if a server designated by a cookie is down, clients may
5551 definitely stick to it because they cannot flush the cookie, so they will not
5552 be able to access the service anymore.
5553
5554 Specifying "redispatch" will allow the proxy to break their persistence and
5555 redistribute them to a working server.
5556
5557 It also allows to retry last connection to another server in case of multiple
5558 connection failures. Of course, it requires having "retries" set to a nonzero
5559 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01005560
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01005561 This form is deprecated, do not use it in any new configuration, use the new
5562 "option redispatch" instead.
5563
5564 See also : "option redispatch"
5565
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005566
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005567reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005568 Add a header at the end of the HTTP request
5569 May be used in sections : defaults | frontend | listen | backend
5570 no | yes | yes | yes
5571 Arguments :
5572 <string> is the complete line to be added. Any space or known delimiter
5573 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005574 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005575
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005576 <cond> is an optional matching condition built from ACLs. It makes it
5577 possible to ignore this rule when other conditions are not met.
5578
Willy Tarreau303c0352008-01-17 19:01:39 +01005579 A new line consisting in <string> followed by a line feed will be added after
5580 the last header of an HTTP request.
5581
5582 Header transformations only apply to traffic which passes through HAProxy,
5583 and not to traffic generated by HAProxy, such as health-checks or error
5584 responses.
5585
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01005586 Example : add "X-Proto: SSL" to requests coming via port 81
5587 acl is-ssl dst_port 81
5588 reqadd X-Proto:\ SSL if is-ssl
5589
5590 See also: "rspadd", section 6 about HTTP header manipulation, and section 7
5591 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005592
5593
Willy Tarreau5321c422010-01-28 20:35:13 +01005594reqallow <search> [{if | unless} <cond>]
5595reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005596 Definitely allow an HTTP request if a line matches a regular expression
5597 May be used in sections : defaults | frontend | listen | backend
5598 no | yes | yes | yes
5599 Arguments :
5600 <search> is the regular expression applied to HTTP headers and to the
5601 request line. This is an extended regular expression. Parenthesis
5602 grouping is supported and no preliminary backslash is required.
5603 Any space or known delimiter must be escaped using a backslash
5604 ('\'). The pattern applies to a full line at a time. The
5605 "reqallow" keyword strictly matches case while "reqiallow"
5606 ignores case.
5607
Willy Tarreau5321c422010-01-28 20:35:13 +01005608 <cond> is an optional matching condition built from ACLs. It makes it
5609 possible to ignore this rule when other conditions are not met.
5610
Willy Tarreau303c0352008-01-17 19:01:39 +01005611 A request containing any line which matches extended regular expression
5612 <search> will mark the request as allowed, even if any later test would
5613 result in a deny. The test applies both to the request line and to request
5614 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005615 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005616
5617 It is easier, faster and more powerful to use ACLs to write access policies.
5618 Reqdeny, reqallow and reqpass should be avoided in new designs.
5619
5620 Example :
5621 # allow www.* but refuse *.local
5622 reqiallow ^Host:\ www\.
5623 reqideny ^Host:\ .*\.local
5624
Willy Tarreau5321c422010-01-28 20:35:13 +01005625 See also: "reqdeny", "block", section 6 about HTTP header manipulation, and
5626 section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005627
5628
Willy Tarreau5321c422010-01-28 20:35:13 +01005629reqdel <search> [{if | unless} <cond>]
5630reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005631 Delete all headers matching a regular expression in an HTTP request
5632 May be used in sections : defaults | frontend | listen | backend
5633 no | yes | yes | yes
5634 Arguments :
5635 <search> is the regular expression applied to HTTP headers and to the
5636 request line. This is an extended regular expression. Parenthesis
5637 grouping is supported and no preliminary backslash is required.
5638 Any space or known delimiter must be escaped using a backslash
5639 ('\'). The pattern applies to a full line at a time. The "reqdel"
5640 keyword strictly matches case while "reqidel" ignores case.
5641
Willy Tarreau5321c422010-01-28 20:35:13 +01005642 <cond> is an optional matching condition built from ACLs. It makes it
5643 possible to ignore this rule when other conditions are not met.
5644
Willy Tarreau303c0352008-01-17 19:01:39 +01005645 Any header line matching extended regular expression <search> in the request
5646 will be completely deleted. Most common use of this is to remove unwanted
5647 and/or dangerous headers or cookies from a request before passing it to the
5648 next servers.
5649
5650 Header transformations only apply to traffic which passes through HAProxy,
5651 and not to traffic generated by HAProxy, such as health-checks or error
5652 responses. Keep in mind that header names are not case-sensitive.
5653
5654 Example :
5655 # remove X-Forwarded-For header and SERVER cookie
5656 reqidel ^X-Forwarded-For:.*
5657 reqidel ^Cookie:.*SERVER=
5658
Willy Tarreau5321c422010-01-28 20:35:13 +01005659 See also: "reqadd", "reqrep", "rspdel", section 6 about HTTP header
5660 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005661
5662
Willy Tarreau5321c422010-01-28 20:35:13 +01005663reqdeny <search> [{if | unless} <cond>]
5664reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005665 Deny an HTTP request if a line matches a regular expression
5666 May be used in sections : defaults | frontend | listen | backend
5667 no | yes | yes | yes
5668 Arguments :
5669 <search> is the regular expression applied to HTTP headers and to the
5670 request line. This is an extended regular expression. Parenthesis
5671 grouping is supported and no preliminary backslash is required.
5672 Any space or known delimiter must be escaped using a backslash
5673 ('\'). The pattern applies to a full line at a time. The
5674 "reqdeny" keyword strictly matches case while "reqideny" ignores
5675 case.
5676
Willy Tarreau5321c422010-01-28 20:35:13 +01005677 <cond> is an optional matching condition built from ACLs. It makes it
5678 possible to ignore this rule when other conditions are not met.
5679
Willy Tarreau303c0352008-01-17 19:01:39 +01005680 A request containing any line which matches extended regular expression
5681 <search> will mark the request as denied, even if any later test would
5682 result in an allow. The test applies both to the request line and to request
5683 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01005684 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01005685
Willy Tarreauced27012008-01-17 20:35:34 +01005686 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005687 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01005688 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01005689
Willy Tarreau303c0352008-01-17 19:01:39 +01005690 It is easier, faster and more powerful to use ACLs to write access policies.
5691 Reqdeny, reqallow and reqpass should be avoided in new designs.
5692
5693 Example :
5694 # refuse *.local, then allow www.*
5695 reqideny ^Host:\ .*\.local
5696 reqiallow ^Host:\ www\.
5697
Willy Tarreau5321c422010-01-28 20:35:13 +01005698 See also: "reqallow", "rspdeny", "block", section 6 about HTTP header
5699 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005700
5701
Willy Tarreau5321c422010-01-28 20:35:13 +01005702reqpass <search> [{if | unless} <cond>]
5703reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005704 Ignore any HTTP request line matching a regular expression in next rules
5705 May be used in sections : defaults | frontend | listen | backend
5706 no | yes | yes | yes
5707 Arguments :
5708 <search> is the regular expression applied to HTTP headers and to the
5709 request line. This is an extended regular expression. Parenthesis
5710 grouping is supported and no preliminary backslash is required.
5711 Any space or known delimiter must be escaped using a backslash
5712 ('\'). The pattern applies to a full line at a time. The
5713 "reqpass" keyword strictly matches case while "reqipass" ignores
5714 case.
5715
Willy Tarreau5321c422010-01-28 20:35:13 +01005716 <cond> is an optional matching condition built from ACLs. It makes it
5717 possible to ignore this rule when other conditions are not met.
5718
Willy Tarreau303c0352008-01-17 19:01:39 +01005719 A request containing any line which matches extended regular expression
5720 <search> will skip next rules, without assigning any deny or allow verdict.
5721 The test applies both to the request line and to request headers. Keep in
5722 mind that URLs in request line are case-sensitive while header names are not.
5723
5724 It is easier, faster and more powerful to use ACLs to write access policies.
5725 Reqdeny, reqallow and reqpass should be avoided in new designs.
5726
5727 Example :
5728 # refuse *.local, then allow www.*, but ignore "www.private.local"
5729 reqipass ^Host:\ www.private\.local
5730 reqideny ^Host:\ .*\.local
5731 reqiallow ^Host:\ www\.
5732
Willy Tarreau5321c422010-01-28 20:35:13 +01005733 See also: "reqallow", "reqdeny", "block", section 6 about HTTP header
5734 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005735
5736
Willy Tarreau5321c422010-01-28 20:35:13 +01005737reqrep <search> <string> [{if | unless} <cond>]
5738reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005739 Replace a regular expression with a string in an HTTP request line
5740 May be used in sections : defaults | frontend | listen | backend
5741 no | yes | yes | yes
5742 Arguments :
5743 <search> is the regular expression applied to HTTP headers and to the
5744 request line. This is an extended regular expression. Parenthesis
5745 grouping is supported and no preliminary backslash is required.
5746 Any space or known delimiter must be escaped using a backslash
5747 ('\'). The pattern applies to a full line at a time. The "reqrep"
5748 keyword strictly matches case while "reqirep" ignores case.
5749
5750 <string> is the complete line to be added. Any space or known delimiter
5751 must be escaped using a backslash ('\'). References to matched
5752 pattern groups are possible using the common \N form, with N
5753 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005754 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005755
Willy Tarreau5321c422010-01-28 20:35:13 +01005756 <cond> is an optional matching condition built from ACLs. It makes it
5757 possible to ignore this rule when other conditions are not met.
5758
Willy Tarreau303c0352008-01-17 19:01:39 +01005759 Any line matching extended regular expression <search> in the request (both
5760 the request line and header lines) will be completely replaced with <string>.
5761 Most common use of this is to rewrite URLs or domain names in "Host" headers.
5762
5763 Header transformations only apply to traffic which passes through HAProxy,
5764 and not to traffic generated by HAProxy, such as health-checks or error
5765 responses. Note that for increased readability, it is suggested to add enough
5766 spaces between the request and the response. Keep in mind that URLs in
5767 request line are case-sensitive while header names are not.
5768
5769 Example :
5770 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04005771 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01005772 # replace "www.mydomain.com" with "www" in the host name.
5773 reqirep ^Host:\ www.mydomain.com Host:\ www
5774
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04005775 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", section 6 about
5776 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005777
5778
Willy Tarreau5321c422010-01-28 20:35:13 +01005779reqtarpit <search> [{if | unless} <cond>]
5780reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005781 Tarpit an HTTP request containing a line matching a regular expression
5782 May be used in sections : defaults | frontend | listen | backend
5783 no | yes | yes | yes
5784 Arguments :
5785 <search> is the regular expression applied to HTTP headers and to the
5786 request line. This is an extended regular expression. Parenthesis
5787 grouping is supported and no preliminary backslash is required.
5788 Any space or known delimiter must be escaped using a backslash
5789 ('\'). The pattern applies to a full line at a time. The
5790 "reqtarpit" keyword strictly matches case while "reqitarpit"
5791 ignores case.
5792
Willy Tarreau5321c422010-01-28 20:35:13 +01005793 <cond> is an optional matching condition built from ACLs. It makes it
5794 possible to ignore this rule when other conditions are not met.
5795
Willy Tarreau303c0352008-01-17 19:01:39 +01005796 A request containing any line which matches extended regular expression
5797 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01005798 be kept open for a pre-defined time, then will return an HTTP error 500 so
5799 that the attacker does not suspect it has been tarpitted. The status 500 will
5800 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01005801 delay is defined by "timeout tarpit", or "timeout connect" if the former is
5802 not set.
5803
5804 The goal of the tarpit is to slow down robots attacking servers with
5805 identifiable requests. Many robots limit their outgoing number of connections
5806 and stay connected waiting for a reply which can take several minutes to
5807 come. Depending on the environment and attack, it may be particularly
5808 efficient at reducing the load on the network and firewalls.
5809
Willy Tarreau5321c422010-01-28 20:35:13 +01005810 Examples :
Willy Tarreau303c0352008-01-17 19:01:39 +01005811 # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but
5812 # block all others.
5813 reqipass ^User-Agent:\.*(Mozilla|MSIE)
5814 reqitarpit ^User-Agent:
5815
Willy Tarreau5321c422010-01-28 20:35:13 +01005816 # block bad guys
5817 acl badguys src 10.1.0.3 172.16.13.20/28
5818 reqitarpit . if badguys
5819
5820 See also: "reqallow", "reqdeny", "reqpass", section 6 about HTTP header
5821 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005822
5823
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02005824retries <value>
5825 Set the number of retries to perform on a server after a connection failure
5826 May be used in sections: defaults | frontend | listen | backend
5827 yes | no | yes | yes
5828 Arguments :
5829 <value> is the number of times a connection attempt should be retried on
5830 a server when a connection either is refused or times out. The
5831 default value is 3.
5832
5833 It is important to understand that this value applies to the number of
5834 connection attempts, not full requests. When a connection has effectively
5835 been established to a server, there will be no more retry.
5836
5837 In order to avoid immediate reconnections to a server which is restarting,
5838 a turn-around timer of 1 second is applied before a retry occurs.
5839
5840 When "option redispatch" is set, the last retry may be performed on another
5841 server even if a cookie references a different server.
5842
5843 See also : "option redispatch"
5844
5845
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005846rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01005847 Add a header at the end of the HTTP response
5848 May be used in sections : defaults | frontend | listen | backend
5849 no | yes | yes | yes
5850 Arguments :
5851 <string> is the complete line to be added. Any space or known delimiter
5852 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005853 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005854
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005855 <cond> is an optional matching condition built from ACLs. It makes it
5856 possible to ignore this rule when other conditions are not met.
5857
Willy Tarreau303c0352008-01-17 19:01:39 +01005858 A new line consisting in <string> followed by a line feed will be added after
5859 the last header of an HTTP response.
5860
5861 Header transformations only apply to traffic which passes through HAProxy,
5862 and not to traffic generated by HAProxy, such as health-checks or error
5863 responses.
5864
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005865 See also: "reqadd", section 6 about HTTP header manipulation, and section 7
5866 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005867
5868
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005869rspdel <search> [{if | unless} <cond>]
5870rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005871 Delete all headers matching a regular expression in an HTTP response
5872 May be used in sections : defaults | frontend | listen | backend
5873 no | yes | yes | yes
5874 Arguments :
5875 <search> is the regular expression applied to HTTP headers and to the
5876 response line. This is an extended regular expression, so
5877 parenthesis grouping is supported and no preliminary backslash
5878 is required. Any space or known delimiter must be escaped using
5879 a backslash ('\'). The pattern applies to a full line at a time.
5880 The "rspdel" keyword strictly matches case while "rspidel"
5881 ignores case.
5882
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005883 <cond> is an optional matching condition built from ACLs. It makes it
5884 possible to ignore this rule when other conditions are not met.
5885
Willy Tarreau303c0352008-01-17 19:01:39 +01005886 Any header line matching extended regular expression <search> in the response
5887 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005888 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01005889 client.
5890
5891 Header transformations only apply to traffic which passes through HAProxy,
5892 and not to traffic generated by HAProxy, such as health-checks or error
5893 responses. Keep in mind that header names are not case-sensitive.
5894
5895 Example :
5896 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02005897 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01005898
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005899 See also: "rspadd", "rsprep", "reqdel", section 6 about HTTP header
5900 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005901
5902
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005903rspdeny <search> [{if | unless} <cond>]
5904rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005905 Block an HTTP response if a line matches a regular expression
5906 May be used in sections : defaults | frontend | listen | backend
5907 no | yes | yes | yes
5908 Arguments :
5909 <search> is the regular expression applied to HTTP headers and to the
5910 response line. This is an extended regular expression, so
5911 parenthesis grouping is supported and no preliminary backslash
5912 is required. Any space or known delimiter must be escaped using
5913 a backslash ('\'). The pattern applies to a full line at a time.
5914 The "rspdeny" keyword strictly matches case while "rspideny"
5915 ignores case.
5916
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005917 <cond> is an optional matching condition built from ACLs. It makes it
5918 possible to ignore this rule when other conditions are not met.
5919
Willy Tarreau303c0352008-01-17 19:01:39 +01005920 A response containing any line which matches extended regular expression
5921 <search> will mark the request as denied. The test applies both to the
5922 response line and to response headers. Keep in mind that header names are not
5923 case-sensitive.
5924
5925 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01005926 block the response before it reaches the client. If a response is denied, it
5927 will be replaced with an HTTP 502 error so that the client never retrieves
5928 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01005929
5930 It is easier, faster and more powerful to use ACLs to write access policies.
5931 Rspdeny should be avoided in new designs.
5932
5933 Example :
5934 # Ensure that no content type matching ms-word will leak
5935 rspideny ^Content-type:\.*/ms-word
5936
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005937 See also: "reqdeny", "acl", "block", section 6 about HTTP header manipulation
5938 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005939
5940
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005941rsprep <search> <string> [{if | unless} <cond>]
5942rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01005943 Replace a regular expression with a string in an HTTP response line
5944 May be used in sections : defaults | frontend | listen | backend
5945 no | yes | yes | yes
5946 Arguments :
5947 <search> is the regular expression applied to HTTP headers and to the
5948 response line. This is an extended regular expression, so
5949 parenthesis grouping is supported and no preliminary backslash
5950 is required. Any space or known delimiter must be escaped using
5951 a backslash ('\'). The pattern applies to a full line at a time.
5952 The "rsprep" keyword strictly matches case while "rspirep"
5953 ignores case.
5954
5955 <string> is the complete line to be added. Any space or known delimiter
5956 must be escaped using a backslash ('\'). References to matched
5957 pattern groups are possible using the common \N form, with N
5958 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005959 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01005960
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005961 <cond> is an optional matching condition built from ACLs. It makes it
5962 possible to ignore this rule when other conditions are not met.
5963
Willy Tarreau303c0352008-01-17 19:01:39 +01005964 Any line matching extended regular expression <search> in the response (both
5965 the response line and header lines) will be completely replaced with
5966 <string>. Most common use of this is to rewrite Location headers.
5967
5968 Header transformations only apply to traffic which passes through HAProxy,
5969 and not to traffic generated by HAProxy, such as health-checks or error
5970 responses. Note that for increased readability, it is suggested to add enough
5971 spaces between the request and the response. Keep in mind that header names
5972 are not case-sensitive.
5973
5974 Example :
5975 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
5976 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
5977
Willy Tarreaufdb563c2010-01-31 15:43:27 +01005978 See also: "rspadd", "rspdel", "reqrep", section 6 about HTTP header
5979 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01005980
5981
David du Colombier486df472011-03-17 10:40:26 +01005982server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005983 Declare a server in a backend
5984 May be used in sections : defaults | frontend | listen | backend
5985 no | no | yes | yes
5986 Arguments :
5987 <name> is the internal name assigned to this server. This name will
Cyril Bonté941a0c62012-10-15 19:44:24 +02005988 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05005989 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01005990
David du Colombier486df472011-03-17 10:40:26 +01005991 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
5992 resolvable hostname is supported, but this name will be resolved
5993 during start-up. Address "0.0.0.0" or "*" has a special meaning.
5994 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02005995 address as the one from the client connection. This is useful in
5996 transparent proxy architectures where the client's connection is
5997 intercepted and haproxy must forward to the original destination
5998 address. This is more or less what the "transparent" keyword does
5999 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01006000 to report statistics. Optionally, an address family prefix may be
6001 used before the address to force the family regardless of the
6002 address format, which can be useful to specify a path to a unix
6003 socket with no slash ('/'). Currently supported prefixes are :
6004 - 'ipv4@' -> address is always IPv4
6005 - 'ipv6@' -> address is always IPv6
6006 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02006007 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01006008 Any part of the address string may reference any number of
6009 environment variables by preceding their name with a dollar
6010 sign ('$') and optionally enclosing them with braces ('{}'),
6011 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006012
Willy Tarreaub6205fd2012-09-24 12:27:33 +02006013 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006014 be sent to this port. If unset, the same port the client
6015 connected to will be used. The port may also be prefixed by a "+"
6016 or a "-". In this case, the server's port will be determined by
6017 adding this value to the client's port.
6018
6019 <param*> is a list of parameters for this server. The "server" keywords
6020 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006021 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006022
6023 Examples :
6024 server first 10.1.1.1:1080 cookie first check inter 1000
6025 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01006026 server transp ipv4@
Willy Tarreaudad36a32013-03-11 01:20:04 +01006027 server backup ${SRV_BACKUP}:1080 backup
6028 server www1_dc1 ${LAN_DC1}.101:80
6029 server www1_dc2 ${LAN_DC2}.101:80
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006030
Mark Lamourinec2247f02012-01-04 13:02:01 -05006031 See also: "default-server", "http-send-name-header" and section 5 about
6032 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006033
6034
6035source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02006036source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006037source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006038 Set the source address for outgoing connections
6039 May be used in sections : defaults | frontend | listen | backend
6040 yes | no | yes | yes
6041 Arguments :
6042 <addr> is the IPv4 address HAProxy will bind to before connecting to a
6043 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01006044
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006045 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01006046 the most appropriate address to reach its destination. Optionally
6047 an address family prefix may be used before the address to force
6048 the family regardless of the address format, which can be useful
6049 to specify a path to a unix socket with no slash ('/'). Currently
6050 supported prefixes are :
6051 - 'ipv4@' -> address is always IPv4
6052 - 'ipv6@' -> address is always IPv6
6053 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02006054 - 'abns@' -> address is in abstract namespace (Linux only)
Willy Tarreaudad36a32013-03-11 01:20:04 +01006055 Any part of the address string may reference any number of
6056 environment variables by preceding their name with a dollar
6057 sign ('$') and optionally enclosing them with braces ('{}'),
6058 similarly to what is done in Bourne shell.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006059
6060 <port> is an optional port. It is normally not needed but may be useful
6061 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02006062 the system will select a free port. Note that port ranges are not
6063 supported in the backend. If you want to force port ranges, you
6064 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006065
6066 <addr2> is the IP address to present to the server when connections are
6067 forwarded in full transparent proxy mode. This is currently only
6068 supported on some patched Linux kernels. When this address is
6069 specified, clients connecting to the server will be presented
6070 with this address, while health checks will still use the address
6071 <addr>.
6072
6073 <port2> is the optional port to present to the server when connections
6074 are forwarded in full transparent proxy mode (see <addr2> above).
6075 The default value of zero means the system will select a free
6076 port.
6077
Willy Tarreaubce70882009-09-07 11:51:47 +02006078 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
6079 This is the name of a comma-separated header list which can
6080 contain multiple IP addresses. By default, the last occurrence is
6081 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01006082 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02006083 by previous proxy, typically Stunnel. In order to use another
6084 occurrence from the last one, please see the <occ> parameter
6085 below. When the header (or occurrence) is not found, no binding
6086 is performed so that the proxy's default IP address is used. Also
6087 keep in mind that the header name is case insensitive, as for any
6088 HTTP header.
6089
6090 <occ> is the occurrence number of a value to be used in a multi-value
6091 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006092 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02006093 address. Positive values indicate a position from the first
6094 occurrence, 1 being the first one. Negative values indicate
6095 positions relative to the last one, -1 being the last one. This
6096 is helpful for situations where an X-Forwarded-For header is set
6097 at the entry point of an infrastructure and must be used several
6098 proxy layers away. When this value is not specified, -1 is
6099 assumed. Passing a zero here disables the feature.
6100
Willy Tarreaud53f96b2009-02-04 18:46:54 +01006101 <name> is an optional interface name to which to bind to for outgoing
6102 traffic. On systems supporting this features (currently, only
6103 Linux), this allows one to bind all traffic to the server to
6104 this interface even if it is not the one the system would select
6105 based on routing tables. This should be used with extreme care.
6106 Note that using this option requires root privileges.
6107
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006108 The "source" keyword is useful in complex environments where a specific
6109 address only is allowed to connect to the servers. It may be needed when a
6110 private address must be used through a public gateway for instance, and it is
6111 known that the system cannot determine the adequate source address by itself.
6112
6113 An extension which is available on certain patched Linux kernels may be used
6114 through the "usesrc" optional keyword. It makes it possible to connect to the
6115 servers with an IP address which does not belong to the system itself. This
6116 is called "full transparent proxy mode". For this to work, the destination
6117 servers have to route their traffic back to this address through the machine
6118 running HAProxy, and IP forwarding must generally be enabled on this machine.
6119
6120 In this "full transparent proxy" mode, it is possible to force a specific IP
6121 address to be presented to the servers. This is not much used in fact. A more
6122 common use is to tell HAProxy to present the client's IP address. For this,
6123 there are two methods :
6124
6125 - present the client's IP and port addresses. This is the most transparent
6126 mode, but it can cause problems when IP connection tracking is enabled on
6127 the machine, because a same connection may be seen twice with different
6128 states. However, this solution presents the huge advantage of not
6129 limiting the system to the 64k outgoing address+port couples, because all
6130 of the client ranges may be used.
6131
6132 - present only the client's IP address and select a spare port. This
6133 solution is still quite elegant but slightly less transparent (downstream
6134 firewalls logs will not match upstream's). It also presents the downside
6135 of limiting the number of concurrent connections to the usual 64k ports.
6136 However, since the upstream and downstream ports are different, local IP
6137 connection tracking on the machine will not be upset by the reuse of the
6138 same session.
6139
6140 Note that depending on the transparent proxy technology used, it may be
6141 required to force the source address. In fact, cttproxy version 2 requires an
6142 IP address in <addr> above, and does not support setting of "0.0.0.0" as the
6143 IP address because it creates NAT entries which much match the exact outgoing
6144 address. Tproxy version 4 and some other kernel patches which work in pure
6145 forwarding mode generally will not have this limitation.
6146
6147 This option sets the default source for all servers in the backend. It may
6148 also be specified in a "defaults" section. Finer source address specification
6149 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006150 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006151
6152 Examples :
6153 backend private
6154 # Connect to the servers using our 192.168.1.200 source address
6155 source 192.168.1.200
6156
6157 backend transparent_ssl1
6158 # Connect to the SSL farm from the client's source address
6159 source 192.168.1.200 usesrc clientip
6160
6161 backend transparent_ssl2
6162 # Connect to the SSL farm from the client's source address and port
6163 # not recommended if IP conntrack is present on the local machine.
6164 source 192.168.1.200 usesrc client
6165
6166 backend transparent_ssl3
6167 # Connect to the SSL farm from the client's source address. It
6168 # is more conntrack-friendly.
6169 source 192.168.1.200 usesrc clientip
6170
6171 backend transparent_smtp
6172 # Connect to the SMTP farm from the client's source address/port
6173 # with Tproxy version 4.
6174 source 0.0.0.0 usesrc clientip
6175
Willy Tarreaubce70882009-09-07 11:51:47 +02006176 backend transparent_http
6177 # Connect to the servers using the client's IP as seen by previous
6178 # proxy.
6179 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
6180
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006181 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006182 the Linux kernel on www.balabit.com, the "bind" keyword.
6183
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006184
Willy Tarreau844e3c52008-01-11 16:28:18 +01006185srvtimeout <timeout> (deprecated)
6186 Set the maximum inactivity time on the server side.
6187 May be used in sections : defaults | frontend | listen | backend
6188 yes | no | yes | yes
6189 Arguments :
6190 <timeout> is the timeout value specified in milliseconds by default, but
6191 can be in any other unit if the number is suffixed by the unit,
6192 as explained at the top of this document.
6193
6194 The inactivity timeout applies when the server is expected to acknowledge or
6195 send data. In HTTP mode, this timeout is particularly important to consider
6196 during the first phase of the server's response, when it has to send the
6197 headers, as it directly represents the server's processing time for the
6198 request. To find out what value to put there, it's often good to start with
6199 what would be considered as unacceptable response times, then check the logs
6200 to observe the response time distribution, and adjust the value accordingly.
6201
6202 The value is specified in milliseconds by default, but can be in any other
6203 unit if the number is suffixed by the unit, as specified at the top of this
6204 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
6205 recommended that the client timeout remains equal to the server timeout in
6206 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006207 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01006208 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreaud72758d2010-01-12 10:42:19 +01006209 seconds (eg: 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01006210
6211 This parameter is specific to backends, but can be specified once for all in
6212 "defaults" sections. This is in fact one of the easiest solutions not to
6213 forget about it. An unspecified timeout results in an infinite timeout, which
6214 is not recommended. Such a usage is accepted and works but reports a warning
6215 during startup because it may results in accumulation of expired sessions in
6216 the system if the system's timeouts are not configured either.
6217
6218 This parameter is provided for compatibility but is currently deprecated.
6219 Please use "timeout server" instead.
6220
Willy Tarreauce887fd2012-05-12 12:50:00 +02006221 See also : "timeout server", "timeout tunnel", "timeout client" and
6222 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01006223
6224
Cyril Bonté66c327d2010-10-12 00:14:37 +02006225stats admin { if | unless } <cond>
6226 Enable statistics admin level if/unless a condition is matched
6227 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006228 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02006229
6230 This statement enables the statistics admin level if/unless a condition is
6231 matched.
6232
6233 The admin level allows to enable/disable servers from the web interface. By
6234 default, statistics page is read-only for security reasons.
6235
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006236 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6237 unless you know what you do : memory is not shared between the
6238 processes, which can result in random behaviours.
6239
Cyril Bonté23b39d92011-02-10 22:54:44 +01006240 Currently, the POST request is limited to the buffer size minus the reserved
6241 buffer space, which means that if the list of servers is too long, the
6242 request won't be processed. It is recommended to alter few servers at a
6243 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006244
6245 Example :
6246 # statistics admin level only for localhost
6247 backend stats_localhost
6248 stats enable
6249 stats admin if LOCALHOST
6250
6251 Example :
6252 # statistics admin level always enabled because of the authentication
6253 backend stats_auth
6254 stats enable
6255 stats auth admin:AdMiN123
6256 stats admin if TRUE
6257
6258 Example :
6259 # statistics admin level depends on the authenticated user
6260 userlist stats-auth
6261 group admin users admin
6262 user admin insecure-password AdMiN123
6263 group readonly users haproxy
6264 user haproxy insecure-password haproxy
6265
6266 backend stats_auth
6267 stats enable
6268 acl AUTH http_auth(stats-auth)
6269 acl AUTH_ADMIN http_auth_group(stats-auth) admin
6270 stats http-request auth unless AUTH
6271 stats admin if AUTH_ADMIN
6272
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006273 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
6274 "bind-process", section 3.4 about userlists and section 7 about
6275 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02006276
6277
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006278stats auth <user>:<passwd>
6279 Enable statistics with authentication and grant access to an account
6280 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006281 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006282 Arguments :
6283 <user> is a user name to grant access to
6284
6285 <passwd> is the cleartext password associated to this user
6286
6287 This statement enables statistics with default settings, and restricts access
6288 to declared users only. It may be repeated as many times as necessary to
6289 allow as many users as desired. When a user tries to access the statistics
6290 without a valid account, a "401 Forbidden" response will be returned so that
6291 the browser asks the user to provide a valid user and password. The real
6292 which will be returned to the browser is configurable using "stats realm".
6293
6294 Since the authentication method is HTTP Basic Authentication, the passwords
6295 circulate in cleartext on the network. Thus, it was decided that the
6296 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006297 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006298
6299 It is also possible to reduce the scope of the proxies which appear in the
6300 report using "stats scope".
6301
6302 Though this statement alone is enough to enable statistics reporting, it is
6303 recommended to set all other settings in order to avoid relying on default
6304 unobvious parameters.
6305
6306 Example :
6307 # public access (limited to this backend only)
6308 backend public_www
6309 server srv1 192.168.0.1:80
6310 stats enable
6311 stats hide-version
6312 stats scope .
6313 stats uri /admin?stats
6314 stats realm Haproxy\ Statistics
6315 stats auth admin1:AdMiN123
6316 stats auth admin2:AdMiN321
6317
6318 # internal monitoring access (unlimited)
6319 backend private_monitoring
6320 stats enable
6321 stats uri /admin?stats
6322 stats refresh 5s
6323
6324 See also : "stats enable", "stats realm", "stats scope", "stats uri"
6325
6326
6327stats enable
6328 Enable statistics reporting with default settings
6329 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006330 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006331 Arguments : none
6332
6333 This statement enables statistics reporting with default settings defined
6334 at build time. Unless stated otherwise, these settings are used :
6335 - stats uri : /haproxy?stats
6336 - stats realm : "HAProxy Statistics"
6337 - stats auth : no authentication
6338 - stats scope : no restriction
6339
6340 Though this statement alone is enough to enable statistics reporting, it is
6341 recommended to set all other settings in order to avoid relying on default
6342 unobvious parameters.
6343
6344 Example :
6345 # public access (limited to this backend only)
6346 backend public_www
6347 server srv1 192.168.0.1:80
6348 stats enable
6349 stats hide-version
6350 stats scope .
6351 stats uri /admin?stats
6352 stats realm Haproxy\ Statistics
6353 stats auth admin1:AdMiN123
6354 stats auth admin2:AdMiN321
6355
6356 # internal monitoring access (unlimited)
6357 backend private_monitoring
6358 stats enable
6359 stats uri /admin?stats
6360 stats refresh 5s
6361
6362 See also : "stats auth", "stats realm", "stats uri"
6363
6364
Willy Tarreaud63335a2010-02-26 12:56:52 +01006365stats hide-version
6366 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006367 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006368 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006369 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006370
Willy Tarreaud63335a2010-02-26 12:56:52 +01006371 By default, the stats page reports some useful status information along with
6372 the statistics. Among them is HAProxy's version. However, it is generally
6373 considered dangerous to report precise version to anyone, as it can help them
6374 target known weaknesses with specific attacks. The "stats hide-version"
6375 statement removes the version from the statistics report. This is recommended
6376 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006377
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006378 Though this statement alone is enough to enable statistics reporting, it is
6379 recommended to set all other settings in order to avoid relying on default
6380 unobvious parameters.
6381
Willy Tarreaud63335a2010-02-26 12:56:52 +01006382 Example :
6383 # public access (limited to this backend only)
6384 backend public_www
6385 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02006386 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006387 stats hide-version
6388 stats scope .
6389 stats uri /admin?stats
6390 stats realm Haproxy\ Statistics
6391 stats auth admin1:AdMiN123
6392 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006393
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006394 # internal monitoring access (unlimited)
6395 backend private_monitoring
6396 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01006397 stats uri /admin?stats
6398 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01006399
Willy Tarreaud63335a2010-02-26 12:56:52 +01006400 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02006401
Willy Tarreau983e01e2010-01-11 18:42:06 +01006402
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02006403stats http-request { allow | deny | auth [realm <realm>] }
6404 [ { if | unless } <condition> ]
6405 Access control for statistics
6406
6407 May be used in sections: defaults | frontend | listen | backend
6408 no | no | yes | yes
6409
6410 As "http-request", these set of options allow to fine control access to
6411 statistics. Each option may be followed by if/unless and acl.
6412 First option with matched condition (or option without condition) is final.
6413 For "deny" a 403 error will be returned, for "allow" normal processing is
6414 performed, for "auth" a 401/407 error code is returned so the client
6415 should be asked to enter a username and password.
6416
6417 There is no fixed limit to the number of http-request statements per
6418 instance.
6419
6420 See also : "http-request", section 3.4 about userlists and section 7
6421 about ACL usage.
6422
6423
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006424stats realm <realm>
6425 Enable statistics and set authentication realm
6426 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006427 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006428 Arguments :
6429 <realm> is the name of the HTTP Basic Authentication realm reported to
6430 the browser. The browser uses it to display it in the pop-up
6431 inviting the user to enter a valid username and password.
6432
6433 The realm is read as a single word, so any spaces in it should be escaped
6434 using a backslash ('\').
6435
6436 This statement is useful only in conjunction with "stats auth" since it is
6437 only related to authentication.
6438
6439 Though this statement alone is enough to enable statistics reporting, it is
6440 recommended to set all other settings in order to avoid relying on default
6441 unobvious parameters.
6442
6443 Example :
6444 # public access (limited to this backend only)
6445 backend public_www
6446 server srv1 192.168.0.1:80
6447 stats enable
6448 stats hide-version
6449 stats scope .
6450 stats uri /admin?stats
6451 stats realm Haproxy\ Statistics
6452 stats auth admin1:AdMiN123
6453 stats auth admin2:AdMiN321
6454
6455 # internal monitoring access (unlimited)
6456 backend private_monitoring
6457 stats enable
6458 stats uri /admin?stats
6459 stats refresh 5s
6460
6461 See also : "stats auth", "stats enable", "stats uri"
6462
6463
6464stats refresh <delay>
6465 Enable statistics with automatic refresh
6466 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006467 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006468 Arguments :
6469 <delay> is the suggested refresh delay, specified in seconds, which will
6470 be returned to the browser consulting the report page. While the
6471 browser is free to apply any delay, it will generally respect it
6472 and refresh the page this every seconds. The refresh interval may
6473 be specified in any other non-default time unit, by suffixing the
6474 unit after the value, as explained at the top of this document.
6475
6476 This statement is useful on monitoring displays with a permanent page
6477 reporting the load balancer's activity. When set, the HTML report page will
6478 include a link "refresh"/"stop refresh" so that the user can select whether
6479 he wants automatic refresh of the page or not.
6480
6481 Though this statement alone is enough to enable statistics reporting, it is
6482 recommended to set all other settings in order to avoid relying on default
6483 unobvious parameters.
6484
6485 Example :
6486 # public access (limited to this backend only)
6487 backend public_www
6488 server srv1 192.168.0.1:80
6489 stats enable
6490 stats hide-version
6491 stats scope .
6492 stats uri /admin?stats
6493 stats realm Haproxy\ Statistics
6494 stats auth admin1:AdMiN123
6495 stats auth admin2:AdMiN321
6496
6497 # internal monitoring access (unlimited)
6498 backend private_monitoring
6499 stats enable
6500 stats uri /admin?stats
6501 stats refresh 5s
6502
6503 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6504
6505
6506stats scope { <name> | "." }
6507 Enable statistics and limit access scope
6508 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006509 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006510 Arguments :
6511 <name> is the name of a listen, frontend or backend section to be
6512 reported. The special name "." (a single dot) designates the
6513 section in which the statement appears.
6514
6515 When this statement is specified, only the sections enumerated with this
6516 statement will appear in the report. All other ones will be hidden. This
6517 statement may appear as many times as needed if multiple sections need to be
6518 reported. Please note that the name checking is performed as simple string
6519 comparisons, and that it is never checked that a give section name really
6520 exists.
6521
6522 Though this statement alone is enough to enable statistics reporting, it is
6523 recommended to set all other settings in order to avoid relying on default
6524 unobvious parameters.
6525
6526 Example :
6527 # public access (limited to this backend only)
6528 backend public_www
6529 server srv1 192.168.0.1:80
6530 stats enable
6531 stats hide-version
6532 stats scope .
6533 stats uri /admin?stats
6534 stats realm Haproxy\ Statistics
6535 stats auth admin1:AdMiN123
6536 stats auth admin2:AdMiN321
6537
6538 # internal monitoring access (unlimited)
6539 backend private_monitoring
6540 stats enable
6541 stats uri /admin?stats
6542 stats refresh 5s
6543
6544 See also : "stats auth", "stats enable", "stats realm", "stats uri"
6545
Willy Tarreaud63335a2010-02-26 12:56:52 +01006546
Willy Tarreauc9705a12010-07-27 20:05:50 +02006547stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01006548 Enable reporting of a description on the statistics page.
6549 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006550 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006551
Willy Tarreauc9705a12010-07-27 20:05:50 +02006552 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01006553 description from global section is automatically used instead.
6554
6555 This statement is useful for users that offer shared services to their
6556 customers, where node or description should be different for each customer.
6557
6558 Though this statement alone is enough to enable statistics reporting, it is
6559 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006560 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006561
6562 Example :
6563 # internal monitoring access (unlimited)
6564 backend private_monitoring
6565 stats enable
6566 stats show-desc Master node for Europe, Asia, Africa
6567 stats uri /admin?stats
6568 stats refresh 5s
6569
6570 See also: "show-node", "stats enable", "stats uri" and "description" in
6571 global section.
6572
6573
6574stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02006575 Enable reporting additional information on the statistics page
6576 May be used in sections : defaults | frontend | listen | backend
6577 yes | yes | yes | yes
6578 Arguments : none
6579
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006580 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01006581 - cap: capabilities (proxy)
6582 - mode: one of tcp, http or health (proxy)
6583 - id: SNMP ID (proxy, socket, server)
6584 - IP (socket, server)
6585 - cookie (backend, server)
6586
6587 Though this statement alone is enough to enable statistics reporting, it is
6588 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006589 unobvious parameters. Default behaviour is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006590
6591 See also: "stats enable", "stats uri".
6592
6593
6594stats show-node [ <name> ]
6595 Enable reporting of a host name on the statistics page.
6596 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006597 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01006598 Arguments:
6599 <name> is an optional name to be reported. If unspecified, the
6600 node name from global section is automatically used instead.
6601
6602 This statement is useful for users that offer shared services to their
6603 customers, where node or description might be different on a stats page
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04006604 provided for each customer. Default behaviour is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006605
6606 Though this statement alone is enough to enable statistics reporting, it is
6607 recommended to set all other settings in order to avoid relying on default
6608 unobvious parameters.
6609
6610 Example:
6611 # internal monitoring access (unlimited)
6612 backend private_monitoring
6613 stats enable
6614 stats show-node Europe-1
6615 stats uri /admin?stats
6616 stats refresh 5s
6617
6618 See also: "show-desc", "stats enable", "stats uri", and "node" in global
6619 section.
6620
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006621
6622stats uri <prefix>
6623 Enable statistics and define the URI prefix to access them
6624 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02006625 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006626 Arguments :
6627 <prefix> is the prefix of any URI which will be redirected to stats. This
6628 prefix may contain a question mark ('?') to indicate part of a
6629 query string.
6630
6631 The statistics URI is intercepted on the relayed traffic, so it appears as a
6632 page within the normal application. It is strongly advised to ensure that the
6633 selected URI will never appear in the application, otherwise it will never be
6634 possible to reach it in the application.
6635
6636 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006637 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006638 It is generally a good idea to include a question mark in the URI so that
6639 intermediate proxies refrain from caching the results. Also, since any string
6640 beginning with the prefix will be accepted as a stats request, the question
6641 mark helps ensuring that no valid URI will begin with the same words.
6642
6643 It is sometimes very convenient to use "/" as the URI prefix, and put that
6644 statement in a "listen" instance of its own. That makes it easy to dedicate
6645 an address or a port to statistics only.
6646
6647 Though this statement alone is enough to enable statistics reporting, it is
6648 recommended to set all other settings in order to avoid relying on default
6649 unobvious parameters.
6650
6651 Example :
6652 # public access (limited to this backend only)
6653 backend public_www
6654 server srv1 192.168.0.1:80
6655 stats enable
6656 stats hide-version
6657 stats scope .
6658 stats uri /admin?stats
6659 stats realm Haproxy\ Statistics
6660 stats auth admin1:AdMiN123
6661 stats auth admin2:AdMiN321
6662
6663 # internal monitoring access (unlimited)
6664 backend private_monitoring
6665 stats enable
6666 stats uri /admin?stats
6667 stats refresh 5s
6668
6669 See also : "stats auth", "stats enable", "stats realm"
6670
6671
Willy Tarreaud63335a2010-02-26 12:56:52 +01006672stick match <pattern> [table <table>] [{if | unless} <cond>]
6673 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006674 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01006675 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006676
6677 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006678 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006679 describes what elements of the incoming request or connection
6680 will be analysed in the hope to find a matching entry in a
6681 stickiness table. This rule is mandatory.
6682
6683 <table> is an optional stickiness table name. If unspecified, the same
6684 backend's table is used. A stickiness table is declared using
6685 the "stick-table" statement.
6686
6687 <cond> is an optional matching condition. It makes it possible to match
6688 on a certain criterion only when other conditions are met (or
6689 not met). For instance, it could be used to match on a source IP
6690 address except when a request passes through a known proxy, in
6691 which case we'd match on a header containing that IP address.
6692
6693 Some protocols or applications require complex stickiness rules and cannot
6694 always simply rely on cookies nor hashing. The "stick match" statement
6695 describes a rule to extract the stickiness criterion from an incoming request
6696 or connection. See section 7 for a complete list of possible patterns and
6697 transformation rules.
6698
6699 The table has to be declared using the "stick-table" statement. It must be of
6700 a type compatible with the pattern. By default it is the one which is present
6701 in the same backend. It is possible to share a table with other backends by
6702 referencing it using the "table" keyword. If another table is referenced,
6703 the server's ID inside the backends are used. By default, all server IDs
6704 start at 1 in each backend, so the server ordering is enough. But in case of
6705 doubt, it is highly recommended to force server IDs using their "id" setting.
6706
6707 It is possible to restrict the conditions where a "stick match" statement
6708 will apply, using "if" or "unless" followed by a condition. See section 7 for
6709 ACL based conditions.
6710
6711 There is no limit on the number of "stick match" statements. The first that
6712 applies and matches will cause the request to be directed to the same server
6713 as was used for the request which created the entry. That way, multiple
6714 matches can be used as fallbacks.
6715
6716 The stick rules are checked after the persistence cookies, so they will not
6717 affect stickiness if a cookie has already been used to select a server. That
6718 way, it becomes very easy to insert cookies and match on IP addresses in
6719 order to maintain stickiness between HTTP and HTTPS.
6720
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006721 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6722 unless you know what you do : memory is not shared between the
6723 processes, which can result in random behaviours.
6724
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006725 Example :
6726 # forward SMTP users to the same server they just used for POP in the
6727 # last 30 minutes
6728 backend pop
6729 mode tcp
6730 balance roundrobin
6731 stick store-request src
6732 stick-table type ip size 200k expire 30m
6733 server s1 192.168.1.1:110
6734 server s2 192.168.1.1:110
6735
6736 backend smtp
6737 mode tcp
6738 balance roundrobin
6739 stick match src table pop
6740 server s1 192.168.1.1:25
6741 server s2 192.168.1.1:25
6742
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006743 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006744 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006745
6746
6747stick on <pattern> [table <table>] [{if | unless} <condition>]
6748 Define a request pattern to associate a user to a server
6749 May be used in sections : defaults | frontend | listen | backend
6750 no | no | yes | yes
6751
6752 Note : This form is exactly equivalent to "stick match" followed by
6753 "stick store-request", all with the same arguments. Please refer
6754 to both keywords for details. It is only provided as a convenience
6755 for writing more maintainable configurations.
