blob: c3aa6fa8439d62558fa439f191f9c3630a0bcb35 [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 Tarreau0e658fb2016-11-25 16:55:50 +01005 version 1.8
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreauf57a29a2017-06-02 15:59:51 +02007 2017/06/02
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
11specified above. It does not provide any hint, example or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013The summary below is meant to help you search sections by name and navigate
14through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
22 sometimes useful to prefix all output lines (logs, console outs) with 3
23 closing angle brackets ('>>>') in order to help get the difference between
24 inputs and outputs when it can become ambiguous. If you add sections,
25 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
341.2.1. The Request line
351.2.2. The request headers
361.3. HTTP response
371.3.1. The Response line
381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
Willy Tarreauc57f0e22009-05-10 13:12:33 +020054
554. Proxies
564.1. Proxy keywords matrix
574.2. Alphabetically sorted keywords reference
58
Willy Tarreau086fbf52012-09-24 20:34:51 +0200595. Bind and Server options
605.1. Bind options
615.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200625.3. Server DNS resolution
635.3.1. Global overview
645.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020065
666. HTTP header manipulation
67
Willy Tarreau74ca5042013-06-11 23:12:07 +0200687. Using ACLs and fetching samples
697.1. ACL basics
707.1.1. Matching booleans
717.1.2. Matching integers
727.1.3. Matching strings
737.1.4. Matching regular expressions (regexes)
747.1.5. Matching arbitrary data blocks
757.1.6. Matching IPv4 and IPv6 addresses
767.2. Using ACLs to form conditions
777.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200787.3.1. Converters
797.3.2. Fetching samples from internal states
807.3.3. Fetching samples at Layer 4
817.3.4. Fetching samples at Layer 5
827.3.5. Fetching samples from buffer contents (Layer 6)
837.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200847.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020085
868. Logging
878.1. Log levels
888.2. Log formats
898.2.1. Default log format
908.2.2. TCP log format
918.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100928.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100938.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200948.3. Advanced logging options
958.3.1. Disabling logging of external tests
968.3.2. Logging before waiting for the session to terminate
978.3.3. Raising log level upon errors
988.3.4. Disabling logging of successful connections
998.4. Timing events
1008.5. Session state at disconnection
1018.6. Non-printable characters
1028.7. Capturing HTTP cookies
1038.8. Capturing HTTP headers
1048.9. Examples of logs
105
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001069. Supported filters
1079.1. Trace
1089.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001099.3. Stream Processing Offload Engine (SPOE)
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200110
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200111
1121. Quick reminder about HTTP
113----------------------------
114
115When haproxy is running in HTTP mode, both the request and the response are
116fully analyzed and indexed, thus it becomes possible to build matching criteria
117on almost anything found in the contents.
118
119However, it is important to understand how HTTP requests and responses are
120formed, and how HAProxy decomposes them. It will then become easier to write
121correct rules and to debug existing configurations.
122
123
1241.1. The HTTP transaction model
125-------------------------------
126
127The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100128to one and only one response. Traditionally, a TCP connection is established
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200129from the client to the server, a request is sent by the client on the
130connection, the server responds and the connection is closed. A new request
131will involve a new connection :
132
133 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
134
135In this mode, called the "HTTP close" mode, there are as many connection
136establishments as there are HTTP transactions. Since the connection is closed
137by the server after the response, the client does not need to know the content
138length.
139
140Due to the transactional nature of the protocol, it was possible to improve it
141to avoid closing a connection between two subsequent transactions. In this mode
142however, it is mandatory that the server indicates the content length for each
143response so that the client does not wait indefinitely. For this, a special
144header is used: "Content-length". This mode is called the "keep-alive" mode :
145
146 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
147
148Its advantages are a reduced latency between transactions, and less processing
149power required on the server side. It is generally better than the close mode,
150but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200151a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200152
153A last improvement in the communications is the pipelining mode. It still uses
154keep-alive, but the client does not wait for the first response to send the
155second request. This is useful for fetching large number of images composing a
156page :
157
158 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
159
160This can obviously have a tremendous benefit on performance because the network
161latency is eliminated between subsequent requests. Many HTTP agents do not
162correctly support pipelining since there is no way to associate a response with
163the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100164server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165
Willy Tarreau70dffda2014-01-30 03:07:23 +0100166By default HAProxy operates in keep-alive mode with regards to persistent
167connections: for each connection it processes each request and response, and
168leaves the connection idle on both sides between the end of a response and the
169start of a new request.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200170
Willy Tarreau70dffda2014-01-30 03:07:23 +0100171HAProxy supports 5 connection modes :
172 - keep alive : all requests and responses are processed (default)
173 - tunnel : only the first request and response are processed,
174 everything else is forwarded with no analysis.
175 - passive close : tunnel with "Connection: close" added in both directions.
176 - server close : the server-facing connection is closed after the response.
177 - forced close : the connection is actively closed after end of response.
178
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200179
1801.2. HTTP request
181-----------------
182
183First, let's consider this HTTP request :
184
185 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100186 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200187 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
188 2 Host: www.mydomain.com
189 3 User-agent: my small browser
190 4 Accept: image/jpeg, image/gif
191 5 Accept: image/png
192
193
1941.2.1. The Request line
195-----------------------
196
197Line 1 is the "request line". It is always composed of 3 fields :
198
199 - a METHOD : GET
200 - a URI : /serv/login.php?lang=en&profile=2
201 - a version tag : HTTP/1.1
202
203All of them are delimited by what the standard calls LWS (linear white spaces),
204which are commonly spaces, but can also be tabs or line feeds/carriage returns
205followed by spaces/tabs. The method itself cannot contain any colon (':') and
206is limited to alphabetic letters. All those various combinations make it
207desirable that HAProxy performs the splitting itself rather than leaving it to
208the user to write a complex or inaccurate regular expression.