6756
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006757 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6758 unless you know what you do : memory is not shared between the
6759 processes, which can result in random behaviours.
6760
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006761 Examples :
6762 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01006763 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006764
6765 # ...is strictly equivalent to this one :
6766 stick match src table pop if !localhost
6767 stick store-request src table pop if !localhost
6768
6769
6770 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
6771 # well as HTTP without cookie. Share the same table between both accesses.
6772 backend http
6773 mode http
6774 balance roundrobin
6775 stick on src table https
6776 cookie SRV insert indirect nocache
6777 server s1 192.168.1.1:80 cookie s1
6778 server s2 192.168.1.1:80 cookie s2
6779
6780 backend https
6781 mode tcp
6782 balance roundrobin
6783 stick-table type ip size 200k expire 30m
6784 stick on src
6785 server s1 192.168.1.1:443
6786 server s2 192.168.1.1:443
6787
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006788 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006789
6790
6791stick store-request <pattern> [table <table>] [{if | unless} <condition>]
6792 Define a request pattern used to create an entry in a stickiness table
6793 May be used in sections : defaults | frontend | listen | backend
6794 no | no | yes | yes
6795
6796 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02006797 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006798 describes what elements of the incoming request or connection
6799 will be analysed, extracted and stored in the table once a
6800 server is selected.
6801
6802 <table> is an optional stickiness table name. If unspecified, the same
6803 backend's table is used. A stickiness table is declared using
6804 the "stick-table" statement.
6805
6806 <cond> is an optional storage condition. It makes it possible to store
6807 certain criteria only when some conditions are met (or not met).
6808 For instance, it could be used to store the source IP address
6809 except when the request passes through a known proxy, in which
6810 case we'd store a converted form of a header containing that IP
6811 address.
6812
6813 Some protocols or applications require complex stickiness rules and cannot
6814 always simply rely on cookies nor hashing. The "stick store-request" statement
6815 describes a rule to decide what to extract from the request and when to do
6816 it, in order to store it into a stickiness table for further requests to
6817 match it using the "stick match" statement. Obviously the extracted part must
6818 make sense and have a chance to be matched in a further request. Storing a
6819 client's IP address for instance often makes sense. Storing an ID found in a
6820 URL parameter also makes sense. Storing a source port will almost never make
6821 any sense because it will be randomly matched. See section 7 for a complete
6822 list of possible patterns and transformation rules.
6823
6824 The table has to be declared using the "stick-table" statement. It must be of
6825 a type compatible with the pattern. By default it is the one which is present
6826 in the same backend. It is possible to share a table with other backends by
6827 referencing it using the "table" keyword. If another table is referenced,
6828 the server's ID inside the backends are used. By default, all server IDs
6829 start at 1 in each backend, so the server ordering is enough. But in case of
6830 doubt, it is highly recommended to force server IDs using their "id" setting.
6831
6832 It is possible to restrict the conditions where a "stick store-request"
6833 statement will apply, using "if" or "unless" followed by a condition. This
6834 condition will be evaluated while parsing the request, so any criteria can be
6835 used. See section 7 for ACL based conditions.
6836
6837 There is no limit on the number of "stick store-request" statements, but
6838 there is a limit of 8 simultaneous stores per request or response. This
6839 makes it possible to store up to 8 criteria, all extracted from either the
6840 request or the response, regardless of the number of rules. Only the 8 first
6841 ones which match will be kept. Using this, it is possible to feed multiple
6842 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01006843 another protocol or access method. Using multiple store-request rules with
6844 the same table is possible and may be used to find the best criterion to rely
6845 on, by arranging the rules by decreasing preference order. Only the first
6846 extracted criterion for a given table will be stored. All subsequent store-
6847 request rules referencing the same table will be skipped and their ACLs will
6848 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006849
6850 The "store-request" rules are evaluated once the server connection has been
6851 established, so that the table will contain the real server that processed
6852 the request.
6853
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006854 Note : Consider not using this feature in multi-process mode (nbproc > 1)
6855 unless you know what you do : memory is not shared between the
6856 processes, which can result in random behaviours.
6857
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006858 Example :
6859 # forward SMTP users to the same server they just used for POP in the
6860 # last 30 minutes
6861 backend pop
6862 mode tcp
6863 balance roundrobin
6864 stick store-request src
6865 stick-table type ip size 200k expire 30m
6866 server s1 192.168.1.1:110
6867 server s2 192.168.1.1:110
6868
6869 backend smtp
6870 mode tcp
6871 balance roundrobin
6872 stick match src table pop
6873 server s1 192.168.1.1:25
6874 server s2 192.168.1.1:25
6875
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006876 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02006877 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006878
6879
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006880stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02006881 size <size> [expire <expire>] [nopurge] [peers <peersect>]
6882 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08006883 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006884 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02006885 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006886
6887 Arguments :
6888 ip a table declared with "type ip" will only store IPv4 addresses.
6889 This form is very compact (about 50 bytes per entry) and allows
6890 very fast entry lookup and stores with almost no overhead. This
6891 is mainly used to store client source IP addresses.
6892
David du Colombier9a6d3c92011-03-17 10:40:24 +01006893 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
6894 This form is very compact (about 60 bytes per entry) and allows
6895 very fast entry lookup and stores with almost no overhead. This
6896 is mainly used to store client source IP addresses.
6897
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006898 integer a table declared with "type integer" will store 32bit integers
6899 which can represent a client identifier found in a request for
6900 instance.
6901
6902 string a table declared with "type string" will store substrings of up
6903 to <len> characters. If the string provided by the pattern
6904 extractor is larger than <len>, it will be truncated before
6905 being stored. During matching, at most <len> characters will be
6906 compared between the string in the table and the extracted
6907 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006908 to 32 characters.
6909
6910 binary a table declared with "type binary" will store binary blocks
6911 of <len> bytes. If the block provided by the pattern
6912 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02006913 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006914 is shorter than <len>, it will be padded by 0. When not
6915 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006916
6917 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02006918 "string" type table (See type "string" above). Or the number
6919 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006920 changing this parameter as memory usage will proportionally
6921 increase.
6922
6923 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01006924 value directly impacts memory usage. Count approximately
6925 50 bytes per entry, plus the size of a string if any. The size
6926 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006927
6928 [nopurge] indicates that we refuse to purge older entries when the table
6929 is full. When not specified and the table is full when haproxy
6930 wants to store an entry in it, it will flush a few of the oldest
6931 entries in order to release some space for the new ones. This is
6932 most often the desired behaviour. In some specific cases, it
6933 be desirable to refuse new entries instead of purging the older
6934 ones. That may be the case when the amount of data to store is
6935 far above the hardware limits and we prefer not to offer access
6936 to new clients than to reject the ones already connected. When
6937 using this parameter, be sure to properly set the "expire"
6938 parameter (see below).
6939
Emeric Brunf099e792010-09-27 12:05:28 +02006940 <peersect> is the name of the peers section to use for replication. Entries
6941 which associate keys to server IDs are kept synchronized with
6942 the remote peers declared in this section. All entries are also
6943 automatically learned from the local peer (old process) during a
6944 soft restart.
6945
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01006946 NOTE : peers can't be used in multi-process mode.
6947
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006948 <expire> defines the maximum duration of an entry in the table since it
6949 was last created, refreshed or matched. The expiration delay is
6950 defined using the standard time format, similarly as the various
6951 timeouts. The maximum duration is slightly above 24 days. See
6952 section 2.2 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02006953 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01006954 be removed once full. Be sure not to use the "nopurge" parameter
6955 if not expiration delay is specified.
6956
Willy Tarreau08d5f982010-06-06 13:34:54 +02006957 <data_type> is used to store additional information in the stick-table. This
6958 may be used by ACLs in order to control various criteria related
6959 to the activity of the client matching the stick-table. For each
6960 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02006961 that the additional data can fit. Several data types may be
6962 stored with an entry. Multiple data types may be specified after
6963 the "store" keyword, as a comma-separated list. Alternatively,
6964 it is possible to repeat the "store" keyword followed by one or
6965 several data types. Except for the "server_id" type which is
6966 automatically detected and enabled, all data types must be
6967 explicitly declared to be stored. If an ACL references a data
6968 type which is not stored, the ACL will simply not match. Some
6969 data types require an argument which must be passed just after
6970 the type between parenthesis. See below for the supported data
6971 types and their arguments.
6972
6973 The data types that can be stored with an entry are the following :
6974 - server_id : this is an integer which holds the numeric ID of the server a
6975 request was assigned to. It is used by the "stick match", "stick store",
6976 and "stick on" rules. It is automatically enabled when referenced.
6977
6978 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
6979 integer which may be used for anything. Most of the time it will be used
6980 to put a special tag on some entries, for instance to note that a
6981 specific behaviour was detected and must be known for future matches.
6982
Willy Tarreauba2ffd12013-05-29 15:54:14 +02006983 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
6984 over a period. It is a positive 32-bit integer integer which may be used
6985 for anything. Just like <gpc0>, it counts events, but instead of keeping
6986 a cumulative count, it maintains the rate at which the counter is
6987 incremented. Most of the time it will be used to measure the frequency of
6988 occurrence of certain events (eg: requests to a specific URL).
6989
Willy Tarreauc9705a12010-07-27 20:05:50 +02006990 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
6991 the absolute number of connections received from clients which matched
6992 this entry. It does not mean the connections were accepted, just that
6993 they were received.
6994
6995 - conn_cur : Current Connections. It is a positive 32-bit integer which
6996 stores the concurrent connection counts for the entry. It is incremented
6997 once an incoming connection matches the entry, and decremented once the
6998 connection leaves. That way it is possible to know at any time the exact
6999 number of concurrent connections for an entry.
7000
7001 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7002 integer parameter <period> which indicates in milliseconds the length
7003 of the period over which the average is measured. It reports the average
7004 incoming connection rate over that period, in connections per period. The
7005 result is an integer which can be matched using ACLs.
7006
7007 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
7008 the absolute number of sessions received from clients which matched this
7009 entry. A session is a connection that was accepted by the layer 4 rules.
7010
7011 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7012 integer parameter <period> which indicates in milliseconds the length
7013 of the period over which the average is measured. It reports the average
7014 incoming session rate over that period, in sessions per period. The
7015 result is an integer which can be matched using ACLs.
7016
7017 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
7018 counts the absolute number of HTTP requests received from clients which
7019 matched this entry. It does not matter whether they are valid requests or
7020 not. Note that this is different from sessions when keep-alive is used on
7021 the client side.
7022
7023 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7024 integer parameter <period> which indicates in milliseconds the length
7025 of the period over which the average is measured. It reports the average
7026 HTTP request rate over that period, in requests per period. The result is
7027 an integer which can be matched using ACLs. It does not matter whether
7028 they are valid requests or not. Note that this is different from sessions
7029 when keep-alive is used on the client side.
7030
7031 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
7032 counts the absolute number of HTTP requests errors induced by clients
7033 which matched this entry. Errors are counted on invalid and truncated
7034 requests, as well as on denied or tarpitted requests, and on failed
7035 authentications. If the server responds with 4xx, then the request is
7036 also counted as an error since it's an error triggered by the client
7037 (eg: vulnerability scan).
7038
7039 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7040 integer parameter <period> which indicates in milliseconds the length
7041 of the period over which the average is measured. It reports the average
7042 HTTP request error rate over that period, in requests per period (see
7043 http_err_cnt above for what is accounted as an error). The result is an
7044 integer which can be matched using ACLs.
7045
7046 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
7047 integer which counts the cumulated amount of bytes received from clients
7048 which matched this entry. Headers are included in the count. This may be
7049 used to limit abuse of upload features on photo or video servers.
7050
7051 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
7052 integer parameter <period> which indicates in milliseconds the length
7053 of the period over which the average is measured. It reports the average
7054 incoming bytes rate over that period, in bytes per period. It may be used
7055 to detect users which upload too much and too fast. Warning: with large
7056 uploads, it is possible that the amount of uploaded data will be counted
7057 once upon termination, thus causing spikes in the average transfer speed
7058 instead of having a smooth one. This may partially be smoothed with
7059 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
7060 recommended for better fairness.
7061
7062 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
7063 integer which counts the cumulated amount of bytes sent to clients which
7064 matched this entry. Headers are included in the count. This may be used
7065 to limit abuse of bots sucking the whole site.
7066
7067 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
7068 an integer parameter <period> which indicates in milliseconds the length
7069 of the period over which the average is measured. It reports the average
7070 outgoing bytes rate over that period, in bytes per period. It may be used
7071 to detect users which download too much and too fast. Warning: with large
7072 transfers, it is possible that the amount of transferred data will be
7073 counted once upon termination, thus causing spikes in the average
7074 transfer speed instead of having a smooth one. This may partially be
7075 smoothed with "option contstats" though this is not perfect yet. Use of
7076 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02007077
Willy Tarreauc00cdc22010-06-06 16:48:26 +02007078 There is only one stick-table per proxy. At the moment of writing this doc,
7079 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007080 to be required, simply create a dummy backend with a stick-table in it and
7081 reference it.
7082
7083 It is important to understand that stickiness based on learning information
7084 has some limitations, including the fact that all learned associations are
7085 lost upon restart. In general it can be good as a complement but not always
7086 as an exclusive stickiness.
7087
Willy Tarreauc9705a12010-07-27 20:05:50 +02007088 Last, memory requirements may be important when storing many data types.
7089 Indeed, storing all indicators above at once in each entry requires 116 bytes
7090 per entry, or 116 MB for a 1-million entries table. This is definitely not
7091 something that can be ignored.
7092
7093 Example:
7094 # Keep track of counters of up to 1 million IP addresses over 5 minutes
7095 # and store a general purpose counter and the average connection rate
7096 # computed over a sliding window of 30 seconds.
7097 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
7098
7099 See also : "stick match", "stick on", "stick store-request", section 2.2
David du Colombiera13d1b92011-03-17 10:40:22 +01007100 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01007101
7102
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007103stick store-response <pattern> [table <table>] [{if | unless} <condition>]
7104 Define a request pattern used to create an entry in a stickiness table
7105 May be used in sections : defaults | frontend | listen | backend
7106 no | no | yes | yes
7107
7108 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007109 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007110 describes what elements of the response or connection will
7111 be analysed, extracted and stored in the table once a
7112 server is selected.
7113
7114 <table> is an optional stickiness table name. If unspecified, the same
7115 backend's table is used. A stickiness table is declared using
7116 the "stick-table" statement.
7117
7118 <cond> is an optional storage condition. It makes it possible to store
7119 certain criteria only when some conditions are met (or not met).
7120 For instance, it could be used to store the SSL session ID only
7121 when the response is a SSL server hello.
7122
7123 Some protocols or applications require complex stickiness rules and cannot
7124 always simply rely on cookies nor hashing. The "stick store-response"
7125 statement describes a rule to decide what to extract from the response and
7126 when to do it, in order to store it into a stickiness table for further
7127 requests to match it using the "stick match" statement. Obviously the
7128 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007129 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007130 See section 7 for a complete list of possible patterns and transformation
7131 rules.
7132
7133 The table has to be declared using the "stick-table" statement. It must be of
7134 a type compatible with the pattern. By default it is the one which is present
7135 in the same backend. It is possible to share a table with other backends by
7136 referencing it using the "table" keyword. If another table is referenced,
7137 the server's ID inside the backends are used. By default, all server IDs
7138 start at 1 in each backend, so the server ordering is enough. But in case of
7139 doubt, it is highly recommended to force server IDs using their "id" setting.
7140
7141 It is possible to restrict the conditions where a "stick store-response"
7142 statement will apply, using "if" or "unless" followed by a condition. This
7143 condition will be evaluated while parsing the response, so any criteria can
7144 be used. See section 7 for ACL based conditions.
7145
7146 There is no limit on the number of "stick store-response" statements, but
7147 there is a limit of 8 simultaneous stores per request or response. This
7148 makes it possible to store up to 8 criteria, all extracted from either the
7149 request or the response, regardless of the number of rules. Only the 8 first
7150 ones which match will be kept. Using this, it is possible to feed multiple
7151 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01007152 another protocol or access method. Using multiple store-response rules with
7153 the same table is possible and may be used to find the best criterion to rely
7154 on, by arranging the rules by decreasing preference order. Only the first
7155 extracted criterion for a given table will be stored. All subsequent store-
7156 response rules referencing the same table will be skipped and their ACLs will
7157 not be evaluated. However, even if a store-request rule references a table, a
7158 store-response rule may also use the same table. This means that each table
7159 may learn exactly one element from the request and one element from the
7160 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007161
7162 The table will contain the real server that processed the request.
7163
7164 Example :
7165 # Learn SSL session ID from both request and response and create affinity.
7166 backend https
7167 mode tcp
7168 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02007169 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007170 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007171
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007172 acl clienthello req_ssl_hello_type 1
7173 acl serverhello rep_ssl_hello_type 2
7174
7175 # use tcp content accepts to detects ssl client and server hello.
7176 tcp-request inspect-delay 5s
7177 tcp-request content accept if clienthello
7178
7179 # no timeout on response inspect delay by default.
7180 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007181
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007182 # SSL session ID (SSLID) may be present on a client or server hello.
7183 # Its length is coded on 1 byte at offset 43 and its value starts
7184 # at offset 44.
7185
7186 # Match and learn on request if client hello.
7187 stick on payload_lv(43,1) if clienthello
7188
7189 # Learn on response if server hello.
7190 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02007191
Emeric Brun6a1cefa2010-09-24 18:15:17 +02007192 server s1 192.168.1.1:443
7193 server s2 192.168.1.1:443
7194
7195 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
7196 extraction.
7197
7198
Willy Tarreau938c7fe2014-04-25 14:21:39 +02007199tcp-check connect [params*]
7200 Opens a new connection
7201 May be used in sections: defaults | frontend | listen | backend
7202 no | no | yes | yes
7203
7204 When an application lies on more than a single TCP port or when HAProxy
7205 load-balance many services in a single backend, it makes sense to probe all
7206 the services individually before considering a server as operational.
7207
7208 When there are no TCP port configured on the server line neither server port
7209 directive, then the 'tcp-check connect port <port>' must be the first step
7210 of the sequence.
7211
7212 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
7213 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
7214 do.
7215
7216 Parameters :
7217 They are optional and can be used to describe how HAProxy should open and
7218 use the TCP connection.
7219
7220 port if not set, check port or server port is used.
7221 It tells HAProxy where to open the connection to.
7222 <port> must be a valid TCP port source integer, from 1 to 65535.
7223
7224 send-proxy send a PROXY protocol string
7225
7226 ssl opens a ciphered connection
7227
7228 Examples:
7229 # check HTTP and HTTPs services on a server.
7230 # first open port 80 thanks to server line port directive, then
7231 # tcp-check opens port 443, ciphered and run a request on it:
7232 option tcp-check
7233 tcp-check connect
7234 tcp-check send GET\ /\ HTTP/1.0\r\n
7235 tcp-check send Host:\ haproxy.1wt.eu\r\n
7236 tcp-check send \r\n
7237 tcp-check expect rstring (2..|3..)
7238 tcp-check connect port 443 ssl
7239 tcp-check send GET\ /\ HTTP/1.0\r\n
7240 tcp-check send Host:\ haproxy.1wt.eu\r\n
7241 tcp-check send \r\n
7242 tcp-check expect rstring (2..|3..)
7243 server www 10.0.0.1 check port 80
7244
7245 # check both POP and IMAP from a single server:
7246 option tcp-check
7247 tcp-check connect port 110
7248 tcp-check expect string +OK\ POP3\ ready
7249 tcp-check connect port 143
7250 tcp-check expect string *\ OK\ IMAP4\ ready
7251 server mail 10.0.0.1 check
7252
7253 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
7254
7255
7256tcp-check expect [!] <match> <pattern>
7257 Specify data to be collected and analysed during a generic health check
7258 May be used in sections: defaults | frontend | listen | backend
7259 no | no | yes | yes
7260
7261 Arguments :
7262 <match> is a keyword indicating how to look for a specific pattern in the
7263 response. The keyword may be one of "string", "rstring" or
7264 binary.
7265 The keyword may be preceded by an exclamation mark ("!") to negate
7266 the match. Spaces are allowed between the exclamation mark and the
7267 keyword. See below for more details on the supported keywords.
7268
7269 <pattern> is the pattern to look for. It may be a string or a regular
7270 expression. If the pattern contains spaces, they must be escaped
7271 with the usual backslash ('\').
7272 If the match is set to binary, then the pattern must be passed as
7273 a serie of hexadecimal digits in an even number. Each sequence of
7274 two digits will represent a byte. The hexadecimal digits may be
7275 used upper or lower case.
7276
7277
7278 The available matches are intentionally similar to their http-check cousins :
7279
7280 string <string> : test the exact string matches in the response buffer.
7281 A health check response will be considered valid if the
7282 response's buffer contains this exact string. If the
7283 "string" keyword is prefixed with "!", then the response
7284 will be considered invalid if the body contains this
7285 string. This can be used to look for a mandatory pattern
7286 in a protocol response, or to detect a failure when a
7287 specific error appears in a protocol banner.
7288
7289 rstring <regex> : test a regular expression on the response buffer.
7290 A health check response will be considered valid if the
7291 response's buffer matches this expression. If the
7292 "rstring" keyword is prefixed with "!", then the response
7293 will be considered invalid if the body matches the
7294 expression.
7295
7296 binary <hexstring> : test the exact string in its hexadecimal form matches
7297 in the response buffer. A health check response will
7298 be considered valid if the response's buffer contains
7299 this exact hexadecimal string.
7300 Purpose is to match data on binary protocols.
7301
7302 It is important to note that the responses will be limited to a certain size
7303 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
7304 Thus, too large responses may not contain the mandatory pattern when using
7305 "string", "rstring" or binary. If a large response is absolutely required, it
7306 is possible to change the default max size by setting the global variable.
7307 However, it is worth keeping in mind that parsing very large responses can
7308 waste some CPU cycles, especially when regular expressions are used, and that
7309 it is always better to focus the checks on smaller resources. Also, in its
7310 current state, the check will not find any string nor regex past a null
7311 character in the response. Similarly it is not possible to request matching
7312 the null character.
7313
7314 Examples :
7315 # perform a POP check
7316 option tcp-check
7317 tcp-check expect string +OK\ POP3\ ready
7318
7319 # perform an IMAP check
7320 option tcp-check
7321 tcp-check expect string *\ OK\ IMAP4\ ready
7322
7323 # look for the redis master server
7324 option tcp-check
7325 tcp-check send PING\r\n
7326 tcp-check expect +PONG
7327 tcp-check send info\ replication\r\n
7328 tcp-check expect string role:master
7329 tcp-check send QUIT\r\n
7330 tcp-check expect string +OK
7331
7332
7333 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
7334 "tcp-check send-binary", "http-check expect", tune.chksize
7335
7336
7337tcp-check send <data>
7338 Specify a string to be sent as a question during a generic health check
7339 May be used in sections: defaults | frontend | listen | backend
7340 no | no | yes | yes
7341
7342 <data> : the data to be sent as a question during a generic health check
7343 session. For now, <data> must be a string.
7344
7345 Examples :
7346 # look for the redis master server
7347 option tcp-check
7348 tcp-check send info\ replication\r\n
7349 tcp-check expect string role:master
7350
7351 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7352 "tcp-check send-binary", tune.chksize
7353
7354
7355tcp-check send-binary <hexastring>
7356 Specify an hexa digits string to be sent as a binary question during a raw
7357 tcp health check
7358 May be used in sections: defaults | frontend | listen | backend
7359 no | no | yes | yes
7360
7361 <data> : the data to be sent as a question during a generic health check
7362 session. For now, <data> must be a string.
7363 <hexastring> : test the exact string in its hexadecimal form matches in the
7364 response buffer. A health check response will be considered
7365 valid if the response's buffer contains this exact
7366 hexadecimal string.
7367 Purpose is to send binary data to ask on binary protocols.
7368
7369 Examples :
7370 # redis check in binary
7371 option tcp-check
7372 tcp-check send-binary 50494e470d0a # PING\r\n
7373 tcp-check expect binary 2b504F4e47 # +PONG
7374
7375
7376 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
7377 "tcp-check send", tune.chksize
7378
7379
Willy Tarreaue9656522010-08-17 15:40:09 +02007380tcp-request connection <action> [{if | unless} <condition>]
7381 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02007382 May be used in sections : defaults | frontend | listen | backend
7383 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02007384 Arguments :
7385 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007386 actions include : "accept", "reject", "track-sc0", "track-sc1",
7387 "track-sc2", and "expect-proxy". See below for more details.
Willy Tarreau1a687942010-05-23 22:40:30 +02007388
Willy Tarreaue9656522010-08-17 15:40:09 +02007389 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007390
7391 Immediately after acceptance of a new incoming connection, it is possible to
7392 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02007393 or dropped or have its counters tracked. Those conditions cannot make use of
7394 any data contents because the connection has not been read from yet, and the
7395 buffers are not yet allocated. This is used to selectively and very quickly
7396 accept or drop connections from various sources with a very low overhead. If
7397 some contents need to be inspected in order to take the decision, the
7398 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007399
Willy Tarreaue9656522010-08-17 15:40:09 +02007400 The "tcp-request connection" rules are evaluated in their exact declaration
7401 order. If no rule matches or if there is no rule, the default action is to
7402 accept the incoming connection. There is no specific limit to the number of
7403 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007404
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007405 Five types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02007406 - accept :
7407 accepts the connection if the condition is true (when used with "if")
7408 or false (when used with "unless"). The first such rule executed ends
7409 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007410
Willy Tarreaue9656522010-08-17 15:40:09 +02007411 - reject :
7412 rejects the connection if the condition is true (when used with "if")
7413 or false (when used with "unless"). The first such rule executed ends
7414 the rules evaluation. Rejected connections do not even become a
7415 session, which is why they are accounted separately for in the stats,
7416 as "denied connections". They are not considered for the session
7417 rate-limit and are not logged either. The reason is that these rules
7418 should only be used to filter extremely high connection rates such as
7419 the ones encountered during a massive DDoS attack. Under these extreme
7420 conditions, the simple action of logging each event would make the
7421 system collapse and would considerably lower the filtering capacity. If
7422 logging is absolutely desired, then "tcp-request content" rules should
7423 be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007424
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007425 - expect-proxy layer4 :
7426 configures the client-facing connection to receive a PROXY protocol
7427 header before any byte is read from the socket. This is equivalent to
7428 having the "accept-proxy" keyword on the "bind" line, except that using
7429 the TCP rule allows the PROXY protocol to be accepted only for certain
7430 IP address ranges using an ACL. This is convenient when multiple layers
7431 of load balancers are passed through by traffic coming from public
7432 hosts.
7433
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007434 - capture <sample> len <length> :
7435 This only applies to "tcp-request content" rules. It captures sample
7436 expression <sample> from the request buffer, and converts it to a
7437 string of at most <len> characters. The resulting string is stored into
7438 the next request "capture" slot, so it will possibly appear next to
7439 some captured HTTP headers. It will then automatically appear in the
7440 logs, and it will be possible to extract it using sample fetch rules to
7441 feed it into headers or anything. The length should be limited given
7442 that this size will be allocated for each capture during the whole
7443 session life. Since it applies to Please check section 7.3 (Fetching
7444 samples) and "capture request header" for more information.
7445
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007446 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02007447 enables tracking of sticky counters from current connection. These
Willy Tarreau09448f72014-06-25 18:12:15 +02007448 rules do not stop evaluation and do not change default action. 3 sets
Willy Tarreaue9656522010-08-17 15:40:09 +02007449 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007450 first "track-sc0" rule executed enables tracking of the counters of the
7451 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02007452 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007453 set. The first "track-sc2" rule executed enables tracking of the
7454 counters of the specified table as the third set. It is a recommended
7455 practice to use the first set of counters for the per-frontend counters
7456 and the second set for the per-backend ones. But this is just a
7457 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007458
Willy Tarreaue9656522010-08-17 15:40:09 +02007459 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02007460 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02007461 in section 7.3. It describes what elements of the incoming
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007462 request or connection will be analysed, extracted, combined,
7463 and used to select which table entry to update the counters.
7464 Note that "tcp-request connection" cannot use content-based
7465 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007466
Willy Tarreaue9656522010-08-17 15:40:09 +02007467 <table> is an optional table to be used instead of the default one,
7468 which is the stick-table declared in the current proxy. All
7469 the counters for the matches and updates for the key will
7470 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007471
Willy Tarreaue9656522010-08-17 15:40:09 +02007472 Once a "track-sc*" rule is executed, the key is looked up in the table
7473 and if it is not found, an entry is allocated for it. Then a pointer to
7474 that entry is kept during all the session's life, and this entry's
7475 counters are updated as often as possible, every time the session's
7476 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007477 Counters are only updated for events that happen after the tracking has
7478 been started. For example, connection counters will not be updated when
7479 tracking layer 7 information, since the connection event happens before
7480 layer7 information is extracted.
7481
Willy Tarreaue9656522010-08-17 15:40:09 +02007482 If the entry tracks concurrent connection counters, one connection is
7483 counted for as long as the entry is tracked, and the entry will not
7484 expire during that time. Tracking counters also provides a performance
7485 advantage over just checking the keys, because only one table lookup is
7486 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007487
Willy Tarreaue9656522010-08-17 15:40:09 +02007488 Note that the "if/unless" condition is optional. If no condition is set on
7489 the action, it is simply performed unconditionally. That can be useful for
7490 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007491
Willy Tarreaue9656522010-08-17 15:40:09 +02007492 Example: accept all connections from white-listed hosts, reject too fast
7493 connection without counting them, and track accepted connections.
7494 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007495
Willy Tarreaue9656522010-08-17 15:40:09 +02007496 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007497 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007498 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007499
Willy Tarreaue9656522010-08-17 15:40:09 +02007500 Example: accept all connections from white-listed hosts, count all other
7501 connections and reject too fast ones. This results in abusive ones
7502 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007503
Willy Tarreaue9656522010-08-17 15:40:09 +02007504 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007505 tcp-request connection track-sc0 src
7506 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007507
Willy Tarreau4f0d9192013-06-11 20:40:55 +02007508 Example: enable the PROXY protocol for traffic coming from all known proxies.
7509
7510 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
7511
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007512 See section 7 about ACL usage.
7513
Willy Tarreaue9656522010-08-17 15:40:09 +02007514 See also : "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007515
7516
Willy Tarreaue9656522010-08-17 15:40:09 +02007517tcp-request content <action> [{if | unless} <condition>]
7518 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02007519 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007520 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02007521 Arguments :
7522 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007523 actions include : "accept", "reject", "track-sc0", "track-sc1",
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007524 "track-sc2" and "capture". See "tcp-request connection" above
7525 for their signification.
Willy Tarreau62644772008-07-16 18:36:06 +02007526
Willy Tarreaue9656522010-08-17 15:40:09 +02007527 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02007528
Willy Tarreaue9656522010-08-17 15:40:09 +02007529 A request's contents can be analysed at an early stage of request processing
7530 called "TCP content inspection". During this stage, ACL-based rules are
7531 evaluated every time the request contents are updated, until either an
7532 "accept" or a "reject" rule matches, or the TCP request inspection delay
7533 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02007534
Willy Tarreaue9656522010-08-17 15:40:09 +02007535 The first difference between these rules and "tcp-request connection" rules
7536 is that "tcp-request content" rules can make use of contents to take a
7537 decision. Most often, these decisions will consider a protocol recognition or
7538 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01007539 both frontends and backends. In case of HTTP keep-alive with the client, all
7540 tcp-request content rules are evaluated again, so haproxy keeps a record of
7541 what sticky counters were assigned by a "tcp-request connection" versus a
7542 "tcp-request content" rule, and flushes all the content-related ones after
7543 processing an HTTP request, so that they may be evaluated again by the rules
7544 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007545 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01007546 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02007547
Willy Tarreaue9656522010-08-17 15:40:09 +02007548 Content-based rules are evaluated in their exact declaration order. If no
7549 rule matches or if there is no rule, the default action is to accept the
7550 contents. There is no specific limit to the number of rules which may be
7551 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02007552
Willy Tarreau18bf01e2014-06-13 16:18:52 +02007553 Four types of actions are supported :
7554 - accept : the request is accepted
7555 - reject : the request is rejected and the connection is closed
7556 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007557 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Willy Tarreau62644772008-07-16 18:36:06 +02007558
Willy Tarreaue9656522010-08-17 15:40:09 +02007559 They have the same meaning as their counter-parts in "tcp-request connection"
7560 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02007561
Willy Tarreauf3338342014-01-28 21:40:28 +01007562 While there is nothing mandatory about it, it is recommended to use the
7563 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
7564 content" rules in the frontend, and track-sc2 for "tcp-request content"
7565 rules in the backend, because that makes the configuration more readable
7566 and easier to troubleshoot, but this is just a guideline and all counters
7567 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02007568
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007569 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02007570 the action, it is simply performed unconditionally. That can be useful for
7571 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02007572
Willy Tarreaue9656522010-08-17 15:40:09 +02007573 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02007574 rules, since HTTP-specific ACL matches are able to preliminarily parse the
7575 contents of a buffer before extracting the required data. If the buffered
7576 contents do not parse as a valid HTTP message, then the ACL does not match.
7577 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01007578 processing, so there is no risk of parsing something differently. In an HTTP
7579 backend connected to from an HTTP frontend, it is guaranteed that HTTP
7580 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02007581
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007582 Tracking layer7 information is also possible provided that the information
7583 are present when the rule is processed. The current solution for making the
7584 rule engine wait for such information is to set an inspect delay and to
7585 condition its execution with an ACL relying on such information.
7586
Willy Tarreau62644772008-07-16 18:36:06 +02007587 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02007588 # Accept HTTP requests containing a Host header saying "example.com"
7589 # and reject everything else.
7590 acl is_host_com hdr(Host) -i example.com
7591 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02007592 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02007593 tcp-request content reject
7594
7595 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02007596 # reject SMTP connection if client speaks first
7597 tcp-request inspect-delay 30s
7598 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007599 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02007600
7601 # Forward HTTPS connection only if client speaks
7602 tcp-request inspect-delay 30s
7603 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02007604 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02007605 tcp-request content reject
7606
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007607 Example:
7608 # Track the last IP from X-Forwarded-For
7609 tcp-request inspect-delay 10s
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007610 tcp-request content track-sc0 hdr(x-forwarded-for,-1) if HTTP
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007611
7612 Example:
7613 # track request counts per "base" (concatenation of Host+URL)
7614 tcp-request inspect-delay 10s
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007615 tcp-request content track-sc0 base table req-rate if HTTP
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01007616
Willy Tarreaue9656522010-08-17 15:40:09 +02007617 Example: track per-frontend and per-backend counters, block abusers at the
7618 frontend when the backend detects abuse.
7619
7620 frontend http
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007621 # Use General Purpose Couter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02007622 # protecting all our sites
7623 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007624 tcp-request connection track-sc0 src
7625 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02007626 ...
7627 use_backend http_dynamic if { path_end .php }
7628
7629 backend http_dynamic
7630 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007631 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02007632 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02007633 acl click_too_fast sc1_http_req_rate gt 10
7634 acl mark_as_abuser sc0_inc_gpc0 gt 0
7635 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02007636 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02007637
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007638 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02007639
Willy Tarreaue9656522010-08-17 15:40:09 +02007640 See also : "tcp-request connection", "tcp-request inspect-delay"
Willy Tarreau62644772008-07-16 18:36:06 +02007641
7642
7643tcp-request inspect-delay <timeout>
7644 Set the maximum allowed time to wait for data during content inspection
7645 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02007646 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02007647 Arguments :
7648 <timeout> is the timeout value specified in milliseconds by default, but
7649 can be in any other unit if the number is suffixed by the unit,
7650 as explained at the top of this document.
7651
7652 People using haproxy primarily as a TCP relay are often worried about the
7653 risk of passing any type of protocol to a server without any analysis. In
7654 order to be able to analyze the request contents, we must first withhold
7655 the data then analyze them. This statement simply enables withholding of
7656 data for at most the specified amount of time.
7657
Willy Tarreaufb356202010-08-03 14:02:05 +02007658 TCP content inspection applies very early when a connection reaches a
7659 frontend, then very early when the connection is forwarded to a backend. This
7660 means that a connection may experience a first delay in the frontend and a
7661 second delay in the backend if both have tcp-request rules.
7662
Willy Tarreau62644772008-07-16 18:36:06 +02007663 Note that when performing content inspection, haproxy will evaluate the whole
7664 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007665 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02007666 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01007667 contents are definitive. If no delay is set, haproxy will not wait at all
7668 and will immediately apply a verdict based on the available information.
7669 Obviously this is unlikely to be very useful and might even be racy, so such
7670 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02007671
7672 As soon as a rule matches, the request is released and continues as usual. If
7673 the timeout is reached and no rule matches, the default policy will be to let
7674 it pass through unaffected.
7675
7676 For most protocols, it is enough to set it to a few seconds, as most clients
7677 send the full request immediately upon connection. Add 3 or more seconds to
7678 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01007679 to use large values, for instance to ensure that the client never talks
Willy Tarreau62644772008-07-16 18:36:06 +02007680 before the server (eg: SMTP), or to wait for a client to talk before passing
7681 data to the server (eg: SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02007682 least the inspection delay, otherwise it will expire first. If the client
7683 closes the connection or if the buffer is full, the delay immediately expires
7684 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02007685
Willy Tarreau55165fe2009-05-10 12:02:55 +02007686 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02007687 "timeout client".
7688
7689
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007690tcp-response content <action> [{if | unless} <condition>]
7691 Perform an action on a session response depending on a layer 4-7 condition
7692 May be used in sections : defaults | frontend | listen | backend
7693 no | no | yes | yes
7694 Arguments :
7695 <action> defines the action to perform if the condition applies. Valid
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007696 actions include : "accept", "close", "reject".
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007697
7698 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
7699
7700 Response contents can be analysed at an early stage of response processing
7701 called "TCP content inspection". During this stage, ACL-based rules are
7702 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007703 "accept", "close" or a "reject" rule matches, or a TCP response inspection
7704 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007705
7706 Most often, these decisions will consider a protocol recognition or validity.
7707
7708 Content-based rules are evaluated in their exact declaration order. If no
7709 rule matches or if there is no rule, the default action is to accept the
7710 contents. There is no specific limit to the number of rules which may be
7711 inserted.
7712
7713 Two types of actions are supported :
7714 - accept :
7715 accepts the response if the condition is true (when used with "if")
7716 or false (when used with "unless"). The first such rule executed ends
7717 the rules evaluation.
7718
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007719 - close :
7720 immediately closes the connection with the server if the condition is
7721 true (when used with "if"), or false (when used with "unless"). The
7722 first such rule executed ends the rules evaluation. The main purpose of
7723 this action is to force a connection to be finished between a client
7724 and a server after an exchange when the application protocol expects
7725 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007726 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02007727 protocols.
7728
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007729 - reject :
7730 rejects the response if the condition is true (when used with "if")
7731 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007732 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007733
7734 Note that the "if/unless" condition is optional. If no condition is set on
7735 the action, it is simply performed unconditionally. That can be useful for
7736 for changing the default action to a reject.
7737
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007738 It is perfectly possible to match layer 7 contents with "tcp-response
7739 content" rules, but then it is important to ensure that a full response has
7740 been buffered, otherwise no contents will match. In order to achieve this,
7741 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02007742 period.
7743
7744 See section 7 about ACL usage.
7745
7746 See also : "tcp-request content", "tcp-response inspect-delay"
7747
7748
7749tcp-response inspect-delay <timeout>
7750 Set the maximum allowed time to wait for a response during content inspection
7751 May be used in sections : defaults | frontend | listen | backend
7752 no | no | yes | yes
7753 Arguments :
7754 <timeout> is the timeout value specified in milliseconds by default, but
7755 can be in any other unit if the number is suffixed by the unit,
7756 as explained at the top of this document.
7757
7758 See also : "tcp-response content", "tcp-request inspect-delay".
7759
7760
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007761timeout check <timeout>
7762 Set additional check timeout, but only after a connection has been already
7763 established.
7764
7765 May be used in sections: defaults | frontend | listen | backend
7766 yes | no | yes | yes
7767 Arguments:
7768 <timeout> is the timeout value specified in milliseconds by default, but
7769 can be in any other unit if the number is suffixed by the unit,
7770 as explained at the top of this document.
7771
7772 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
7773 for check and "timeout check" as an additional read timeout. The "min" is
7774 used so that people running with *very* long "timeout connect" (eg. those
7775 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01007776 (Please also note that there is no valid reason to have such long connect
7777 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
7778 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007779
7780 If "timeout check" is not set haproxy uses "inter" for complete check
7781 timeout (connect + read) exactly like all <1.3.15 version.
7782
7783 In most cases check request is much simpler and faster to handle than normal
7784 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01007785 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007786
7787 This parameter is specific to backends, but can be specified once for all in
7788 "defaults" sections. This is in fact one of the easiest solutions not to
7789 forget about it.
7790
Willy Tarreau41a340d2008-01-22 12:25:31 +01007791 See also: "timeout connect", "timeout queue", "timeout server",
7792 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007793
7794
Willy Tarreau0ba27502007-12-24 16:55:16 +01007795timeout client <timeout>
7796timeout clitimeout <timeout> (deprecated)
7797 Set the maximum inactivity time on the client side.
7798 May be used in sections : defaults | frontend | listen | backend
7799 yes | yes | yes | no
7800 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007801 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007802 can be in any other unit if the number is suffixed by the unit,
7803 as explained at the top of this document.
7804
7805 The inactivity timeout applies when the client is expected to acknowledge or
7806 send data. In HTTP mode, this timeout is particularly important to consider
7807 during the first phase, when the client sends the request, and during the
7808 response while it is reading data sent by the server. The value is specified
7809 in milliseconds by default, but can be in any other unit if the number is
7810 suffixed by the unit, as specified at the top of this document. In TCP mode
7811 (and to a lesser extent, in HTTP mode), it is highly recommended that the
7812 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007813 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01007814 losses by specifying timeouts that are slightly above multiples of 3 seconds
Willy Tarreauce887fd2012-05-12 12:50:00 +02007815 (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
7816 sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02007817 which overrides "timeout client" and "timeout server" for tunnels, as well as
7818 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007819
7820 This parameter is specific to frontends, but can be specified once for all in
7821 "defaults" sections. This is in fact one of the easiest solutions not to
7822 forget about it. An unspecified timeout results in an infinite timeout, which
7823 is not recommended. Such a usage is accepted and works but reports a warning
7824 during startup because it may results in accumulation of expired sessions in
7825 the system if the system's timeouts are not configured either.
7826
7827 This parameter replaces the old, deprecated "clitimeout". It is recommended
7828 to use it to write new configurations. The form "timeout clitimeout" is
7829 provided only by backwards compatibility but its use is strongly discouraged.
7830
Willy Tarreauce887fd2012-05-12 12:50:00 +02007831 See also : "clitimeout", "timeout server", "timeout tunnel".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007832
7833
Willy Tarreau05cdd962014-05-10 14:30:07 +02007834timeout client-fin <timeout>
7835 Set the inactivity timeout on the client side for half-closed connections.
7836 May be used in sections : defaults | frontend | listen | backend
7837 yes | yes | yes | no
7838 Arguments :
7839 <timeout> is the timeout value specified in milliseconds by default, but
7840 can be in any other unit if the number is suffixed by the unit,
7841 as explained at the top of this document.
7842
7843 The inactivity timeout applies when the client is expected to acknowledge or
7844 send data while one direction is already shut down. This timeout is different
7845 from "timeout client" in that it only applies to connections which are closed
7846 in one direction. This is particularly useful to avoid keeping connections in
7847 FIN_WAIT state for too long when clients do not disconnect cleanly. This
7848 problem is particularly common long connections such as RDP or WebSocket.
7849 Note that this timeout can override "timeout tunnel" when a connection shuts
7850 down in one direction.
7851
7852 This parameter is specific to frontends, but can be specified once for all in
7853 "defaults" sections. By default it is not set, so half-closed connections
7854 will use the other timeouts (timeout.client or timeout.tunnel).
7855
7856 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
7857
7858
Willy Tarreau0ba27502007-12-24 16:55:16 +01007859timeout connect <timeout>
7860timeout contimeout <timeout> (deprecated)
7861 Set the maximum time to wait for a connection attempt to a server to succeed.
7862 May be used in sections : defaults | frontend | listen | backend
7863 yes | no | yes | yes
7864 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007865 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01007866 can be in any other unit if the number is suffixed by the unit,
7867 as explained at the top of this document.
7868
7869 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007870 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01007871 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01007872 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01007873 connect timeout also presets both queue and tarpit timeouts to the same value
7874 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007875
7876 This parameter is specific to backends, but can be specified once for all in
7877 "defaults" sections. This is in fact one of the easiest solutions not to
7878 forget about it. An unspecified timeout results in an infinite timeout, which
7879 is not recommended. Such a usage is accepted and works but reports a warning
7880 during startup because it may results in accumulation of failed sessions in
7881 the system if the system's timeouts are not configured either.
7882
7883 This parameter replaces the old, deprecated "contimeout". It is recommended
7884 to use it to write new configurations. The form "timeout contimeout" is
7885 provided only by backwards compatibility but its use is strongly discouraged.
7886
Willy Tarreau41a340d2008-01-22 12:25:31 +01007887 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
7888 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01007889
7890
Willy Tarreaub16a5742010-01-10 14:46:16 +01007891timeout http-keep-alive <timeout>
7892 Set the maximum allowed time to wait for a new HTTP request to appear
7893 May be used in sections : defaults | frontend | listen | backend
7894 yes | yes | yes | yes
7895 Arguments :
7896 <timeout> is the timeout value specified in milliseconds by default, but
7897 can be in any other unit if the number is suffixed by the unit,
7898 as explained at the top of this document.
7899
7900 By default, the time to wait for a new request in case of keep-alive is set
7901 by "timeout http-request". However this is not always convenient because some
7902 people want very short keep-alive timeouts in order to release connections
7903 faster, and others prefer to have larger ones but still have short timeouts
7904 once the request has started to present itself.
7905
7906 The "http-keep-alive" timeout covers these needs. It will define how long to
7907 wait for a new HTTP request to start coming after a response was sent. Once
7908 the first byte of request has been seen, the "http-request" timeout is used
7909 to wait for the complete request to come. Note that empty lines prior to a
7910 new request do not refresh the timeout and are not counted as a new request.
7911
7912 There is also another difference between the two timeouts : when a connection
7913 expires during timeout http-keep-alive, no error is returned, the connection
7914 just closes. If the connection expires in "http-request" while waiting for a
7915 connection to complete, a HTTP 408 error is returned.
7916
7917 In general it is optimal to set this value to a few tens to hundreds of
7918 milliseconds, to allow users to fetch all objects of a page at once but
7919 without waiting for further clicks. Also, if set to a very small value (eg:
7920 1 millisecond) it will probably only accept pipelined requests but not the
7921 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +02007922 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +01007923
7924 If this parameter is not set, the "http-request" timeout applies, and if both
7925 are not set, "timeout client" still applies at the lower level. It should be
7926 set in the frontend to take effect, unless the frontend is in TCP mode, in
7927 which case the HTTP backend's timeout will be used.
7928
7929 See also : "timeout http-request", "timeout client".
7930
7931
Willy Tarreau036fae02008-01-06 13:24:40 +01007932timeout http-request <timeout>
7933 Set the maximum allowed time to wait for a complete HTTP request
7934 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007935 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +01007936 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01007937 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +01007938 can be in any other unit if the number is suffixed by the unit,
7939 as explained at the top of this document.
7940
7941 In order to offer DoS protection, it may be required to lower the maximum
7942 accepted time to receive a complete HTTP request without affecting the client
7943 timeout. This helps protecting against established connections on which
7944 nothing is sent. The client timeout cannot offer a good protection against
7945 this abuse because it is an inactivity timeout, which means that if the
7946 attacker sends one character every now and then, the timeout will not
7947 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +02007948 types, the request will be aborted if it does not complete in time. When the
7949 timeout expires, an HTTP 408 response is sent to the client to inform it
7950 about the problem, and the connection is closed. The logs will report
7951 termination codes "cR". Some recent browsers are having problems with this
7952 standard, well-documented behaviour, so it might be needed to hide the 408
7953 code using "errorfile 408 /dev/null". See more details in the explanations of
7954 the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +01007955
7956 Note that this timeout only applies to the header part of the request, and
7957 not to any data. As soon as the empty line is received, this timeout is not
Willy Tarreaub16a5742010-01-10 14:46:16 +01007958 used anymore. It is used again on keep-alive connections to wait for a second
7959 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +01007960
7961 Generally it is enough to set it to a few seconds, as most clients send the
7962 full request immediately upon connection. Add 3 or more seconds to cover TCP
7963 retransmits but that's all. Setting it to very low values (eg: 50 ms) will
7964 generally work on local networks as long as there are no packet losses. This
7965 will prevent people from sending bare HTTP requests using telnet.
7966
7967 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +02007968 chunk of the incoming request. It should be set in the frontend to take
7969 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
7970 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +01007971
Willy Tarreau2705a612014-05-23 17:38:34 +02007972 See also : "errorfile", "timeout http-keep-alive", "timeout client".
Willy Tarreau036fae02008-01-06 13:24:40 +01007973
Willy Tarreau844e3c52008-01-11 16:28:18 +01007974
7975timeout queue <timeout>
7976 Set the maximum time to wait in the queue for a connection slot to be free
7977 May be used in sections : defaults | frontend | listen | backend
7978 yes | no | yes | yes
7979 Arguments :
7980 <timeout> is the timeout value specified in milliseconds by default, but
7981 can be in any other unit if the number is suffixed by the unit,
7982 as explained at the top of this document.
7983
7984 When a server's maxconn is reached, connections are left pending in a queue
7985 which may be server-specific or global to the backend. In order not to wait
7986 indefinitely, a timeout is applied to requests pending in the queue. If the
7987 timeout is reached, it is considered that the request will almost never be
7988 served, so it is dropped and a 503 error is returned to the client.
7989
7990 The "timeout queue" statement allows to fix the maximum time for a request to
7991 be left pending in a queue. If unspecified, the same value as the backend's
7992 connection timeout ("timeout connect") is used, for backwards compatibility
7993 with older versions with no "timeout queue" parameter.
7994
7995 See also : "timeout connect", "contimeout".
7996
7997
7998timeout server <timeout>
7999timeout srvtimeout <timeout> (deprecated)
8000 Set the maximum inactivity time on the server side.
8001 May be used in sections : defaults | frontend | listen | backend
8002 yes | no | yes | yes
8003 Arguments :
8004 <timeout> is the timeout value specified in milliseconds by default, but
8005 can be in any other unit if the number is suffixed by the unit,
8006 as explained at the top of this document.
8007
8008 The inactivity timeout applies when the server is expected to acknowledge or
8009 send data. In HTTP mode, this timeout is particularly important to consider
8010 during the first phase of the server's response, when it has to send the
8011 headers, as it directly represents the server's processing time for the
8012 request. To find out what value to put there, it's often good to start with
8013 what would be considered as unacceptable response times, then check the logs
8014 to observe the response time distribution, and adjust the value accordingly.
8015
8016 The value is specified in milliseconds by default, but can be in any other
8017 unit if the number is suffixed by the unit, as specified at the top of this
8018 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8019 recommended that the client timeout remains equal to the server timeout in
8020 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008021 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008022 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreauce887fd2012-05-12 12:50:00 +02008023 seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed
8024 with short-lived sessions (eg: WebSocket and HTTP), it's worth considering
8025 "timeout tunnel", which overrides "timeout client" and "timeout server" for
8026 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008027
8028 This parameter is specific to backends, but can be specified once for all in
8029 "defaults" sections. This is in fact one of the easiest solutions not to
8030 forget about it. An unspecified timeout results in an infinite timeout, which
8031 is not recommended. Such a usage is accepted and works but reports a warning
8032 during startup because it may results in accumulation of expired sessions in
8033 the system if the system's timeouts are not configured either.
8034
8035 This parameter replaces the old, deprecated "srvtimeout". It is recommended
8036 to use it to write new configurations. The form "timeout srvtimeout" is
8037 provided only by backwards compatibility but its use is strongly discouraged.
8038
Willy Tarreauce887fd2012-05-12 12:50:00 +02008039 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008040
Willy Tarreau05cdd962014-05-10 14:30:07 +02008041
8042timeout server-fin <timeout>
8043 Set the inactivity timeout on the server side for half-closed connections.
8044 May be used in sections : defaults | frontend | listen | backend
8045 yes | no | yes | yes
8046 Arguments :
8047 <timeout> is the timeout value specified in milliseconds by default, but
8048 can be in any other unit if the number is suffixed by the unit,
8049 as explained at the top of this document.
8050
8051 The inactivity timeout applies when the server is expected to acknowledge or
8052 send data while one direction is already shut down. This timeout is different
8053 from "timeout server" in that it only applies to connections which are closed
8054 in one direction. This is particularly useful to avoid keeping connections in
8055 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
8056 This problem is particularly common long connections such as RDP or WebSocket.
8057 Note that this timeout can override "timeout tunnel" when a connection shuts
8058 down in one direction. This setting was provided for completeness, but in most
8059 situations, it should not be needed.
8060
8061 This parameter is specific to backends, but can be specified once for all in
8062 "defaults" sections. By default it is not set, so half-closed connections
8063 will use the other timeouts (timeout.server or timeout.tunnel).
8064
8065 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
8066
Willy Tarreau844e3c52008-01-11 16:28:18 +01008067
8068timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +01008069 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +01008070 May be used in sections : defaults | frontend | listen | backend
8071 yes | yes | yes | yes
8072 Arguments :
8073 <timeout> is the tarpit duration specified in milliseconds by default, but
8074 can be in any other unit if the number is suffixed by the unit,
8075 as explained at the top of this document.
8076
8077 When a connection is tarpitted using "reqtarpit", it is maintained open with
8078 no activity for a certain amount of time, then closed. "timeout tarpit"
8079 defines how long it will be maintained open.
8080
8081 The value is specified in milliseconds by default, but can be in any other
8082 unit if the number is suffixed by the unit, as specified at the top of this
8083 document. If unspecified, the same value as the backend's connection timeout
8084 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +01008085 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008086
8087 See also : "timeout connect", "contimeout".
8088
8089
Willy Tarreauce887fd2012-05-12 12:50:00 +02008090timeout tunnel <timeout>
8091 Set the maximum inactivity time on the client and server side for tunnels.
8092 May be used in sections : defaults | frontend | listen | backend
8093 yes | no | yes | yes
8094 Arguments :
8095 <timeout> is the timeout value specified in milliseconds by default, but
8096 can be in any other unit if the number is suffixed by the unit,
8097 as explained at the top of this document.
8098
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008099 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +02008100 between a client and a server, and the connection remains inactive in both
8101 directions. This timeout supersedes both the client and server timeouts once
8102 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
8103 analyser remains attached to either connection (eg: tcp content rules are
8104 accepted). In HTTP, this timeout is used when a connection is upgraded (eg:
8105 when switching to the WebSocket protocol, or forwarding a CONNECT request
8106 to a proxy), or after the first response when no keepalive/close option is
8107 specified.
8108
Willy Tarreau05cdd962014-05-10 14:30:07 +02008109 Since this timeout is usually used in conjunction with long-lived connections,
8110 it usually is a good idea to also set "timeout client-fin" to handle the
8111 situation where a client suddenly disappears from the net and does not
8112 acknowledge a close, or sends a shutdown and does not acknowledge pending
8113 data anymore. This can happen in lossy networks where firewalls are present,
8114 and is detected by the presence of large amounts of sessions in a FIN_WAIT
8115 state.
8116
Willy Tarreauce887fd2012-05-12 12:50:00 +02008117 The value is specified in milliseconds by default, but can be in any other
8118 unit if the number is suffixed by the unit, as specified at the top of this
8119 document. Whatever the expected normal idle time, it is a good practice to
8120 cover at least one or several TCP packet losses by specifying timeouts that
8121 are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum).
8122
8123 This parameter is specific to backends, but can be specified once for all in
8124 "defaults" sections. This is in fact one of the easiest solutions not to
8125 forget about it.
8126
8127 Example :
8128 defaults http
8129 option http-server-close
8130 timeout connect 5s
8131 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +02008132 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +02008133 timeout server 30s
8134 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
8135
Willy Tarreau05cdd962014-05-10 14:30:07 +02008136 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +02008137
8138
Willy Tarreau844e3c52008-01-11 16:28:18 +01008139transparent (deprecated)
8140 Enable client-side transparent proxying
8141 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01008142 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01008143 Arguments : none
8144
8145 This keyword was introduced in order to provide layer 7 persistence to layer
8146 3 load balancers. The idea is to use the OS's ability to redirect an incoming
8147 connection for a remote address to a local process (here HAProxy), and let
8148 this process know what address was initially requested. When this option is
8149 used, sessions without cookies will be forwarded to the original destination
8150 IP address of the incoming request (which should match that of another
8151 equipment), while requests with cookies will still be forwarded to the
8152 appropriate server.
8153
8154 The "transparent" keyword is deprecated, use "option transparent" instead.
8155
8156 Note that contrary to a common belief, this option does NOT make HAProxy
8157 present the client's IP to the server when establishing the connection.
8158
Willy Tarreau844e3c52008-01-11 16:28:18 +01008159 See also: "option transparent"
8160
William Lallemanda73203e2012-03-12 12:48:57 +01008161unique-id-format <string>
8162 Generate a unique ID for each request.
8163 May be used in sections : defaults | frontend | listen | backend
8164 yes | yes | yes | no
8165 Arguments :
8166 <string> is a log-format string.
8167
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008168 This keyword creates a ID for each request using the custom log format. A
8169 unique ID is useful to trace a request passing through many components of
8170 a complex infrastructure. The newly created ID may also be logged using the
8171 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +01008172
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008173 The format should be composed from elements that are guaranteed to be
8174 unique when combined together. For instance, if multiple haproxy instances
8175 are involved, it might be important to include the node name. It is often
8176 needed to log the incoming connection's source and destination addresses
8177 and ports. Note that since multiple requests may be performed over the same
8178 connection, including a request counter may help differentiate them.
8179 Similarly, a timestamp may protect against a rollover of the counter.
8180 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +01008181
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008182 It is recommended to use hexadecimal notation for many fields since it
8183 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +01008184
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008185 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008186
Julien Vehentf21be322014-03-07 08:27:34 -05008187 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008188
8189 will generate:
8190
8191 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8192
8193 See also: "unique-id-header"
8194
8195unique-id-header <name>
8196 Add a unique ID header in the HTTP request.
8197 May be used in sections : defaults | frontend | listen | backend
8198 yes | yes | yes | no
8199 Arguments :
8200 <name> is the name of the header.
8201
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008202 Add a unique-id header in the HTTP request sent to the server, using the
8203 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +01008204
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008205 Example:
William Lallemanda73203e2012-03-12 12:48:57 +01008206
Julien Vehentf21be322014-03-07 08:27:34 -05008207 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +01008208 unique-id-header X-Unique-ID
8209
8210 will generate:
8211
8212 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
8213
8214 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +01008215
Willy Tarreauf51658d2014-04-23 01:21:56 +02008216use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008217 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008218 May be used in sections : defaults | frontend | listen | backend
8219 no | yes | yes | no
8220 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008221 <backend> is the name of a valid backend or "listen" section, or a
8222 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008223
Willy Tarreauf51658d2014-04-23 01:21:56 +02008224 <condition> is a condition composed of ACLs, as described in section 7. If
8225 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008226
8227 When doing content-switching, connections arrive on a frontend and are then
8228 dispatched to various backends depending on a number of conditions. The
8229 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +02008230 "use_backend" keyword. While it is normally used with HTTP processing, it can
8231 also be used in pure TCP, either without content using stateless ACLs (eg:
8232 source address validation) or combined with a "tcp-request" rule to wait for
8233 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +01008234
8235 There may be as many "use_backend" rules as desired. All of these rules are
8236 evaluated in their declaration order, and the first one which matches will
8237 assign the backend.
8238
8239 In the first form, the backend will be used if the condition is met. In the
8240 second form, the backend will be used if the condition is not met. If no
8241 condition is valid, the backend defined with "default_backend" will be used.
8242 If no default backend is defined, either the servers in the same section are
8243 used (in case of a "listen" section) or, in case of a frontend, no server is
8244 used and a 503 service unavailable response is returned.
8245
Willy Tarreau51aecc72009-07-12 09:47:04 +02008246 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008247 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +02008248 and backend processing will immediately follow, or the backend will wait for
8249 a complete HTTP request to get in. This feature is useful when a frontend
8250 must decode several protocols on a unique port, one of them being HTTP.
8251
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008252 When <backend> is a simple name, it is resolved at configuration time, and an
8253 error is reported if the specified backend does not exist. If <backend> is
8254 a log-format string instead, no check may be done at configuration time, so
8255 the backend name is resolved dynamically at run time. If the resulting
8256 backend name does not correspond to any valid backend, no other rule is
8257 evaluated, and the default_backend directive is applied instead. Note that
8258 when using dynamic backend names, it is highly recommended to use a prefix
8259 that no other backend uses in order to ensure that an unauthorized backend
8260 cannot be forced from the request.
8261
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008262 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01008263 used to detect the association between frontends and backends to compute the
8264 backend's "fullconn" setting. This cannot be done for dynamic names.
8265
8266 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
8267 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008268
Willy Tarreau036fae02008-01-06 13:24:40 +01008269
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008270use-server <server> if <condition>
8271use-server <server> unless <condition>
8272 Only use a specific server if/unless an ACL-based condition is matched.
8273 May be used in sections : defaults | frontend | listen | backend
8274 no | no | yes | yes
8275 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008276 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008277
8278 <condition> is a condition composed of ACLs, as described in section 7.
8279
8280 By default, connections which arrive to a backend are load-balanced across
8281 the available servers according to the configured algorithm, unless a
8282 persistence mechanism such as a cookie is used and found in the request.
8283
8284 Sometimes it is desirable to forward a particular request to a specific
8285 server without having to declare a dedicated backend for this server. This
8286 can be achieved using the "use-server" rules. These rules are evaluated after
8287 the "redirect" rules and before evaluating cookies, and they have precedence
8288 on them. There may be as many "use-server" rules as desired. All of these
8289 rules are evaluated in their declaration order, and the first one which
8290 matches will assign the server.
8291
8292 If a rule designates a server which is down, and "option persist" is not used
8293 and no force-persist rule was validated, it is ignored and evaluation goes on
8294 with the next rules until one matches.
8295
8296 In the first form, the server will be used if the condition is met. In the
8297 second form, the server will be used if the condition is not met. If no
8298 condition is valid, the processing continues and the server will be assigned
8299 according to other persistence mechanisms.
8300
8301 Note that even if a rule is matched, cookie processing is still performed but
8302 does not assign the server. This allows prefixed cookies to have their prefix
8303 stripped.
8304
8305 The "use-server" statement works both in HTTP and TCP mode. This makes it
8306 suitable for use with content-based inspection. For instance, a server could
8307 be selected in a farm according to the TLS SNI field. And if these servers
8308 have their weight set to zero, they will not be used for other traffic.
8309
8310 Example :
8311 # intercept incoming TLS requests based on the SNI field
8312 use-server www if { req_ssl_sni -i www.example.com }
8313 server www 192.168.0.1:443 weight 0
8314 use-server mail if { req_ssl_sni -i mail.example.com }
8315 server mail 192.168.0.1:587 weight 0
8316 use-server imap if { req_ssl_sni -i imap.example.com }
8317 server mail 192.168.0.1:993 weight 0
8318 # all the rest is forwarded to this server
8319 server default 192.168.0.2:443 check
8320
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008321 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008322
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008323
83245. Bind and Server options
8325--------------------------
8326
8327The "bind", "server" and "default-server" keywords support a number of settings
8328depending on some build options and on the system HAProxy was built on. These
8329settings generally each consist in one word sometimes followed by a value,
8330written on the same line as the "bind" or "server" line. All these options are
8331described in this section.
8332
8333
83345.1. Bind options
8335-----------------
8336
8337The "bind" keyword supports a certain number of settings which are all passed
8338as arguments on the same line. The order in which those arguments appear makes
8339no importance, provided that they appear after the bind address. All of these
8340parameters are optional. Some of them consist in a single words (booleans),
8341while other ones expect a value after them. In this case, the value must be
8342provided immediately after the setting name.
8343
8344The currently supported settings are the following ones.
8345
8346accept-proxy
8347 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +02008348 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
8349 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008350 3/4 addresses of the incoming connection to be used everywhere an address is
8351 used, with the only exception of "tcp-request connection" rules which will
8352 only see the real connection address. Logs will reflect the addresses
8353 indicated in the protocol, unless it is violated, in which case the real
8354 address will still be used. This keyword combined with support from external
8355 components can be used as an efficient and reliable alternative to the
8356 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +02008357 usable. See also "tcp-request connection expect-proxy" for a finer-grained
8358 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008359
Willy Tarreauab861d32013-04-02 02:30:41 +02008360alpn <protocols>
8361 This enables the TLS ALPN extension and advertises the specified protocol
8362 list as supported on top of ALPN. The protocol list consists in a comma-
8363 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
8364 quotes). This requires that the SSL library is build with support for TLS
8365 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
8366 initial NPN extension.
8367
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008368backlog <backlog>
8369 Sets the socket's backlog to this value. If unspecified, the frontend's
8370 backlog is used instead, which generally defaults to the maxconn value.
8371
Emeric Brun7fb34422012-09-28 15:26:15 +02008372ecdhe <named curve>
8373 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +01008374 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
8375 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +02008376
Emeric Brunfd33a262012-10-11 16:28:27 +02008377ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008378 This setting is only available when support for OpenSSL was built in. It
8379 designates a PEM file from which to load CA certificates used to verify
8380 client's certificate.
8381
Emeric Brunb6dc9342012-09-28 17:55:37 +02008382ca-ignore-err [all|<errorID>,...]
8383 This setting is only available when support for OpenSSL was built in.
8384 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
8385 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
8386 error is ignored.
8387
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008388ciphers <ciphers>
8389 This setting is only available when support for OpenSSL was built in. It sets
8390 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008391 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008392 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
8393 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
8394
Emeric Brunfd33a262012-10-11 16:28:27 +02008395crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +02008396 This setting is only available when support for OpenSSL was built in. It
8397 designates a PEM file from which to load certificate revocation list used
8398 to verify client's certificate.
8399
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008400crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +00008401 This setting is only available when support for OpenSSL was built in. It
8402 designates a PEM file containing both the required certificates and any
8403 associated private keys. This file can be built by concatenating multiple
8404 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
8405 requires an intermediate certificate, this can also be concatenated into this
8406 file.
8407
8408 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
8409 are loaded.
8410
8411 If a directory name is used instead of a PEM file, then all files found in
Emeric Brun4147b2e2014-06-16 18:36:30 +02008412 that directory will be loaded unless their name ends with '.issuer' or
8413 '.ocsp' (reserved extensions). This directive may be specified multiple times
Alex Davies0fbf0162013-03-02 16:04:50 +00008414 in order to load certificates from multiple files or directories. The
8415 certificates will be presented to clients who provide a valid TLS Server Name
8416 Indication field matching one of their CN or alt subjects. Wildcards are
8417 supported, where a wildcard character '*' is used instead of the first
8418 hostname component (eg: *.example.org matches www.example.org but not
8419 www.sub.example.org).
8420
8421 If no SNI is provided by the client or if the SSL library does not support
8422 TLS extensions, or if the client provides an SNI hostname which does not
8423 match any certificate, then the first loaded certificate will be presented.
8424 This means that when loading certificates from a directory, it is highly
8425 recommended to load the default one first as a file.
8426
Emeric Brune032bfa2012-09-28 13:01:45 +02008427 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008428
Alex Davies0fbf0162013-03-02 16:04:50 +00008429 Some CAs (such as Godaddy) offer a drop down list of server types that do not
8430 include HAProxy when obtaining a certificate. If this happens be sure to
Godbach8bf60a12014-04-21 21:42:41 +08008431 choose a webserver that the CA believes requires an intermediate CA (for
Alex Davies0fbf0162013-03-02 16:04:50 +00008432 Godaddy, selection Apache Tomcat will get the correct bundle, but many
8433 others, e.g. nginx, result in a wrong bundle that will not work for some
8434 clients).
8435
Emeric Brun4147b2e2014-06-16 18:36:30 +02008436 For each PEM file, haproxy checks for the presence of file at the same path
8437 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
8438 Status Request extension (also known as "OCSP stapling") is automatically
8439 enabled. The content of this file is optional. If not empty, it must contain
8440 a valid OCSP Response in DER format. In order to be valid an OCSP Response
8441 must comply with the following rules: it has to indicate a good status,
8442 it has to be a single response for the certificate of the PEM file, and it
8443 has to be valid at the moment of addition. If these rules are not respected
8444 the OCSP Response is ignored and a warning is emitted. In order to identify
8445 which certificate an OCSP Response applies to, the issuer's certificate is
8446 necessary. If the issuer's certificate is not found in the PEM file, it will
8447 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
8448 if it exists otherwise it will fail with an error.
8449
Emeric Brunb6dc9342012-09-28 17:55:37 +02008450crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +00008451 This setting is only available when support for OpenSSL was built in. Sets a
8452 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008453 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +00008454 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +02008455
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008456crt-list <file>
8457 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008458 designates a list of PEM file with an optional list of SNI filter per
8459 certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008460
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008461 <crtfile> [[!]<snifilter> ...]
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008462
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +02008463 Wildcards are supported in the SNI filter. Negative filter are also supported,
8464 only useful in combination with a wildcard filter to exclude a particular SNI.
8465 The certificates will be presented to clients who provide a valid TLS Server
8466 Name Indication field matching one of the SNI filters. If no SNI filter is
8467 specified, the CN and alt subjects are used. This directive may be specified
8468 multiple times. See the "crt" option for more information. The default
8469 certificate is still needed to meet OpenSSL expectations. If it is not used,
8470 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +01008471
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008472defer-accept
8473 Is an optional keyword which is supported only on certain Linux kernels. It
8474 states that a connection will only be accepted once some data arrive on it,
8475 or at worst after the first retransmit. This should be used only on protocols
8476 for which the client talks first (eg: HTTP). It can slightly improve
8477 performance by ensuring that most of the request is already available when
8478 the connection is accepted. On the other hand, it will not be able to detect
8479 connections which don't talk. It is important to note that this option is
8480 broken in all kernels up to 2.6.31, as the connection is never accepted until
8481 the client talks. This can cause issues with front firewalls which would see
8482 an established connection while the proxy will only see it in SYN_RECV. This
8483 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
8484
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008485force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008486 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008487 this listener. SSLv3 is generally less expensive than the TLS counterparts
8488 for high connection rates. See also "force-tls*", "no-sslv3", and "no-tls*".
8489
8490force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008491 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008492 this listener. See also "force-tls*", "no-sslv3", and "no-tls*".
8493
8494force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008495 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008496 this listener. See also "force-tls*", "no-sslv3", and "no-tls*".
8497
8498force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008499 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008500 this listener. See also "force-tls*", "no-sslv3", and "no-tls*".
8501
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008502gid <gid>
8503 Sets the group of the UNIX sockets to the designated system gid. It can also
8504 be set by default in the global section's "unix-bind" statement. Note that
8505 some platforms simply ignore this. This setting is equivalent to the "group"
8506 setting except that the group ID is used instead of its name. This setting is
8507 ignored by non UNIX sockets.
8508
8509group <group>
8510 Sets the group of the UNIX sockets to the designated system group. It can
8511 also be set by default in the global section's "unix-bind" statement. Note
8512 that some platforms simply ignore this. This setting is equivalent to the
8513 "gid" setting except that the group name is used instead of its gid. This
8514 setting is ignored by non UNIX sockets.
8515
8516id <id>
8517 Fixes the socket ID. By default, socket IDs are automatically assigned, but
8518 sometimes it is more convenient to fix them to ease monitoring. This value
8519 must be strictly positive and unique within the listener/frontend. This
8520 option can only be used when defining only a single socket.
8521
8522interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +01008523 Restricts the socket to a specific interface. When specified, only packets
8524 received from that particular interface are processed by the socket. This is
8525 currently only supported on Linux. The interface must be a primary system
8526 interface, not an aliased interface. It is also possible to bind multiple
8527 frontends to the same address if they are bound to different interfaces. Note
8528 that binding to a network interface requires root privileges. This parameter
8529 is only compatible with TCPv4/TCPv6 sockets.
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008530
Willy Tarreauabb175f2012-09-24 12:43:26 +02008531level <level>
8532 This setting is used with the stats sockets only to restrict the nature of
8533 the commands that can be issued on the socket. It is ignored by other
8534 sockets. <level> can be one of :
8535 - "user" is the least privileged level ; only non-sensitive stats can be
8536 read, and no change is allowed. It would make sense on systems where it
8537 is not easy to restrict access to the socket.
8538 - "operator" is the default level and fits most common uses. All data can
8539 be read, and only non-sensitive changes are permitted (eg: clear max
8540 counters).
8541 - "admin" should be used with care, as everything is permitted (eg: clear
8542 all counters).
8543
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008544maxconn <maxconn>
8545 Limits the sockets to this number of concurrent connections. Extraneous
8546 connections will remain in the system's backlog until a connection is
8547 released. If unspecified, the limit will be the same as the frontend's
8548 maxconn. Note that in case of port ranges or multiple addresses, the same
8549 value will be applied to each socket. This setting enables different
8550 limitations on expensive sockets, for instance SSL entries which may easily
8551 eat all memory.
8552
8553mode <mode>
8554 Sets the octal mode used to define access permissions on the UNIX socket. It
8555 can also be set by default in the global section's "unix-bind" statement.
8556 Note that some platforms simply ignore this. This setting is ignored by non
8557 UNIX sockets.
8558
8559mss <maxseg>
8560 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
8561 connections. This can be used to force a lower MSS for certain specific
8562 ports, for instance for connections passing through a VPN. Note that this
8563 relies on a kernel feature which is theoretically supported under Linux but
8564 was buggy in all versions prior to 2.6.28. It may or may not work on other
8565 operating systems. It may also not change the advertised value but change the
8566 effective size of outgoing segments. The commonly advertised value for TCPv4
8567 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
8568 positive, it will be used as the advertised MSS. If it is negative, it will
8569 indicate by how much to reduce the incoming connection's advertised MSS for
8570 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
8571
8572name <name>
8573 Sets an optional name for these sockets, which will be reported on the stats
8574 page.
8575
8576nice <nice>
8577 Sets the 'niceness' of connections initiated from the socket. Value must be
8578 in the range -1024..1024 inclusive, and defaults to zero. Positive values
8579 means that such connections are more friendly to others and easily offer
8580 their place in the scheduler. On the opposite, negative values mean that
8581 connections want to run with a higher priority than others. The difference
8582 only happens under high loads when the system is close to saturation.
8583 Negative values are appropriate for low-latency or administration services,
8584 and high values are generally recommended for CPU intensive tasks such as SSL
8585 processing or bulk transfers which are less sensible to latency. For example,
8586 it may make sense to use a positive value for an SMTP socket and a negative
8587 one for an RDP socket.
8588
Emeric Brun9b3009b2012-10-05 11:55:06 +02008589no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008590 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008591 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008592 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008593 be enabled using any configuration option. See also "force-tls*",
8594 and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008595
Emeric Brun90ad8722012-10-02 14:00:59 +02008596no-tls-tickets
8597 This setting is only available when support for OpenSSL was built in. It
8598 disables the stateless session resumption (RFC 5077 TLS Ticket
8599 extension) and force to use stateful session resumption. Stateless
8600 session resumption is more expensive in CPU usage.
8601
Emeric Brun9b3009b2012-10-05 11:55:06 +02008602no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008603 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008604 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008605 when SSL is supported. Note that SSLv2 is forced disabled in the code and
8606 cannot be enabled using any configuration option. See also "force-tls*",
8607 and "force-sslv3".
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008608
Emeric Brun9b3009b2012-10-05 11:55:06 +02008609no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +02008610 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008611 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008612 when SSL is supported. Note that SSLv2 is forced disabled in the code and
8613 cannot be enabled using any configuration option. See also "force-tls*",
8614 and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008615
Emeric Brun9b3009b2012-10-05 11:55:06 +02008616no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +02008617 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008618 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +02008619 when SSL is supported. Note that SSLv2 is forced disabled in the code and
8620 cannot be enabled using any configuration option. See also "force-tls*",
8621 and "force-sslv3".
Emeric Brunf5da4932012-09-28 19:42:54 +02008622
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008623npn <protocols>
8624 This enables the NPN TLS extension and advertises the specified protocol list
8625 as supported on top of NPN. The protocol list consists in a comma-delimited
8626 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
8627 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +02008628 enabled (check with haproxy -vv). Note that the NPN extension has been
8629 replaced with the ALPN extension (see the "alpn" keyword).
Willy Tarreau6c9a3d52012-10-18 18:57:14 +02008630
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008631process [ all | odd | even | <number 1-64>[-<number 1-64>] ]
8632 This restricts the list of processes on which this listener is allowed to
8633 run. It does not enforce any process but eliminates those which do not match.
8634 If the frontend uses a "bind-process" setting, the intersection between the
8635 two is applied. If in the end the listener is not allowed to run on any
8636 remaining process, a warning is emitted, and the listener will either run on
8637 the first process of the listener if a single process was specified, or on
8638 all of its processes if multiple processes were specified. For the unlikely
Willy Tarreauae302532014-05-07 19:22:24 +02008639 case where several ranges are needed, this directive may be repeated. The
8640 main purpose of this directive is to be used with the stats sockets and have
8641 one different socket per process. The second purpose is to have multiple bind
8642 lines sharing the same IP:port but not the same process in a listener, so
8643 that the system can distribute the incoming connections into multiple queues
8644 and allow a smoother inter-process load balancing. Currently Linux 3.9 and
8645 above is known for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02008646
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008647ssl
8648 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008649 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008650 certificate is necessary (see "crt" above). All contents in the buffers will
8651 appear in clear text, so that ACLs and HTTP processing will only have access
8652 to deciphered contents.
8653
Emmanuel Hocdet65623372013-01-24 17:17:15 +01008654strict-sni
8655 This setting is only available when support for OpenSSL was built in. The
8656 SSL/TLS negotiation is allow only if the client provided an SNI which match
8657 a certificate. The default certificate is not used.
8658 See the "crt" option for more information.
8659
Willy Tarreau1c862c52012-10-05 16:21:00 +02008660tfo
Lukas Tribus0defb902013-02-13 23:35:39 +01008661 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +02008662 enables TCP Fast Open on the listening socket, which means that clients which
8663 support this feature will be able to send a request and receive a response
8664 during the 3-way handshake starting from second connection, thus saving one
8665 round-trip after the first connection. This only makes sense with protocols
8666 that use high connection rates and where each round trip matters. This can
8667 possibly cause issues with many firewalls which do not accept data on SYN
8668 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +02008669 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
8670 need to build HAProxy with USE_TFO=1 if your libc doesn't define
8671 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +02008672
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008673transparent
8674 Is an optional keyword which is supported only on certain Linux kernels. It
8675 indicates that the addresses will be bound even if they do not belong to the
8676 local machine, and that packets targeting any of these addresses will be
8677 intercepted just as if the addresses were locally configured. This normally
8678 requires that IP forwarding is enabled. Caution! do not use this with the
8679 default address '*', as it would redirect any traffic for the specified port.
8680 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
8681 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
8682 kernel version. Some distribution kernels include backports of the feature,
8683 so check for support with your vendor.
8684
Willy Tarreau77e3af92012-11-24 15:07:23 +01008685v4v6
8686 Is an optional keyword which is supported only on most recent systems
8687 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
8688 and IPv6 when it uses the default address. Doing so is sometimes necessary
8689 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008690 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +01008691
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008692v6only
8693 Is an optional keyword which is supported only on most recent systems
8694 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
8695 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +01008696 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
8697 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +01008698
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008699uid <uid>
8700 Sets the owner of the UNIX sockets to the designated system uid. It can also
8701 be set by default in the global section's "unix-bind" statement. Note that
8702 some platforms simply ignore this. This setting is equivalent to the "user"
8703 setting except that the user numeric ID is used instead of its name. This
8704 setting is ignored by non UNIX sockets.
8705
8706user <user>
8707 Sets the owner of the UNIX sockets to the designated system user. It can also
8708 be set by default in the global section's "unix-bind" statement. Note that
8709 some platforms simply ignore this. This setting is equivalent to the "uid"
8710 setting except that the user name is used instead of its uid. This setting is
8711 ignored by non UNIX sockets.
8712
Emeric Brun1a073b42012-09-28 17:07:34 +02008713verify [none|optional|required]
8714 This setting is only available when support for OpenSSL was built in. If set
8715 to 'none', client certificate is not requested. This is the default. In other
8716 cases, a client certificate is requested. If the client does not provide a
8717 certificate after the request and if 'verify' is set to 'required', then the
8718 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +02008719 certificate provided by the client is always verified using CAs from
8720 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
8721 is aborted, regardless of the 'verify' option, unless the error code exactly
8722 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +02008723
Willy Tarreaub6205fd2012-09-24 12:27:33 +020087245.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +01008725------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02008726
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008727The "server" and "default-server" keywords support a certain number of settings
8728which are all passed as arguments on the server line. The order in which those
8729arguments appear does not count, and they are all optional. Some of those
8730settings are single words (booleans) while others expect one or several values
8731after them. In this case, the values must immediately follow the setting name.
8732Except default-server, all those settings must be specified after the server's
8733address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +02008734
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008735 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01008736 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +02008737
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008738The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008739
Willy Tarreauceb4ac92012-04-28 00:41:46 +02008740addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008741 Using the "addr" parameter, it becomes possible to use a different IP address
8742 to send health-checks. On some servers, it may be desirable to dedicate an IP
8743 address to specific component able to perform complex tests which are more
8744 suitable to health-checks than the application. This parameter is ignored if
8745 the "check" parameter is not set. See also the "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008746
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008747 Supported in default-server: No
8748
Simon Hormand60d6912013-11-25 10:46:36 +09008749agent-check
8750 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +01008751 health check. An agent health check is performed by making a TCP connection
8752 to the port set by the "agent-port" parameter and reading an ASCII string.
8753 The string is made of a series of words delimited by spaces, tabs or commas
8754 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +09008755
Willy Tarreau81f5d942013-12-09 20:51:51 +01008756 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +09008757 Values in this format will set the weight proportional to the initial
8758 weight of a server as configured when haproxy starts.
8759
Willy Tarreau81f5d942013-12-09 20:51:51 +01008760 - The word "ready". This will turn the server's administrative state to the
8761 READY mode, thus cancelling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +09008762
Willy Tarreau81f5d942013-12-09 20:51:51 +01008763 - The word "drain". This will turn the server's administrative state to the
8764 DRAIN mode, thus it will not accept any new connections other than those
8765 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +09008766
Willy Tarreau81f5d942013-12-09 20:51:51 +01008767 - The word "maint". This will turn the server's administrative state to the
8768 MAINT mode, thus it will not accept any new connections at all, and health
8769 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +09008770
Willy Tarreau81f5d942013-12-09 20:51:51 +01008771 - The words "down", "failed", or "stopped", optionally followed by a
8772 description string after a sharp ('#'). All of these mark the server's
8773 operating state as DOWN, but since the word itself is reported on the stats
8774 page, the difference allows an administrator to know if the situation was
8775 expected or not : the service may intentionally be stopped, may appear up
8776 but fail some validity tests, or may be seen as down (eg: missing process,
8777 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +09008778
Willy Tarreau81f5d942013-12-09 20:51:51 +01008779 - The word "up" sets back the server's operating state as UP if health checks
8780 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +09008781
Willy Tarreau81f5d942013-12-09 20:51:51 +01008782 Parameters which are not advertised by the agent are not changed. For
8783 example, an agent might be designed to monitor CPU usage and only report a
8784 relative weight and never interact with the operating status. Similarly, an
8785 agent could be designed as an end-user interface with 3 radio buttons
8786 allowing an administrator to change only the administrative state. However,
8787 it is important to consider that only the agent may revert its own actions,
8788 so if a server is set to DRAIN mode or to DOWN state using the agent, the
8789 agent must implement the other equivalent actions to bring the service into
8790 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +09008791
Simon Horman2f1f9552013-11-25 10:46:37 +09008792 Failure to connect to the agent is not considered an error as connectivity
8793 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +01008794 parameter. Warning though, it is not a good idea to stop an agent after it
8795 reports "down", since only an agent reporting "up" will be able to turn the
8796 server up again. Note that the CLI on the Unix stats socket is also able to
8797 force an agent's result in order to workaround a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +09008798
Willy Tarreau81f5d942013-12-09 20:51:51 +01008799 Requires the "agent-port" parameter to be set. See also the "agent-inter"
8800 parameter.
Simon Hormand60d6912013-11-25 10:46:36 +09008801
8802 Supported in default-server: No
8803
8804agent-inter <delay>
8805 The "agent-inter" parameter sets the interval between two agent checks
8806 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8807
8808 Just as with every other time-based parameter, it may be entered in any
8809 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
8810 parameter also serves as a timeout for agent checks "timeout check" is
8811 not set. In order to reduce "resonance" effects when multiple servers are
8812 hosted on the same hardware, the agent and health checks of all servers
8813 are started with a small time offset between them. It is also possible to
8814 add some random noise in the agent and health checks interval using the
8815 global "spread-checks" keyword. This makes sense for instance when a lot
8816 of backends use the same servers.
8817
8818 See also the "agent-check" and "agent-port" parameters.
8819
8820 Supported in default-server: Yes
8821
8822agent-port <port>
8823 The "agent-port" parameter sets the TCP port used for agent checks.
8824
8825 See also the "agent-check" and "agent-inter" parameters.
8826
8827 Supported in default-server: Yes
8828
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008829backup
8830 When "backup" is present on a server line, the server is only used in load
8831 balancing when all other non-backup servers are unavailable. Requests coming
8832 with a persistence cookie referencing the server will always be served
8833 though. By default, only the first operational backup server is used, unless
8834 the "allbackups" option is set in the backend. See also the "allbackups"
8835 option.
Willy Tarreau6a06a402007-07-15 20:15:28 +02008836
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008837 Supported in default-server: No
8838
Emeric Brunef42d922012-10-11 16:11:36 +02008839ca-file <cafile>
8840 This setting is only available when support for OpenSSL was built in. It
8841 designates a PEM file from which to load CA certificates used to verify
8842 server's certificate.
8843
8844 Supported in default-server: No
8845
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008846check
8847 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +01008848 always considered available. If "check" is set, the server is available when
8849 accepting periodic TCP connections, to ensure that it is really able to serve
8850 requests. The default address and port to send the tests to are those of the
8851 server, and the default source is the same as the one defined in the
8852 backend. It is possible to change the address using the "addr" parameter, the
8853 port using the "port" parameter, the source address using the "source"
8854 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +09008855 parameters. The request method is define in the backend using the "httpchk",
8856 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
8857 refer to those options and parameters for more information.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008858
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008859 Supported in default-server: No
8860
Willy Tarreau6c16adc2012-10-05 00:04:16 +02008861check-send-proxy
8862 This option forces emission of a PROXY protocol line with outgoing health
8863 checks, regardless of whether the server uses send-proxy or not for the
8864 normal traffic. By default, the PROXY protocol is enabled for health checks
8865 if it is already enabled for normal traffic and if no "port" nor "addr"
8866 directive is present. However, if such a directive is present, the
8867 "check-send-proxy" option needs to be used to force the use of the
8868 protocol. See also the "send-proxy" option for more information.
8869
8870 Supported in default-server: No
8871
Willy Tarreau763a95b2012-10-04 23:15:39 +02008872check-ssl
8873 This option forces encryption of all health checks over SSL, regardless of
8874 whether the server uses SSL or not for the normal traffic. This is generally
8875 used when an explicit "port" or "addr" directive is specified and SSL health
8876 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008877 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +02008878 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
8879 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
8880 All SSL settings are common to health checks and traffic (eg: ciphers).
8881 See the "ssl" option for more information.
8882
8883 Supported in default-server: No
8884
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008885ciphers <ciphers>
8886 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008887 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02008888 string is defined in "man 1 ciphers". When SSL is used to communicate with
8889 servers on the local network, it is common to see a weaker set of algorithms
8890 than what is used over the internet. Doing so reduces CPU usage on both the
8891 server and haproxy while still keeping it compatible with deployed software.
8892 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
8893 is needed and just connectivity, using DES can be appropriate.
8894
Willy Tarreau763a95b2012-10-04 23:15:39 +02008895 Supported in default-server: No
8896
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008897cookie <value>
8898 The "cookie" parameter sets the cookie value assigned to the server to
8899 <value>. This value will be checked in incoming requests, and the first
8900 operational server possessing the same value will be selected. In return, in
8901 cookie insertion or rewrite modes, this value will be assigned to the cookie
8902 sent to the client. There is nothing wrong in having several servers sharing
8903 the same cookie value, and it is in fact somewhat common between normal and
8904 backup servers. See also the "cookie" keyword in backend section.
8905
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008906 Supported in default-server: No
8907
Emeric Brunef42d922012-10-11 16:11:36 +02008908crl-file <crlfile>
8909 This setting is only available when support for OpenSSL was built in. It
8910 designates a PEM file from which to load certificate revocation list used
8911 to verify server's certificate.
8912
8913 Supported in default-server: No
8914
Emeric Bruna7aa3092012-10-26 12:58:00 +02008915crt <cert>
8916 This setting is only available when support for OpenSSL was built in.
8917 It designates a PEM file from which to load both a certificate and the
8918 associated private key. This file can be built by concatenating both PEM
8919 files into one. This certificate will be sent if the server send a client
8920 certificate request.
8921
8922 Supported in default-server: No
8923
Willy Tarreau96839092010-03-29 10:02:24 +02008924disabled
8925 The "disabled" keyword starts the server in the "disabled" state. That means
8926 that it is marked down in maintenance mode, and no connection other than the
8927 ones allowed by persist mode will reach it. It is very well suited to setup
8928 new servers, because normal traffic will never reach them, while it is still
8929 possible to test the service by making use of the force-persist mechanism.
8930
8931 Supported in default-server: No
8932
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008933error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +01008934 If health observing is enabled, the "error-limit" parameter specifies the
8935 number of consecutive errors that triggers event selected by the "on-error"
8936 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008937
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008938 Supported in default-server: Yes
8939
8940 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01008941
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008942fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008943 The "fall" parameter states that a server will be considered as dead after
8944 <count> consecutive unsuccessful health checks. This value defaults to 3 if
8945 unspecified. See also the "check", "inter" and "rise" parameters.
8946
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008947 Supported in default-server: Yes
8948
Emeric Brun8694b9a2012-10-05 14:39:07 +02008949force-sslv3
8950 This option enforces use of SSLv3 only when SSL is used to communicate with
8951 the server. SSLv3 is generally less expensive than the TLS counterparts for
8952 high connection rates. See also "no-tlsv*", "no-sslv3".
8953
8954 Supported in default-server: No
8955
8956force-tlsv10
8957 This option enforces use of TLSv1.0 only when SSL is used to communicate with
8958 the server. See also "no-tlsv*", "no-sslv3".
8959
8960 Supported in default-server: No
8961
8962force-tlsv11
8963 This option enforces use of TLSv1.1 only when SSL is used to communicate with
8964 the server. See also "no-tlsv*", "no-sslv3".
8965
8966 Supported in default-server: No
8967
8968force-tlsv12
8969 This option enforces use of TLSv1.2 only when SSL is used to communicate with
8970 the server. See also "no-tlsv*", "no-sslv3".
8971
8972 Supported in default-server: No
8973
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008974id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02008975 Set a persistent ID for the server. This ID must be positive and unique for
8976 the proxy. An unused ID will automatically be assigned if unset. The first
8977 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008978
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01008979 Supported in default-server: No
8980
8981inter <delay>
8982fastinter <delay>
8983downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008984 The "inter" parameter sets the interval between two consecutive health checks
8985 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
8986 It is also possible to use "fastinter" and "downinter" to optimize delays
8987 between checks depending on the server state :
8988
8989 Server state | Interval used
8990 ---------------------------------+-----------------------------------------
8991 UP 100% (non-transitional) | "inter"
8992 ---------------------------------+-----------------------------------------
8993 Transitionally UP (going down), |
8994 Transitionally DOWN (going up), | "fastinter" if set, "inter" otherwise.
8995 or yet unchecked. |
8996 ---------------------------------+-----------------------------------------
8997 DOWN 100% (non-transitional) | "downinter" if set, "inter" otherwise.
8998 ---------------------------------+-----------------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +01008999
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009000 Just as with every other time-based parameter, they can be entered in any
9001 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
9002 serves as a timeout for health checks sent to servers if "timeout check" is
9003 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +09009004 hosted on the same hardware, the agent and health checks of all servers
9005 are started with a small time offset between them. It is also possible to
9006 add some random noise in the agent and health checks interval using the
9007 global "spread-checks" keyword. This makes sense for instance when a lot
9008 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009009
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009010 Supported in default-server: Yes
9011
9012maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009013 The "maxconn" parameter specifies the maximal number of concurrent
9014 connections that will be sent to this server. If the number of incoming
9015 concurrent requests goes higher than this value, they will be queued, waiting
9016 for a connection to be released. This parameter is very important as it can
9017 save fragile servers from going down under extreme loads. If a "minconn"
9018 parameter is specified, the limit becomes dynamic. The default value is "0"
9019 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
9020 the backend's "fullconn" keyword.
9021
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009022 Supported in default-server: Yes
9023
9024maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009025 The "maxqueue" parameter specifies the maximal number of connections which
9026 will wait in the queue for this server. If this limit is reached, next
9027 requests will be redispatched to other servers instead of indefinitely
9028 waiting to be served. This will break persistence but may allow people to
9029 quickly re-log in when the server they try to connect to is dying. The
9030 default value is "0" which means the queue is unlimited. See also the
9031 "maxconn" and "minconn" parameters.
9032
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009033 Supported in default-server: Yes
9034
9035minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009036 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
9037 limit following the backend's load. The server will always accept at least
9038 <minconn> connections, never more than <maxconn>, and the limit will be on
9039 the ramp between both values when the backend has less than <fullconn>
9040 concurrent connections. This makes it possible to limit the load on the
9041 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009042 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009043 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009044
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009045 Supported in default-server: Yes
9046
Emeric Brun9b3009b2012-10-05 11:55:06 +02009047no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009048 This option disables support for SSLv3 when SSL is used to communicate with
9049 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emeric Brun8694b9a2012-10-05 14:39:07 +02009050 using any configuration option. See also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009051
Willy Tarreau763a95b2012-10-04 23:15:39 +02009052 Supported in default-server: No
9053
Emeric Brunf9c5c472012-10-11 15:28:34 +02009054no-tls-tickets
9055 This setting is only available when support for OpenSSL was built in. It
9056 disables the stateless session resumption (RFC 5077 TLS Ticket
9057 extension) and force to use stateful session resumption. Stateless
9058 session resumption is more expensive in CPU usage for servers.
9059
9060 Supported in default-server: No
9061
Emeric Brun9b3009b2012-10-05 11:55:06 +02009062no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +02009063 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009064 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9065 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun8694b9a2012-10-05 14:39:07 +02009066 often makes sense to disable it when communicating with local servers. See
9067 also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009068
Willy Tarreau763a95b2012-10-04 23:15:39 +02009069 Supported in default-server: No
9070
Emeric Brun9b3009b2012-10-05 11:55:06 +02009071no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +02009072 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +02009073 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9074 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun8694b9a2012-10-05 14:39:07 +02009075 often makes sense to disable it when communicating with local servers. See
9076 also "force-sslv3", "force-tlsv*".
Emeric Brunf5da4932012-09-28 19:42:54 +02009077
Willy Tarreau763a95b2012-10-04 23:15:39 +02009078 Supported in default-server: No
9079
Emeric Brun9b3009b2012-10-05 11:55:06 +02009080no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +02009081 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009082 the server. Note that SSLv2 is disabled in the code and cannot be enabled
9083 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun8694b9a2012-10-05 14:39:07 +02009084 often makes sense to disable it when communicating with local servers. See
9085 also "force-sslv3", "force-tlsv*".
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009086
Willy Tarreau763a95b2012-10-04 23:15:39 +02009087 Supported in default-server: No
9088
Simon Hormanfa461682011-06-25 09:39:49 +09009089non-stick
9090 Never add connections allocated to this sever to a stick-table.
9091 This may be used in conjunction with backup to ensure that
9092 stick-table persistence is disabled for backup servers.
9093
Willy Tarreau763a95b2012-10-04 23:15:39 +02009094 Supported in default-server: No
9095
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009096observe <mode>
9097 This option enables health adjusting based on observing communication with
9098 the server. By default this functionality is disabled and enabling it also
9099 requires to enable health checks. There are two supported modes: "layer4" and
9100 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
9101 significant. In layer7, which is only allowed for http proxies, responses
9102 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +01009103 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009104
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009105 Supported in default-server: No
9106
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009107 See also the "check", "on-error" and "error-limit".
9108
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009109on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009110 Select what should happen when enough consecutive errors are detected.
9111 Currently, four modes are available:
9112 - fastinter: force fastinter
9113 - fail-check: simulate a failed check, also forces fastinter (default)
9114 - sudden-death: simulate a pre-fatal failed health check, one more failed
9115 check will mark a server down, forces fastinter
9116 - mark-down: mark the server immediately down and force fastinter
9117
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009118 Supported in default-server: Yes
9119
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +01009120 See also the "check", "observe" and "error-limit".
9121
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009122on-marked-down <action>
9123 Modify what occurs when a server is marked down.
9124 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009125 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
9126 all connections to the server are immediately terminated when the server
9127 goes down. It might be used if the health check detects more complex cases
9128 than a simple connection status, and long timeouts would cause the service
9129 to remain unresponsive for too long a time. For instance, a health check
9130 might detect that a database is stuck and that there's no chance to reuse
9131 existing connections anymore. Connections killed this way are logged with
9132 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +09009133
9134 Actions are disabled by default
9135
9136 Supported in default-server: Yes
9137
Justin Karnegeseb2c24a2012-05-24 15:28:52 -07009138on-marked-up <action>
9139 Modify what occurs when a server is marked up.
9140 Currently one action is available:
9141 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
9142 done only if the server is not in backup state and if it is not disabled
9143 (it must have an effective weight > 0). This can be used sometimes to force
9144 an active server to take all the traffic back after recovery when dealing
9145 with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble
9146 than it tries to solve (eg: incomplete transactions), so use this feature
9147 with extreme care. Sessions killed because a server comes up are logged
9148 with an 'U' termination code (for "Up").
9149
9150 Actions are disabled by default
9151
9152 Supported in default-server: Yes
9153
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009154port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009155 Using the "port" parameter, it becomes possible to use a different port to
9156 send health-checks. On some servers, it may be desirable to dedicate a port
9157 to a specific component able to perform complex tests which are more suitable
9158 to health-checks than the application. It is common to run a simple script in
9159 inetd for instance. This parameter is ignored if the "check" parameter is not
9160 set. See also the "addr" parameter.
9161
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009162 Supported in default-server: Yes
9163
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009164redir <prefix>
9165 The "redir" parameter enables the redirection mode for all GET and HEAD
9166 requests addressing this server. This means that instead of having HAProxy
9167 forward the request to the server, it will send an "HTTP 302" response with
9168 the "Location" header composed of this prefix immediately followed by the
9169 requested URI beginning at the leading '/' of the path component. That means
9170 that no trailing slash should be used after <prefix>. All invalid requests
9171 will be rejected, and all non-GET or HEAD requests will be normally served by
9172 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009173 mangling nor cookie insertion is possible in the response. However, cookies in
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009174 requests are still analysed, making this solution completely usable to direct
9175 users to a remote location in case of local disaster. Main use consists in
9176 increasing bandwidth for static servers by having the clients directly
9177 connect to them. Note: never use a relative location here, it would cause a
9178 loop between the client and HAProxy!
9179
9180 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
9181
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009182 Supported in default-server: No
9183
9184rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009185 The "rise" parameter states that a server will be considered as operational
9186 after <count> consecutive successful health checks. This value defaults to 2
9187 if unspecified. See also the "check", "inter" and "fall" parameters.
9188
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009189 Supported in default-server: Yes
9190
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009191send-proxy
9192 The "send-proxy" parameter enforces use of the PROXY protocol over any
9193 connection established to this server. The PROXY protocol informs the other
9194 end about the layer 3/4 addresses of the incoming connection, so that it can
9195 know the client's address or the public address it accessed to, whatever the
9196 upper layer protocol. For connections accepted by an "accept-proxy" listener,
9197 the advertised address will be used. Only TCPv4 and TCPv6 address families
9198 are supported. Other families such as Unix sockets, will report an UNKNOWN
9199 family. Servers using this option can fully be chained to another instance of
9200 haproxy listening with an "accept-proxy" setting. This setting must not be
Willy Tarreau6c16adc2012-10-05 00:04:16 +02009201 used if the server isn't aware of the protocol. When health checks are sent
9202 to the server, the PROXY protocol is automatically used when this option is
9203 set, unless there is an explicit "port" or "addr" directive, in which case an
9204 explicit "check-send-proxy" directive would also be needed to use the PROXY
9205 protocol. See also the "accept-proxy" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +01009206
9207 Supported in default-server: No
9208
David Safb76832014-05-08 23:42:08 -04009209send-proxy-v2
9210 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
9211 over any connection established to this server. The PROXY protocol informs
9212 the other end about the layer 3/4 addresses of the incoming connection, so
9213 that it can know the client's address or the public address it accessed to,
9214 whatever the upper layer protocol. This setting must not be used if the
9215 server isn't aware of this version of the protocol. See also the "send-proxy"
9216 option of the "bind" keyword.
9217
9218 Supported in default-server: No
9219
9220send-proxy-v2-ssl
9221 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9222 2 over any connection established to this server. The PROXY protocol informs
9223 the other end about the layer 3/4 addresses of the incoming connection, so
9224 that it can know the client's address or the public address it accessed to,
9225 whatever the upper layer protocol. In addition, the SSL information extension
9226 of the PROXY protocol is added to the PROXY protocol header. This setting
9227 must not be used if the server isn't aware of this version of the protocol.
9228 See also the "send-proxy-v2" option of the "bind" keyword.
9229
9230 Supported in default-server: No
9231
9232send-proxy-v2-ssl-cn
9233 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
9234 2 over any connection established to this server. The PROXY protocol informs
9235 the other end about the layer 3/4 addresses of the incoming connection, so
9236 that it can know the client's address or the public address it accessed to,
9237 whatever the upper layer protocol. In addition, the SSL information extension
9238 of the PROXY protocol, along along with the Common Name from the subject of
9239 the client certificate (if any), is added to the PROXY protocol header. This
9240 setting must not be used if the server isn't aware of this version of the
9241 protocol. See also the "send-proxy-v2" option of the "bind" keyword.
9242
9243 Supported in default-server: No
9244
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009245slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009246 The "slowstart" parameter for a server accepts a value in milliseconds which
9247 indicates after how long a server which has just come back up will run at
9248 full speed. Just as with every other time-based parameter, it can be entered
9249 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
9250 linearly from 0 to 100% during this time. The limitation applies to two
9251 parameters :
9252
9253 - maxconn: the number of connections accepted by the server will grow from 1
9254 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
9255
9256 - weight: when the backend uses a dynamic weighted algorithm, the weight
9257 grows linearly from 1 to 100%. In this case, the weight is updated at every
9258 health-check. For this reason, it is important that the "inter" parameter
9259 is smaller than the "slowstart", in order to maximize the number of steps.
9260
9261 The slowstart never applies when haproxy starts, otherwise it would cause
9262 trouble to running servers. It only applies when a server has been previously
9263 seen as failed.
9264
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009265 Supported in default-server: Yes
9266
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009267source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009268source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009269source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009270 The "source" parameter sets the source address which will be used when
9271 connecting to the server. It follows the exact same parameters and principle
9272 as the backend "source" keyword, except that it only applies to the server
9273 referencing it. Please consult the "source" keyword for details.
9274
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009275 Additionally, the "source" statement on a server line allows one to specify a
9276 source port range by indicating the lower and higher bounds delimited by a
9277 dash ('-'). Some operating systems might require a valid IP address when a
9278 source port range is specified. It is permitted to have the same IP/range for
9279 several servers. Doing so makes it possible to bypass the maximum of 64k
9280 total concurrent connections. The limit will then reach 64k connections per
9281 server.
9282
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009283 Supported in default-server: No
9284
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009285ssl
Willy Tarreau44f65392013-06-25 07:56:20 +02009286 This option enables SSL ciphering on outgoing connections to the server. It
9287 is critical to verify server certificates using "verify" when using SSL to
9288 connect to servers, otherwise the communication is prone to trivial man in
9289 the-middle attacks rendering SSL useless. When this option is used, health
9290 checks are automatically sent in SSL too unless there is a "port" or an
9291 "addr" directive indicating the check should be sent to a different location.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009292 See the "check-ssl" option to force SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +02009293
9294 Supported in default-server: No
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02009295
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009296track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +02009297 This option enables ability to set the current state of the server by tracking
9298 another one. It is possible to track a server which itself tracks another
9299 server, provided that at the end of the chain, a server has health checks
9300 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009301 used, it has to be enabled on both proxies.
9302
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009303 Supported in default-server: No
9304
Emeric Brunef42d922012-10-11 16:11:36 +02009305verify [none|required]
9306 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +01009307 to 'none', server certificate is not verified. In the other case, The
9308 certificate provided by the server is verified using CAs from 'ca-file'
9309 and optional CRLs from 'crl-file'. If 'ssl_server_verify' is not specified
9310 in global section, this is the default. On verify failure the handshake
Willy Tarreau44f65392013-06-25 07:56:20 +02009311 is aborted. It is critically important to verify server certificates when
9312 using SSL to connect to servers, otherwise the communication is prone to
9313 trivial man-in-the-middle attacks rendering SSL totally useless.
Emeric Brunef42d922012-10-11 16:11:36 +02009314
9315 Supported in default-server: No
9316
Evan Broderbe554312013-06-27 00:05:25 -07009317verifyhost <hostname>
9318 This setting is only available when support for OpenSSL was built in, and
9319 only takes effect if 'verify required' is also specified. When set, the
9320 hostnames in the subject and subjectAlternateNames of the certificate
9321 provided by the server are checked. If none of the hostnames in the
9322 certificate match the specified hostname, the handshake is aborted. The
9323 hostnames in the server-provided certificate may include wildcards.
9324
9325 Supported in default-server: No
9326
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009327weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009328 The "weight" parameter is used to adjust the server's weight relative to
9329 other servers. All servers will receive a load proportional to their weight
9330 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +02009331 load. The default weight is 1, and the maximal value is 256. A value of 0
9332 means the server will not participate in load-balancing but will still accept
9333 persistent connections. If this parameter is used to distribute the load
9334 according to server's capacity, it is recommended to start with values which
9335 can both grow and shrink, for instance between 10 and 100 to leave enough
9336 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009337
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +01009338 Supported in default-server: Yes
9339
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009340
93416. HTTP header manipulation
9342---------------------------
9343
9344In HTTP mode, it is possible to rewrite, add or delete some of the request and
9345response headers based on regular expressions. It is also possible to block a
9346request or a response if a particular header matches a regular expression,
9347which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +01009348against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009349
Willy Tarreau70dffda2014-01-30 03:07:23 +01009350If HAProxy encounters an "Informational Response" (status code 1xx), it is able
9351to process all rsp* rules which can allow, deny, rewrite or delete a header,
9352but it will refuse to add a header to any such messages as this is not
9353HTTP-compliant. The reason for still processing headers in such responses is to
9354stop and/or fix any possible information leak which may happen, for instance
9355because another downstream equipment would unconditionally add a header, or if
9356a server name appears there. When such messages are seen, normal processing
9357still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +02009358
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009359This section covers common usage of the following keywords, described in detail
9360in section 4.2 :
9361
9362 - reqadd <string>
9363 - reqallow <search>
9364 - reqiallow <search>
9365 - reqdel <search>
9366 - reqidel <search>
9367 - reqdeny <search>
9368 - reqideny <search>
9369 - reqpass <search>
9370 - reqipass <search>
9371 - reqrep <search> <replace>
9372 - reqirep <search> <replace>
9373 - reqtarpit <search>
9374 - reqitarpit <search>
9375 - rspadd <string>
9376 - rspdel <search>
9377 - rspidel <search>
9378 - rspdeny <search>
9379 - rspideny <search>
9380 - rsprep <search> <replace>
9381 - rspirep <search> <replace>
9382
9383With all these keywords, the same conventions are used. The <search> parameter
9384is a POSIX extended regular expression (regex) which supports grouping through
9385parenthesis (without the backslash). Spaces and other delimiters must be
9386prefixed with a backslash ('\') to avoid confusion with a field delimiter.
9387Other characters may be prefixed with a backslash to change their meaning :
9388
9389 \t for a tab
9390 \r for a carriage return (CR)
9391 \n for a new line (LF)
9392 \ to mark a space and differentiate it from a delimiter
9393 \# to mark a sharp and differentiate it from a comment
9394 \\ to use a backslash in a regex
9395 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
9396 \xXX to write the ASCII hex code XX as in the C language
9397
9398The <replace> parameter contains the string to be used to replace the largest
9399portion of text matching the regex. It can make use of the special characters
9400above, and can reference a substring which is delimited by parenthesis in the
9401regex, by writing a backslash ('\') immediately followed by one digit from 0 to
94029 indicating the group position (0 designating the entire line). This practice
9403is very common to users of the "sed" program.
9404
9405The <string> parameter represents the string which will systematically be added
9406after the last header line. It can also use special character sequences above.
9407
9408Notes related to these keywords :
9409---------------------------------
9410 - these keywords are not always convenient to allow/deny based on header
9411 contents. It is strongly recommended to use ACLs with the "block" keyword
9412 instead, resulting in far more flexible and manageable rules.
9413
9414 - lines are always considered as a whole. It is not possible to reference
9415 a header name only or a value only. This is important because of the way
9416 headers are written (notably the number of spaces after the colon).
9417
9418 - the first line is always considered as a header, which makes it possible to
9419 rewrite or filter HTTP requests URIs or response codes, but in turn makes
9420 it harder to distinguish between headers and request line. The regex prefix
9421 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
9422 ^[^ \t:]*: matches any header name followed by a colon.
9423
9424 - for performances reasons, the number of characters added to a request or to
9425 a response is limited at build time to values between 1 and 4 kB. This
9426 should normally be far more than enough for most usages. If it is too short
9427 on occasional usages, it is possible to gain some space by removing some
9428 useless headers before adding new ones.
9429
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009430 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009431 without the 'i' letter except that they ignore case when matching patterns.
9432
9433 - when a request passes through a frontend then a backend, all req* rules
9434 from the frontend will be evaluated, then all req* rules from the backend
9435 will be evaluated. The reverse path is applied to responses.
9436
9437 - req* statements are applied after "block" statements, so that "block" is
9438 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +01009439 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009440
9441
Willy Tarreau74ca5042013-06-11 23:12:07 +020094427. Using ACLs and fetching samples
9443----------------------------------
9444
9445Haproxy is capable of extracting data from request or response streams, from
9446client or server information, from tables, environmental information etc...
9447The action of extracting such data is called fetching a sample. Once retrieved,
9448these samples may be used for various purposes such as a key to a stick-table,
9449but most common usages consist in matching them against predefined constant
9450data called patterns.
9451
9452
94537.1. ACL basics
9454---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009455
9456The use of Access Control Lists (ACL) provides a flexible solution to perform
9457content switching and generally to take decisions based on content extracted
9458from the request, the response or any environmental status. The principle is
9459simple :
9460
Willy Tarreau74ca5042013-06-11 23:12:07 +02009461 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009462 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +02009463 - apply one or multiple pattern matching methods on this sample
9464 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009465
Willy Tarreau74ca5042013-06-11 23:12:07 +02009466The actions generally consist in blocking a request, selecting a backend, or
9467adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009468
9469In order to define a test, the "acl" keyword is used. The syntax is :
9470
Willy Tarreau74ca5042013-06-11 23:12:07 +02009471 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009472
9473This creates a new ACL <aclname> or completes an existing one with new tests.
9474Those tests apply to the portion of request/response specified in <criterion>
9475and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009476an operator which may be specified before the set of values. Optionally some
9477conversion operators may be applied to the sample, and they will be specified
9478as a comma-delimited list of keywords just after the first keyword. The values
9479are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009480
9481ACL names must be formed from upper and lower case letters, digits, '-' (dash),
9482'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
9483which means that "my_acl" and "My_Acl" are two different ACLs.
9484
9485There is no enforced limit to the number of ACLs. The unused ones do not affect
9486performance, they just consume a small amount of memory.
9487
Willy Tarreau74ca5042013-06-11 23:12:07 +02009488The criterion generally is the name of a sample fetch method, or one of its ACL
9489specific declinations. The default test method is implied by the output type of
9490this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009491methods of a same sample fetch method. The sample fetch methods are the only
9492ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009493
9494Sample fetch methods return data which can be of the following types :
9495 - boolean
9496 - integer (signed or unsigned)
9497 - IPv4 or IPv6 address
9498 - string
9499 - data block
9500
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009501Converters transform any of these data into any of these. For example, some
9502converters might convert a string to a lower-case string while other ones
9503would turn a string to an IPv4 address, or apply a netmask to an IP address.
9504The resulting sample is of the type of the last converter applied to the list,
9505which defaults to the type of the sample fetch method.
9506
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009507Each sample or converter returns data of a specific type, specified with its
9508keyword in this documentation. When an ACL is declared using a standard sample
9509fetch method, certain types automatically involved a default matching method
9510which are summarized in the table below :
9511
9512 +---------------------+-----------------+
9513 | Sample or converter | Default |
9514 | output type | matching method |
9515 +---------------------+-----------------+
9516 | boolean | bool |
9517 +---------------------+-----------------+
9518 | integer | int |
9519 +---------------------+-----------------+
9520 | ip | ip |
9521 +---------------------+-----------------+
9522 | string | str |
9523 +---------------------+-----------------+
9524 | binary | none, use "-m" |
9525 +---------------------+-----------------+
9526
9527Note that in order to match a binary samples, it is mandatory to specify a
9528matching method, see below.
9529
Willy Tarreau74ca5042013-06-11 23:12:07 +02009530The ACL engine can match these types against patterns of the following types :
9531 - boolean
9532 - integer or integer range
9533 - IP address / network
9534 - string (exact, substring, suffix, prefix, subdir, domain)
9535 - regular expression
9536 - hex block
9537
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009538The following ACL flags are currently supported :
9539
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009540 -i : ignore case during matching of all subsequent patterns.
9541 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009542 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009543 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009544 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009545 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +02009546 -- : force end of flags. Useful when a string looks like one of the flags.
9547
Willy Tarreau74ca5042013-06-11 23:12:07 +02009548The "-f" flag is followed by the name of a file from which all lines will be
9549read as individual values. It is even possible to pass multiple "-f" arguments
9550if the patterns are to be loaded from multiple files. Empty lines as well as
9551lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
9552will be stripped. If it is absolutely necessary to insert a valid pattern
9553beginning with a sharp, just prefix it with a space so that it is not taken for
9554a comment. Depending on the data type and match method, haproxy may load the
9555lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
9556exact string matching. In this case, duplicates will automatically be removed.
9557
Thierry FOURNIER9860c412014-01-29 14:23:29 +01009558The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
9559parsed as two column file. The first column contains the patterns used by the
9560ACL, and the second column contain the samples. The sample can be used later by
9561a map. This can be useful in some rare cases where an ACL would just be used to
9562check for the existence of a pattern in a map before a mapping is applied.
9563
Thierry FOURNIER3534d882014-01-20 17:01:44 +01009564The "-u" flag forces the unique id of the ACL. This unique id is used with the
9565socket interface to identify ACL and dynamically change its values. Note that a
9566file is always identified by its name even if an id is set.
9567
Willy Tarreau74ca5042013-06-11 23:12:07 +02009568Also, note that the "-i" flag applies to subsequent entries and not to entries
9569loaded from files preceding it. For instance :
9570
9571 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
9572
9573In this example, each line of "exact-ua.lst" will be exactly matched against
9574the "user-agent" header of the request. Then each line of "generic-ua" will be
9575case-insensitively matched. Then the word "test" will be insensitively matched
9576as well.
9577
9578The "-m" flag is used to select a specific pattern matching method on the input
9579sample. All ACL-specific criteria imply a pattern matching method and generally
9580do not need this flag. However, this flag is useful with generic sample fetch
9581methods to describe how they're going to be matched against the patterns. This
9582is required for sample fetches which return data type for which there is no
9583obvious matching method (eg: string or binary). When "-m" is specified and
9584followed by a pattern matching method name, this method is used instead of the
9585default one for the criterion. This makes it possible to match contents in ways
9586that were not initially planned, or with sample fetch methods which return a
9587string. The matching method also affects the way the patterns are parsed.
9588
Thierry FOURNIERb7729c92014-02-11 16:24:41 +01009589The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
9590By default, if the parser cannot parse ip address it considers that the parsed
9591string is maybe a domain name and try dns resolution. The flag "-n" disable this
9592resolution. It is useful for detecting malformed ip lists. Note that if the DNS
9593server is not reachable, the haproxy configuration parsing may last many minutes
9594waiting fir the timeout. During this time no error messages are displayed. The
9595flag "-n" disable this behavior. Note also that during the runtime, this
9596function is disabled for the dynamic acl modifications.
9597
Willy Tarreau74ca5042013-06-11 23:12:07 +02009598There are some restrictions however. Not all methods can be used with all
9599sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
9600be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +02009601
9602 - "found" : only check if the requested sample could be found in the stream,
9603 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +02009604 to pass any pattern to avoid confusion. This matching method is
9605 particularly useful to detect presence of certain contents such
9606 as headers, cookies, etc... even if they are empty and without
9607 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009608
9609 - "bool" : check the value as a boolean. It can only be applied to fetches
9610 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +02009611 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009612
9613 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +02009614 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009615
9616 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +02009617 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009618
9619 - "bin" : match the contents against an hexadecimal string representing a
9620 binary sequence. This may be used with binary or string samples.
9621
9622 - "len" : match the sample's length as an integer. This may be used with
9623 binary or string samples.
9624
Willy Tarreau74ca5042013-06-11 23:12:07 +02009625 - "str" : exact match : match the contents against a string. This may be
9626 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009627
Willy Tarreau74ca5042013-06-11 23:12:07 +02009628 - "sub" : substring match : check that the contents contain at least one of
9629 the provided string patterns. This may be used with binary or
9630 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009631
Willy Tarreau74ca5042013-06-11 23:12:07 +02009632 - "reg" : regex match : match the contents against a list of regular
9633 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009634
Willy Tarreau74ca5042013-06-11 23:12:07 +02009635 - "beg" : prefix match : check that the contents begin like the provided
9636 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009637
Willy Tarreau74ca5042013-06-11 23:12:07 +02009638 - "end" : suffix match : check that the contents end like the provided
9639 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009640
Willy Tarreau74ca5042013-06-11 23:12:07 +02009641 - "dir" : subdir match : check that a slash-delimited portion of the
9642 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009643 This may be used with binary or string samples.
9644
Willy Tarreau74ca5042013-06-11 23:12:07 +02009645 - "dom" : domain match : check that a dot-delimited portion of the contents
9646 exactly match one of the provided string patterns. This may be
9647 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +02009648
9649For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
9650request, it is possible to do :
9651
9652 acl jsess_present cook(JSESSIONID) -m found
9653
9654In order to apply a regular expression on the 500 first bytes of data in the
9655buffer, one would use the following acl :
9656
9657 acl script_tag payload(0,500) -m reg -i <script>
9658
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009659On systems where the regex library is much slower when using "-i", it is
9660possible to convert the sample to lowercase before matching, like this :
9661
9662 acl script_tag payload(0,500),lower -m reg <script>
9663
Willy Tarreau74ca5042013-06-11 23:12:07 +02009664All ACL-specific criteria imply a default matching method. Most often, these
9665criteria are composed by concatenating the name of the original sample fetch
9666method and the matching method. For example, "hdr_beg" applies the "beg" match
9667to samples retrieved using the "hdr" fetch method. Since all ACL-specific
9668criteria rely on a sample fetch method, it is always possible instead to use
9669the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009670
Willy Tarreau74ca5042013-06-11 23:12:07 +02009671If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009672the matching method is simply applied to the underlying sample fetch method.
9673For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009674
Willy Tarreau74ca5042013-06-11 23:12:07 +02009675 acl short_form hdr_beg(host) www.
9676 acl alternate1 hdr_beg(host) -m beg www.
9677 acl alternate2 hdr_dom(host) -m beg www.
9678 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009679
Willy Tarreau2b5285d2010-05-09 23:45:24 +02009680
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009681The table below summarizes the compatibility matrix between sample or converter
9682types and the pattern types to fetch against. It indicates for each compatible
9683combination the name of the matching method to be used, surrounded with angle
9684brackets ">" and "<" when the method is the default one and will work by
9685default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009686
Willy Tarreau74ca5042013-06-11 23:12:07 +02009687 +-------------------------------------------------+
9688 | Input sample type |
9689 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009690 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009691 +----------------------+---------+---------+---------+---------+---------+
9692 | none (presence only) | found | found | found | found | found |
9693 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009694 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009695 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009696 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009697 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009698 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009699 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009700 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009701 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +02009702 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009703 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009704 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009705 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009706 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009707 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009708 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009709 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009710 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009711 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009712 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009713 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +01009714 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +02009715 +----------------------+---------+---------+---------+---------+---------+
9716 | hex block | | | | bin | bin |
9717 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +02009718
9719
Willy Tarreau74ca5042013-06-11 23:12:07 +020097207.1.1. Matching booleans
9721------------------------
9722
9723In order to match a boolean, no value is needed and all values are ignored.
9724Boolean matching is used by default for all fetch methods of type "boolean".
9725When boolean matching is used, the fetched value is returned as-is, which means
9726that a boolean "true" will always match and a boolean "false" will never match.
9727
9728Boolean matching may also be enforced using "-m bool" on fetch methods which
9729return an integer value. Then, integer value 0 is converted to the boolean
9730"false" and all other values are converted to "true".
9731
Willy Tarreau6a06a402007-07-15 20:15:28 +02009732
Willy Tarreau74ca5042013-06-11 23:12:07 +020097337.1.2. Matching integers
9734------------------------
9735
9736Integer matching applies by default to integer fetch methods. It can also be
9737enforced on boolean fetches using "-m int". In this case, "false" is converted
9738to the integer 0, and "true" is converted to the integer 1.
9739
9740Integer matching also supports integer ranges and operators. Note that integer
9741matching only applies to positive values. A range is a value expressed with a
9742lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009743
9744For instance, "1024:65535" is a valid range to represent a range of
9745unprivileged ports, and "1024:" would also work. "0:1023" is a valid
9746representation of privileged ports, and ":1023" would also work.
9747
Willy Tarreau62644772008-07-16 18:36:06 +02009748As a special case, some ACL functions support decimal numbers which are in fact
9749two integers separated by a dot. This is used with some version checks for
9750instance. All integer properties apply to those decimal numbers, including
9751ranges and operators.
9752
Willy Tarreau6a06a402007-07-15 20:15:28 +02009753For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +01009754operators with ranges does not make much sense and is strongly discouraged.
9755Similarly, it does not make much sense to perform order comparisons with a set
9756of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009757
Willy Tarreau0ba27502007-12-24 16:55:16 +01009758Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009759
9760 eq : true if the tested value equals at least one value
9761 ge : true if the tested value is greater than or equal to at least one value
9762 gt : true if the tested value is greater than at least one value
9763 le : true if the tested value is less than or equal to at least one value
9764 lt : true if the tested value is less than at least one value
9765
Willy Tarreau0ba27502007-12-24 16:55:16 +01009766For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +02009767
9768 acl negative-length hdr_val(content-length) lt 0
9769
Willy Tarreau62644772008-07-16 18:36:06 +02009770This one matches SSL versions between 3.0 and 3.1 (inclusive) :
9771
9772 acl sslv3 req_ssl_ver 3:3.1
9773
Willy Tarreau6a06a402007-07-15 20:15:28 +02009774
Willy Tarreau74ca5042013-06-11 23:12:07 +020097757.1.3. Matching strings
9776-----------------------
9777
9778String matching applies to string or binary fetch methods, and exists in 6
9779different forms :
9780
9781 - exact match (-m str) : the extracted string must exactly match the
9782 patterns ;
9783
9784 - substring match (-m sub) : the patterns are looked up inside the
9785 extracted string, and the ACL matches if any of them is found inside ;
9786
9787 - prefix match (-m beg) : the patterns are compared with the beginning of
9788 the extracted string, and the ACL matches if any of them matches.
9789
9790 - suffix match (-m end) : the patterns are compared with the end of the
9791 extracted string, and the ACL matches if any of them matches.
9792
9793 - subdir match (-m sub) : the patterns are looked up inside the extracted
9794 string, delimited with slashes ("/"), and the ACL matches if any of them
9795 matches.
9796
9797 - domain match (-m dom) : the patterns are looked up inside the extracted
9798 string, delimited with dots ("."), and the ACL matches if any of them
9799 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009800
9801String matching applies to verbatim strings as they are passed, with the
9802exception of the backslash ("\") which makes it possible to escape some
9803characters such as the space. If the "-i" flag is passed before the first
9804string, then the matching will be performed ignoring the case. In order
9805to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +01009806before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009807
9808
Willy Tarreau74ca5042013-06-11 23:12:07 +020098097.1.4. Matching regular expressions (regexes)
9810---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009811
9812Just like with string matching, regex matching applies to verbatim strings as
9813they are passed, with the exception of the backslash ("\") which makes it
9814possible to escape some characters such as the space. If the "-i" flag is
9815passed before the first regex, then the matching will be performed ignoring
9816the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +01009817the "--" flag before the first string. Same principle applies of course to
9818match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +02009819
9820
Willy Tarreau74ca5042013-06-11 23:12:07 +020098217.1.5. Matching arbitrary data blocks
9822-------------------------------------
9823
9824It is possible to match some extracted samples against a binary block which may
9825not safely be represented as a string. For this, the patterns must be passed as
9826a series of hexadecimal digits in an even number, when the match method is set
9827to binary. Each sequence of two digits will represent a byte. The hexadecimal
9828digits may be used upper or lower case.
9829
9830Example :
9831 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
9832 acl hello payload(0,6) -m bin 48656c6c6f0a
9833
9834
98357.1.6. Matching IPv4 and IPv6 addresses
9836---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02009837
9838IPv4 addresses values can be specified either as plain addresses or with a
9839netmask appended, in which case the IPv4 address matches whenever it is
9840within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009841host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +01009842difficult to read and debug configurations. If hostnames are used, you should
9843at least ensure that they are present in /etc/hosts so that the configuration
9844does not depend on any random DNS match at the moment the configuration is
9845parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +02009846
Willy Tarreauceb4ac92012-04-28 00:41:46 +02009847IPv6 may be entered in their usual form, with or without a netmask appended.
9848Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
9849trouble with randomly resolved IP addresses, host names are never allowed in
9850IPv6 patterns.
9851
9852HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
9853following situations :
9854 - tested address is IPv4, pattern address is IPv4, the match applies
9855 in IPv4 using the supplied mask if any.
9856 - tested address is IPv6, pattern address is IPv6, the match applies
9857 in IPv6 using the supplied mask if any.
9858 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
9859 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
9860 ::IPV4 or ::ffff:IPV4, otherwise it fails.
9861 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
9862 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
9863 applied in IPv6 using the supplied IPv6 mask.
9864
Willy Tarreau74ca5042013-06-11 23:12:07 +02009865
98667.2. Using ACLs to form conditions
9867----------------------------------
9868
9869Some actions are only performed upon a valid condition. A condition is a
9870combination of ACLs with operators. 3 operators are supported :
9871
9872 - AND (implicit)
9873 - OR (explicit with the "or" keyword or the "||" operator)
9874 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +02009875
Willy Tarreau74ca5042013-06-11 23:12:07 +02009876A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +02009877
Willy Tarreau74ca5042013-06-11 23:12:07 +02009878 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +02009879
Willy Tarreau74ca5042013-06-11 23:12:07 +02009880Such conditions are generally used after an "if" or "unless" statement,
9881indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +02009882
Willy Tarreau74ca5042013-06-11 23:12:07 +02009883For instance, to block HTTP requests to the "*" URL with methods other than
9884"OPTIONS", as well as POST requests without content-length, and GET or HEAD
9885requests with a content-length greater than 0, and finally every request which
9886is not either GET/HEAD/POST/OPTIONS !
9887
9888 acl missing_cl hdr_cnt(Content-length) eq 0
9889 block if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
9890 block if METH_GET HTTP_CONTENT
9891 block unless METH_GET or METH_POST or METH_OPTIONS
9892
9893To select a different backend for requests to static contents on the "www" site
9894and to every request on the "img", "video", "download" and "ftp" hosts :
9895
9896 acl url_static path_beg /static /images /img /css
9897 acl url_static path_end .gif .png .jpg .css .js
9898 acl host_www hdr_beg(host) -i www
9899 acl host_static hdr_beg(host) -i img. video. download. ftp.
9900
9901 # now use backend "static" for all static-only hosts, and for static urls
9902 # of host "www". Use backend "www" for the rest.
9903 use_backend static if host_static or host_www url_static
9904 use_backend www if host_www
9905
9906It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
9907expressions that are built on the fly without needing to be declared. They must
9908be enclosed between braces, with a space before and after each brace (because
9909the braces must be seen as independent words). Example :
9910
9911 The following rule :
9912
9913 acl missing_cl hdr_cnt(Content-length) eq 0
9914 block if METH_POST missing_cl
9915
9916 Can also be written that way :
9917
9918 block if METH_POST { hdr_cnt(Content-length) eq 0 }
9919
9920It is generally not recommended to use this construct because it's a lot easier
9921to leave errors in the configuration when written that way. However, for very
9922simple rules matching only one source IP address for instance, it can make more
9923sense to use them than to declare ACLs with random names. Another example of
9924good use is the following :
9925
9926 With named ACLs :
9927
9928 acl site_dead nbsrv(dynamic) lt 2
9929 acl site_dead nbsrv(static) lt 2
9930 monitor fail if site_dead
9931
9932 With anonymous ACLs :
9933
9934 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
9935
9936See section 4.2 for detailed help on the "block" and "use_backend" keywords.
9937
9938
99397.3. Fetching samples
9940---------------------
9941
9942Historically, sample fetch methods were only used to retrieve data to match
9943against patterns using ACLs. With the arrival of stick-tables, a new class of
9944sample fetch methods was created, most often sharing the same syntax as their
9945ACL counterpart. These sample fetch methods are also known as "fetches". As
9946of now, ACLs and fetches have converged. All ACL fetch methods have been made
9947available as fetch methods, and ACLs may use any sample fetch method as well.
9948
9949This section details all available sample fetch methods and their output type.
9950Some sample fetch methods have deprecated aliases that are used to maintain
9951compatibility with existing configurations. They are then explicitly marked as
9952deprecated and should not be used in new setups.
9953
9954The ACL derivatives are also indicated when available, with their respective
9955matching methods. These ones all have a well defined default pattern matching
9956method, so it is never necessary (though allowed) to pass the "-m" option to
9957indicate how the sample will be matched using ACLs.
9958
9959As indicated in the sample type versus matching compatibility matrix above,
9960when using a generic sample fetch method in an ACL, the "-m" option is
9961mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
9962the same keyword exists as an ACL keyword and as a standard fetch method, the
9963ACL engine will automatically pick the ACL-only one by default.
9964
9965Some of these keywords support one or multiple mandatory arguments, and one or
9966multiple optional arguments. These arguments are strongly typed and are checked
9967when the configuration is parsed so that there is no risk of running with an
9968incorrect argument (eg: an unresolved backend name). Fetch function arguments
9969are passed between parenthesis and are delimited by commas. When an argument
9970is optional, it will be indicated below between square brackets ('[ ]'). When
9971all arguments are optional, the parenthesis may be omitted.
9972
9973Thus, the syntax of a standard sample fetch method is one of the following :
9974 - name
9975 - name(arg1)
9976 - name(arg1,arg2)
9977
Thierry FOURNIER060762e2014-04-23 13:29:15 +02009978
99797.3.1. Converters
9980-----------------
9981
Willy Tarreaue6b11e42013-11-26 19:02:32 +01009982Sample fetch methods may be combined with transformations to be applied on top
9983of the fetched sample (also called "converters"). These combinations form what
9984is called "sample expressions" and the result is a "sample". Initially this
9985was only supported by "stick on" and "stick store-request" directives but this
9986has now be extended to all places where samples may be used (acls, log-format,
9987unique-id-format, add-header, ...).
9988
9989These transformations are enumerated as a series of specific keywords after the
9990sample fetch method. These keywords may equally be appended immediately after
9991the fetch keyword's argument, delimited by a comma. These keywords can also
9992support some arguments (eg: a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009993
Willy Tarreau74ca5042013-06-11 23:12:07 +02009994The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +01009995
Emeric Brun53d1a982014-04-30 18:21:37 +02009996base64
9997 Converts a binary input sample to a base64 string. It is used to log or
9998 transfer binary content in a way that can be reliably transferred (eg:
9999 an SSL ID can be copied in a header).
10000
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020010001djb2([<avalanche>])
10002 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
10003 hash function. Optionally, it is possible to apply a full avalanche hash
10004 function to the output if the optional <avalanche> argument equals 1. This
10005 converter uses the same functions as used by the various hash-based load
10006 balancing algorithms, so it will provide exactly the same results. It is
10007 mostly intended for debugging, but can be used as a stick-table entry to
10008 collect rough statistics. It must not be used for security purposes as a
10009 32-bit hash is trivial to break. See also "sdbm", "wt6" and the "hash-type"
10010 directive.
10011
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010012hex
10013 Converts a binary input sample to an hex string containing two hex digits per
10014 input byte. It is used to log or transfer hex dumps of some binary input data
10015 in a way that can be reliably transferred (eg: an SSL ID can be copied in a
10016 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010010017
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010018http_date([<offset>])
10019 Converts an integer supposed to contain a date since epoch to a string
10020 representing this date in a format suitable for use in HTTP header fields. If
10021 an offset value is specified, then it is a number of seconds that is added to
10022 the date before the conversion is operated. This is particularly useful to
10023 emit Date header fields, Expires values in responses when combined with a
10024 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010025
Willy Tarreaud9f316a2014-07-10 14:03:38 +020010026in_table(<table>)
10027 Uses the string representation of the input sample to perform a look up in
10028 the specified table. If the key is not found in the table, a boolean false
10029 is returned. Otherwise a boolean true is returned. This can be used to verify
10030 the presence of a certain key in a table tracking some elements (eg: whether
10031 or not a source IP address or an Authorization header was already seen).
10032
Willy Tarreauffcb2e42014-07-10 16:29:08 +020010033ipmask(<mask>)
10034 Apply a mask to an IPv4 address, and use the result for lookups and storage.
10035 This can be used to make all hosts within a certain mask to share the same
10036 table entries and as such use the same server. The mask can be passed in
10037 dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24).
10038
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010039language(<value>[,<default>])
10040 Returns the value with the highest q-factor from a list as extracted from the
10041 "accept-language" header using "req.fhdr". Values with no q-factor have a
10042 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
10043 belong to the list of semi-colon delimited <values> will be considered. The
10044 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
10045 given list and a default value is provided, it is returned. Note that language
10046 names may have a variant after a dash ('-'). If this variant is present in the
10047 list, it will be matched, but if it is not, only the base language is checked.
10048 The match is case-sensitive, and the output string is always one of those
10049 provided in arguments. The ordering of arguments is meaningless, only the
10050 ordering of the values in the request counts, as the first value among
10051 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020010052
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010053 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010054
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010055 # this configuration switches to the backend matching a
10056 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020010057
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010058 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
10059 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
10060 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
10061 use_backend spanish if es
10062 use_backend french if fr
10063 use_backend english if en
10064 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020010065
Willy Tarreauffcb2e42014-07-10 16:29:08 +020010066lower
10067 Convert a string sample to lower case. This can only be placed after a string
10068 sample fetch function or after a transformation keyword returning a string
10069 type. The result is of type string.
10070
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020010071ltime(<format>[,<offset>])
10072 Converts an integer supposed to contain a date since epoch to a string
10073 representing this date in local time using a format defined by the <format>
10074 string using strftime(3). The purpose is to allow any date format to be used
10075 in logs. An optional <offset> in seconds may be applied to the input date
10076 (positive or negative). See the strftime() man page for the format supported
10077 by your operating system. See also the utime converter.
10078
10079 Example :
10080
10081 # Emit two colons, one with the local time and another with ip:port
10082 # Eg: 20140710162350 127.0.0.1:57325
10083 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
10084
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010085map(<map_file>[,<default_value>])
10086map_<match_type>(<map_file>[,<default_value>])
10087map_<match_type>_<output_type>(<map_file>[,<default_value>])
10088 Search the input value from <map_file> using the <match_type> matching method,
10089 and return the associated value converted to the type <output_type>. If the
10090 input value cannot be found in the <map_file>, the converter returns the
10091 <default_value>. If the <default_value> is not set, the converter fails and
10092 acts as if no input value could be fetched. If the <match_type> is not set, it
10093 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
10094 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
10095 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010096
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010097 It is important to avoid overlapping between the keys : IP addresses and
10098 strings are stored in trees, so the first of the finest match will be used.
10099 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010100
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010101 The following array contains the list of all map functions avalaible sorted by
10102 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010103
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010104 input type | match method | output type str | output type int | output type ip
10105 -----------+--------------+-----------------+-----------------+---------------
10106 str | str | map_str | map_str_int | map_str_ip
10107 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020010108 str | beg | map_beg | map_beg_int | map_end_ip
10109 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010110 str | sub | map_sub | map_sub_int | map_sub_ip
10111 -----------+--------------+-----------------+-----------------+---------------
10112 str | dir | map_dir | map_dir_int | map_dir_ip
10113 -----------+--------------+-----------------+-----------------+---------------
10114 str | dom | map_dom | map_dom_int | map_dom_ip
10115 -----------+--------------+-----------------+-----------------+---------------
10116 str | end | map_end | map_end_int | map_end_ip
10117 -----------+--------------+-----------------+-----------------+---------------
10118 str | reg | map_reg | map_reg_int | map_reg_ip
10119 -----------+--------------+-----------------+-----------------+---------------
10120 int | int | map_int | map_int_int | map_int_ip
10121 -----------+--------------+-----------------+-----------------+---------------
10122 ip | ip | map_ip | map_ip_int | map_ip_ip
10123 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010124
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010125 The file contains one key + value per line. Lines which start with '#' are
10126 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
10127 is then the first "word" (series of non-space/tabs characters), and the value
10128 is what follows this series of space/tab till the end of the line excluding
10129 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010130
Thierry FOURNIER060762e2014-04-23 13:29:15 +020010131 Example :
10132
10133 # this is a comment and is ignored
10134 2.22.246.0/23 United Kingdom \n
10135 <-><-----------><--><------------><---->
10136 | | | | `- trailing spaces ignored
10137 | | | `---------- value
10138 | | `-------------------- middle spaces ignored
10139 | `---------------------------- key
10140 `------------------------------------ leading spaces ignored
10141
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020010142sdbm([<avalanche>])
10143 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
10144 hash function. Optionally, it is possible to apply a full avalanche hash
10145 function to the output if the optional <avalanche> argument equals 1. This
10146 converter uses the same functions as used by the various hash-based load
10147 balancing algorithms, so it will provide exactly the same results. It is
10148 mostly intended for debugging, but can be used as a stick-table entry to
10149 collect rough statistics. It must not be used for security purposes as a
10150 32-bit hash is trivial to break. See also "djb2", "wt6" and the "hash-type"
10151 directive.
10152
Willy Tarreaud9f316a2014-07-10 14:03:38 +020010153table_bytes_in_rate(<table>)
10154 Uses the string representation of the input sample to perform a look up in
10155 the specified table. If the key is not found in the table, integer value zero
10156 is returned. Otherwise the converter returns the average client-to-server
10157 bytes rate associated with the input sample in the designated table, measured
10158 in amount of bytes over the period configured in the table. See also the
10159 sc_bytes_in_rate sample fetch keyword.
10160
10161
10162table_bytes_out_rate(<table>)
10163 Uses the string representation of the input sample to perform a look up in
10164 the specified table. If the key is not found in the table, integer value zero
10165 is returned. Otherwise the converter returns the average server-to-client
10166 bytes rate associated with the input sample in the designated table, measured
10167 in amount of bytes over the period configured in the table. See also the
10168 sc_bytes_out_rate sample fetch keyword.
10169
10170table_conn_cnt(<table>)
10171 Uses the string representation of the input sample to perform a look up in
10172 the specified table. If the key is not found in the table, integer value zero
10173 is returned. Otherwise the converter returns the cumulated amount of incoming
10174 connections associated with the input sample in the designated table. See
10175 also the sc_conn_cnt sample fetch keyword.
10176
10177table_conn_cur(<table>)
10178 Uses the string representation of the input sample to perform a look up in
10179 the specified table. If the key is not found in the table, integer value zero
10180 is returned. Otherwise the converter returns the current amount of concurrent
10181 tracked connections associated with the input sample in the designated table.
10182 See also the sc_conn_cur sample fetch keyword.
10183
10184table_conn_rate(<table>)
10185 Uses the string representation of the input sample to perform a look up in
10186 the specified table. If the key is not found in the table, integer value zero
10187 is returned. Otherwise the converter returns the average incoming connection
10188 rate associated with the input sample in the designated table. See also the
10189 sc_conn_rate sample fetch keyword.
10190
10191table_gpc0(<table>)
10192 Uses the string representation of the input sample to perform a look up in
10193 the specified table. If the key is not found in the table, integer value zero
10194 is returned. Otherwise the converter returns the current value of the first
10195 general purpose counter associated with the input sample in the designated
10196 table. See also the sc_get_gpc0 sample fetch keyword.
10197
10198table_gpc0_rate(<table>)
10199 Uses the string representation of the input sample to perform a look up in
10200 the specified table. If the key is not found in the table, integer value zero
10201 is returned. Otherwise the converter returns the frequency which the gpc0
10202 counter was incremented over the configured period in the table, associated
10203 with the input sample in the designated table. See also the sc_get_gpc0_rate
10204 sample fetch keyword.
10205
10206table_http_err_cnt(<table>)
10207 Uses the string representation of the input sample to perform a look up in
10208 the specified table. If the key is not found in the table, integer value zero
10209 is returned. Otherwise the converter returns the cumulated amount of HTTP
10210 errors associated with the input sample in the designated table. See also the
10211 sc_http_err_cnt sample fetch keyword.
10212
10213table_http_err_rate(<table>)
10214 Uses the string representation of the input sample to perform a look up in
10215 the specified table. If the key is not found in the table, integer value zero
10216 is returned. Otherwise the average rate of HTTP errors associated with the
10217 input sample in the designated table, measured in amount of errors over the
10218 period configured in the table. See also the sc_http_err_rate sample fetch
10219 keyword.
10220
10221table_http_req_cnt(<table>)
10222 Uses the string representation of the input sample to perform a look up in
10223 the specified table. If the key is not found in the table, integer value zero
10224 is returned. Otherwise the converter returns the cumulated amount of HTTP
10225 requests associated with the input sample in the designated table. See also
10226 the sc_http_req_cnt sample fetch keyword.
10227
10228table_http_req_rate(<table>)
10229 Uses the string representation of the input sample to perform a look up in
10230 the specified table. If the key is not found in the table, integer value zero
10231 is returned. Otherwise the average rate of HTTP requests associated with the
10232 input sample in the designated table, measured in amount of requests over the
10233 period configured in the table. See also the sc_http_req_rate sample fetch
10234 keyword.
10235
10236table_kbytes_in(<table>)
10237 Uses the string representation of the input sample to perform a look up in
10238 the specified table. If the key is not found in the table, integer value zero
10239 is returned. Otherwise the converter returns the cumulated amount of client-
10240 to-server data associated with the input sample in the designated table,
10241 measured in kilobytes. The test is currently performed on 32-bit integers,
10242 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
10243 keyword.
10244
10245table_kbytes_out(<table>)
10246 Uses the string representation of the input sample to perform a look up in
10247 the specified table. If the key is not found in the table, integer value zero
10248 is returned. Otherwise the converter returns the cumulated amount of server-
10249 to-client data associated with the input sample in the designated table,
10250 measured in kilobytes. The test is currently performed on 32-bit integers,
10251 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
10252 keyword.
10253
10254table_server_id(<table>)
10255 Uses the string representation of the input sample to perform a look up in
10256 the specified table. If the key is not found in the table, integer value zero
10257 is returned. Otherwise the converter returns the server ID associated with
10258 the input sample in the designated table. A server ID is associated to a
10259 sample by a "stick" rule when a connection to a server succeeds. A server ID
10260 zero means that no server is associated with this key.
10261
10262table_sess_cnt(<table>)
10263 Uses the string representation of the input sample to perform a look up in
10264 the specified table. If the key is not found in the table, integer value zero
10265 is returned. Otherwise the converter returns the cumulated amount of incoming
10266 sessions associated with the input sample in the designated table. Note that
10267 a session here refers to an incoming connection being accepted by the
10268 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
10269 keyword.
10270
10271table_sess_rate(<table>)
10272 Uses the string representation of the input sample to perform a look up in
10273 the specified table. If the key is not found in the table, integer value zero
10274 is returned. Otherwise the converter returns the average incoming session
10275 rate associated with the input sample in the designated table. Note that a
10276 session here refers to an incoming connection being accepted by the
10277 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
10278 keyword.
10279
10280table_trackers(<table>)
10281 Uses the string representation of the input sample to perform a look up in
10282 the specified table. If the key is not found in the table, integer value zero
10283 is returned. Otherwise the converter returns the current amount of concurrent
10284 connections tracking the same key as the input sample in the designated
10285 table. It differs from table_conn_cur in that it does not rely on any stored
10286 information but on the table's reference count (the "use" value which is
10287 returned by "show table" on the CLI). This may sometimes be more suited for
10288 layer7 tracking. It can be used to tell a server how many concurrent
10289 connections there are from a given address for example. See also the
10290 sc_trackers sample fetch keyword.
10291
Willy Tarreauffcb2e42014-07-10 16:29:08 +020010292upper
10293 Convert a string sample to upper case. This can only be placed after a string
10294 sample fetch function or after a transformation keyword returning a string
10295 type. The result is of type string.
10296
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020010297utime(<format>[,<offset>])
10298 Converts an integer supposed to contain a date since epoch to a string
10299 representing this date in UTC time using a format defined by the <format>
10300 string using strftime(3). The purpose is to allow any date format to be used
10301 in logs. An optional <offset> in seconds may be applied to the input date
10302 (positive or negative). See the strftime() man page for the format supported
10303 by your operating system. See also the ltime converter.
10304
10305 Example :
10306
10307 # Emit two colons, one with the UTC time and another with ip:port
10308 # Eg: 20140710162350 127.0.0.1:57325
10309 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
10310
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020010311wt6([<avalanche>])
10312 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
10313 hash function. Optionally, it is possible to apply a full avalanche hash
10314 function to the output if the optional <avalanche> argument equals 1. This
10315 converter uses the same functions as used by the various hash-based load
10316 balancing algorithms, so it will provide exactly the same results. It is
10317 mostly intended for debugging, but can be used as a stick-table entry to
10318 collect rough statistics. It must not be used for security purposes as a
10319 32-bit hash is trivial to break. See also "djb2", "sdbm", and the "hash-type"
10320 directive.
10321
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010322
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200103237.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020010324--------------------------------------------
10325
10326A first set of sample fetch methods applies to internal information which does
10327not even relate to any client information. These ones are sometimes used with
10328"monitor-fail" directives to report an internal status to external watchers.
10329The sample fetch methods described in this section are usable anywhere.
10330
10331always_false : boolean
10332 Always returns the boolean "false" value. It may be used with ACLs as a
10333 temporary replacement for another one when adjusting configurations.
10334
10335always_true : boolean
10336 Always returns the boolean "true" value. It may be used with ACLs as a
10337 temporary replacement for another one when adjusting configurations.
10338
10339avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010340 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020010341 divided by the number of active servers. The current backend is used if no
10342 backend is specified. This is very similar to "queue" except that the size of
10343 the farm is considered, in order to give a more accurate measurement of the
10344 time it may take for a new connection to be processed. The main usage is with
10345 ACL to return a sorry page to new users when it becomes certain they will get
10346 a degraded service, or to pass to the backend servers in a header so that
10347 they decide to work in degraded mode or to disable some functions to speed up
10348 the processing a bit. Note that in the event there would not be any active
10349 server anymore, twice the number of queued connections would be considered as
10350 the measured value. This is a fair estimate, as we expect one server to get
10351 back soon anyway, but we still prefer to send new traffic to another backend
10352 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
10353 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010010354
Willy Tarreau74ca5042013-06-11 23:12:07 +020010355be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020010356 Applies to the number of currently established connections on the backend,
10357 possibly including the connection being evaluated. If no backend name is
10358 specified, the current one is used. But it is also possible to check another
10359 backend. It can be used to use a specific farm when the nominal one is full.
10360 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020010361
Willy Tarreau74ca5042013-06-11 23:12:07 +020010362be_sess_rate([<backend>]) : integer
10363 Returns an integer value corresponding to the sessions creation rate on the
10364 backend, in number of new sessions per second. This is used with ACLs to
10365 switch to an alternate backend when an expensive or fragile one reaches too
10366 high a session rate, or to limit abuse of service (eg. prevent sucking of an
10367 online dictionary). It can also be useful to add this element to logs using a
10368 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010369
10370 Example :
10371 # Redirect to an error page if the dictionary is requested too often
10372 backend dynamic
10373 mode http
10374 acl being_scanned be_sess_rate gt 100
10375 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010010376
Willy Tarreau74ca5042013-06-11 23:12:07 +020010377connslots([<backend>]) : integer
10378 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010379 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020010380 connections on all servers and the maximum queue size. This is probably only
10381 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050010382
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010383 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020010384 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010385 usage; see "use_backend" keyword) can be redirected to a different backend.
10386
Willy Tarreau55165fe2009-05-10 12:02:55 +020010387 'connslots' = number of available server connection slots, + number of
10388 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010389
Willy Tarreaua36af912009-10-10 12:02:45 +020010390 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020010391 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020010392 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020010393 you want to be able to differentiate between different backends, and their
10394 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020010395 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020010396 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010397
Willy Tarreau55165fe2009-05-10 12:02:55 +020010398 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
10399 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020010400 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020010401 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080010402
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010403date([<offset>]) : integer
10404 Returns the current date as the epoch (number of seconds since 01/01/1970).
10405 If an offset value is specified, then it is a number of seconds that is added
10406 to the current date before returning the value. This is particularly useful
10407 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020010408 It is useful combined with the http_date converter.
10409
10410 Example :
10411
10412 # set an expires header to now+1 hour in every response
10413 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020010414
Willy Tarreau595ec542013-06-12 21:34:28 +020010415env(<name>) : string
10416 Returns a string containing the value of environment variable <name>. As a
10417 reminder, environment variables are per-process and are sampled when the
10418 process starts. This can be useful to pass some information to a next hop
10419 server, or with ACLs to take specific action when the process is started a
10420 certain way.
10421
10422 Examples :
10423 # Pass the Via header to next hop with the local hostname in it
10424 http-request add-header Via 1.1\ %[env(HOSTNAME)]
10425
10426 # reject cookie-less requests when the STOP environment variable is set
10427 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
10428
Willy Tarreau74ca5042013-06-11 23:12:07 +020010429fe_conn([<frontend>]) : integer
10430 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010010431 possibly including the connection being evaluated. If no frontend name is
10432 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010433 frontend. It can be used to return a sorry page before hard-blocking, or to
10434 use a specific backend to drain new requests when the farm is considered
10435 full. This is mostly used with ACLs but can also be used to pass some
10436 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
10437 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020010438
Willy Tarreau74ca5042013-06-11 23:12:07 +020010439fe_sess_rate([<frontend>]) : integer
10440 Returns an integer value corresponding to the sessions creation rate on the
10441 frontend, in number of new sessions per second. This is used with ACLs to
10442 limit the incoming session rate to an acceptable range in order to prevent
10443 abuse of service at the earliest moment, for example when combined with other
10444 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
10445 down below the limit. It can also be useful to add this element to logs using
10446 a log-format directive. See also the "rate-limit sessions" directive for use
10447 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010448
10449 Example :
10450 # This frontend limits incoming mails to 10/s with a max of 100
10451 # concurrent connections. We accept any connection below 10/s, and
10452 # force excess clients to wait for 100 ms. Since clients are limited to
10453 # 100 max, there cannot be more than 10 incoming mails per second.
10454 frontend mail
10455 bind :25
10456 mode tcp
10457 maxconn 100
10458 acl too_fast fe_sess_rate ge 10
10459 tcp-request inspect-delay 100ms
10460 tcp-request content accept if ! too_fast
10461 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010010462
Willy Tarreau74ca5042013-06-11 23:12:07 +020010463nbsrv([<backend>]) : integer
10464 Returns an integer value corresponding to the number of usable servers of
10465 either the current backend or the named backend. This is mostly used with
10466 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010010467 switch to an alternate backend when the number of servers is too low to
10468 to handle some load. It is useful to report a failure when combined with
10469 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010470
Willy Tarreau74ca5042013-06-11 23:12:07 +020010471queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010010472 Returns the total number of queued connections of the designated backend,
10473 including all the connections in server queues. If no backend name is
10474 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020010475 one. This is useful with ACLs or to pass statistics to backend servers. This
10476 can be used to take actions when queuing goes above a known level, generally
10477 indicating a surge of traffic or a massive slowdown on the servers. One
10478 possible action could be to reject new users but still accept old ones. See
10479 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
10480
Willy Tarreau84310e22014-02-14 11:59:04 +010010481rand([<range>]) : integer
10482 Returns a random integer value within a range of <range> possible values,
10483 starting at zero. If the range is not specified, it defaults to 2^32, which
10484 gives numbers between 0 and 4294967295. It can be useful to pass some values
10485 needed to take some routing decisions for example, or just for debugging
10486 purposes. This random must not be used for security purposes.
10487
Willy Tarreau74ca5042013-06-11 23:12:07 +020010488srv_conn([<backend>/]<server>) : integer
10489 Returns an integer value corresponding to the number of currently established
10490 connections on the designated server, possibly including the connection being
10491 evaluated. If <backend> is omitted, then the server is looked up in the
10492 current backend. It can be used to use a specific farm when one server is
10493 full, or to inform the server about our view of the number of active
10494 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
10495 methods.
10496
10497srv_is_up([<backend>/]<server>) : boolean
10498 Returns true when the designated server is UP, and false when it is either
10499 DOWN or in maintenance mode. If <backend> is omitted, then the server is
10500 looked up in the current backend. It is mainly used to take action based on
10501 an external status reported via a health check (eg: a geographical site's
10502 availability). Another possible use which is more of a hack consists in
10503 using dummy servers as boolean variables that can be enabled or disabled from
10504 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
10505
10506srv_sess_rate([<backend>/]<server>) : integer
10507 Returns an integer corresponding to the sessions creation rate on the
10508 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010509 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020010510 used with ACLs but can make sense with logs too. This is used to switch to an
10511 alternate backend when an expensive or fragile one reaches too high a session
10512 rate, or to limit abuse of service (eg. prevent latent requests from
10513 overloading servers).
10514
10515 Example :
10516 # Redirect to a separate back
10517 acl srv1_full srv_sess_rate(be1/srv1) gt 50
10518 acl srv2_full srv_sess_rate(be1/srv2) gt 50
10519 use_backend be2 if srv1_full or srv2_full
10520
10521table_avl([<table>]) : integer
10522 Returns the total number of available entries in the current proxy's
10523 stick-table or in the designated stick-table. See also table_cnt.
10524
10525table_cnt([<table>]) : integer
10526 Returns the total number of entries currently in use in the current proxy's
10527 stick-table or in the designated stick-table. See also src_conn_cnt and
10528 table_avl for other entry counting methods.
10529
10530
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200105317.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020010532----------------------------------
10533
10534The layer 4 usually describes just the transport layer which in haproxy is
10535closest to the connection, where no content is yet made available. The fetch
10536methods described here are usable as low as the "tcp-request connection" rule
10537sets unless they require some future information. Those generally include
10538TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010539the incoming connection. For retrieving a value from a sticky counters, the
10540counter number can be explicitly set as 0, 1, or 2 using the pre-defined
10541"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010542argument when using the "sc_" prefix. An optional table may be specified with
10543the "sc*" form, in which case the currently tracked key will be looked up into
10544this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010545
10546be_id : integer
10547 Returns an integer containing the current backend's id. It can be used in
10548 frontends with responses to check which backend processed the request.
10549
10550dst : ip
10551 This is the destination IPv4 address of the connection on the client side,
10552 which is the address the client connected to. It can be useful when running
10553 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
10554 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
10555 RFC 4291.
10556
10557dst_conn : integer
10558 Returns an integer value corresponding to the number of currently established
10559 connections on the same socket including the one being evaluated. It is
10560 normally used with ACLs but can as well be used to pass the information to
10561 servers in an HTTP header or in logs. It can be used to either return a sorry
10562 page before hard-blocking, or to use a specific backend to drain new requests
10563 when the socket is considered saturated. This offers the ability to assign
10564 different limits to different listening ports or addresses. See also the
10565 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010566
Willy Tarreau74ca5042013-06-11 23:12:07 +020010567dst_port : integer
10568 Returns an integer value corresponding to the destination TCP port of the
10569 connection on the client side, which is the port the client connected to.
10570 This might be used when running in transparent mode, when assigning dynamic
10571 ports to some clients for a whole application session, to stick all users to
10572 a same server, or to pass the destination port information to a server using
10573 an HTTP header.
10574
10575fe_id : integer
10576 Returns an integer containing the current frontend's id. It can be used in
10577 backends to check from which backend it was called, or to stick all users
10578 coming via a same frontend to the same server.
10579
Cyril Bonté62ba8702014-04-22 23:52:25 +020010580sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010581sc0_bytes_in_rate([<table>]) : integer
10582sc1_bytes_in_rate([<table>]) : integer
10583sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010584 Returns the average client-to-server bytes rate from the currently tracked
10585 counters, measured in amount of bytes over the period configured in the
10586 table. See also src_bytes_in_rate.
10587
Cyril Bonté62ba8702014-04-22 23:52:25 +020010588sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010589sc0_bytes_out_rate([<table>]) : integer
10590sc1_bytes_out_rate([<table>]) : integer
10591sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010592 Returns the average server-to-client bytes rate from the currently tracked
10593 counters, measured in amount of bytes over the period configured in the
10594 table. See also src_bytes_out_rate.
10595
Cyril Bonté62ba8702014-04-22 23:52:25 +020010596sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010597sc0_clr_gpc0([<table>]) : integer
10598sc1_clr_gpc0([<table>]) : integer
10599sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020010600 Clears the first General Purpose Counter associated to the currently tracked
10601 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010010602 stored value is zero, so first invocation will always return zero. This is
10603 typically used as a second ACL in an expression in order to mark a connection
10604 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010605
10606 # block if 5 consecutive requests continue to come faster than 10 sess
10607 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010608 acl abuse sc0_http_req_rate gt 10
10609 acl kill sc0_inc_gpc0 gt 5
10610 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010611 tcp-request connection accept if !abuse save
10612 tcp-request connection reject if abuse kill
10613
Cyril Bonté62ba8702014-04-22 23:52:25 +020010614sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010615sc0_conn_cnt([<table>]) : integer
10616sc1_conn_cnt([<table>]) : integer
10617sc2_conn_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010618 Returns the cumulated number of incoming connections from currently tracked
10619 counters. See also src_conn_cnt.
10620
Cyril Bonté62ba8702014-04-22 23:52:25 +020010621sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010622sc0_conn_cur([<table>]) : integer
10623sc1_conn_cur([<table>]) : integer
10624sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010625 Returns the current amount of concurrent connections tracking the same
10626 tracked counters. This number is automatically incremented when tracking
10627 begins and decremented when tracking stops. See also src_conn_cur.
10628
Cyril Bonté62ba8702014-04-22 23:52:25 +020010629sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010630sc0_conn_rate([<table>]) : integer
10631sc1_conn_rate([<table>]) : integer
10632sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010633 Returns the average connection rate from the currently tracked counters,
10634 measured in amount of connections over the period configured in the table.
10635 See also src_conn_rate.
10636
Cyril Bonté62ba8702014-04-22 23:52:25 +020010637sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010638sc0_get_gpc0([<table>]) : integer
10639sc1_get_gpc0([<table>]) : integer
10640sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010641 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010642 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010643
Cyril Bonté62ba8702014-04-22 23:52:25 +020010644sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010645sc0_gpc0_rate([<table>]) : integer
10646sc1_gpc0_rate([<table>]) : integer
10647sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010648 Returns the average increment rate of the first General Purpose Counter
10649 associated to the currently tracked counters. It reports the frequency
10650 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010651 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10652 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10653 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020010654
Cyril Bonté62ba8702014-04-22 23:52:25 +020010655sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010656sc0_http_err_cnt([<table>]) : integer
10657sc1_http_err_cnt([<table>]) : integer
10658sc2_http_err_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010659 Returns the cumulated number of HTTP errors from the currently tracked
10660 counters. This includes the both request errors and 4xx error responses.
10661 See also src_http_err_cnt.
10662
Cyril Bonté62ba8702014-04-22 23:52:25 +020010663sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010664sc0_http_err_rate([<table>]) : integer
10665sc1_http_err_rate([<table>]) : integer
10666sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010667 Returns the average rate of HTTP errors from the currently tracked counters,
10668 measured in amount of errors over the period configured in the table. This
10669 includes the both request errors and 4xx error responses. See also
10670 src_http_err_rate.
10671
Cyril Bonté62ba8702014-04-22 23:52:25 +020010672sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010673sc0_http_req_cnt([<table>]) : integer
10674sc1_http_req_cnt([<table>]) : integer
10675sc2_http_req_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010676 Returns the cumulated number of HTTP requests from the currently tracked
10677 counters. This includes every started request, valid or not. See also
10678 src_http_req_cnt.
10679
Cyril Bonté62ba8702014-04-22 23:52:25 +020010680sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010681sc0_http_req_rate([<table>]) : integer
10682sc1_http_req_rate([<table>]) : integer
10683sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010684 Returns the average rate of HTTP requests from the currently tracked
10685 counters, measured in amount of requests over the period configured in
10686 the table. This includes every started request, valid or not. See also
10687 src_http_req_rate.
10688
Cyril Bonté62ba8702014-04-22 23:52:25 +020010689sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010690sc0_inc_gpc0([<table>]) : integer
10691sc1_inc_gpc0([<table>]) : integer
10692sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010693 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010010694 tracked counters, and returns its new value. Before the first invocation,
10695 the stored value is zero, so first invocation will increase it to 1 and will
10696 return 1. This is typically used as a second ACL in an expression in order
10697 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020010698
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010699 acl abuse sc0_http_req_rate gt 10
10700 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010701 tcp-request connection reject if abuse kill
10702
Cyril Bonté62ba8702014-04-22 23:52:25 +020010703sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010704sc0_kbytes_in([<table>]) : integer
10705sc1_kbytes_in([<table>]) : integer
10706sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020010707 Returns the total amount of client-to-server data from the currently tracked
10708 counters, measured in kilobytes. The test is currently performed on 32-bit
10709 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020010710
Cyril Bonté62ba8702014-04-22 23:52:25 +020010711sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010712sc0_kbytes_out([<table>]) : integer
10713sc1_kbytes_out([<table>]) : integer
10714sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020010715 Returns the total amount of server-to-client data from the currently tracked
10716 counters, measured in kilobytes. The test is currently performed on 32-bit
10717 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020010718
Cyril Bonté62ba8702014-04-22 23:52:25 +020010719sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010720sc0_sess_cnt([<table>]) : integer
10721sc1_sess_cnt([<table>]) : integer
10722sc2_sess_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010723 Returns the cumulated number of incoming connections that were transformed
10724 into sessions, which means that they were accepted by a "tcp-request
10725 connection" rule, from the currently tracked counters. A backend may count
10726 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010727 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020010728 with the client. See also src_sess_cnt.
10729
Cyril Bonté62ba8702014-04-22 23:52:25 +020010730sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010731sc0_sess_rate([<table>]) : integer
10732sc1_sess_rate([<table>]) : integer
10733sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020010734 Returns the average session rate from the currently tracked counters,
10735 measured in amount of sessions over the period configured in the table. A
10736 session is a connection that got past the early "tcp-request connection"
10737 rules. A backend may count more sessions than connections because each
10738 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010739 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020010740
Cyril Bonté62ba8702014-04-22 23:52:25 +020010741sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020010742sc0_tracked([<table>]) : boolean
10743sc1_tracked([<table>]) : boolean
10744sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020010745 Returns true if the designated session counter is currently being tracked by
10746 the current session. This can be useful when deciding whether or not we want
10747 to set some values in a header passed to the server.
10748
Cyril Bonté62ba8702014-04-22 23:52:25 +020010749sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020010750sc0_trackers([<table>]) : integer
10751sc1_trackers([<table>]) : integer
10752sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010010753 Returns the current amount of concurrent connections tracking the same
10754 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010755 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010010756 that it does not rely on any stored information but on the table's reference
10757 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020010758 may sometimes be more suited for layer7 tracking. It can be used to tell a
10759 server how many concurrent connections there are from a given address for
10760 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010010761
Willy Tarreau74ca5042013-06-11 23:12:07 +020010762so_id : integer
10763 Returns an integer containing the current listening socket's id. It is useful
10764 in frontends involving many "bind" lines, or to stick all users coming via a
10765 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010766
Willy Tarreau74ca5042013-06-11 23:12:07 +020010767src : ip
10768 This is the source IPv4 address of the client of the session. It is of type
10769 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
10770 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
10771 TCP-level source address which is used, and not the address of a client
10772 behind a proxy. However if the "accept-proxy" bind directive is used, it can
10773 be the address of a client behind another PROXY-protocol compatible component
10774 for all rule sets except "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010775
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010010776 Example:
10777 # add an HTTP header in requests with the originating address' country
10778 http-request set-header X-Country %[src,map_ip(geoip.lst)]
10779
Willy Tarreau74ca5042013-06-11 23:12:07 +020010780src_bytes_in_rate([<table>]) : integer
10781 Returns the average bytes rate from the incoming connection's source address
10782 in the current proxy's stick-table or in the designated stick-table, measured
10783 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010784 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010785
Willy Tarreau74ca5042013-06-11 23:12:07 +020010786src_bytes_out_rate([<table>]) : integer
10787 Returns the average bytes rate to the incoming connection's source address in
10788 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010789 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010790 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010791
Willy Tarreau74ca5042013-06-11 23:12:07 +020010792src_clr_gpc0([<table>]) : integer
10793 Clears the first General Purpose Counter associated to the incoming
10794 connection's source address in the current proxy's stick-table or in the
10795 designated stick-table, and returns its previous value. If the address is not
10796 found, an entry is created and 0 is returned. This is typically used as a
10797 second ACL in an expression in order to mark a connection when a first ACL
10798 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020010799
10800 # block if 5 consecutive requests continue to come faster than 10 sess
10801 # per second, and reset the counter as soon as the traffic slows down.
10802 acl abuse src_http_req_rate gt 10
10803 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010010804 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020010805 tcp-request connection accept if !abuse save
10806 tcp-request connection reject if abuse kill
10807
Willy Tarreau74ca5042013-06-11 23:12:07 +020010808src_conn_cnt([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010809 Returns the cumulated number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020010810 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010811 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010812 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010813
Willy Tarreau74ca5042013-06-11 23:12:07 +020010814src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010815 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010816 current incoming connection's source address in the current proxy's
10817 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010818 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010819
Willy Tarreau74ca5042013-06-11 23:12:07 +020010820src_conn_rate([<table>]) : integer
10821 Returns the average connection rate from the incoming connection's source
10822 address in the current proxy's stick-table or in the designated stick-table,
10823 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010824 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010825
Willy Tarreau74ca5042013-06-11 23:12:07 +020010826src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020010827 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020010828 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020010829 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010830 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010831
Willy Tarreau74ca5042013-06-11 23:12:07 +020010832src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010833 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020010834 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010835 stick-table or in the designated stick-table. It reports the frequency
10836 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010837 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
10838 that the "gpc0_rate" counter must be stored in the stick-table for a value to
10839 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010840
Willy Tarreau74ca5042013-06-11 23:12:07 +020010841src_http_err_cnt([<table>]) : integer
10842 Returns the cumulated number of HTTP errors from the incoming connection's
10843 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020010844 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010845 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020010846 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010847
Willy Tarreau74ca5042013-06-11 23:12:07 +020010848src_http_err_rate([<table>]) : integer
10849 Returns the average rate of HTTP errors from the incoming connection's source
10850 address in the current proxy's stick-table or in the designated stick-table,
10851 measured in amount of errors over the period configured in the table. This
10852 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010853 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010854
Willy Tarreau74ca5042013-06-11 23:12:07 +020010855src_http_req_cnt([<table>]) : integer
10856 Returns the cumulated number of HTTP requests from the incoming connection's
10857 source address in the current proxy's stick-table or in the designated stick-
10858 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010859 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010860
Willy Tarreau74ca5042013-06-11 23:12:07 +020010861src_http_req_rate([<table>]) : integer
10862 Returns the average rate of HTTP requests from the incoming connection's
10863 source address in the current proxy's stick-table or in the designated stick-
10864 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020010865 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010866 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010867
Willy Tarreau74ca5042013-06-11 23:12:07 +020010868src_inc_gpc0([<table>]) : integer
10869 Increments the first General Purpose Counter associated to the incoming
10870 connection's source address in the current proxy's stick-table or in the
10871 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010872 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020010873 This is typically used as a second ACL in an expression in order to mark a
10874 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020010875
10876 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010010877 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020010878 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020010879
Willy Tarreau74ca5042013-06-11 23:12:07 +020010880src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020010881 Returns the total amount of data received from the incoming connection's
10882 source address in the current proxy's stick-table or in the designated
10883 stick-table, measured in kilobytes. If the address is not found, zero is
10884 returned. The test is currently performed on 32-bit integers, which limits
10885 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010886
Willy Tarreau74ca5042013-06-11 23:12:07 +020010887src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020010888 Returns the total amount of data sent to the incoming connection's source
10889 address in the current proxy's stick-table or in the designated stick-table,
10890 measured in kilobytes. If the address is not found, zero is returned. The
10891 test is currently performed on 32-bit integers, which limits values to 4
10892 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010893
Willy Tarreau74ca5042013-06-11 23:12:07 +020010894src_port : integer
10895 Returns an integer value corresponding to the TCP source port of the
10896 connection on the client side, which is the port the client connected from.
10897 Usage of this function is very limited as modern protocols do not care much
10898 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010010899
Willy Tarreau74ca5042013-06-11 23:12:07 +020010900src_sess_cnt([<table>]) : integer
10901 Returns the cumulated number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020010902 connection's source IPv4 address in the current proxy's stick-table or in the
10903 designated stick-table, that were transformed into sessions, which means that
10904 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010905 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010906
Willy Tarreau74ca5042013-06-11 23:12:07 +020010907src_sess_rate([<table>]) : integer
10908 Returns the average session rate from the incoming connection's source
10909 address in the current proxy's stick-table or in the designated stick-table,
10910 measured in amount of sessions over the period configured in the table. A
10911 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020010912 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010913
Willy Tarreau74ca5042013-06-11 23:12:07 +020010914src_updt_conn_cnt([<table>]) : integer
10915 Creates or updates the entry associated to the incoming connection's source
10916 address in the current proxy's stick-table or in the designated stick-table.
10917 This table must be configured to store the "conn_cnt" data type, otherwise
10918 the match will be ignored. The current count is incremented by one, and the
10919 expiration timer refreshed. The updated count is returned, so this match
10920 can't return zero. This was used to reject service abusers based on their
10921 source address. Note: it is recommended to use the more complete "track-sc*"
10922 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010923
10924 Example :
10925 # This frontend limits incoming SSH connections to 3 per 10 second for
10926 # each source address, and rejects excess connections until a 10 second
10927 # silence is observed. At most 20 addresses are tracked.
10928 listen ssh
10929 bind :22
10930 mode tcp
10931 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020010932 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020010933 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020010934 server local 127.0.0.1:22
10935
Willy Tarreau74ca5042013-06-11 23:12:07 +020010936srv_id : integer
10937 Returns an integer containing the server's id when processing the response.
10938 While it's almost only used with ACLs, it may be used for logging or
10939 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020010940
Hervé COMMOWICK35ed8012010-12-15 14:04:51 +010010941
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200109427.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020010943----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020010944
Willy Tarreau74ca5042013-06-11 23:12:07 +020010945The layer 5 usually describes just the session layer which in haproxy is
10946closest to the session once all the connection handshakes are finished, but
10947when no content is yet made available. The fetch methods described here are
10948usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010949future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020010950
Emeric Brun645ae792014-04-30 14:21:06 +020010951ssl_bc : boolean
10952 Returns true when the back connection was made via an SSL/TLS transport
10953 layer and is locally deciphered. This means the outgoing connection was made
10954 other a server with the "ssl" option.
10955
10956ssl_bc_alg_keysize : integer
10957 Returns the symmetric cipher key size supported in bits when the outgoing
10958 connection was made over an SSL/TLS transport layer.
10959
10960ssl_bc_cipher : string
10961 Returns the name of the used cipher when the outgoing connection was made
10962 over an SSL/TLS transport layer.
10963
10964ssl_bc_protocol : string
10965 Returns the name of the used protocol when the outgoing connection was made
10966 over an SSL/TLS transport layer.
10967
Emeric Brunb73a9b02014-04-30 18:49:19 +020010968ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020010969 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020010970 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
10971 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020010972
10973ssl_bc_session_id : binary
10974 Returns the SSL ID of the back connection when the outgoing connection was
10975 made over an SSL/TLS transport layer. It is useful to log if we want to know
10976 if session was reused or not.
10977
10978ssl_bc_use_keysize : integer
10979 Returns the symmetric cipher key size used in bits when the outgoing
10980 connection was made over an SSL/TLS transport layer.
10981
Willy Tarreau74ca5042013-06-11 23:12:07 +020010982ssl_c_ca_err : integer
10983 When the incoming connection was made over an SSL/TLS transport layer,
10984 returns the ID of the first error detected during verification of the client
10985 certificate at depth > 0, or 0 if no error was encountered during this
10986 verification process. Please refer to your SSL library's documentation to
10987 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020010988
Willy Tarreau74ca5042013-06-11 23:12:07 +020010989ssl_c_ca_err_depth : integer
10990 When the incoming connection was made over an SSL/TLS transport layer,
10991 returns the depth in the CA chain of the first error detected during the
10992 verification of the client certificate. If no error is encountered, 0 is
10993 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010994
Willy Tarreau74ca5042013-06-11 23:12:07 +020010995ssl_c_err : integer
10996 When the incoming connection was made over an SSL/TLS transport layer,
10997 returns the ID of the first error detected during verification at depth 0, or
10998 0 if no error was encountered during this verification process. Please refer
10999 to your SSL library's documentation to find the exhaustive list of error
11000 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020011001
Willy Tarreau74ca5042013-06-11 23:12:07 +020011002ssl_c_i_dn([<entry>[,<occ>]]) : string
11003 When the incoming connection was made over an SSL/TLS transport layer,
11004 returns the full distinguished name of the issuer of the certificate
11005 presented by the client when no <entry> is specified, or the value of the
11006 first given entry found from the beginning of the DN. If a positive/negative
11007 occurrence number is specified as the optional second argument, it returns
11008 the value of the nth given entry value from the beginning/end of the DN.
11009 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
11010 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020011011
Willy Tarreau74ca5042013-06-11 23:12:07 +020011012ssl_c_key_alg : string
11013 Returns the name of the algorithm used to generate the key of the certificate
11014 presented by the client when the incoming connection was made over an SSL/TLS
11015 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020011016
Willy Tarreau74ca5042013-06-11 23:12:07 +020011017ssl_c_notafter : string
11018 Returns the end date presented by the client as a formatted string
11019 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
11020 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020011021
Willy Tarreau74ca5042013-06-11 23:12:07 +020011022ssl_c_notbefore : string
11023 Returns the start date presented by the client as a formatted string
11024 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
11025 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010011026
Willy Tarreau74ca5042013-06-11 23:12:07 +020011027ssl_c_s_dn([<entry>[,<occ>]]) : string
11028 When the incoming connection was made over an SSL/TLS transport layer,
11029 returns the full distinguished name of the subject of the certificate
11030 presented by the client when no <entry> is specified, or the value of the
11031 first given entry found from the beginning of the DN. If a positive/negative
11032 occurrence number is specified as the optional second argument, it returns
11033 the value of the nth given entry value from the beginning/end of the DN.
11034 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
11035 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010011036
Willy Tarreau74ca5042013-06-11 23:12:07 +020011037ssl_c_serial : binary
11038 Returns the serial of the certificate presented by the client when the
11039 incoming connection was made over an SSL/TLS transport layer. When used for
11040 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020011041
Willy Tarreau74ca5042013-06-11 23:12:07 +020011042ssl_c_sha1 : binary
11043 Returns the SHA-1 fingerprint of the certificate presented by the client when
11044 the incoming connection was made over an SSL/TLS transport layer. This can be
11045 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020011046 Note that the output is binary, so if you want to pass that signature to the
11047 server, you need to encode it in hex or base64, such as in the example below:
11048
11049 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020011050
Willy Tarreau74ca5042013-06-11 23:12:07 +020011051ssl_c_sig_alg : string
11052 Returns the name of the algorithm used to sign the certificate presented by
11053 the client when the incoming connection was made over an SSL/TLS transport
11054 layer.
Emeric Brun87855892012-10-17 17:39:35 +020011055
Willy Tarreau74ca5042013-06-11 23:12:07 +020011056ssl_c_used : boolean
11057 Returns true if current SSL session uses a client certificate even if current
11058 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020011059
Willy Tarreau74ca5042013-06-11 23:12:07 +020011060ssl_c_verify : integer
11061 Returns the verify result error ID when the incoming connection was made over
11062 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
11063 refer to your SSL library's documentation for an exhaustive list of error
11064 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020011065
Willy Tarreau74ca5042013-06-11 23:12:07 +020011066ssl_c_version : integer
11067 Returns the version of the certificate presented by the client when the
11068 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020011069
Willy Tarreau74ca5042013-06-11 23:12:07 +020011070ssl_f_i_dn([<entry>[,<occ>]]) : string
11071 When the incoming connection was made over an SSL/TLS transport layer,
11072 returns the full distinguished name of the issuer of the certificate
11073 presented by the frontend when no <entry> is specified, or the value of the
11074 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020011075 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020011076 the value of the nth given entry value from the beginning/end of the DN.
11077 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
11078 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020011079
Willy Tarreau74ca5042013-06-11 23:12:07 +020011080ssl_f_key_alg : string
11081 Returns the name of the algorithm used to generate the key of the certificate
11082 presented by the frontend when the incoming connection was made over an
11083 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020011084
Willy Tarreau74ca5042013-06-11 23:12:07 +020011085ssl_f_notafter : string
11086 Returns the end date presented by the frontend as a formatted string
11087 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
11088 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020011089
Willy Tarreau74ca5042013-06-11 23:12:07 +020011090ssl_f_notbefore : string
11091 Returns the start date presented by the frontend as a formatted string
11092 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
11093 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020011094
Willy Tarreau74ca5042013-06-11 23:12:07 +020011095ssl_f_s_dn([<entry>[,<occ>]]) : string
11096 When the incoming connection was made over an SSL/TLS transport layer,
11097 returns the full distinguished name of the subject of the certificate
11098 presented by the frontend when no <entry> is specified, or the value of the
11099 first given entry found from the beginning of the DN. If a positive/negative
11100 occurrence number is specified as the optional second argument, it returns
11101 the value of the nth given entry value from the beginning/end of the DN.
11102 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
11103 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020011104
Willy Tarreau74ca5042013-06-11 23:12:07 +020011105ssl_f_serial : binary
11106 Returns the serial of the certificate presented by the frontend when the
11107 incoming connection was made over an SSL/TLS transport layer. When used for
11108 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020011109
Emeric Brun55f4fa82014-04-30 17:11:25 +020011110ssl_f_sha1 : binary
11111 Returns the SHA-1 fingerprint of the certificate presented by the frontend
11112 when the incoming connection was made over an SSL/TLS transport layer. This
11113 can be used to know which certificate was chosen using SNI.
11114
Willy Tarreau74ca5042013-06-11 23:12:07 +020011115ssl_f_sig_alg : string
11116 Returns the name of the algorithm used to sign the certificate presented by
11117 the frontend when the incoming connection was made over an SSL/TLS transport
11118 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020011119
Willy Tarreau74ca5042013-06-11 23:12:07 +020011120ssl_f_version : integer
11121 Returns the version of the certificate presented by the frontend when the
11122 incoming connection was made over an SSL/TLS transport layer.
11123
11124ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020011125 Returns true when the front connection was made via an SSL/TLS transport
11126 layer and is locally deciphered. This means it has matched a socket declared
11127 with a "bind" line having the "ssl" option.
11128
Willy Tarreau74ca5042013-06-11 23:12:07 +020011129 Example :
11130 # This passes "X-Proto: https" to servers when client connects over SSL
11131 listen http-https
11132 bind :80
11133 bind :443 ssl crt /etc/haproxy.pem
11134 http-request add-header X-Proto https if { ssl_fc }
11135
11136ssl_fc_alg_keysize : integer
11137 Returns the symmetric cipher key size supported in bits when the incoming
11138 connection was made over an SSL/TLS transport layer.
11139
11140ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011141 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020011142 incoming connection made via a TLS transport layer and locally deciphered by
11143 haproxy. The result is a string containing the protocol name advertised by
11144 the client. The SSL library must have been built with support for TLS
11145 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
11146 not advertised unless the "alpn" keyword on the "bind" line specifies a
11147 protocol list. Also, nothing forces the client to pick a protocol from this
11148 list, any other one may be requested. The TLS ALPN extension is meant to
11149 replace the TLS NPN extension. See also "ssl_fc_npn".
11150
Willy Tarreau74ca5042013-06-11 23:12:07 +020011151ssl_fc_cipher : string
11152 Returns the name of the used cipher when the incoming connection was made
11153 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020011154
Willy Tarreau74ca5042013-06-11 23:12:07 +020011155ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020011156 Returns true if a client certificate is present in an incoming connection over
11157 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010011158 Note: on SSL session resumption with Session ID or TLS ticket, client
11159 certificate is not present in the current connection but may be retrieved
11160 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
11161 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020011162
Willy Tarreau74ca5042013-06-11 23:12:07 +020011163ssl_fc_has_sni : boolean
11164 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020011165 in an incoming connection was made over an SSL/TLS transport layer. Returns
11166 true when the incoming connection presents a TLS SNI field. This requires
11167 that the SSL library is build with support for TLS extensions enabled (check
11168 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020011169
Willy Tarreau74ca5042013-06-11 23:12:07 +020011170ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011171 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020011172 made via a TLS transport layer and locally deciphered by haproxy. The result
11173 is a string containing the protocol name advertised by the client. The SSL
11174 library must have been built with support for TLS extensions enabled (check
11175 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
11176 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
11177 forces the client to pick a protocol from this list, any other one may be
11178 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020011179
Willy Tarreau74ca5042013-06-11 23:12:07 +020011180ssl_fc_protocol : string
11181 Returns the name of the used protocol when the incoming connection was made
11182 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020011183
Emeric Brunb73a9b02014-04-30 18:49:19 +020011184ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040011185 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020011186 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
11187 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040011188
Willy Tarreau74ca5042013-06-11 23:12:07 +020011189ssl_fc_session_id : binary
11190 Returns the SSL ID of the front connection when the incoming connection was
11191 made over an SSL/TLS transport layer. It is useful to stick a given client to
11192 a server. It is important to note that some browsers refresh their session ID
11193 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020011194
Willy Tarreau74ca5042013-06-11 23:12:07 +020011195ssl_fc_sni : string
11196 This extracts the Server Name Indication TLS extension (SNI) field from an
11197 incoming connection made via an SSL/TLS transport layer and locally
11198 deciphered by haproxy. The result (when present) typically is a string
11199 matching the HTTPS host name (253 chars or less). The SSL library must have
11200 been built with support for TLS extensions enabled (check haproxy -vv).
11201
11202 This fetch is different from "req_ssl_sni" above in that it applies to the
11203 connection being deciphered by haproxy and not to SSL contents being blindly
11204 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020011205 requires that the SSL library is build with support for TLS extensions
11206 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020011207
Willy Tarreau74ca5042013-06-11 23:12:07 +020011208 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020011209 ssl_fc_sni_end : suffix match
11210 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020011211
Willy Tarreau74ca5042013-06-11 23:12:07 +020011212ssl_fc_use_keysize : integer
11213 Returns the symmetric cipher key size used in bits when the incoming
11214 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011215
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011216
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200112177.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011218------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020011219
Willy Tarreau74ca5042013-06-11 23:12:07 +020011220Fetching samples from buffer contents is a bit different from the previous
11221sample fetches above because the sampled data are ephemeral. These data can
11222only be used when they're available and will be lost when they're forwarded.
11223For this reason, samples fetched from buffer contents during a request cannot
11224be used in a response for example. Even while the data are being fetched, they
11225can change. Sometimes it is necessary to set some delays or combine multiple
11226sample fetch methods to ensure that the expected data are complete and usable,
11227for example through TCP request content inspection. Please see the "tcp-request
11228content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020011229
Willy Tarreau74ca5042013-06-11 23:12:07 +020011230payload(<offset>,<length>) : binary (deprecated)
11231 This is an alias for "req.payload" when used in the context of a request (eg:
11232 "stick on", "stick match"), and for "res.payload" when used in the context of
11233 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011234
Willy Tarreau74ca5042013-06-11 23:12:07 +020011235payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
11236 This is an alias for "req.payload_lv" when used in the context of a request
11237 (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the
11238 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010011239
Willy Tarreau74ca5042013-06-11 23:12:07 +020011240req.len : integer
11241req_len : integer (deprecated)
11242 Returns an integer value corresponding to the number of bytes present in the
11243 request buffer. This is mostly used in ACL. It is important to understand
11244 that this test does not return false as long as the buffer is changing. This
11245 means that a check with equality to zero will almost always immediately match
11246 at the beginning of the session, while a test for more data will wait for
11247 that data to come in and return false only when haproxy is certain that no
11248 more data will come in. This test was designed to be used with TCP request
11249 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011250
Willy Tarreau74ca5042013-06-11 23:12:07 +020011251req.payload(<offset>,<length>) : binary
11252 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011253 in the request buffer. As a special case, if the <length> argument is zero,
11254 the the whole buffer from <offset> to the end is extracted. This can be used
11255 with ACLs in order to check for the presence of some content in a buffer at
11256 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011257
Willy Tarreau74ca5042013-06-11 23:12:07 +020011258 ACL alternatives :
11259 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011260
Willy Tarreau74ca5042013-06-11 23:12:07 +020011261req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11262 This extracts a binary block whose size is specified at <offset1> for <length>
11263 bytes, and which starts at <offset2> if specified or just after the length in
11264 the request buffer. The <offset2> parameter also supports relative offsets if
11265 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011266
Willy Tarreau74ca5042013-06-11 23:12:07 +020011267 ACL alternatives :
11268 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011269
Willy Tarreau74ca5042013-06-11 23:12:07 +020011270 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011271
Willy Tarreau74ca5042013-06-11 23:12:07 +020011272req.proto_http : boolean
11273req_proto_http : boolean (deprecated)
11274 Returns true when data in the request buffer look like HTTP and correctly
11275 parses as such. It is the same parser as the common HTTP request parser which
11276 is used so there should be no surprises. The test does not match until the
11277 request is complete, failed or timed out. This test may be used to report the
11278 protocol in TCP logs, but the biggest use is to block TCP request analysis
11279 until a complete HTTP request is present in the buffer, for example to track
11280 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011281
Willy Tarreau74ca5042013-06-11 23:12:07 +020011282 Example:
11283 # track request counts per "base" (concatenation of Host+URL)
11284 tcp-request inspect-delay 10s
11285 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011286 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020011287
Willy Tarreau74ca5042013-06-11 23:12:07 +020011288req.rdp_cookie([<name>]) : string
11289rdp_cookie([<name>]) : string (deprecated)
11290 When the request buffer looks like the RDP protocol, extracts the RDP cookie
11291 <name>, or any cookie if unspecified. The parser only checks for the first
11292 cookie, as illustrated in the RDP protocol specification. The cookie name is
11293 case insensitive. Generally the "MSTS" cookie name will be used, as it can
11294 contain the user name of the client connecting to the server if properly
11295 configured on the client. The "MSTSHASH" cookie is often used as well for
11296 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011297
Willy Tarreau74ca5042013-06-11 23:12:07 +020011298 This differs from "balance rdp-cookie" in that any balancing algorithm may be
11299 used and thus the distribution of clients to backend servers is not linked to
11300 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
11301 such as "balance roundrobin" or "balance leastconn" will lead to a more even
11302 distribution of clients to backend servers than the hash used by "balance
11303 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011304
Willy Tarreau74ca5042013-06-11 23:12:07 +020011305 ACL derivatives :
11306 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011307
Willy Tarreau74ca5042013-06-11 23:12:07 +020011308 Example :
11309 listen tse-farm
11310 bind 0.0.0.0:3389
11311 # wait up to 5s for an RDP cookie in the request
11312 tcp-request inspect-delay 5s
11313 tcp-request content accept if RDP_COOKIE
11314 # apply RDP cookie persistence
11315 persist rdp-cookie
11316 # Persist based on the mstshash cookie
11317 # This is only useful makes sense if
11318 # balance rdp-cookie is not used
11319 stick-table type string size 204800
11320 stick on req.rdp_cookie(mstshash)
11321 server srv1 1.1.1.1:3389
11322 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011323
Willy Tarreau74ca5042013-06-11 23:12:07 +020011324 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
11325 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011326
Willy Tarreau74ca5042013-06-11 23:12:07 +020011327req.rdp_cookie_cnt([name]) : integer
11328rdp_cookie_cnt([name]) : integer (deprecated)
11329 Tries to parse the request buffer as RDP protocol, then returns an integer
11330 corresponding to the number of RDP cookies found. If an optional cookie name
11331 is passed, only cookies matching this name are considered. This is mostly
11332 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011333
Willy Tarreau74ca5042013-06-11 23:12:07 +020011334 ACL derivatives :
11335 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011336
Willy Tarreau74ca5042013-06-11 23:12:07 +020011337req.ssl_hello_type : integer
11338req_ssl_hello_type : integer (deprecated)
11339 Returns an integer value containing the type of the SSL hello message found
11340 in the request buffer if the buffer contains data that parse as a complete
11341 SSL (v3 or superior) client hello message. Note that this only applies to raw
11342 contents found in the request buffer and not to contents deciphered via an
11343 SSL data layer, so this will not work with "bind" lines having the "ssl"
11344 option. This is mostly used in ACL to detect presence of an SSL hello message
11345 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011346
Willy Tarreau74ca5042013-06-11 23:12:07 +020011347req.ssl_sni : string
11348req_ssl_sni : string (deprecated)
11349 Returns a string containing the value of the Server Name TLS extension sent
11350 by a client in a TLS stream passing through the request buffer if the buffer
11351 contains data that parse as a complete SSL (v3 or superior) client hello
11352 message. Note that this only applies to raw contents found in the request
11353 buffer and not to contents deciphered via an SSL data layer, so this will not
11354 work with "bind" lines having the "ssl" option. SNI normally contains the
11355 name of the host the client tries to connect to (for recent browsers). SNI is
11356 useful for allowing or denying access to certain hosts when SSL/TLS is used
11357 by the client. This test was designed to be used with TCP request content
11358 inspection. If content switching is needed, it is recommended to first wait
11359 for a complete client hello (type 1), like in the example below. See also
11360 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011361
Willy Tarreau74ca5042013-06-11 23:12:07 +020011362 ACL derivatives :
11363 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011364
Willy Tarreau74ca5042013-06-11 23:12:07 +020011365 Examples :
11366 # Wait for a client hello for at most 5 seconds
11367 tcp-request inspect-delay 5s
11368 tcp-request content accept if { req_ssl_hello_type 1 }
11369 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
11370 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020011371
Willy Tarreau74ca5042013-06-11 23:12:07 +020011372res.ssl_hello_type : integer
11373rep_ssl_hello_type : integer (deprecated)
11374 Returns an integer value containing the type of the SSL hello message found
11375 in the response buffer if the buffer contains data that parses as a complete
11376 SSL (v3 or superior) hello message. Note that this only applies to raw
11377 contents found in the response buffer and not to contents deciphered via an
11378 SSL data layer, so this will not work with "server" lines having the "ssl"
11379 option. This is mostly used in ACL to detect presence of an SSL hello message
11380 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau51539362012-05-08 12:46:28 +020011381
Willy Tarreau74ca5042013-06-11 23:12:07 +020011382req.ssl_ver : integer
11383req_ssl_ver : integer (deprecated)
11384 Returns an integer value containing the version of the SSL/TLS protocol of a
11385 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
11386 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
11387 composed of the major version multiplied by 65536, added to the minor
11388 version. Note that this only applies to raw contents found in the request
11389 buffer and not to contents deciphered via an SSL data layer, so this will not
11390 work with "bind" lines having the "ssl" option. The ACL version of the test
11391 matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This
11392 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011393
Willy Tarreau74ca5042013-06-11 23:12:07 +020011394 ACL derivatives :
11395 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011396
Willy Tarreau47e8eba2013-09-11 23:28:46 +020011397res.len : integer
11398 Returns an integer value corresponding to the number of bytes present in the
11399 response buffer. This is mostly used in ACL. It is important to understand
11400 that this test does not return false as long as the buffer is changing. This
11401 means that a check with equality to zero will almost always immediately match
11402 at the beginning of the session, while a test for more data will wait for
11403 that data to come in and return false only when haproxy is certain that no
11404 more data will come in. This test was designed to be used with TCP response
11405 content inspection.
11406
Willy Tarreau74ca5042013-06-11 23:12:07 +020011407res.payload(<offset>,<length>) : binary
11408 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020011409 in the response buffer. As a special case, if the <length> argument is zero,
11410 the the whole buffer from <offset> to the end is extracted. This can be used
11411 with ACLs in order to check for the presence of some content in a buffer at
11412 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011413
Willy Tarreau74ca5042013-06-11 23:12:07 +020011414res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
11415 This extracts a binary block whose size is specified at <offset1> for <length>
11416 bytes, and which starts at <offset2> if specified or just after the length in
11417 the response buffer. The <offset2> parameter also supports relative offsets
11418 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011419
Willy Tarreau74ca5042013-06-11 23:12:07 +020011420 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011421
Willy Tarreau74ca5042013-06-11 23:12:07 +020011422wait_end : boolean
11423 This fetch either returns true when the inspection period is over, or does
11424 not fetch. It is only used in ACLs, in conjunction with content analysis to
11425 avoid returning a wrong verdict early. It may also be used to delay some
11426 actions, such as a delayed reject for some special addresses. Since it either
11427 stops the rules evaluation or immediately returns true, it is recommended to
11428 use this acl as the last one in a rule. Please note that the default ACL
11429 "WAIT_END" is always usable without prior declaration. This test was designed
11430 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011431
Willy Tarreau74ca5042013-06-11 23:12:07 +020011432 Examples :
11433 # delay every incoming request by 2 seconds
11434 tcp-request inspect-delay 2s
11435 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010011436
Willy Tarreau74ca5042013-06-11 23:12:07 +020011437 # don't immediately tell bad guys they are rejected
11438 tcp-request inspect-delay 10s
11439 acl goodguys src 10.0.0.0/24
11440 acl badguys src 10.0.1.0/24
11441 tcp-request content accept if goodguys
11442 tcp-request content reject if badguys WAIT_END
11443 tcp-request content reject
11444
11445
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200114467.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020011447--------------------------------------
11448
11449It is possible to fetch samples from HTTP contents, requests and responses.
11450This application layer is also called layer 7. It is only possible to fetch the
11451data in this section when a full HTTP request or response has been parsed from
11452its respective request or response buffer. This is always the case with all
11453HTTP specific rules and for sections running with "mode http". When using TCP
11454content inspection, it may be necessary to support an inspection delay in order
11455to let the request or response come in first. These fetches may require a bit
11456more CPU resources than the layer 4 ones, but not much since the request and
11457response are indexed.
11458
11459base : string
11460 This returns the concatenation of the first Host header and the path part of
11461 the request, which starts at the first slash and ends before the question
11462 mark. It can be useful in virtual hosted environments to detect URL abuses as
11463 well as to improve shared caches efficiency. Using this with a limited size
11464 stick table also allows one to collect statistics about most commonly
11465 requested objects by host/path. With ACLs it can allow simple content
11466 switching rules involving the host and the path at the same time, such as
11467 "www.example.com/favicon.ico". See also "path" and "uri".
11468
11469 ACL derivatives :
11470 base : exact string match
11471 base_beg : prefix match
11472 base_dir : subdir match
11473 base_dom : domain match
11474 base_end : suffix match
11475 base_len : length match
11476 base_reg : regex match
11477 base_sub : substring match
11478
11479base32 : integer
11480 This returns a 32-bit hash of the value returned by the "base" fetch method
11481 above. This is useful to track per-URL activity on high traffic sites without
11482 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020011483 memory. The output type is an unsigned integer. The hash function used is
11484 SDBM with full avalanche on the output. Technically, base32 is exactly equal
11485 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020011486
11487base32+src : binary
11488 This returns the concatenation of the base32 fetch above and the src fetch
11489 below. The resulting type is of type binary, with a size of 8 or 20 bytes
11490 depending on the source address family. This can be used to track per-IP,
11491 per-URL counters.
11492
William Lallemand65ad6e12014-01-31 15:08:02 +010011493capture.req.hdr(<idx>) : string
11494 This extracts the content of the header captured by the "capture request
11495 header", idx is the position of the capture keyword in the configuration.
11496 The first entry is an index of 0. See also: "capture request header".
11497
11498capture.req.method : string
11499 This extracts the METHOD of an HTTP request. It can be used in both request
11500 and response. Unlike "method", it can be used in both request and response
11501 because it's allocated.
11502
11503capture.req.uri : string
11504 This extracts the request's URI, which starts at the first slash and ends
11505 before the first space in the request (without the host part). Unlike "path"
11506 and "url", it can be used in both request and response because it's
11507 allocated.
11508
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011509capture.req.ver : string
11510 This extracts the request's HTTP version and returns either "HTTP/1.0" or
11511 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
11512 logs because it relies on a persistent flag.
11513
William Lallemand65ad6e12014-01-31 15:08:02 +010011514capture.res.hdr(<idx>) : string
11515 This extracts the content of the header captured by the "capture response
11516 header", idx is the position of the capture keyword in the configuration.
11517 The first entry is an index of 0.
11518 See also: "capture response header"
11519
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020011520capture.res.ver : string
11521 This extracts the response's HTTP version and returns either "HTTP/1.0" or
11522 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
11523 persistent flag.
11524
Willy Tarreau74ca5042013-06-11 23:12:07 +020011525req.cook([<name>]) : string
11526cook([<name>]) : string (deprecated)
11527 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11528 header line from the request, and returns its value as string. If no name is
11529 specified, the first cookie value is returned. When used with ACLs, all
11530 matching cookies are evaluated. Spaces around the name and the value are
11531 ignored as requested by the Cookie header specification (RFC6265). The cookie
11532 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
11533 well return an empty value if it is present. Use the "found" match to detect
11534 presence. Use the res.cook() variant for response cookies sent by the server.
11535
11536 ACL derivatives :
11537 cook([<name>]) : exact string match
11538 cook_beg([<name>]) : prefix match
11539 cook_dir([<name>]) : subdir match
11540 cook_dom([<name>]) : domain match
11541 cook_end([<name>]) : suffix match
11542 cook_len([<name>]) : length match
11543 cook_reg([<name>]) : regex match
11544 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010011545
Willy Tarreau74ca5042013-06-11 23:12:07 +020011546req.cook_cnt([<name>]) : integer
11547cook_cnt([<name>]) : integer (deprecated)
11548 Returns an integer value representing the number of occurrences of the cookie
11549 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011550
Willy Tarreau74ca5042013-06-11 23:12:07 +020011551req.cook_val([<name>]) : integer
11552cook_val([<name>]) : integer (deprecated)
11553 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11554 header line from the request, and converts its value to an integer which is
11555 returned. If no name is specified, the first cookie value is returned. When
11556 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020011557
Willy Tarreau74ca5042013-06-11 23:12:07 +020011558cookie([<name>]) : string (deprecated)
11559 This extracts the last occurrence of the cookie name <name> on a "Cookie"
11560 header line from the request, or a "Set-Cookie" header from the response, and
11561 returns its value as a string. A typical use is to get multiple clients
11562 sharing a same profile use the same server. This can be similar to what
11563 "appsession" does with the "request-learn" statement, but with support for
11564 multi-peer synchronization and state keeping across restarts. If no name is
11565 specified, the first cookie value is returned. This fetch should not be used
11566 anymore and should be replaced by req.cook() or res.cook() instead as it
11567 ambiguously uses the direction based on the context where it is used.
11568 See also : "appsession".
Willy Tarreaud63335a2010-02-26 12:56:52 +010011569
Willy Tarreau74ca5042013-06-11 23:12:07 +020011570hdr([<name>[,<occ>]]) : string
11571 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
11572 used on responses. Please refer to these respective fetches for more details.
11573 In case of doubt about the fetch direction, please use the explicit ones.
11574 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011575 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011576
Willy Tarreau74ca5042013-06-11 23:12:07 +020011577req.fhdr(<name>[,<occ>]) : string
11578 This extracts the last occurrence of header <name> in an HTTP request. When
11579 used from an ACL, all occurrences are iterated over until a match is found.
11580 Optionally, a specific occurrence might be specified as a position number.
11581 Positive values indicate a position from the first occurrence, with 1 being
11582 the first one. Negative values indicate positions relative to the last one,
11583 with -1 being the last one. It differs from req.hdr() in that any commas
11584 present in the value are returned and are not used as delimiters. This is
11585 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011586
Willy Tarreau74ca5042013-06-11 23:12:07 +020011587req.fhdr_cnt([<name>]) : integer
11588 Returns an integer value representing the number of occurrences of request
11589 header field name <name>, or the total number of header fields if <name> is
11590 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
11591 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011592
Willy Tarreau74ca5042013-06-11 23:12:07 +020011593req.hdr([<name>[,<occ>]]) : string
11594 This extracts the last occurrence of header <name> in an HTTP request. When
11595 used from an ACL, all occurrences are iterated over until a match is found.
11596 Optionally, a specific occurrence might be specified as a position number.
11597 Positive values indicate a position from the first occurrence, with 1 being
11598 the first one. Negative values indicate positions relative to the last one,
11599 with -1 being the last one. A typical use is with the X-Forwarded-For header
11600 once converted to IP, associated with an IP stick-table. The function
11601 considers any comma as a delimiter for distinct values. If full-line headers
11602 are desired instead, use req.fhdr(). Please carefully check RFC2616 to know
11603 how certain headers are supposed to be parsed. Also, some of them are case
11604 insensitive (eg: Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010011605
Willy Tarreau74ca5042013-06-11 23:12:07 +020011606 ACL derivatives :
11607 hdr([<name>[,<occ>]]) : exact string match
11608 hdr_beg([<name>[,<occ>]]) : prefix match
11609 hdr_dir([<name>[,<occ>]]) : subdir match
11610 hdr_dom([<name>[,<occ>]]) : domain match
11611 hdr_end([<name>[,<occ>]]) : suffix match
11612 hdr_len([<name>[,<occ>]]) : length match
11613 hdr_reg([<name>[,<occ>]]) : regex match
11614 hdr_sub([<name>[,<occ>]]) : substring match
11615
11616req.hdr_cnt([<name>]) : integer
11617hdr_cnt([<header>]) : integer (deprecated)
11618 Returns an integer value representing the number of occurrences of request
11619 header field name <name>, or the total number of header field values if
11620 <name> is not specified. It is important to remember that one header line may
11621 count as several headers if it has several values. The function considers any
11622 comma as a delimiter for distinct values. If full-line headers are desired
11623 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
11624 detect presence, absence or abuse of a specific header, as well as to block
11625 request smuggling attacks by rejecting requests which contain more than one
11626 of certain headers. See "req.hdr" for more information on header matching.
11627
11628req.hdr_ip([<name>[,<occ>]]) : ip
11629hdr_ip([<name>[,<occ>]]) : ip (deprecated)
11630 This extracts the last occurrence of header <name> in an HTTP request,
11631 converts it to an IPv4 or IPv6 address and returns this address. When used
11632 with ACLs, all occurrences are checked, and if <name> is omitted, every value
11633 of every header is checked. Optionally, a specific occurrence might be
11634 specified as a position number. Positive values indicate a position from the
11635 first occurrence, with 1 being the first one. Negative values indicate
11636 positions relative to the last one, with -1 being the last one. A typical use
11637 is with the X-Forwarded-For and X-Client-IP headers.
11638
11639req.hdr_val([<name>[,<occ>]]) : integer
11640hdr_val([<name>[,<occ>]]) : integer (deprecated)
11641 This extracts the last occurrence of header <name> in an HTTP request, and
11642 converts it to an integer value. When used with ACLs, all occurrences are
11643 checked, and if <name> is omitted, every value of every header is checked.
11644 Optionally, a specific occurrence might be specified as a position number.
11645 Positive values indicate a position from the first occurrence, with 1 being
11646 the first one. Negative values indicate positions relative to the last one,
11647 with -1 being the last one. A typical use is with the X-Forwarded-For header.
11648
11649http_auth(<userlist>) : boolean
11650 Returns a boolean indicating whether the authentication data received from
11651 the client match a username & password stored in the specified userlist. This
11652 fetch function is not really useful outside of ACLs. Currently only http
11653 basic auth is supported.
11654
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011655http_auth_group(<userlist>) : string
11656 Returns a string corresponding to the user name found in the authentication
11657 data received from the client if both the user name and password are valid
11658 according to the specified userlist. The main purpose is to use it in ACLs
11659 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011660 This fetch function is not really useful outside of ACLs. Currently only http
11661 basic auth is supported.
11662
11663 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010011664 http_auth_group(<userlist>) : group ...
11665 Returns true when the user extracted from the request and whose password is
11666 valid according to the specified userlist belongs to at least one of the
11667 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011668
11669http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020011670 Returns true when the request being processed is the first one of the
11671 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020011672 from some requests when a request is not the first one, or to help grouping
11673 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020011674
Willy Tarreau74ca5042013-06-11 23:12:07 +020011675method : integer + string
11676 Returns an integer value corresponding to the method in the HTTP request. For
11677 example, "GET" equals 1 (check sources to establish the matching). Value 9
11678 means "other method" and may be converted to a string extracted from the
11679 stream. This should not be used directly as a sample, this is only meant to
11680 be used from ACLs, which transparently convert methods from patterns to these
11681 integer + string values. Some predefined ACL already check for most common
11682 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011683
Willy Tarreau74ca5042013-06-11 23:12:07 +020011684 ACL derivatives :
11685 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011686
Willy Tarreau74ca5042013-06-11 23:12:07 +020011687 Example :
11688 # only accept GET and HEAD requests
11689 acl valid_method method GET HEAD
11690 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020011691
Willy Tarreau74ca5042013-06-11 23:12:07 +020011692path : string
11693 This extracts the request's URL path, which starts at the first slash and
11694 ends before the question mark (without the host part). A typical use is with
11695 prefetch-capable caches, and with portals which need to aggregate multiple
11696 information from databases and keep them in caches. Note that with outgoing
11697 caches, it would be wiser to use "url" instead. With ACLs, it's typically
11698 used to match exact file names (eg: "/login.php"), or directory parts using
11699 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011700
Willy Tarreau74ca5042013-06-11 23:12:07 +020011701 ACL derivatives :
11702 path : exact string match
11703 path_beg : prefix match
11704 path_dir : subdir match
11705 path_dom : domain match
11706 path_end : suffix match
11707 path_len : length match
11708 path_reg : regex match
11709 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020011710
Willy Tarreau74ca5042013-06-11 23:12:07 +020011711req.ver : string
11712req_ver : string (deprecated)
11713 Returns the version string from the HTTP request, for example "1.1". This can
11714 be useful for logs, but is mostly there for ACL. Some predefined ACL already
11715 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011716
Willy Tarreau74ca5042013-06-11 23:12:07 +020011717 ACL derivatives :
11718 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020011719
Willy Tarreau74ca5042013-06-11 23:12:07 +020011720res.comp : boolean
11721 Returns the boolean "true" value if the response has been compressed by
11722 HAProxy, otherwise returns boolean "false". This may be used to add
11723 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011724
Willy Tarreau74ca5042013-06-11 23:12:07 +020011725res.comp_algo : string
11726 Returns a string containing the name of the algorithm used if the response
11727 was compressed by HAProxy, for example : "deflate". This may be used to add
11728 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011729
Willy Tarreau74ca5042013-06-11 23:12:07 +020011730res.cook([<name>]) : string
11731scook([<name>]) : string (deprecated)
11732 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11733 header line from the response, and returns its value as string. If no name is
11734 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011735
Willy Tarreau74ca5042013-06-11 23:12:07 +020011736 ACL derivatives :
11737 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020011738
Willy Tarreau74ca5042013-06-11 23:12:07 +020011739res.cook_cnt([<name>]) : integer
11740scook_cnt([<name>]) : integer (deprecated)
11741 Returns an integer value representing the number of occurrences of the cookie
11742 <name> in the response, or all cookies if <name> is not specified. This is
11743 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011744
Willy Tarreau74ca5042013-06-11 23:12:07 +020011745res.cook_val([<name>]) : integer
11746scook_val([<name>]) : integer (deprecated)
11747 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11748 header line from the response, and converts its value to an integer which is
11749 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010011750
Willy Tarreau74ca5042013-06-11 23:12:07 +020011751res.fhdr([<name>[,<occ>]]) : string
11752 This extracts the last occurrence of header <name> in an HTTP response, or of
11753 the last header if no <name> is specified. Optionally, a specific occurrence
11754 might be specified as a position number. Positive values indicate a position
11755 from the first occurrence, with 1 being the first one. Negative values
11756 indicate positions relative to the last one, with -1 being the last one. It
11757 differs from res.hdr() in that any commas present in the value are returned
11758 and are not used as delimiters. If this is not desired, the res.hdr() fetch
11759 should be used instead. This is sometimes useful with headers such as Date or
11760 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011761
Willy Tarreau74ca5042013-06-11 23:12:07 +020011762res.fhdr_cnt([<name>]) : integer
11763 Returns an integer value representing the number of occurrences of response
11764 header field name <name>, or the total number of header fields if <name> is
11765 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
11766 the number of full line headers and does not stop on commas. If this is not
11767 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011768
Willy Tarreau74ca5042013-06-11 23:12:07 +020011769res.hdr([<name>[,<occ>]]) : string
11770shdr([<name>[,<occ>]]) : string (deprecated)
11771 This extracts the last occurrence of header <name> in an HTTP response, or of
11772 the last header if no <name> is specified. Optionally, a specific occurrence
11773 might be specified as a position number. Positive values indicate a position
11774 from the first occurrence, with 1 being the first one. Negative values
11775 indicate positions relative to the last one, with -1 being the last one. This
11776 can be useful to learn some data into a stick-table. The function considers
11777 any comma as a delimiter for distinct values. If this is not desired, the
11778 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011779
Willy Tarreau74ca5042013-06-11 23:12:07 +020011780 ACL derivatives :
11781 shdr([<name>[,<occ>]]) : exact string match
11782 shdr_beg([<name>[,<occ>]]) : prefix match
11783 shdr_dir([<name>[,<occ>]]) : subdir match
11784 shdr_dom([<name>[,<occ>]]) : domain match
11785 shdr_end([<name>[,<occ>]]) : suffix match
11786 shdr_len([<name>[,<occ>]]) : length match
11787 shdr_reg([<name>[,<occ>]]) : regex match
11788 shdr_sub([<name>[,<occ>]]) : substring match
11789
11790res.hdr_cnt([<name>]) : integer
11791shdr_cnt([<name>]) : integer (deprecated)
11792 Returns an integer value representing the number of occurrences of response
11793 header field name <name>, or the total number of header fields if <name> is
11794 not specified. The function considers any comma as a delimiter for distinct
11795 values. If this is not desired, the res.fhdr_cnt() fetch should be used
11796 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011797
Willy Tarreau74ca5042013-06-11 23:12:07 +020011798res.hdr_ip([<name>[,<occ>]]) : ip
11799shdr_ip([<name>[,<occ>]]) : ip (deprecated)
11800 This extracts the last occurrence of header <name> in an HTTP response,
11801 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
11802 specific occurrence might be specified as a position number. Positive values
11803 indicate a position from the first occurrence, with 1 being the first one.
11804 Negative values indicate positions relative to the last one, with -1 being
11805 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011806
Willy Tarreau74ca5042013-06-11 23:12:07 +020011807res.hdr_val([<name>[,<occ>]]) : integer
11808shdr_val([<name>[,<occ>]]) : integer (deprecated)
11809 This extracts the last occurrence of header <name> in an HTTP response, and
11810 converts it to an integer value. Optionally, a specific occurrence might be
11811 specified as a position number. Positive values indicate a position from the
11812 first occurrence, with 1 being the first one. Negative values indicate
11813 positions relative to the last one, with -1 being the last one. This can be
11814 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011815
Willy Tarreau74ca5042013-06-11 23:12:07 +020011816res.ver : string
11817resp_ver : string (deprecated)
11818 Returns the version string from the HTTP response, for example "1.1". This
11819 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020011820
Willy Tarreau74ca5042013-06-11 23:12:07 +020011821 ACL derivatives :
11822 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010011823
Willy Tarreau74ca5042013-06-11 23:12:07 +020011824set-cookie([<name>]) : string (deprecated)
11825 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
11826 header line from the response and uses the corresponding value to match. This
11827 can be comparable to what "appsession" does with default options, but with
11828 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011829
Willy Tarreau74ca5042013-06-11 23:12:07 +020011830 This fetch function is deprecated and has been superseded by the "res.cook"
11831 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010011832
Willy Tarreau74ca5042013-06-11 23:12:07 +020011833 See also : "appsession"
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011834
Willy Tarreau74ca5042013-06-11 23:12:07 +020011835status : integer
11836 Returns an integer containing the HTTP status code in the HTTP response, for
11837 example, 302. It is mostly used within ACLs and integer ranges, for example,
11838 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011839
Willy Tarreau74ca5042013-06-11 23:12:07 +020011840url : string
11841 This extracts the request's URL as presented in the request. A typical use is
11842 with prefetch-capable caches, and with portals which need to aggregate
11843 multiple information from databases and keep them in caches. With ACLs, using
11844 "path" is preferred over using "url", because clients may send a full URL as
11845 is normally done with proxies. The only real use is to match "*" which does
11846 not match in "path", and for which there is already a predefined ACL. See
11847 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011848
Willy Tarreau74ca5042013-06-11 23:12:07 +020011849 ACL derivatives :
11850 url : exact string match
11851 url_beg : prefix match
11852 url_dir : subdir match
11853 url_dom : domain match
11854 url_end : suffix match
11855 url_len : length match
11856 url_reg : regex match
11857 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011858
Willy Tarreau74ca5042013-06-11 23:12:07 +020011859url_ip : ip
11860 This extracts the IP address from the request's URL when the host part is
11861 presented as an IP address. Its use is very limited. For instance, a
11862 monitoring system might use this field as an alternative for the source IP in
11863 order to test what path a given source address would follow, or to force an
11864 entry in a table for a given source address. With ACLs it can be used to
11865 restrict access to certain systems through a proxy, for example when combined
11866 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011867
Willy Tarreau74ca5042013-06-11 23:12:07 +020011868url_port : integer
11869 This extracts the port part from the request's URL. Note that if the port is
11870 not specified in the request, port 80 is assumed. With ACLs it can be used to
11871 restrict access to certain systems through a proxy, for example when combined
11872 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011873
Willy Tarreau74ca5042013-06-11 23:12:07 +020011874urlp(<name>[,<delim>]) : string
11875url_param(<name>[,<delim>]) : string
11876 This extracts the first occurrence of the parameter <name> in the query
11877 string, which begins after either '?' or <delim>, and which ends before '&',
11878 ';' or <delim>. The parameter name is case-sensitive. The result is a string
11879 corresponding to the value of the parameter <name> as presented in the
11880 request (no URL decoding is performed). This can be used for session
11881 stickiness based on a client ID, to extract an application cookie passed as a
11882 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
11883 this fetch do not iterate over multiple parameters and stop at the first one
11884 as well.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011885
Willy Tarreau74ca5042013-06-11 23:12:07 +020011886 ACL derivatives :
11887 urlp(<name>[,<delim>]) : exact string match
11888 urlp_beg(<name>[,<delim>]) : prefix match
11889 urlp_dir(<name>[,<delim>]) : subdir match
11890 urlp_dom(<name>[,<delim>]) : domain match
11891 urlp_end(<name>[,<delim>]) : suffix match
11892 urlp_len(<name>[,<delim>]) : length match
11893 urlp_reg(<name>[,<delim>]) : regex match
11894 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011895
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011896
Willy Tarreau74ca5042013-06-11 23:12:07 +020011897 Example :
11898 # match http://example.com/foo?PHPSESSIONID=some_id
11899 stick on urlp(PHPSESSIONID)
11900 # match http://example.com/foo;JSESSIONID=some_id
11901 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020011902
Willy Tarreau74ca5042013-06-11 23:12:07 +020011903urlp_val(<name>[,<delim>]) : integer
11904 See "urlp" above. This one extracts the URL parameter <name> in the request
11905 and converts it to an integer value. This can be used for session stickiness
11906 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020011907
Willy Tarreau198a7442008-01-17 12:05:32 +010011908
Willy Tarreau74ca5042013-06-11 23:12:07 +0200119097.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011910---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011911
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011912Some predefined ACLs are hard-coded so that they do not have to be declared in
11913every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020011914order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010011915
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011916ACL name Equivalent to Usage
11917---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011918FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020011919HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011920HTTP_1.0 req_ver 1.0 match HTTP version 1.0
11921HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010011922HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
11923HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
11924HTTP_URL_SLASH url_beg / match URL beginning with "/"
11925HTTP_URL_STAR url * match URL equal to "*"
11926LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011927METH_CONNECT method CONNECT match HTTP CONNECT method
11928METH_GET method GET HEAD match HTTP GET or HEAD method
11929METH_HEAD method HEAD match HTTP HEAD method
11930METH_OPTIONS method OPTIONS match HTTP OPTIONS method
11931METH_POST method POST match HTTP POST method
11932METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020011933RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011934REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010011935TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011936WAIT_END wait_end wait for end of content analysis
11937---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010011938
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011939
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200119408. Logging
11941----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010011942
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011943One of HAProxy's strong points certainly lies is its precise logs. It probably
11944provides the finest level of information available for such a product, which is
11945very important for troubleshooting complex environments. Standard information
11946provided in logs include client ports, TCP/HTTP state timers, precise session
11947state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011948to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011949headers.
11950
11951In order to improve administrators reactivity, it offers a great transparency
11952about encountered problems, both internal and external, and it is possible to
11953send logs to different sources at the same time with different level filters :
11954
11955 - global process-level logs (system errors, start/stop, etc..)
11956 - per-instance system and internal errors (lack of resource, bugs, ...)
11957 - per-instance external troubles (servers up/down, max connections)
11958 - per-instance activity (client connections), either at the establishment or
11959 at the termination.
11960
11961The ability to distribute different levels of logs to different log servers
11962allow several production teams to interact and to fix their problems as soon
11963as possible. For example, the system team might monitor system-wide errors,
11964while the application team might be monitoring the up/down for their servers in
11965real time, and the security team might analyze the activity logs with one hour
11966delay.
11967
11968
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200119698.1. Log levels
11970---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011971
Simon Hormandf791f52011-05-29 15:01:10 +090011972TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011973source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090011974HTTP request, HTTP return code, number of bytes transmitted, conditions
11975in which the session ended, and even exchanged cookies values. For example
11976track a particular user's problems. All messages may be sent to up to two
11977syslog servers. Check the "log" keyword in section 4.2 for more information
11978about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011979
11980
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200119818.2. Log formats
11982----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011983
William Lallemand48940402012-01-30 16:47:22 +010011984HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090011985and will be detailed in the following sections. A few of them may vary
11986slightly with the configuration, due to indicators specific to certain
11987options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010011988
11989 - the default format, which is very basic and very rarely used. It only
11990 provides very basic information about the incoming connection at the moment
11991 it is accepted : source IP:port, destination IP:port, and frontend-name.
11992 This mode will eventually disappear so it will not be described to great
11993 extents.
11994
11995 - the TCP format, which is more advanced. This format is enabled when "option
11996 tcplog" is set on the frontend. HAProxy will then usually wait for the
11997 connection to terminate before logging. This format provides much richer
11998 information, such as timers, connection counts, queue size, etc... This
11999 format is recommended for pure TCP proxies.
12000
12001 - the HTTP format, which is the most advanced for HTTP proxying. This format
12002 is enabled when "option httplog" is set on the frontend. It provides the
12003 same information as the TCP format with some HTTP-specific fields such as
12004 the request, the status code, and captures of headers and cookies. This
12005 format is recommended for HTTP proxies.
12006
Emeric Brun3a058f32009-06-30 18:26:00 +020012007 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
12008 fields arranged in the same order as the CLF format. In this mode, all
12009 timers, captures, flags, etc... appear one per field after the end of the
12010 common fields, in the same order they appear in the standard HTTP format.
12011
William Lallemand48940402012-01-30 16:47:22 +010012012 - the custom log format, allows you to make your own log line.
12013
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012014Next sections will go deeper into details for each of these formats. Format
12015specification will be performed on a "field" basis. Unless stated otherwise, a
12016field is a portion of text delimited by any number of spaces. Since syslog
12017servers are susceptible of inserting fields at the beginning of a line, it is
12018always assumed that the first field is the one containing the process name and
12019identifier.
12020
12021Note : Since log lines may be quite long, the log examples in sections below
12022 might be broken into multiple lines. The example log lines will be
12023 prefixed with 3 closing angle brackets ('>>>') and each time a log is
12024 broken into multiple lines, each non-final line will end with a
12025 backslash ('\') and the next line will start indented by two characters.
12026
12027
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200120288.2.1. Default log format
12029-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012030
12031This format is used when no specific option is set. The log is emitted as soon
12032as the connection is accepted. One should note that this currently is the only
12033format which logs the request's destination IP and ports.
12034
12035 Example :
12036 listen www
12037 mode http
12038 log global
12039 server srv1 127.0.0.1:8000
12040
12041 >>> Feb 6 12:12:09 localhost \
12042 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
12043 (www/HTTP)
12044
12045 Field Format Extract from the example above
12046 1 process_name '[' pid ']:' haproxy[14385]:
12047 2 'Connect from' Connect from
12048 3 source_ip ':' source_port 10.0.1.2:33312
12049 4 'to' to
12050 5 destination_ip ':' destination_port 10.0.3.31:8012
12051 6 '(' frontend_name '/' mode ')' (www/HTTP)
12052
12053Detailed fields description :
12054 - "source_ip" is the IP address of the client which initiated the connection.
12055 - "source_port" is the TCP port of the client which initiated the connection.
12056 - "destination_ip" is the IP address the client connected to.
12057 - "destination_port" is the TCP port the client connected to.
12058 - "frontend_name" is the name of the frontend (or listener) which received
12059 and processed the connection.
12060 - "mode is the mode the frontend is operating (TCP or HTTP).
12061
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012062In case of a UNIX socket, the source and destination addresses are marked as
12063"unix:" and the ports reflect the internal ID of the socket which accepted the
12064connection (the same ID as reported in the stats).
12065
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012066It is advised not to use this deprecated format for newer installations as it
12067will eventually disappear.
12068
12069
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200120708.2.2. TCP log format
12071---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012072
12073The TCP format is used when "option tcplog" is specified in the frontend, and
12074is the recommended format for pure TCP proxies. It provides a lot of precious
12075information for troubleshooting. Since this format includes timers and byte
12076counts, the log is normally emitted at the end of the session. It can be
12077emitted earlier if "option logasap" is specified, which makes sense in most
12078environments with long sessions such as remote terminals. Sessions which match
12079the "monitor" rules are never logged. It is also possible not to emit logs for
12080sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012081specifying "option dontlognull" in the frontend. Successful connections will
12082not be logged if "option dontlog-normal" is specified in the frontend. A few
12083fields may slightly vary depending on some configuration options, those are
12084marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012085
12086 Example :
12087 frontend fnt
12088 mode tcp
12089 option tcplog
12090 log global
12091 default_backend bck
12092
12093 backend bck
12094 server srv1 127.0.0.1:8000
12095
12096 >>> Feb 6 12:12:56 localhost \
12097 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
12098 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
12099
12100 Field Format Extract from the example above
12101 1 process_name '[' pid ']:' haproxy[14387]:
12102 2 client_ip ':' client_port 10.0.1.2:33313
12103 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
12104 4 frontend_name fnt
12105 5 backend_name '/' server_name bck/srv1
12106 6 Tw '/' Tc '/' Tt* 0/0/5007
12107 7 bytes_read* 212
12108 8 termination_state --
12109 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
12110 10 srv_queue '/' backend_queue 0/0
12111
12112Detailed fields description :
12113 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012114 connection to haproxy. If the connection was accepted on a UNIX socket
12115 instead, the IP address would be replaced with the word "unix". Note that
12116 when the connection is accepted on a socket configured with "accept-proxy"
12117 and the PROXY protocol is correctly used, then the logs will reflect the
12118 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012119
12120 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012121 If the connection was accepted on a UNIX socket instead, the port would be
12122 replaced with the ID of the accepting socket, which is also reported in the
12123 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012124
12125 - "accept_date" is the exact date when the connection was received by haproxy
12126 (which might be very slightly different from the date observed on the
12127 network if there was some queuing in the system's backlog). This is usually
12128 the same date which may appear in any upstream firewall's log.
12129
12130 - "frontend_name" is the name of the frontend (or listener) which received
12131 and processed the connection.
12132
12133 - "backend_name" is the name of the backend (or listener) which was selected
12134 to manage the connection to the server. This will be the same as the
12135 frontend if no switching rule has been applied, which is common for TCP
12136 applications.
12137
12138 - "server_name" is the name of the last server to which the connection was
12139 sent, which might differ from the first one if there were connection errors
12140 and a redispatch occurred. Note that this server belongs to the backend
12141 which processed the request. If the connection was aborted before reaching
12142 a server, "<NOSRV>" is indicated instead of a server name.
12143
12144 - "Tw" is the total time in milliseconds spent waiting in the various queues.
12145 It can be "-1" if the connection was aborted before reaching the queue.
12146 See "Timers" below for more details.
12147
12148 - "Tc" is the total time in milliseconds spent waiting for the connection to
12149 establish to the final server, including retries. It can be "-1" if the
12150 connection was aborted before a connection could be established. See
12151 "Timers" below for more details.
12152
12153 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012154 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012155 "option logasap" was specified, then the time counting stops at the moment
12156 the log is emitted. In this case, a '+' sign is prepended before the value,
12157 indicating that the final one will be larger. See "Timers" below for more
12158 details.
12159
12160 - "bytes_read" is the total number of bytes transmitted from the server to
12161 the client when the log is emitted. If "option logasap" is specified, the
12162 this value will be prefixed with a '+' sign indicating that the final one
12163 may be larger. Please note that this value is a 64-bit counter, so log
12164 analysis tools must be able to handle it without overflowing.
12165
12166 - "termination_state" is the condition the session was in when the session
12167 ended. This indicates the session state, which side caused the end of
12168 session to happen, and for what reason (timeout, error, ...). The normal
12169 flags should be "--", indicating the session was closed by either end with
12170 no data remaining in buffers. See below "Session state at disconnection"
12171 for more details.
12172
12173 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012174 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012175 limits have been reached. For instance, if actconn is close to 512 when
12176 multiple connection errors occur, chances are high that the system limits
12177 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012178 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012179
12180 - "feconn" is the total number of concurrent connections on the frontend when
12181 the session was logged. It is useful to estimate the amount of resource
12182 required to sustain high loads, and to detect when the frontend's "maxconn"
12183 has been reached. Most often when this value increases by huge jumps, it is
12184 because there is congestion on the backend servers, but sometimes it can be
12185 caused by a denial of service attack.
12186
12187 - "beconn" is the total number of concurrent connections handled by the
12188 backend when the session was logged. It includes the total number of
12189 concurrent connections active on servers as well as the number of
12190 connections pending in queues. It is useful to estimate the amount of
12191 additional servers needed to support high loads for a given application.
12192 Most often when this value increases by huge jumps, it is because there is
12193 congestion on the backend servers, but sometimes it can be caused by a
12194 denial of service attack.
12195
12196 - "srv_conn" is the total number of concurrent connections still active on
12197 the server when the session was logged. It can never exceed the server's
12198 configured "maxconn" parameter. If this value is very often close or equal
12199 to the server's "maxconn", it means that traffic regulation is involved a
12200 lot, meaning that either the server's maxconn value is too low, or that
12201 there aren't enough servers to process the load with an optimal response
12202 time. When only one of the server's "srv_conn" is high, it usually means
12203 that this server has some trouble causing the connections to take longer to
12204 be processed than on other servers.
12205
12206 - "retries" is the number of connection retries experienced by this session
12207 when trying to connect to the server. It must normally be zero, unless a
12208 server is being stopped at the same moment the connection was attempted.
12209 Frequent retries generally indicate either a network problem between
12210 haproxy and the server, or a misconfigured system backlog on the server
12211 preventing new connections from being queued. This field may optionally be
12212 prefixed with a '+' sign, indicating that the session has experienced a
12213 redispatch after the maximal retry count has been reached on the initial
12214 server. In this case, the server name appearing in the log is the one the
12215 connection was redispatched to, and not the first one, though both may
12216 sometimes be the same in case of hashing for instance. So as a general rule
12217 of thumb, when a '+' is present in front of the retry count, this count
12218 should not be attributed to the logged server.
12219
12220 - "srv_queue" is the total number of requests which were processed before
12221 this one in the server queue. It is zero when the request has not gone
12222 through the server queue. It makes it possible to estimate the approximate
12223 server's response time by dividing the time spent in queue by the number of
12224 requests in the queue. It is worth noting that if a session experiences a
12225 redispatch and passes through two server queues, their positions will be
12226 cumulated. A request should not pass through both the server queue and the
12227 backend queue unless a redispatch occurs.
12228
12229 - "backend_queue" is the total number of requests which were processed before
12230 this one in the backend's global queue. It is zero when the request has not
12231 gone through the global queue. It makes it possible to estimate the average
12232 queue length, which easily translates into a number of missing servers when
12233 divided by a server's "maxconn" parameter. It is worth noting that if a
12234 session experiences a redispatch, it may pass twice in the backend's queue,
12235 and then both positions will be cumulated. A request should not pass
12236 through both the server queue and the backend queue unless a redispatch
12237 occurs.
12238
12239
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122408.2.3. HTTP log format
12241----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012242
12243The HTTP format is the most complete and the best suited for HTTP proxies. It
12244is enabled by when "option httplog" is specified in the frontend. It provides
12245the same level of information as the TCP format with additional features which
12246are specific to the HTTP protocol. Just like the TCP format, the log is usually
12247emitted at the end of the session, unless "option logasap" is specified, which
12248generally only makes sense for download sites. A session which matches the
12249"monitor" rules will never logged. It is also possible not to log sessions for
12250which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012251frontend. Successful connections will not be logged if "option dontlog-normal"
12252is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012253
12254Most fields are shared with the TCP log, some being different. A few fields may
12255slightly vary depending on some configuration options. Those ones are marked
12256with a star ('*') after the field name below.
12257
12258 Example :
12259 frontend http-in
12260 mode http
12261 option httplog
12262 log global
12263 default_backend bck
12264
12265 backend static
12266 server srv1 127.0.0.1:8000
12267
12268 >>> Feb 6 12:14:14 localhost \
12269 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
12270 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 +010012271 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012272
12273 Field Format Extract from the example above
12274 1 process_name '[' pid ']:' haproxy[14389]:
12275 2 client_ip ':' client_port 10.0.1.2:33317
12276 3 '[' accept_date ']' [06/Feb/2009:12:14:14.655]
12277 4 frontend_name http-in
12278 5 backend_name '/' server_name static/srv1
12279 6 Tq '/' Tw '/' Tc '/' Tr '/' Tt* 10/0/30/69/109
12280 7 status_code 200
12281 8 bytes_read* 2750
12282 9 captured_request_cookie -
12283 10 captured_response_cookie -
12284 11 termination_state ----
12285 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
12286 13 srv_queue '/' backend_queue 0/0
12287 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
12288 15 '{' captured_response_headers* '}' {}
12289 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010012290
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012291
12292Detailed fields description :
12293 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012294 connection to haproxy. If the connection was accepted on a UNIX socket
12295 instead, the IP address would be replaced with the word "unix". Note that
12296 when the connection is accepted on a socket configured with "accept-proxy"
12297 and the PROXY protocol is correctly used, then the logs will reflect the
12298 forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012299
12300 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010012301 If the connection was accepted on a UNIX socket instead, the port would be
12302 replaced with the ID of the accepting socket, which is also reported in the
12303 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012304
12305 - "accept_date" is the exact date when the TCP connection was received by
12306 haproxy (which might be very slightly different from the date observed on
12307 the network if there was some queuing in the system's backlog). This is
12308 usually the same date which may appear in any upstream firewall's log. This
12309 does not depend on the fact that the client has sent the request or not.
12310
12311 - "frontend_name" is the name of the frontend (or listener) which received
12312 and processed the connection.
12313
12314 - "backend_name" is the name of the backend (or listener) which was selected
12315 to manage the connection to the server. This will be the same as the
12316 frontend if no switching rule has been applied.
12317
12318 - "server_name" is the name of the last server to which the connection was
12319 sent, which might differ from the first one if there were connection errors
12320 and a redispatch occurred. Note that this server belongs to the backend
12321 which processed the request. If the request was aborted before reaching a
12322 server, "<NOSRV>" is indicated instead of a server name. If the request was
12323 intercepted by the stats subsystem, "<STATS>" is indicated instead.
12324
12325 - "Tq" is the total time in milliseconds spent waiting for the client to send
12326 a full HTTP request, not counting data. It can be "-1" if the connection
12327 was aborted before a complete request could be received. It should always
12328 be very small because a request generally fits in one single packet. Large
12329 times here generally indicate network trouble between the client and
12330 haproxy. See "Timers" below for more details.
12331
12332 - "Tw" is the total time in milliseconds spent waiting in the various queues.
12333 It can be "-1" if the connection was aborted before reaching the queue.
12334 See "Timers" below for more details.
12335
12336 - "Tc" is the total time in milliseconds spent waiting for the connection to
12337 establish to the final server, including retries. It can be "-1" if the
12338 request was aborted before a connection could be established. See "Timers"
12339 below for more details.
12340
12341 - "Tr" is the total time in milliseconds spent waiting for the server to send
12342 a full HTTP response, not counting data. It can be "-1" if the request was
12343 aborted before a complete response could be received. It generally matches
12344 the server's processing time for the request, though it may be altered by
12345 the amount of data sent by the client to the server. Large times here on
12346 "GET" requests generally indicate an overloaded server. See "Timers" below
12347 for more details.
12348
12349 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012350 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012351 "option logasap" was specified, then the time counting stops at the moment
12352 the log is emitted. In this case, a '+' sign is prepended before the value,
12353 indicating that the final one will be larger. See "Timers" below for more
12354 details.
12355
12356 - "status_code" is the HTTP status code returned to the client. This status
12357 is generally set by the server, but it might also be set by haproxy when
12358 the server cannot be reached or when its response is blocked by haproxy.
12359
12360 - "bytes_read" is the total number of bytes transmitted to the client when
12361 the log is emitted. This does include HTTP headers. If "option logasap" is
12362 specified, the this value will be prefixed with a '+' sign indicating that
12363 the final one may be larger. Please note that this value is a 64-bit
12364 counter, so log analysis tools must be able to handle it without
12365 overflowing.
12366
12367 - "captured_request_cookie" is an optional "name=value" entry indicating that
12368 the client had this cookie in the request. The cookie name and its maximum
12369 length are defined by the "capture cookie" statement in the frontend
12370 configuration. The field is a single dash ('-') when the option is not
12371 set. Only one cookie may be captured, it is generally used to track session
12372 ID exchanges between a client and a server to detect session crossing
12373 between clients due to application bugs. For more details, please consult
12374 the section "Capturing HTTP headers and cookies" below.
12375
12376 - "captured_response_cookie" is an optional "name=value" entry indicating
12377 that the server has returned a cookie with its response. The cookie name
12378 and its maximum length are defined by the "capture cookie" statement in the
12379 frontend configuration. The field is a single dash ('-') when the option is
12380 not set. Only one cookie may be captured, it is generally used to track
12381 session ID exchanges between a client and a server to detect session
12382 crossing between clients due to application bugs. For more details, please
12383 consult the section "Capturing HTTP headers and cookies" below.
12384
12385 - "termination_state" is the condition the session was in when the session
12386 ended. This indicates the session state, which side caused the end of
12387 session to happen, for what reason (timeout, error, ...), just like in TCP
12388 logs, and information about persistence operations on cookies in the last
12389 two characters. The normal flags should begin with "--", indicating the
12390 session was closed by either end with no data remaining in buffers. See
12391 below "Session state at disconnection" for more details.
12392
12393 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012394 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012395 limits have been reached. For instance, if actconn is close to 512 or 1024
12396 when multiple connection errors occur, chances are high that the system
12397 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012398 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012399 system.
12400
12401 - "feconn" is the total number of concurrent connections on the frontend when
12402 the session was logged. It is useful to estimate the amount of resource
12403 required to sustain high loads, and to detect when the frontend's "maxconn"
12404 has been reached. Most often when this value increases by huge jumps, it is
12405 because there is congestion on the backend servers, but sometimes it can be
12406 caused by a denial of service attack.
12407
12408 - "beconn" is the total number of concurrent connections handled by the
12409 backend when the session was logged. It includes the total number of
12410 concurrent connections active on servers as well as the number of
12411 connections pending in queues. It is useful to estimate the amount of
12412 additional servers needed to support high loads for a given application.
12413 Most often when this value increases by huge jumps, it is because there is
12414 congestion on the backend servers, but sometimes it can be caused by a
12415 denial of service attack.
12416
12417 - "srv_conn" is the total number of concurrent connections still active on
12418 the server when the session was logged. It can never exceed the server's
12419 configured "maxconn" parameter. If this value is very often close or equal
12420 to the server's "maxconn", it means that traffic regulation is involved a
12421 lot, meaning that either the server's maxconn value is too low, or that
12422 there aren't enough servers to process the load with an optimal response
12423 time. When only one of the server's "srv_conn" is high, it usually means
12424 that this server has some trouble causing the requests to take longer to be
12425 processed than on other servers.
12426
12427 - "retries" is the number of connection retries experienced by this session
12428 when trying to connect to the server. It must normally be zero, unless a
12429 server is being stopped at the same moment the connection was attempted.
12430 Frequent retries generally indicate either a network problem between
12431 haproxy and the server, or a misconfigured system backlog on the server
12432 preventing new connections from being queued. This field may optionally be
12433 prefixed with a '+' sign, indicating that the session has experienced a
12434 redispatch after the maximal retry count has been reached on the initial
12435 server. In this case, the server name appearing in the log is the one the
12436 connection was redispatched to, and not the first one, though both may
12437 sometimes be the same in case of hashing for instance. So as a general rule
12438 of thumb, when a '+' is present in front of the retry count, this count
12439 should not be attributed to the logged server.
12440
12441 - "srv_queue" is the total number of requests which were processed before
12442 this one in the server queue. It is zero when the request has not gone
12443 through the server queue. It makes it possible to estimate the approximate
12444 server's response time by dividing the time spent in queue by the number of
12445 requests in the queue. It is worth noting that if a session experiences a
12446 redispatch and passes through two server queues, their positions will be
12447 cumulated. A request should not pass through both the server queue and the
12448 backend queue unless a redispatch occurs.
12449
12450 - "backend_queue" is the total number of requests which were processed before
12451 this one in the backend's global queue. It is zero when the request has not
12452 gone through the global queue. It makes it possible to estimate the average
12453 queue length, which easily translates into a number of missing servers when
12454 divided by a server's "maxconn" parameter. It is worth noting that if a
12455 session experiences a redispatch, it may pass twice in the backend's queue,
12456 and then both positions will be cumulated. A request should not pass
12457 through both the server queue and the backend queue unless a redispatch
12458 occurs.
12459
12460 - "captured_request_headers" is a list of headers captured in the request due
12461 to the presence of the "capture request header" statement in the frontend.
12462 Multiple headers can be captured, they will be delimited by a vertical bar
12463 ('|'). When no capture is enabled, the braces do not appear, causing a
12464 shift of remaining fields. It is important to note that this field may
12465 contain spaces, and that using it requires a smarter log parser than when
12466 it's not used. Please consult the section "Capturing HTTP headers and
12467 cookies" below for more details.
12468
12469 - "captured_response_headers" is a list of headers captured in the response
12470 due to the presence of the "capture response header" statement in the
12471 frontend. Multiple headers can be captured, they will be delimited by a
12472 vertical bar ('|'). When no capture is enabled, the braces do not appear,
12473 causing a shift of remaining fields. It is important to note that this
12474 field may contain spaces, and that using it requires a smarter log parser
12475 than when it's not used. Please consult the section "Capturing HTTP headers
12476 and cookies" below for more details.
12477
12478 - "http_request" is the complete HTTP request line, including the method,
12479 request and HTTP version string. Non-printable characters are encoded (see
12480 below the section "Non-printable characters"). This is always the last
12481 field, and it is always delimited by quotes and is the only one which can
12482 contain quotes. If new fields are added to the log format, they will be
12483 added before this field. This field might be truncated if the request is
12484 huge and does not fit in the standard syslog buffer (1024 characters). This
12485 is the reason why this field must always remain the last one.
12486
12487
Cyril Bontédc4d9032012-04-08 21:57:39 +0200124888.2.4. Custom log format
12489------------------------
William Lallemand48940402012-01-30 16:47:22 +010012490
Willy Tarreau2beef582012-12-20 17:22:52 +010012491The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010012492mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010012493
12494HAproxy understands some log format variables. % precedes log format variables.
12495Variables can take arguments using braces ('{}'), and multiple arguments are
12496separated by commas within the braces. Flags may be added or removed by
12497prefixing them with a '+' or '-' sign.
12498
12499Special variable "%o" may be used to propagate its flags to all other
12500variables on the same format string. This is particularly handy with quoted
12501string formats ("Q").
12502
Willy Tarreauc8368452012-12-21 00:09:23 +010012503If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020012504as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010012505less common information such as the client's SSL certificate's DN, or to log
12506the key that would be used to store an entry into a stick table.
12507
William Lallemand48940402012-01-30 16:47:22 +010012508Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012509In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010012510in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010012511
12512Flags are :
12513 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012514 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
William Lallemand48940402012-01-30 16:47:22 +010012515
12516 Example:
12517
12518 log-format %T\ %t\ Some\ Text
12519 log-format %{+Q}o\ %t\ %s\ %{-Q}r
12520
12521At the moment, the default HTTP format is defined this way :
12522
Willy Tarreau2beef582012-12-20 17:22:52 +010012523 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tq/%Tw/%Tc/%Tr/%Tt\ %ST\ %B\ %CC\ \
12524 %CS\ %tsc\ %ac/%fc/%bc/%sc/%rc\ %sq/%bq\ %hr\ %hs\ %{+Q}r
William Lallemand48940402012-01-30 16:47:22 +010012525
William Lallemandbddd4fd2012-02-27 11:23:10 +010012526the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010012527
Willy Tarreau2beef582012-12-20 17:22:52 +010012528 log-format %{+Q}o\ %{-Q}ci\ -\ -\ [%T]\ %r\ %ST\ %B\ \"\"\ \"\"\ %cp\ \
Willy Tarreau773d65f2012-10-12 14:56:11 +020012529 %ms\ %ft\ %b\ %s\ \%Tq\ %Tw\ %Tc\ %Tr\ %Tt\ %tsc\ %ac\ %fc\ \
Willy Tarreau2beef582012-12-20 17:22:52 +010012530 %bc\ %sc\ %rc\ %sq\ %bq\ %CC\ %CS\ \%hrl\ %hsl
William Lallemand48940402012-01-30 16:47:22 +010012531
William Lallemandbddd4fd2012-02-27 11:23:10 +010012532and the default TCP format is defined this way :
12533
Willy Tarreau2beef582012-12-20 17:22:52 +010012534 log-format %ci:%cp\ [%t]\ %ft\ %b/%s\ %Tw/%Tc/%Tt\ %B\ %ts\ \
William Lallemandbddd4fd2012-02-27 11:23:10 +010012535 %ac/%fc/%bc/%sc/%rc\ %sq/%bq
12536
William Lallemand48940402012-01-30 16:47:22 +010012537Please refer to the table below for currently defined variables :
12538
William Lallemandbddd4fd2012-02-27 11:23:10 +010012539 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012540 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012541 +---+------+-----------------------------------------------+-------------+
12542 | | %o | special variable, apply flags on all next var | |
12543 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010012544 | | %B | bytes_read (from server to client) | numeric |
12545 | H | %CC | captured_request_cookie | string |
12546 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020012547 | | %H | hostname | string |
William Lallemanda73203e2012-03-12 12:48:57 +010012548 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020012549 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012550 | | %T | gmt_date_time | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012551 | | %Tc | Tc | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080012552 | | %Tl | local_date_time | date |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012553 | H | %Tq | Tq | numeric |
12554 | H | %Tr | Tr | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012555 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012556 | | %Tt | Tt | numeric |
12557 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012558 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012559 | | %ac | actconn | numeric |
12560 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012561 | | %bc | beconn (backend concurrent connections) | numeric |
12562 | | %bi | backend_source_ip (connecting address) | IP |
12563 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012564 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010012565 | | %ci | client_ip (accepted address) | IP |
12566 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012567 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012568 | | %fc | feconn (frontend concurrent connections) | numeric |
12569 | | %fi | frontend_ip (accepting address) | IP |
12570 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020012571 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020012572 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020012573 | | %hr | captured_request_headers default style | string |
12574 | | %hrl | captured_request_headers CLF style | string list |
12575 | | %hs | captured_response_headers default style | string |
12576 | | %hsl | captured_response_headers CLF style | string list |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012577 | | %ms | accept date milliseconds | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020012578 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012579 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012580 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010012581 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012582 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012583 | | %sc | srv_conn (server concurrent connections) | numeric |
12584 | | %si | server_IP (target address) | IP |
12585 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012586 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012587 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
12588 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010012589 | | %t | date_time (with millisecond resolution) | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012590 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012591 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010012592 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010012593
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020012594 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010012595
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010012596
125978.2.5. Error log format
12598-----------------------
12599
12600When an incoming connection fails due to an SSL handshake or an invalid PROXY
12601protocol header, haproxy will log the event using a shorter, fixed line format.
12602By default, logs are emitted at the LOG_INFO level, unless the option
12603"log-separate-errors" is set in the backend, in which case the LOG_ERR level
12604will be used. Connections on which no data are exchanged (eg: probes) are not
12605logged if the "dontlognull" option is set.
12606
12607The format looks like this :
12608
12609 >>> Dec 3 18:27:14 localhost \
12610 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
12611 Connection error during SSL handshake
12612
12613 Field Format Extract from the example above
12614 1 process_name '[' pid ']:' haproxy[6103]:
12615 2 client_ip ':' client_port 127.0.0.1:56059
12616 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
12617 4 frontend_name "/" bind_name ":" frt/f1:
12618 5 message Connection error during SSL handshake
12619
12620These fields just provide minimal information to help debugging connection
12621failures.
12622
12623
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126248.3. Advanced logging options
12625-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012626
12627Some advanced logging options are often looked for but are not easy to find out
12628just by looking at the various options. Here is an entry point for the few
12629options which can enable better logging. Please refer to the keywords reference
12630for more information about their usage.
12631
12632
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126338.3.1. Disabling logging of external tests
12634------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012635
12636It is quite common to have some monitoring tools perform health checks on
12637haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
12638commercial load-balancer, and sometimes it will simply be a more complete
12639monitoring system such as Nagios. When the tests are very frequent, users often
12640ask how to disable logging for those checks. There are three possibilities :
12641
12642 - if connections come from everywhere and are just TCP probes, it is often
12643 desired to simply disable logging of connections without data exchange, by
12644 setting "option dontlognull" in the frontend. It also disables logging of
12645 port scans, which may or may not be desired.
12646
12647 - if the connection come from a known source network, use "monitor-net" to
12648 declare this network as monitoring only. Any host in this network will then
12649 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012650 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012651 such as other load-balancers.
12652
12653 - if the tests are performed on a known URI, use "monitor-uri" to declare
12654 this URI as dedicated to monitoring. Any host sending this request will
12655 only get the result of a health-check, and the request will not be logged.
12656
12657
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126588.3.2. Logging before waiting for the session to terminate
12659----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012660
12661The problem with logging at end of connection is that you have no clue about
12662what is happening during very long sessions, such as remote terminal sessions
12663or large file downloads. This problem can be worked around by specifying
12664"option logasap" in the frontend. Haproxy will then log as soon as possible,
12665just before data transfer begins. This means that in case of TCP, it will still
12666log the connection status to the server, and in case of HTTP, it will log just
12667after processing the server headers. In this case, the number of bytes reported
12668is the number of header bytes sent to the client. In order to avoid confusion
12669with normal logs, the total time field and the number of bytes are prefixed
12670with a '+' sign which means that real numbers are certainly larger.
12671
12672
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126738.3.3. Raising log level upon errors
12674------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012675
12676Sometimes it is more convenient to separate normal traffic from errors logs,
12677for instance in order to ease error monitoring from log files. When the option
12678"log-separate-errors" is used, connections which experience errors, timeouts,
12679retries, redispatches or HTTP status codes 5xx will see their syslog level
12680raised from "info" to "err". This will help a syslog daemon store the log in
12681a separate file. It is very important to keep the errors in the normal traffic
12682file too, so that log ordering is not altered. You should also be careful if
12683you already have configured your syslog daemon to store all logs higher than
12684"notice" in an "admin" file, because the "err" level is higher than "notice".
12685
12686
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200126878.3.4. Disabling logging of successful connections
12688--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020012689
12690Although this may sound strange at first, some large sites have to deal with
12691multiple thousands of logs per second and are experiencing difficulties keeping
12692them intact for a long time or detecting errors within them. If the option
12693"dontlog-normal" is set on the frontend, all normal connections will not be
12694logged. In this regard, a normal connection is defined as one without any
12695error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
12696and a response with a status 5xx is not considered normal and will be logged
12697too. Of course, doing is is really discouraged as it will remove most of the
12698useful information from the logs. Do this only if you have no other
12699alternative.
12700
12701
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200127028.4. Timing events
12703------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012704
12705Timers provide a great help in troubleshooting network problems. All values are
12706reported in milliseconds (ms). These timers should be used in conjunction with
12707the session termination flags. In TCP mode with "option tcplog" set on the
12708frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
12709mode, 5 control points are reported under the form "Tq/Tw/Tc/Tr/Tt" :
12710
12711 - Tq: total time to get the client request (HTTP mode only). It's the time
12712 elapsed between the moment the client connection was accepted and the
12713 moment the proxy received the last HTTP header. The value "-1" indicates
12714 that the end of headers (empty line) has never been seen. This happens when
12715 the client closes prematurely or times out.
12716
12717 - Tw: total time spent in the queues waiting for a connection slot. It
12718 accounts for backend queue as well as the server queues, and depends on the
12719 queue size, and the time needed for the server to complete previous
12720 requests. The value "-1" means that the request was killed before reaching
12721 the queue, which is generally what happens with invalid or denied requests.
12722
12723 - Tc: total time to establish the TCP connection to the server. It's the time
12724 elapsed between the moment the proxy sent the connection request, and the
12725 moment it was acknowledged by the server, or between the TCP SYN packet and
12726 the matching SYN/ACK packet in return. The value "-1" means that the
12727 connection never established.
12728
12729 - Tr: server response time (HTTP mode only). It's the time elapsed between
12730 the moment the TCP connection was established to the server and the moment
12731 the server sent its complete response headers. It purely shows its request
12732 processing time, without the network overhead due to the data transmission.
12733 It is worth noting that when the client has data to send to the server, for
12734 instance during a POST request, the time already runs, and this can distort
12735 apparent response time. For this reason, it's generally wise not to trust
12736 too much this field for POST requests initiated from clients behind an
12737 untrusted network. A value of "-1" here means that the last the response
12738 header (empty line) was never seen, most likely because the server timeout
12739 stroke before the server managed to process the request.
12740
12741 - Tt: total session duration time, between the moment the proxy accepted it
12742 and the moment both ends were closed. The exception is when the "logasap"
12743 option is specified. In this case, it only equals (Tq+Tw+Tc+Tr), and is
12744 prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012745 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012746
12747 Td = Tt - (Tq + Tw + Tc + Tr)
12748
12749 Timers with "-1" values have to be excluded from this equation. In TCP
12750 mode, "Tq" and "Tr" have to be excluded too. Note that "Tt" can never be
12751 negative.
12752
12753These timers provide precious indications on trouble causes. Since the TCP
12754protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
12755that timers close to multiples of 3s are nearly always related to lost packets
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012756due to network problems (wires, negotiation, congestion). Moreover, if "Tt" is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012757close to a timeout value specified in the configuration, it often means that a
12758session has been aborted on timeout.
12759
12760Most common cases :
12761
12762 - If "Tq" is close to 3000, a packet has probably been lost between the
12763 client and the proxy. This is very rare on local networks but might happen
12764 when clients are on far remote networks and send large requests. It may
12765 happen that values larger than usual appear here without any network cause.
12766 Sometimes, during an attack or just after a resource starvation has ended,
12767 haproxy may accept thousands of connections in a few milliseconds. The time
12768 spent accepting these connections will inevitably slightly delay processing
12769 of other connections, and it can happen that request times in the order of
12770 a few tens of milliseconds are measured after a few thousands of new
Patrick Mezard105faca2010-06-12 17:02:46 +020012771 connections have been accepted at once. Setting "option http-server-close"
12772 may display larger request times since "Tq" also measures the time spent
12773 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012774
12775 - If "Tc" is close to 3000, a packet has probably been lost between the
12776 server and the proxy during the server connection phase. This value should
12777 always be very low, such as 1 ms on local networks and less than a few tens
12778 of ms on remote networks.
12779
Willy Tarreau55165fe2009-05-10 12:02:55 +020012780 - If "Tr" is nearly always lower than 3000 except some rare values which seem
12781 to be the average majored by 3000, there are probably some packets lost
12782 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012783
12784 - If "Tt" is large even for small byte counts, it generally is because
12785 neither the client nor the server decides to close the connection, for
12786 instance because both have agreed on a keep-alive connection mode. In order
12787 to solve this issue, it will be needed to specify "option httpclose" on
12788 either the frontend or the backend. If the problem persists, it means that
12789 the server ignores the "close" connection mode and expects the client to
12790 close. Then it will be required to use "option forceclose". Having the
12791 smallest possible 'Tt' is important when connection regulation is used with
12792 the "maxconn" option on the servers, since no new connection will be sent
12793 to the server until another one is released.
12794
12795Other noticeable HTTP log cases ('xx' means any value to be ignored) :
12796
12797 Tq/Tw/Tc/Tr/+Tt The "option logasap" is present on the frontend and the log
12798 was emitted before the data phase. All the timers are valid
12799 except "Tt" which is shorter than reality.
12800
12801 -1/xx/xx/xx/Tt The client was not able to send a complete request in time
12802 or it aborted too early. Check the session termination flags
12803 then "timeout http-request" and "timeout client" settings.
12804
12805 Tq/-1/xx/xx/Tt It was not possible to process the request, maybe because
12806 servers were out of order, because the request was invalid
12807 or forbidden by ACL rules. Check the session termination
12808 flags.
12809
12810 Tq/Tw/-1/xx/Tt The connection could not establish on the server. Either it
12811 actively refused it or it timed out after Tt-(Tq+Tw) ms.
12812 Check the session termination flags, then check the
12813 "timeout connect" setting. Note that the tarpit action might
12814 return similar-looking patterns, with "Tw" equal to the time
12815 the client connection was maintained open.
12816
12817 Tq/Tw/Tc/-1/Tt The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012818 a complete response in time, or it closed its connection
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012819 unexpectedly after Tt-(Tq+Tw+Tc) ms. Check the session
12820 termination flags, then check the "timeout server" setting.
12821
12822
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200128238.5. Session state at disconnection
12824-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012825
12826TCP and HTTP logs provide a session termination indicator in the
12827"termination_state" field, just before the number of active connections. It is
128282-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
12829each of which has a special meaning :
12830
12831 - On the first character, a code reporting the first event which caused the
12832 session to terminate :
12833
12834 C : the TCP session was unexpectedly aborted by the client.
12835
12836 S : the TCP session was unexpectedly aborted by the server, or the
12837 server explicitly refused it.
12838
12839 P : the session was prematurely aborted by the proxy, because of a
12840 connection limit enforcement, because a DENY filter was matched,
12841 because of a security check which detected and blocked a dangerous
12842 error in server response which might have caused information leak
Willy Tarreau570f2212013-06-10 16:42:09 +020012843 (eg: cacheable cookie).
12844
12845 L : the session was locally processed by haproxy and was not passed to
12846 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012847
12848 R : a resource on the proxy has been exhausted (memory, sockets, source
12849 ports, ...). Usually, this appears during the connection phase, and
12850 system logs should contain a copy of the precise error. If this
12851 happens, it must be considered as a very serious anomaly which
12852 should be fixed as soon as possible by any means.
12853
12854 I : an internal error was identified by the proxy during a self-check.
12855 This should NEVER happen, and you are encouraged to report any log
12856 containing this, because this would almost certainly be a bug. It
12857 would be wise to preventively restart the process after such an
12858 event too, in case it would be caused by memory corruption.
12859
Simon Horman752dc4a2011-06-21 14:34:59 +090012860 D : the session was killed by haproxy because the server was detected
12861 as down and was configured to kill all connections when going down.
12862
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012863 U : the session was killed by haproxy on this backup server because an
12864 active server was detected as up and was configured to kill all
12865 backup connections when going up.
12866
Willy Tarreaua2a64e92011-09-07 23:01:56 +020012867 K : the session was actively killed by an admin operating on haproxy.
12868
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012869 c : the client-side timeout expired while waiting for the client to
12870 send or receive data.
12871
12872 s : the server-side timeout expired while waiting for the server to
12873 send or receive data.
12874
12875 - : normal session completion, both the client and the server closed
12876 with nothing left in the buffers.
12877
12878 - on the second character, the TCP or HTTP session state when it was closed :
12879
Willy Tarreauf7b30a92010-12-06 22:59:17 +010012880 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012881 (HTTP mode only). Nothing was sent to any server.
12882
12883 Q : the proxy was waiting in the QUEUE for a connection slot. This can
12884 only happen when servers have a 'maxconn' parameter set. It can
12885 also happen in the global queue after a redispatch consecutive to
12886 a failed attempt to connect to a dying server. If no redispatch is
12887 reported, then no connection attempt was made to any server.
12888
12889 C : the proxy was waiting for the CONNECTION to establish on the
12890 server. The server might at most have noticed a connection attempt.
12891
12892 H : the proxy was waiting for complete, valid response HEADERS from the
12893 server (HTTP only).
12894
12895 D : the session was in the DATA phase.
12896
12897 L : the proxy was still transmitting LAST data to the client while the
12898 server had already finished. This one is very rare as it can only
12899 happen when the client dies while receiving the last packets.
12900
12901 T : the request was tarpitted. It has been held open with the client
12902 during the whole "timeout tarpit" duration or until the client
12903 closed, both of which will be reported in the "Tw" timer.
12904
12905 - : normal session completion after end of data transfer.
12906
12907 - the third character tells whether the persistence cookie was provided by
12908 the client (only in HTTP mode) :
12909
12910 N : the client provided NO cookie. This is usually the case for new
12911 visitors, so counting the number of occurrences of this flag in the
12912 logs generally indicate a valid trend for the site frequentation.
12913
12914 I : the client provided an INVALID cookie matching no known server.
12915 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020012916 cookies between HTTP/HTTPS sites, persistence conditionally
12917 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012918
12919 D : the client provided a cookie designating a server which was DOWN,
12920 so either "option persist" was used and the client was sent to
12921 this server, or it was not set and the client was redispatched to
12922 another server.
12923
Willy Tarreau996a92c2010-10-13 19:30:47 +020012924 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012925 server.
12926
Willy Tarreau996a92c2010-10-13 19:30:47 +020012927 E : the client provided a valid cookie, but with a last date which was
12928 older than what is allowed by the "maxidle" cookie parameter, so
12929 the cookie is consider EXPIRED and is ignored. The request will be
12930 redispatched just as if there was no cookie.
12931
12932 O : the client provided a valid cookie, but with a first date which was
12933 older than what is allowed by the "maxlife" cookie parameter, so
12934 the cookie is consider too OLD and is ignored. The request will be
12935 redispatched just as if there was no cookie.
12936
Willy Tarreauc89ccb62012-04-05 21:18:22 +020012937 U : a cookie was present but was not used to select the server because
12938 some other server selection mechanism was used instead (typically a
12939 "use-server" rule).
12940
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012941 - : does not apply (no cookie set in configuration).
12942
12943 - the last character reports what operations were performed on the persistence
12944 cookie returned by the server (only in HTTP mode) :
12945
12946 N : NO cookie was provided by the server, and none was inserted either.
12947
12948 I : no cookie was provided by the server, and the proxy INSERTED one.
12949 Note that in "cookie insert" mode, if the server provides a cookie,
12950 it will still be overwritten and reported as "I" here.
12951
Willy Tarreau996a92c2010-10-13 19:30:47 +020012952 U : the proxy UPDATED the last date in the cookie that was presented by
12953 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012954 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020012955 date indicated in the cookie. If any other change happens, such as
12956 a redispatch, then the cookie will be marked as inserted instead.
12957
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012958 P : a cookie was PROVIDED by the server and transmitted as-is.
12959
12960 R : the cookie provided by the server was REWRITTEN by the proxy, which
12961 happens in "cookie rewrite" or "cookie prefix" modes.
12962
12963 D : the cookie provided by the server was DELETED by the proxy.
12964
12965 - : does not apply (no cookie set in configuration).
12966
Willy Tarreau996a92c2010-10-13 19:30:47 +020012967The combination of the two first flags gives a lot of information about what
12968was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012969helpful to detect server saturation, network troubles, local system resource
12970starvation, attacks, etc...
12971
12972The most common termination flags combinations are indicated below. They are
12973alphabetically sorted, with the lowercase set just after the upper case for
12974easier finding and understanding.
12975
12976 Flags Reason
12977
12978 -- Normal termination.
12979
12980 CC The client aborted before the connection could be established to the
12981 server. This can happen when haproxy tries to connect to a recently
12982 dead (or unchecked) server, and the client aborts while haproxy is
12983 waiting for the server to respond or for "timeout connect" to expire.
12984
12985 CD The client unexpectedly aborted during data transfer. This can be
12986 caused by a browser crash, by an intermediate equipment between the
12987 client and haproxy which decided to actively break the connection,
12988 by network routing issues between the client and haproxy, or by a
12989 keep-alive session between the server and the client terminated first
12990 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012991
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012992 cD The client did not send nor acknowledge any data for as long as the
12993 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020012994 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010012995
12996 CH The client aborted while waiting for the server to start responding.
12997 It might be the server taking too long to respond or the client
12998 clicking the 'Stop' button too fast.
12999
13000 cH The "timeout client" stroke while waiting for client data during a
13001 POST request. This is sometimes caused by too large TCP MSS values
13002 for PPPoE networks which cannot transport full-sized packets. It can
13003 also happen when client timeout is smaller than server timeout and
13004 the server takes too long to respond.
13005
13006 CQ The client aborted while its session was queued, waiting for a server
13007 with enough empty slots to accept it. It might be that either all the
13008 servers were saturated or that the assigned server was taking too
13009 long a time to respond.
13010
13011 CR The client aborted before sending a full HTTP request. Most likely
13012 the request was typed by hand using a telnet client, and aborted
13013 too early. The HTTP status code is likely a 400 here. Sometimes this
13014 might also be caused by an IDS killing the connection between haproxy
13015 and the client.
13016
13017 cR The "timeout http-request" stroke before the client sent a full HTTP
13018 request. This is sometimes caused by too large TCP MSS values on the
13019 client side for PPPoE networks which cannot transport full-sized
13020 packets, or by clients sending requests by hand and not typing fast
13021 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020013022 request. The HTTP status code is likely a 408 here. Note: recently,
13023 some browsers such as Google Chrome started to break the deployed Web
13024 infrastructure by aggressively implementing a new "pre-connect"
13025 feature, consisting in sending connections to sites recently visited
13026 without sending any request on them until the user starts to browse
13027 the site. This mechanism causes massive disruption among resource-
13028 limited servers, and causes a lot of 408 errors in HAProxy logs.
13029 Worse, some people report that sometimes the browser displays the 408
13030 error when the user expects to see the actual content (Mozilla fixed
13031 this bug in 2004, while Chrome users continue to report it in 2014),
13032 so in this case, using "errorfile 408 /dev/null" can be used as a
13033 workaround. More information on the subject is available here :
13034 https://bugzilla.mozilla.org/show_bug.cgi?id=248827
13035 https://code.google.com/p/chromium/issues/detail?id=85229
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013036
13037 CT The client aborted while its session was tarpitted. It is important to
13038 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020013039 wrong tarpit rules have been written. If a lot of them happen, it
13040 might make sense to lower the "timeout tarpit" value to something
13041 closer to the average reported "Tw" timer, in order not to consume
13042 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013043
Willy Tarreau570f2212013-06-10 16:42:09 +020013044 LR The request was intercepted and locally handled by haproxy. Generally
13045 it means that this was a redirect or a stats request.
13046
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013047 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013048 the TCP connection (the proxy received a TCP RST or an ICMP message
13049 in return). Under some circumstances, it can also be the network
13050 stack telling the proxy that the server is unreachable (eg: no route,
13051 or no ARP response on local network). When this happens in HTTP mode,
13052 the status code is likely a 502 or 503 here.
13053
13054 sC The "timeout connect" stroke before a connection to the server could
13055 complete. When this happens in HTTP mode, the status code is likely a
13056 503 or 504 here.
13057
13058 SD The connection to the server died with an error during the data
13059 transfer. This usually means that haproxy has received an RST from
13060 the server or an ICMP message from an intermediate equipment while
13061 exchanging data with the server. This can be caused by a server crash
13062 or by a network issue on an intermediate equipment.
13063
13064 sD The server did not send nor acknowledge any data for as long as the
13065 "timeout server" setting during the data phase. This is often caused
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013066 by too short timeouts on L4 equipments before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013067 load-balancers, ...), as well as keep-alive sessions maintained
13068 between the client and the server expiring first on haproxy.
13069
13070 SH The server aborted before sending its full HTTP response headers, or
13071 it crashed while processing the request. Since a server aborting at
13072 this moment is very rare, it would be wise to inspect its logs to
13073 control whether it crashed and why. The logged request may indicate a
13074 small set of faulty requests, demonstrating bugs in the application.
13075 Sometimes this might also be caused by an IDS killing the connection
13076 between haproxy and the server.
13077
13078 sH The "timeout server" stroke before the server could return its
13079 response headers. This is the most common anomaly, indicating too
13080 long transactions, probably caused by server or database saturation.
13081 The immediate workaround consists in increasing the "timeout server"
13082 setting, but it is important to keep in mind that the user experience
13083 will suffer from these long response times. The only long term
13084 solution is to fix the application.
13085
13086 sQ The session spent too much time in queue and has been expired. See
13087 the "timeout queue" and "timeout connect" settings to find out how to
13088 fix this if it happens too often. If it often happens massively in
13089 short periods, it may indicate general problems on the affected
13090 servers due to I/O or database congestion, or saturation caused by
13091 external attacks.
13092
13093 PC The proxy refused to establish a connection to the server because the
13094 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020013095 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013096 so that it does not happen anymore. This status is very rare and
13097 might happen when the global "ulimit-n" parameter is forced by hand.
13098
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010013099 PD The proxy blocked an incorrectly formatted chunked encoded message in
13100 a request or a response, after the server has emitted its headers. In
13101 most cases, this will indicate an invalid message from the server to
Willy Tarreauf3a3e132013-08-31 08:16:26 +020013102 the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647
13103 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010013104
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013105 PH The proxy blocked the server's response, because it was invalid,
13106 incomplete, dangerous (cache control), or matched a security filter.
13107 In any case, an HTTP 502 error is sent to the client. One possible
13108 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010013109 containing unauthorized characters. It is also possible but quite
13110 rare, that the proxy blocked a chunked-encoding request from the
13111 client due to an invalid syntax, before the server responded. In this
13112 case, an HTTP 400 error is sent to the client and reported in the
13113 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013114
13115 PR The proxy blocked the client's HTTP request, either because of an
13116 invalid HTTP syntax, in which case it returned an HTTP 400 error to
13117 the client, or because a deny filter matched, in which case it
13118 returned an HTTP 403 error.
13119
13120 PT The proxy blocked the client's request and has tarpitted its
13121 connection before returning it a 500 server error. Nothing was sent
13122 to the server. The connection was maintained open for as long as
13123 reported by the "Tw" timer field.
13124
13125 RC A local resource has been exhausted (memory, sockets, source ports)
13126 preventing the connection to the server from establishing. The error
13127 logs will tell precisely what was missing. This is very rare and can
13128 only be solved by proper system tuning.
13129
Willy Tarreau996a92c2010-10-13 19:30:47 +020013130The combination of the two last flags gives a lot of information about how
13131persistence was handled by the client, the server and by haproxy. This is very
13132important to troubleshoot disconnections, when users complain they have to
13133re-authenticate. The commonly encountered flags are :
13134
13135 -- Persistence cookie is not enabled.
13136
13137 NN No cookie was provided by the client, none was inserted in the
13138 response. For instance, this can be in insert mode with "postonly"
13139 set on a GET request.
13140
13141 II A cookie designating an invalid server was provided by the client,
13142 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040013143 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020013144 value can be presented by a client when no other server knows it.
13145
13146 NI No cookie was provided by the client, one was inserted in the
13147 response. This typically happens for first requests from every user
13148 in "insert" mode, which makes it an easy way to count real users.
13149
13150 VN A cookie was provided by the client, none was inserted in the
13151 response. This happens for most responses for which the client has
13152 already got a cookie.
13153
13154 VU A cookie was provided by the client, with a last visit date which is
13155 not completely up-to-date, so an updated cookie was provided in
13156 response. This can also happen if there was no date at all, or if
13157 there was a date but the "maxidle" parameter was not set, so that the
13158 cookie can be switched to unlimited time.
13159
13160 EI A cookie was provided by the client, with a last visit date which is
13161 too old for the "maxidle" parameter, so the cookie was ignored and a
13162 new cookie was inserted in the response.
13163
13164 OI A cookie was provided by the client, with a first visit date which is
13165 too old for the "maxlife" parameter, so the cookie was ignored and a
13166 new cookie was inserted in the response.
13167
13168 DI The server designated by the cookie was down, a new server was
13169 selected and a new cookie was emitted in the response.
13170
13171 VI The server designated by the cookie was not marked dead but could not
13172 be reached. A redispatch happened and selected another one, which was
13173 then advertised in the response.
13174
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013175
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131768.6. Non-printable characters
13177-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013178
13179In order not to cause trouble to log analysis tools or terminals during log
13180consulting, non-printable characters are not sent as-is into log files, but are
13181converted to the two-digits hexadecimal representation of their ASCII code,
13182prefixed by the character '#'. The only characters that can be logged without
13183being escaped are comprised between 32 and 126 (inclusive). Obviously, the
13184escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
13185is the same for the character '"' which becomes "#22", as well as '{', '|' and
13186'}' when logging headers.
13187
13188Note that the space character (' ') is not encoded in headers, which can cause
13189issues for tools relying on space count to locate fields. A typical header
13190containing spaces is "User-Agent".
13191
13192Last, it has been observed that some syslog daemons such as syslog-ng escape
13193the quote ('"') with a backslash ('\'). The reverse operation can safely be
13194performed since no quote may appear anywhere else in the logs.
13195
13196
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131978.7. Capturing HTTP cookies
13198---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013199
13200Cookie capture simplifies the tracking a complete user session. This can be
13201achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013202section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013203cookie will simultaneously be checked in the request ("Cookie:" header) and in
13204the response ("Set-Cookie:" header). The respective values will be reported in
13205the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013206locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013207not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
13208user switches to a new session for example, because the server will reassign it
13209a new cookie. It is also possible to detect if a server unexpectedly sets a
13210wrong cookie to a client, leading to session crossing.
13211
13212 Examples :
13213 # capture the first cookie whose name starts with "ASPSESSION"
13214 capture cookie ASPSESSION len 32
13215
13216 # capture the first cookie whose name is exactly "vgnvisitor"
13217 capture cookie vgnvisitor= len 32
13218
13219
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200132208.8. Capturing HTTP headers
13221---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013222
13223Header captures are useful to track unique request identifiers set by an upper
13224proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
13225the response, one can search for information about the response length, how the
13226server asked the cache to behave, or an object location during a redirection.
13227
13228Header captures are performed using the "capture request header" and "capture
13229response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013230section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013231
13232It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013233time. Non-existent headers are logged as empty strings, and if one header
13234appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013235are grouped within braces '{' and '}' in the same order as they were declared,
13236and delimited with a vertical bar '|' without any space. Response headers
13237follow the same representation, but are displayed after a space following the
13238request headers block. These blocks are displayed just before the HTTP request
13239in the logs.
13240
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020013241As a special case, it is possible to specify an HTTP header capture in a TCP
13242frontend. The purpose is to enable logging of headers which will be parsed in
13243an HTTP backend if the request is then switched to this HTTP backend.
13244
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013245 Example :
13246 # This instance chains to the outgoing proxy
13247 listen proxy-out
13248 mode http
13249 option httplog
13250 option logasap
13251 log global
13252 server cache1 192.168.1.1:3128
13253
13254 # log the name of the virtual server
13255 capture request header Host len 20
13256
13257 # log the amount of data uploaded during a POST
13258 capture request header Content-Length len 10
13259
13260 # log the beginning of the referrer
13261 capture request header Referer len 20
13262
13263 # server name (useful for outgoing proxies only)
13264 capture response header Server len 20
13265
13266 # logging the content-length is useful with "option logasap"
13267 capture response header Content-Length len 10
13268
13269 # log the expected cache behaviour on the response
13270 capture response header Cache-Control len 8
13271
13272 # the Via header will report the next proxy's name
13273 capture response header Via len 20
13274
13275 # log the URL location during a redirection
13276 capture response header Location len 20
13277
13278 >>> Aug 9 20:26:09 localhost \
13279 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
13280 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
13281 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
13282 "GET http://fr.adserver.yahoo.com/"
13283
13284 >>> Aug 9 20:30:46 localhost \
13285 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
13286 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
13287 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013288 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013289
13290 >>> Aug 9 20:30:46 localhost \
13291 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
13292 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
13293 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
13294 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013295 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013296
13297
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200132988.9. Examples of logs
13299---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013300
13301These are real-world examples of logs accompanied with an explanation. Some of
13302them have been made up by hand. The syslog part has been removed for better
13303reading. Their sole purpose is to explain how to decipher them.
13304
13305 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
13306 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
13307 "HEAD / HTTP/1.0"
13308
13309 => long request (6.5s) entered by hand through 'telnet'. The server replied
13310 in 147 ms, and the session ended normally ('----')
13311
13312 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
13313 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
13314 0/9 "HEAD / HTTP/1.0"
13315
13316 => Idem, but the request was queued in the global queue behind 9 other
13317 requests, and waited there for 1230 ms.
13318
13319 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
13320 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
13321 "GET /image.iso HTTP/1.0"
13322
13323 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013324 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013325 14 ms, 243 bytes of headers were sent to the client, and total time from
13326 accept to first data byte is 30 ms.
13327
13328 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
13329 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
13330 "GET /cgi-bin/bug.cgi? HTTP/1.0"
13331
13332 => the proxy blocked a server response either because of an "rspdeny" or
13333 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020013334 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013335 risked being cached. In this case, the response is replaced with a "502
13336 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
13337 to return the 502 and not the server.
13338
13339 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013340 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 +010013341
13342 => the client never completed its request and aborted itself ("C---") after
13343 8.5s, while the proxy was waiting for the request headers ("-R--").
13344 Nothing was sent to any server.
13345
13346 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
13347 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
13348
13349 => The client never completed its request, which was aborted by the
13350 time-out ("c---") after 50s, while the proxy was waiting for the request
13351 headers ("-R--"). Nothing was sent to any server, but the proxy could
13352 send a 408 return code to the client.
13353
13354 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
13355 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
13356
13357 => This log was produced with "option tcplog". The client timed out after
13358 5 seconds ("c----").
13359
13360 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
13361 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010013362 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013363
13364 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013365 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010013366 (config says 'retries 3'), and no redispatch (otherwise we would have
13367 seen "/+3"). Status code 503 was returned to the client. There were 115
13368 connections on this server, 202 connections on this proxy, and 205 on
13369 the global process. It is possible that the server refused the
13370 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013371
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013372
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200133739. Statistics and monitoring
13374----------------------------
13375
13376It is possible to query HAProxy about its status. The most commonly used
13377mechanism is the HTTP statistics page. This page also exposes an alternative
13378CSV output format for monitoring tools. The same format is provided on the
13379Unix socket.
13380
13381
133829.1. CSV format
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013383---------------
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +010013384
Willy Tarreau7f062c42009-03-05 18:43:00 +010013385The statistics may be consulted either from the unix socket or from the HTTP
Willy Tarreaua3310dc2014-06-16 15:43:21 +020013386page. Both means provide a CSV format whose fields follow. The first line
13387begins with a sharp ('#') and has one word per comma-delimited field which
13388represents the title of the column. All other lines starting at the second one
13389use a classical CSV format using a comma as the delimiter, and the double quote
13390('"') as an optional text delimiter, but only if the enclosed text is ambiguous
13391(if it contains a quote or a comma). The double-quote character ('"') in the
13392text is doubled ('""'), which is the format that most tools recognize. Please
13393do not insert any column before these ones in order not to break tools which
13394use hard-coded column positions.
Willy Tarreau7f062c42009-03-05 18:43:00 +010013395
James Westbyebe62d62014-07-08 10:14:57 -040013396In brackets after each field name are the types which may have a value for
13397that field. The types are L (Listeners), F (Frontends), B (Backends), and
13398S (Servers).
13399
13400 0. pxname [LFBS]: proxy name
13401 1. svname [LFBS]: service name (FRONTEND for frontend, BACKEND for backend,
13402 any name for server/listener)
13403 2. qcur [..BS]: current queued requests. For the backend this reports the
13404 number queued without a server assigned.
13405 3. qmax [..BS]: max value of qcur
13406 4. scur [LFBS]: current sessions
13407 5. smax [LFBS]: max sessions
13408 6. slim [LFBS]: configured session limit
13409 7. stot [LFBS]: cumulative number of connections
13410 8. bin [LFBS]: bytes in
13411 9. bout [LFBS]: bytes out
13412 10. dreq [LFB.]: requests denied because of security concerns.
13413 - For tcp this is because of a matched tcp-request content rule.
13414 - For http this is because of a matched http-request or tarpit rule.
13415 11. dresp [LFBS]: responses denied because of security concerns.
13416 - For http this is because of a matched http-request rule, or
13417 "option checkcache".
13418 12. ereq [LF..]: request errors. Some of the possible causes are:
13419 - early termination from the client, before the request has been sent.
13420 - read error from the client
13421 - client timeout
13422 - client closed connection
13423 - various bad requests from the client.
13424 - request was tarpitted.
13425 13. econ [..BS]: number of requests that encountered an error trying to
13426 connect to a backend server. The backend stat is the sum of the stat
13427 for all servers of that backend, plus any connection errors not
13428 associated with a particular server (such as the backend having no
13429 active servers).
13430 14. eresp [..BS]: response errors. srv_abrt will be counted here also.
13431 Some other errors are:
13432 - write error on the client socket (won't be counted for the server stat)
13433 - failure applying filters to the response.
13434 15. wretr [..BS]: number of times a connection to a server was retried.
13435 16. wredis [..BS]: number of times a request was redispatched to another
13436 server. The server value counts the number of times that server was
13437 switched away from.
13438 17. status [LFBS]: status (UP/DOWN/NOLB/MAINT/MAINT(via)...)
13439 18. weight [..BS]: server weight (server), total weight (backend)
13440 19. act [..BS]: server is active (server), number of active servers (backend)
13441 20. bck [..BS]: server is backup (server), number of backup servers (backend)
13442 21. chkfail [...S]: number of failed checks. (Only counts checks failed when
13443 the server is up.)
13444 22. chkdown [..BS]: number of UP->DOWN transitions. The backend counter counts
13445 transitions to the whole backend being down, rather than the sum of the
13446 counters for each server.
13447 23. lastchg [..BS]: number of seconds since the last UP<->DOWN transition
13448 24. downtime [..BS]: total downtime (in seconds). The value for the backend
13449 is the downtime for the whole backend, not the sum of the server downtime.
13450 25. qlimit [...S]: configured maxqueue for the server, or nothing in the
13451 value is 0 (default, meaning no limit)
13452 26. pid [LFBS]: process id (0 for first instance, 1 for second, ...)
13453 27. iid [LFBS]: unique proxy id
13454 28. sid [L..S]: server id (unique inside a proxy)
13455 29. throttle [...S]: current throttle percentage for the server, when
13456 slowstart is active, or no value if not in slowstart.
13457 30. lbtot [..BS]: total number of times a server was selected, either for new
13458 sessions, or when re-dispatching. The server counter is the number
13459 of times that server was selected.
13460 31. tracked [...S]: id of proxy/server if tracking is enabled.
13461 32. type [LFBS]: (0=frontend, 1=backend, 2=server, 3=socket/listener)
13462 33. rate [.FBS]: number of sessions per second over last elapsed second
13463 34. rate_lim [.F..]: configured limit on new sessions per second
13464 35. rate_max [.FBS]: max number of new sessions per second
13465 36. check_status [...S]: status of last health check, one of:
Cyril Bontéf0c60612010-02-06 14:44:47 +010013466 UNK -> unknown
13467 INI -> initializing
13468 SOCKERR -> socket error
13469 L4OK -> check passed on layer 4, no upper layers testing enabled
13470 L4TMOUT -> layer 1-4 timeout
13471 L4CON -> layer 1-4 connection problem, for example
13472 "Connection refused" (tcp rst) or "No route to host" (icmp)
13473 L6OK -> check passed on layer 6
13474 L6TOUT -> layer 6 (SSL) timeout
13475 L6RSP -> layer 6 invalid response - protocol error
13476 L7OK -> check passed on layer 7
13477 L7OKC -> check conditionally passed on layer 7, for example 404 with
13478 disable-on-404
13479 L7TOUT -> layer 7 (HTTP/SMTP) timeout
13480 L7RSP -> layer 7 invalid response - protocol error
13481 L7STS -> layer 7 response error, for example HTTP 5xx
James Westbyebe62d62014-07-08 10:14:57 -040013482 37. check_code [...S]: layer5-7 code, if available
13483 38. check_duration [...S]: time in ms took to finish last health check
13484 39. hrsp_1xx [.FBS]: http responses with 1xx code
13485 40. hrsp_2xx [.FBS]: http responses with 2xx code
13486 41. hrsp_3xx [.FBS]: http responses with 3xx code
13487 42. hrsp_4xx [.FBS]: http responses with 4xx code
13488 43. hrsp_5xx [.FBS]: http responses with 5xx code
13489 44. hrsp_other [.FBS]: http responses with other codes (protocol error)
13490 45. hanafail [...S]: failed health checks details
13491 46. req_rate [.F..]: HTTP requests per second over last elapsed second
13492 47. req_rate_max [.F..]: max number of HTTP requests per second observed
13493 48. req_tot [.F..]: total number of HTTP requests received
13494 49. cli_abrt [..BS]: number of data transfers aborted by the client
13495 50. srv_abrt [..BS]: number of data transfers aborted by the server
13496 (inc. in eresp)
13497 51. comp_in [.FB.]: number of HTTP response bytes fed to the compressor
13498 52. comp_out [.FB.]: number of HTTP response bytes emitted by the compressor
13499 53. comp_byp [.FB.]: number of bytes that bypassed the HTTP compressor
13500 (CPU/BW limit)
13501 54. comp_rsp [.FB.]: number of HTTP responses that were compressed
13502 55. lastsess [..BS]: number of seconds since last session assigned to
13503 server/backend
13504 56. last_chk [...S]: last health check contents or textual error
13505 57. last_agt [...S]: last agent check contents or textual error
13506 58. qtime [..BS]: the average queue time in ms over the 1024 last requests
13507 59. ctime [..BS]: the average connect time in ms over the 1024 last requests
13508 60. rtime [..BS]: the average response time in ms over the 1024 last requests
13509 (0 for TCP)
13510 61. ttime [..BS]: the average total session time in ms over the 1024 last
13511 requests
Willy Tarreau844e3c52008-01-11 16:28:18 +010013512
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013513
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200135149.2. Unix Socket commands
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013515-------------------------
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013516
Willy Tarreau468f4932013-08-01 16:50:16 +020013517The stats socket is not enabled by default. In order to enable it, it is
13518necessary to add one line in the global section of the haproxy configuration.
13519A second line is recommended to set a larger timeout, always appreciated when
13520issuing commands by hand :
13521
13522 global
13523 stats socket /var/run/haproxy.sock mode 600 level admin
13524 stats timeout 2m
13525
13526It is also possible to add multiple instances of the stats socket by repeating
13527the line, and make them listen to a TCP port instead of a UNIX socket. This is
13528never done by default because this is dangerous, but can be handy in some
13529situations :
13530
13531 global
13532 stats socket /var/run/haproxy.sock mode 600 level admin
13533 stats socket ipv4@192.168.0.1:9999 level admin
13534 stats timeout 2m
13535
13536To access the socket, an external utility such as "socat" is required. Socat is a
13537swiss-army knife to connect anything to anything. We use it to connect terminals
13538to the socket, or a couple of stdin/stdout pipes to it for scripts. The two main
13539syntaxes we'll use are the following :
13540
13541 # socat /var/run/haproxy.sock stdio
13542 # socat /var/run/haproxy.sock readline
13543
13544The first one is used with scripts. It is possible to send the output of a
13545script to haproxy, and pass haproxy's output to another script. That's useful
13546for retrieving counters or attack traces for example.
13547
13548The second one is only useful for issuing commands by hand. It has the benefit
13549that the terminal is handled by the readline library which supports line
13550editing and history, which is very convenient when issuing repeated commands
13551(eg: watch a counter).
13552
13553The socket supports two operation modes :
13554 - interactive
13555 - non-interactive
13556
13557The non-interactive mode is the default when socat connects to the socket. In
13558this mode, a single line may be sent. It is processed as a whole, responses are
13559sent back, and the connection closes after the end of the response. This is the
13560mode that scripts and monitoring tools use. It is possible to send multiple
13561commands in this mode, they need to be delimited by a semi-colon (';'). For
13562example :
13563
13564 # echo "show info;show stat;show table" | socat /var/run/haproxy stdio
13565
13566The interactive mode displays a prompt ('>') and waits for commands to be
13567entered on the line, then processes them, and displays the prompt again to wait
13568for a new command. This mode is entered via the "prompt" command which must be
13569sent on the first line in non-interactive mode. The mode is a flip switch, if
13570"prompt" is sent in interactive mode, it is disabled and the connection closes
13571after processing the last command of the same line.
13572
13573For this reason, when debugging by hand, it's quite common to start with the
13574"prompt" command :
13575
13576 # socat /var/run/haproxy readline
13577 prompt
13578 > show info
13579 ...
13580 >
13581
13582Since multiple commands may be issued at once, haproxy uses the empty line as a
13583delimiter to mark an end of output for each command, and takes care of ensuring
13584that no command can emit an empty line on output. A script can thus easily
13585parse the output even when multiple commands were pipelined on a single line.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013586
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013587It is important to understand that when multiple haproxy processes are started
13588on the same sockets, any process may pick up the request and will output its
13589own stats.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013590
Willy Tarreau468f4932013-08-01 16:50:16 +020013591The list of commands currently supported on the stats socket is provided below.
13592If an unknown command is sent, haproxy displays the usage message which reminds
13593all supported commands. Some commands support a more complex syntax, generally
13594it will explain what part of the command is invalid when this happens.
13595
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013596add acl <acl> <pattern>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013597 Add an entry into the acl <acl>. <acl> is the #<id> or the <file> returned by
13598 "show acl". This command does not verify if the entry already exists. This
13599 command cannot be used if the reference <acl> is a file also used with a map.
13600 In this case, you must use the command "add map" in place of "add acl".
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013601
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013602add map <map> <key> <value>
13603 Add an entry into the map <map> to associate the value <value> to the key
13604 <key>. This command does not verify if the entry already exists. It is
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013605 mainly used to fill a map after a clear operation. Note that if the reference
13606 <map> is a file and is shared with a map, this map will contain also a new
13607 pattern entry.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013608
Willy Tarreaud63335a2010-02-26 12:56:52 +010013609clear counters
13610 Clear the max values of the statistics counters in each proxy (frontend &
13611 backend) and in each server. The cumulated counters are not affected. This
13612 can be used to get clean counters after an incident, without having to
13613 restart nor to clear traffic counters. This command is restricted and can
13614 only be issued on sockets configured for levels "operator" or "admin".
13615
13616clear counters all
13617 Clear all statistics counters in each proxy (frontend & backend) and in each
13618 server. This has the same effect as restarting. This command is restricted
13619 and can only be issued on sockets configured for level "admin".
13620
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013621clear acl <acl>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013622 Remove all entries from the acl <acl>. <acl> is the #<id> or the <file>
13623 returned by "show acl". Note that if the reference <acl> is a file and is
13624 shared with a map, this map will be also cleared.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013625
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013626clear map <map>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013627 Remove all entries from the map <map>. <map> is the #<id> or the <file>
13628 returned by "show map". Note that if the reference <map> is a file and is
13629 shared with a acl, this acl will be also cleared.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013630
Simon Hormanc88b8872011-06-15 15:18:49 +090013631clear table <table> [ data.<type> <operator> <value> ] | [ key <key> ]
13632 Remove entries from the stick-table <table>.
13633
13634 This is typically used to unblock some users complaining they have been
13635 abusively denied access to a service, but this can also be used to clear some
13636 stickiness entries matching a server that is going to be replaced (see "show
13637 table" below for details). Note that sometimes, removal of an entry will be
13638 refused because it is currently tracked by a session. Retrying a few seconds
13639 later after the session ends is usual enough.
13640
13641 In the case where no options arguments are given all entries will be removed.
13642
13643 When the "data." form is used entries matching a filter applied using the
13644 stored data (see "stick-table" in section 4.2) are removed. A stored data
13645 type must be specified in <type>, and this data type must be stored in the
13646 table otherwise an error is reported. The data is compared according to
13647 <operator> with the 64-bit integer <value>. Operators are the same as with
13648 the ACLs :
13649
13650 - eq : match entries whose data is equal to this value
13651 - ne : match entries whose data is not equal to this value
13652 - le : match entries whose data is less than or equal to this value
13653 - ge : match entries whose data is greater than or equal to this value
13654 - lt : match entries whose data is less than this value
13655 - gt : match entries whose data is greater than this value
13656
13657 When the key form is used the entry <key> is removed. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090013658 same type as the table, which currently is limited to IPv4, IPv6, integer and
13659 string.
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013660
13661 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013662 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013663 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020013664 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
13665 bytes_out_rate(60000)=187
13666 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13667 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013668
13669 $ echo "clear table http_proxy key 127.0.0.1" | socat stdio /tmp/sock1
13670
13671 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Emeric Brun7c6b82e2010-09-24 16:34:28 +020013672 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau62a36c42010-08-17 15:53:10 +020013673 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
13674 bytes_out_rate(60000)=191
Simon Hormanc88b8872011-06-15 15:18:49 +090013675 $ echo "clear table http_proxy data.gpc0 eq 1" | socat stdio /tmp/sock1
13676 $ echo "show table http_proxy" | socat stdio /tmp/sock1
13677 >>> # table: http_proxy, type: ip, size:204800, used:1
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020013678
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013679del acl <acl> [<key>|#<ref>]
13680 Delete all the acl entries from the acl <acl> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013681 <acl> is the #<id> or the <file> returned by "show acl". If the <ref> is used,
13682 this command delete only the listed reference. The reference can be found with
13683 listing the content of the acl. Note that if the reference <acl> is a file and
13684 is shared with a map, the entry will be also deleted in the map.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013685
13686del map <map> [<key>|#<ref>]
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013687 Delete all the map entries from the map <map> corresponding to the key <key>.
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013688 <map> is the #<id> or the <file> returned by "show map". If the <ref> is used,
13689 this command delete only the listed reference. The reference can be found with
13690 listing the content of the map. Note that if the reference <map> is a file and
13691 is shared with a acl, the entry will be also deleted in the map.
Thierry FOURNIER0b90f312014-01-29 20:40:18 +010013692
13693disable agent <backend>/<server>
Simon Horman671b6f02013-11-25 10:46:39 +090013694 Mark the auxiliary agent check as temporarily stopped.
13695
13696 In the case where an agent check is being run as a auxiliary check, due
13697 to the agent-check parameter of a server directive, new checks are only
13698 initialised when the agent is in the enabled. Thus, disable agent will
13699 prevent any new agent checks from begin initiated until the agent
13700 re-enabled using enable agent.
13701
13702 When an agent is disabled the processing of an auxiliary agent check that
13703 was initiated while the agent was set as enabled is as follows: All
13704 results that would alter the weight, specifically "drain" or a weight
13705 returned by the agent, are ignored. The processing of agent check is
13706 otherwise unchanged.
13707
13708 The motivation for this feature is to allow the weight changing effects
13709 of the agent checks to be paused to allow the weight of a server to be
13710 configured using set weight without being overridden by the agent.
13711
13712 This command is restricted and can only be issued on sockets configured for
13713 level "admin".
13714
Willy Tarreau532a4502011-09-07 22:37:44 +020013715disable frontend <frontend>
13716 Mark the frontend as temporarily stopped. This corresponds to the mode which
13717 is used during a soft restart : the frontend releases the port but can be
13718 enabled again if needed. This should be used with care as some non-Linux OSes
13719 are unable to enable it back. This is intended to be used in environments
13720 where stopping a proxy is not even imaginable but a misconfigured proxy must
13721 be fixed. That way it's possible to release the port and bind it into another
13722 process to restore operations. The frontend will appear with status "STOP"
13723 on the stats page.
13724
13725 The frontend may be specified either by its name or by its numeric ID,
13726 prefixed with a sharp ('#').
13727
13728 This command is restricted and can only be issued on sockets configured for
13729 level "admin".
13730
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013731disable health <backend>/<server>
13732 Mark the primary health check as temporarily stopped. This will disable
13733 sending of health checks, and the last health check result will be ignored.
13734 The server will be in unchecked state and considered UP unless an auxiliary
13735 agent check forces it down.
13736
13737 This command is restricted and can only be issued on sockets configured for
13738 level "admin".
13739
Willy Tarreaud63335a2010-02-26 12:56:52 +010013740disable server <backend>/<server>
13741 Mark the server DOWN for maintenance. In this mode, no more checks will be
13742 performed on the server until it leaves maintenance.
13743 If the server is tracked by other servers, those servers will be set to DOWN
13744 during the maintenance.
13745
13746 In the statistics page, a server DOWN for maintenance will appear with a
13747 "MAINT" status, its tracking servers with the "MAINT(via)" one.
13748
13749 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013750 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013751
13752 This command is restricted and can only be issued on sockets configured for
13753 level "admin".
13754
Simon Horman671b6f02013-11-25 10:46:39 +090013755enable agent <backend>/<server>
13756 Resume auxiliary agent check that was temporarily stopped.
13757
13758 See "disable agent" for details of the effect of temporarily starting
13759 and stopping an auxiliary agent.
13760
13761 This command is restricted and can only be issued on sockets configured for
13762 level "admin".
13763
Willy Tarreau532a4502011-09-07 22:37:44 +020013764enable frontend <frontend>
13765 Resume a frontend which was temporarily stopped. It is possible that some of
13766 the listening ports won't be able to bind anymore (eg: if another process
13767 took them since the 'disable frontend' operation). If this happens, an error
13768 is displayed. Some operating systems might not be able to resume a frontend
13769 which was disabled.
13770
13771 The frontend may be specified either by its name or by its numeric ID,
13772 prefixed with a sharp ('#').
13773
13774 This command is restricted and can only be issued on sockets configured for
13775 level "admin".
13776
Willy Tarreau9b5aecd2014-05-23 11:53:10 +020013777enable health <backend>/<server>
13778 Resume a primary health check that was temporarily stopped. This will enable
13779 sending of health checks again. Please see "disable health" for details.
13780
13781 This command is restricted and can only be issued on sockets configured for
13782 level "admin".
13783
Willy Tarreaud63335a2010-02-26 12:56:52 +010013784enable server <backend>/<server>
13785 If the server was previously marked as DOWN for maintenance, this marks the
13786 server UP and checks are re-enabled.
13787
13788 Both the backend and the server may be specified either by their name or by
Willy Tarreauf5f31922011-08-02 11:32:07 +020013789 their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013790
13791 This command is restricted and can only be issued on sockets configured for
13792 level "admin".
13793
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013794get map <map> <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013795get acl <acl> <value>
13796 Lookup the value <value> in the map <map> or in the ACL <acl>. <map> or <acl>
13797 are the #<id> or the <file> returned by "show map" or "show acl". This command
13798 returns all the matching patterns associated with this map. This is useful for
13799 debugging maps and ACLs. The output format is composed by one line par
13800 matching type. Each line is composed by space-delimited series of words.
Thierry FOURNIER5b16df72014-02-26 18:07:38 +010013801
13802 The first two words are:
13803
13804 <match method>: The match method applied. It can be "found", "bool",
13805 "int", "ip", "bin", "len", "str", "beg", "sub", "dir",
13806 "dom", "end" or "reg".
13807
13808 <match result>: The result. Can be "match" or "no-match".
13809
13810 The following words are returned only if the pattern matches an entry.
13811
13812 <index type>: "tree" or "list". The internal lookup algorithm.
13813
13814 <case>: "case-insensitive" or "case-sensitive". The
13815 interpretation of the case.
13816
13817 <entry matched>: match="<entry>". Return the matched pattern. It is
13818 useful with regular expressions.
13819
13820 The two last word are used to show the returned value and its type. With the
13821 "acl" case, the pattern doesn't exist.
13822
13823 return=nothing: No return because there are no "map".
13824 return="<value>": The value returned in the string format.
13825 return=cannot-display: The value cannot be converted as string.
13826
13827 type="<type>": The type of the returned sample.
13828
Willy Tarreaud63335a2010-02-26 12:56:52 +010013829get weight <backend>/<server>
13830 Report the current weight and the initial weight of server <server> in
13831 backend <backend> or an error if either doesn't exist. The initial weight is
13832 the one that appears in the configuration file. Both are normally equal
13833 unless the current weight has been changed. Both the backend and the server
13834 may be specified either by their name or by their numeric ID, prefixed with a
Willy Tarreauf5f31922011-08-02 11:32:07 +020013835 sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013836
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013837help
13838 Print the list of known keywords and their basic usage. The same help screen
13839 is also displayed for unknown commands.
Willy Tarreau3dfe6cd2008-12-07 22:29:48 +010013840
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020013841prompt
13842 Toggle the prompt at the beginning of the line and enter or leave interactive
13843 mode. In interactive mode, the connection is not closed after a command
13844 completes. Instead, the prompt will appear again, indicating the user that
13845 the interpreter is waiting for a new command. The prompt consists in a right
13846 angle bracket followed by a space "> ". This mode is particularly convenient
13847 when one wants to periodically check information such as stats or errors.
13848 It is also a good idea to enter interactive mode before issuing a "help"
13849 command.
13850
13851quit
13852 Close the connection when in interactive mode.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010013853
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010013854set map <map> [<key>|#<ref>] <value>
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010013855 Modify the value corresponding to each key <key> in a map <map>. <map> is the
13856 #<id> or <file> returned by "show map". If the <ref> is used in place of
13857 <key>, only the entry pointed by <ref> is changed. The new value is <value>.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010013858
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013859set maxconn frontend <frontend> <value>
Willy Tarreau3c7a79d2012-09-26 21:07:15 +020013860 Dynamically change the specified frontend's maxconn setting. Any positive
13861 value is allowed including zero, but setting values larger than the global
13862 maxconn does not make much sense. If the limit is increased and connections
13863 were pending, they will immediately be accepted. If it is lowered to a value
13864 below the current number of connections, new connections acceptation will be
Willy Tarreau2a0f4d22011-08-02 11:49:05 +020013865 delayed until the threshold is reached. The frontend might be specified by
13866 either its name or its numeric ID prefixed with a sharp ('#').
13867
Willy Tarreau91886b62011-09-07 14:38:31 +020013868set maxconn global <maxconn>
13869 Dynamically change the global maxconn setting within the range defined by the
13870 initial global maxconn setting. If it is increased and connections were
13871 pending, they will immediately be accepted. If it is lowered to a value below
13872 the current number of connections, new connections acceptation will be
13873 delayed until the threshold is reached. A value of zero restores the initial
13874 setting.
13875
Willy Tarreauf5b22872011-09-07 16:13:44 +020013876set rate-limit connections global <value>
13877 Change the process-wide connection rate limit, which is set by the global
13878 'maxconnrate' setting. A value of zero disables the limitation. This limit
13879 applies to all frontends and the change has an immediate effect. The value
13880 is passed in number of connections per second.
13881
William Lallemandd85f9172012-11-09 17:05:39 +010013882set rate-limit http-compression global <value>
13883 Change the maximum input compression rate, which is set by the global
13884 'maxcomprate' setting. A value of zero disables the limitation. The value is
William Lallemand096f5542012-11-19 17:26:05 +010013885 passed in number of kilobytes per second. The value is available in the "show
13886 info" on the line "CompressBpsRateLim" in bytes.
William Lallemandd85f9172012-11-09 17:05:39 +010013887
Willy Tarreau93e7c002013-10-07 18:51:07 +020013888set rate-limit sessions global <value>
13889 Change the process-wide session rate limit, which is set by the global
13890 'maxsessrate' setting. A value of zero disables the limitation. This limit
13891 applies to all frontends and the change has an immediate effect. The value
13892 is passed in number of sessions per second.
13893
Willy Tarreaue43d5322013-10-07 20:01:52 +020013894set rate-limit ssl-sessions global <value>
13895 Change the process-wide SSL session rate limit, which is set by the global
13896 'maxsslrate' setting. A value of zero disables the limitation. This limit
13897 applies to all frontends and the change has an immediate effect. The value
13898 is passed in number of sessions per second sent to the SSL stack. It applies
13899 before the handshake in order to protect the stack against handshake abuses.
13900
Willy Tarreau2a4b70f2014-05-22 18:42:35 +020013901set server <backend>/<server> agent [ up | down ]
13902 Force a server's agent to a new state. This can be useful to immediately
13903 switch a server's state regardless of some slow agent checks for example.
13904 Note that the change is propagated to tracking servers if any.
13905
13906set server <backend>/<server> health [ up | stopping | down ]
13907 Force a server's health to a new state. This can be useful to immediately
13908 switch a server's state regardless of some slow health checks for example.
13909 Note that the change is propagated to tracking servers if any.
13910
13911set server <backend>/<server> state [ ready | drain | maint ]
13912 Force a server's administrative state to a new state. This can be useful to
13913 disable load balancing and/or any traffic to a server. Setting the state to
13914 "ready" puts the server in normal mode, and the command is the equivalent of
13915 the "enable server" command. Setting the state to "maint" disables any traffic
13916 to the server as well as any health checks. This is the equivalent of the
13917 "disable server" command. Setting the mode to "drain" only removes the server
13918 from load balancing but still allows it to be checked and to accept new
13919 persistent connections. Changes are propagated to tracking servers if any.
13920
13921set server <backend>/<server> weight <weight>[%]
13922 Change a server's weight to the value passed in argument. This is the exact
13923 equivalent of the "set weight" command below.
13924
Emeric Brun4147b2e2014-06-16 18:36:30 +020013925set ssl ocsp-response <response>
13926 This command is used to update an OCSP Response for a certificate (see "crt"
13927 on "bind" lines). Same controls are performed as during the initial loading of
13928 the response. The <response> must be passed as a base64 encoded string of the
13929 DER encoded response from the OCSP server.
13930
13931 Example:
13932 openssl ocsp -issuer issuer.pem -cert server.pem \
13933 -host ocsp.issuer.com:80 -respout resp.der
13934 echo "set ssl ocsp-response $(base64 -w 10000 resp.der)" | \
13935 socat stdio /var/run/haproxy.stat
13936
Willy Tarreau47060b62013-08-01 21:11:42 +020013937set table <table> key <key> [data.<data_type> <value>]*
Willy Tarreau654694e2012-06-07 01:03:16 +020013938 Create or update a stick-table entry in the table. If the key is not present,
13939 an entry is inserted. See stick-table in section 4.2 to find all possible
13940 values for <data_type>. The most likely use consists in dynamically entering
13941 entries for source IP addresses, with a flag in gpc0 to dynamically block an
Willy Tarreau47060b62013-08-01 21:11:42 +020013942 IP address or affect its quality of service. It is possible to pass multiple
13943 data_types in a single call.
Willy Tarreau654694e2012-06-07 01:03:16 +020013944
Willy Tarreaud63335a2010-02-26 12:56:52 +010013945set timeout cli <delay>
13946 Change the CLI interface timeout for current connection. This can be useful
13947 during long debugging sessions where the user needs to constantly inspect
13948 some indicators without being disconnected. The delay is passed in seconds.
13949
13950set weight <backend>/<server> <weight>[%]
13951 Change a server's weight to the value passed in argument. If the value ends
13952 with the '%' sign, then the new weight will be relative to the initially
Simon Horman58b5d292013-02-12 10:45:52 +090013953 configured weight. Absolute weights are permitted between 0 and 256.
13954 Relative weights must be positive with the resulting absolute weight is
13955 capped at 256. Servers which are part of a farm running a static
13956 load-balancing algorithm have stricter limitations because the weight
13957 cannot change once set. Thus for these servers, the only accepted values
13958 are 0 and 100% (or 0 and the initial weight). Changes take effect
13959 immediately, though certain LB algorithms require a certain amount of
13960 requests to consider changes. A typical usage of this command is to
13961 disable a server during an update by setting its weight to zero, then to
13962 enable it again after the update by setting it back to 100%. This command
13963 is restricted and can only be issued on sockets configured for level
13964 "admin". Both the backend and the server may be specified either by their
13965 name or by their numeric ID, prefixed with a sharp ('#').
Willy Tarreaud63335a2010-02-26 12:56:52 +010013966
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013967show errors [<iid>]
13968 Dump last known request and response errors collected by frontends and
13969 backends. If <iid> is specified, the limit the dump to errors concerning
Willy Tarreau6162db22009-10-10 17:13:00 +020013970 either frontend or backend whose ID is <iid>. This command is restricted
13971 and can only be issued on sockets configured for levels "operator" or
13972 "admin".
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010013973
13974 The errors which may be collected are the last request and response errors
13975 caused by protocol violations, often due to invalid characters in header
13976 names. The report precisely indicates what exact character violated the
13977 protocol. Other important information such as the exact date the error was
13978 detected, frontend and backend names, the server name (when known), the
13979 internal session ID and the source address which has initiated the session
13980 are reported too.
13981
13982 All characters are returned, and non-printable characters are encoded. The
13983 most common ones (\t = 9, \n = 10, \r = 13 and \e = 27) are encoded as one
13984 letter following a backslash. The backslash itself is encoded as '\\' to
13985 avoid confusion. Other non-printable characters are encoded '\xNN' where
13986 NN is the two-digits hexadecimal representation of the character's ASCII
13987 code.
13988
13989 Lines are prefixed with the position of their first character, starting at 0
13990 for the beginning of the buffer. At most one input line is printed per line,
13991 and large lines will be broken into multiple consecutive output lines so that
13992 the output never goes beyond 79 characters wide. It is easy to detect if a
13993 line was broken, because it will not end with '\n' and the next line's offset
13994 will be followed by a '+' sign, indicating it is a continuation of previous
13995 line.
13996
13997 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020013998 $ echo "show errors" | socat stdio /tmp/sock1
13999 >>> [04/Mar/2009:15:46:56.081] backend http-in (#2) : invalid response
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010014000 src 127.0.0.1, session #54, frontend fe-eth0 (#1), server s2 (#1)
14001 response length 213 bytes, error at position 23:
14002
14003 00000 HTTP/1.0 200 OK\r\n
14004 00017 header/bizarre:blah\r\n
14005 00038 Location: blah\r\n
14006 00054 Long-line: this is a very long line which should b
14007 00104+ e broken into multiple lines on the output buffer,
14008 00154+ otherwise it would be too large to print in a ter
14009 00204+ minal\r\n
14010 00211 \r\n
14011
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014012 In the example above, we see that the backend "http-in" which has internal
Willy Tarreaue0c8a1a2009-03-04 16:33:10 +010014013 ID 2 has blocked an invalid response from its server s2 which has internal
14014 ID 1. The request was on session 54 initiated by source 127.0.0.1 and
14015 received by frontend fe-eth0 whose ID is 1. The total response length was
14016 213 bytes when the error was detected, and the error was at byte 23. This
14017 is the slash ('/') in header name "header/bizarre", which is not a valid
14018 HTTP character for a header name.
Krzysztof Piotr Oledzki2c6962c2008-03-02 02:42:14 +010014019
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020014020show info
14021 Dump info about haproxy status on current process.
14022
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010014023show map [<map>]
14024 Dump info about map converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010014025 maps is returned. If a <map> is specified, its contents are dumped. <map> is
14026 the #<id> or <file>. The first column is a unique identifier. It can be used
14027 as reference for the operation "del map" and "set map". The second column is
14028 the pattern and the third column is the sample if available. The data returned
14029 are not directly a list of available maps, but are the list of all patterns
14030 composing any map. Many of these patterns can be shared with ACL.
Thierry FOURNIERd32079e2014-01-29 20:02:04 +010014031
14032show acl [<acl>]
14033 Dump info about acl converters. Without argument, the list of all available
Thierry FOURNIER65ce6132014-03-20 11:42:45 +010014034 acls is returned. If a <acl> is specified, its contents are dumped. <acl> if
14035 the #<id> or <file>. The dump format is the same than the map even for the
14036 sample value. The data returned are not a list of available ACL, but are the
14037 list of all patterns composing any ACL. Many of these patterns can be shared
14038 with maps.
Thierry FOURNIERc0e0d7b2013-12-11 16:55:52 +010014039
Willy Tarreau12833bb2014-01-28 16:49:56 +010014040show pools
14041 Dump the status of internal memory pools. This is useful to track memory
14042 usage when suspecting a memory leak for example. It does exactly the same
14043 as the SIGQUIT when running in foreground except that it does not flush
14044 the pools.
14045
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020014046show sess
14047 Dump all known sessions. Avoid doing this on slow connections as this can
Willy Tarreau6162db22009-10-10 17:13:00 +020014048 be huge. This command is restricted and can only be issued on sockets
14049 configured for levels "operator" or "admin".
14050
Willy Tarreau66dc20a2010-03-05 17:53:32 +010014051show sess <id>
14052 Display a lot of internal information about the specified session identifier.
14053 This identifier is the first field at the beginning of the lines in the dumps
14054 of "show sess" (it corresponds to the session pointer). Those information are
14055 useless to most users but may be used by haproxy developers to troubleshoot a
14056 complex bug. The output format is intentionally not documented so that it can
Willy Tarreau76153662012-11-26 01:16:39 +010014057 freely evolve depending on demands. The special id "all" dumps the states of
14058 all sessions, which can be avoided as much as possible as it is highly CPU
14059 intensive and can take a lot of time.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020014060
14061show stat [<iid> <type> <sid>]
14062 Dump statistics in the CSV format. By passing <id>, <type> and <sid>, it is
14063 possible to dump only selected items :
14064 - <iid> is a proxy ID, -1 to dump everything
14065 - <type> selects the type of dumpable objects : 1 for frontends, 2 for
14066 backends, 4 for servers, -1 for everything. These values can be ORed,
14067 for example:
14068 1 + 2 = 3 -> frontend + backend.
14069 1 + 2 + 4 = 7 -> frontend + backend + server.
14070 - <sid> is a server ID, -1 to dump everything from the selected proxy.
14071
14072 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020014073 $ echo "show info;show stat" | socat stdio unix-connect:/tmp/sock1
14074 >>> Name: HAProxy
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020014075 Version: 1.4-dev2-49
14076 Release_date: 2009/09/23
14077 Nbproc: 1
14078 Process_num: 1
14079 (...)
14080
14081 # pxname,svname,qcur,qmax,scur,smax,slim,stot,bin,bout,dreq, (...)
14082 stats,FRONTEND,,,0,0,1000,0,0,0,0,0,0,,,,,OPEN,,,,,,,,,1,1,0, (...)
14083 stats,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,0,0,0,,0,250,(...)
14084 (...)
14085 www1,BACKEND,0,0,0,0,1000,0,0,0,0,0,,0,0,0,0,UP,1,1,0,,0,250, (...)
14086
14087 $
14088
14089 Here, two commands have been issued at once. That way it's easy to find
14090 which process the stats apply to in multi-process mode. Notice the empty
14091 line after the information output which marks the end of the first block.
14092 A similar empty line appears at the end of the second block (stats) so that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014093 the reader knows the output has not been truncated.
Willy Tarreau9a42c0d2009-09-22 19:31:03 +020014094
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014095show table
14096 Dump general information on all known stick-tables. Their name is returned
14097 (the name of the proxy which holds them), their type (currently zero, always
14098 IP), their size in maximum possible number of entries, and the number of
14099 entries currently in use.
14100
14101 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020014102 $ echo "show table" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090014103 >>> # table: front_pub, type: ip, size:204800, used:171454
14104 >>> # table: back_rdp, type: ip, size:204800, used:0
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014105
Simon Horman17bce342011-06-15 15:18:47 +090014106show table <name> [ data.<type> <operator> <value> ] | [ key <key> ]
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014107 Dump contents of stick-table <name>. In this mode, a first line of generic
14108 information about the table is reported as with "show table", then all
14109 entries are dumped. Since this can be quite heavy, it is possible to specify
Simon Horman17bce342011-06-15 15:18:47 +090014110 a filter in order to specify what entries to display.
14111
14112 When the "data." form is used the filter applies to the stored data (see
14113 "stick-table" in section 4.2). A stored data type must be specified
14114 in <type>, and this data type must be stored in the table otherwise an
14115 error is reported. The data is compared according to <operator> with the
14116 64-bit integer <value>. Operators are the same as with the ACLs :
14117
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014118 - eq : match entries whose data is equal to this value
14119 - ne : match entries whose data is not equal to this value
14120 - le : match entries whose data is less than or equal to this value
14121 - ge : match entries whose data is greater than or equal to this value
14122 - lt : match entries whose data is less than this value
14123 - gt : match entries whose data is greater than this value
14124
Simon Hormanc88b8872011-06-15 15:18:49 +090014125
14126 When the key form is used the entry <key> is shown. The key must be of the
Simon Horman619e3cc2011-06-15 15:18:52 +090014127 same type as the table, which currently is limited to IPv4, IPv6, integer,
14128 and string.
Simon Horman17bce342011-06-15 15:18:47 +090014129
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014130 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020014131 $ echo "show table http_proxy" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090014132 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020014133 >>> 0x80e6a4c: key=127.0.0.1 use=0 exp=3594729 gpc0=0 conn_rate(30000)=1 \
14134 bytes_out_rate(60000)=187
14135 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
14136 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014137
Willy Tarreau62a36c42010-08-17 15:53:10 +020014138 $ echo "show table http_proxy data.gpc0 gt 0" | socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090014139 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020014140 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
14141 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014142
Willy Tarreau62a36c42010-08-17 15:53:10 +020014143 $ echo "show table http_proxy data.conn_rate gt 5" | \
14144 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090014145 >>> # table: http_proxy, type: ip, size:204800, used:2
Willy Tarreau62a36c42010-08-17 15:53:10 +020014146 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
14147 bytes_out_rate(60000)=191
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014148
Simon Horman17bce342011-06-15 15:18:47 +090014149 $ echo "show table http_proxy key 127.0.0.2" | \
14150 socat stdio /tmp/sock1
Simon Horman64b28d02011-08-13 08:03:50 +090014151 >>> # table: http_proxy, type: ip, size:204800, used:2
Simon Horman17bce342011-06-15 15:18:47 +090014152 >>> 0x80e6a80: key=127.0.0.2 use=0 exp=3594740 gpc0=1 conn_rate(30000)=10 \
14153 bytes_out_rate(60000)=191
14154
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014155 When the data criterion applies to a dynamic value dependent on time such as
14156 a bytes rate, the value is dynamically computed during the evaluation of the
14157 entry in order to decide whether it has to be dumped or not. This means that
14158 such a filter could match for some time then not match anymore because as
14159 time goes, the average event rate drops.
14160
14161 It is possible to use this to extract lists of IP addresses abusing the
14162 service, in order to monitor them or even blacklist them in a firewall.
14163 Example :
Willy Tarreau62a36c42010-08-17 15:53:10 +020014164 $ echo "show table http_proxy data.gpc0 gt 0" \
14165 | socat stdio /tmp/sock1 \
Willy Tarreau88bc4ec2010-08-01 07:58:48 +020014166 | fgrep 'key=' | cut -d' ' -f2 | cut -d= -f2 > abusers-ip.txt
14167 ( or | awk '/key/{ print a[split($2,a,"=")]; }' )
Krzysztof Piotr Oledzki719e7262009-10-04 15:02:46 +020014168
Willy Tarreau532a4502011-09-07 22:37:44 +020014169shutdown frontend <frontend>
14170 Completely delete the specified frontend. All the ports it was bound to will
14171 be released. It will not be possible to enable the frontend anymore after
14172 this operation. This is intended to be used in environments where stopping a
14173 proxy is not even imaginable but a misconfigured proxy must be fixed. That
14174 way it's possible to release the port and bind it into another process to
14175 restore operations. The frontend will not appear at all on the stats page
14176 once it is terminated.
14177
14178 The frontend may be specified either by its name or by its numeric ID,
14179 prefixed with a sharp ('#').
14180
14181 This command is restricted and can only be issued on sockets configured for
14182 level "admin".
14183
Willy Tarreaua295edc2011-09-07 23:21:03 +020014184shutdown session <id>
14185 Immediately terminate the session matching the specified session identifier.
14186 This identifier is the first field at the beginning of the lines in the dumps
14187 of "show sess" (it corresponds to the session pointer). This can be used to
14188 terminate a long-running session without waiting for a timeout or when an
14189 endless transfer is ongoing. Such terminated sessions are reported with a 'K'
14190 flag in the logs.
14191
Cyril Bontée63a1eb2014-07-12 18:22:42 +020014192shutdown sessions server <backend>/<server>
Willy Tarreau52b2d222011-09-07 23:48:48 +020014193 Immediately terminate all the sessions attached to the specified server. This
14194 can be used to terminate long-running sessions after a server is put into
14195 maintenance mode, for instance. Such terminated sessions are reported with a
14196 'K' flag in the logs.
14197
Willy Tarreau0ba27502007-12-24 16:55:16 +010014198/*
14199 * Local variables:
14200 * fill-column: 79
14201 * End:
14202 */