209
210The URI itself can have several forms :
211
212 - A "relative URI" :
213
214 /serv/login.php?lang=en&profile=2
215
216 It is a complete URL without the host part. This is generally what is
217 received by servers, reverse proxies and transparent proxies.
218
219 - An "absolute URI", also called a "URL" :
220
221 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
222
223 It is composed of a "scheme" (the protocol name followed by '://'), a host
224 name or address, optionally a colon (':') followed by a port number, then
225 a relative URI beginning at the first slash ('/') after the address part.
226 This is generally what proxies receive, but a server supporting HTTP/1.1
227 must accept this form too.
228
229 - a star ('*') : this form is only accepted in association with the OPTIONS
230 method and is not relayable. It is used to inquiry a next hop's
231 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100232
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200233 - an address:port combination : 192.168.0.12:80
234 This is used with the CONNECT method, which is used to establish TCP
235 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
236 other protocols too.
237
238In a relative URI, two sub-parts are identified. The part before the question
239mark is called the "path". It is typically the relative path to static objects
240on the server. The part after the question mark is called the "query string".
241It is mostly used with GET requests sent to dynamic scripts and is very
242specific to the language, framework or application in use.
243
244
2451.2.2. The request headers
246--------------------------
247
248The headers start at the second line. They are composed of a name at the
249beginning of the line, immediately followed by a colon (':'). Traditionally,
250an LWS is added after the colon but that's not required. Then come the values.
251Multiple identical headers may be folded into one single line, delimiting the
252values with commas, provided that their order is respected. This is commonly
253encountered in the "Cookie:" field. A header may span over multiple lines if
254the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
255define a total of 3 values for the "Accept:" header.
256
257Contrary to a common mis-conception, header names are not case-sensitive, and
258their values are not either if they refer to other header names (such as the
259"Connection:" header).
260
261The end of the headers is indicated by the first empty line. People often say
262that it's a double line feed, which is not exact, even if a double line feed
263is one valid form of empty line.
264
265Fortunately, HAProxy takes care of all these complex combinations when indexing
266headers, checking values and counting them, so there is no reason to worry
267about the way they could be written, but it is important not to accuse an
268application of being buggy if it does unusual, valid things.
269
270Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000271 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200272 in the middle of headers by LWS in order to join multi-line headers. This
273 is necessary for proper analysis and helps less capable HTTP parsers to work
274 correctly and not to be fooled by such complex constructs.
275
276
2771.3. HTTP response
278------------------
279
280An HTTP response looks very much like an HTTP request. Both are called HTTP
281messages. Let's consider this HTTP response :
282
283 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100284 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200285 1 HTTP/1.1 200 OK
286 2 Content-length: 350
287 3 Content-Type: text/html
288
Willy Tarreau816b9792009-09-15 21:25:21 +0200289As a special case, HTTP supports so called "Informational responses" as status
290codes 1xx. These messages are special in that they don't convey any part of the
291response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100292continue to post its request for instance. In the case of a status 100 response
293the requested information will be carried by the next non-100 response message
294following the informational one. This implies that multiple responses may be
295sent to a single request, and that this only works when keep-alive is enabled
296(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
297correctly forward and skip them, and only process the next non-100 response. As
298such, these messages are neither logged nor transformed, unless explicitly
299state otherwise. Status 101 messages indicate that the protocol is changing
300over the same connection and that haproxy must switch to tunnel mode, just as
301if a CONNECT had occurred. Then the Upgrade header would contain additional
302information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200303
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200304
3051.3.1. The Response line
306------------------------
307
308Line 1 is the "response line". It is always composed of 3 fields :
309
310 - a version tag : HTTP/1.1
311 - a status code : 200
312 - a reason : OK
313
314The status code is always 3-digit. The first digit indicates a general status :
Willy Tarreau816b9792009-09-15 21:25:21 +0200315 - 1xx = informational message to be skipped (eg: 100, 101)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200316 - 2xx = OK, content is following (eg: 200, 206)
317 - 3xx = OK, no content following (eg: 302, 304)
318 - 4xx = error caused by the client (eg: 401, 403, 404)
319 - 5xx = error caused by the server (eg: 500, 502, 503)
320
Lukas Tribus23953682017-04-28 13:24:30 +0000321Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100322"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200323found there, but it's a common practice to respect the well-established
324messages. It can be composed of one or multiple words, such as "OK", "Found",
325or "Authentication Required".
326
327Haproxy may emit the following status codes by itself :
328
329 Code When / reason
330 200 access to stats page, and when replying to monitoring requests
331 301 when performing a redirection, depending on the configured code
332 302 when performing a redirection, depending on the configured code
333 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100334 307 when performing a redirection, depending on the configured code
335 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200336 400 for an invalid or too large request
337 401 when an authentication is required to perform the action (when
338 accessing the stats page)
339 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
340 408 when the request timeout strikes before the request is complete
341 500 when haproxy encounters an unrecoverable internal error, such as a
342 memory allocation failure, which should never happen
343 502 when the server returns an empty, invalid or incomplete response, or
344 when an "rspdeny" filter blocks the response.
345 503 when no server was available to handle the request, or in response to
346 monitoring requests which match the "monitor fail" condition
347 504 when the response timeout strikes before the server responds
348
349The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3504.2).
351
352
3531.3.2. The response headers
354---------------------------
355
356Response headers work exactly like request headers, and as such, HAProxy uses
357the same parsing function for both. Please refer to paragraph 1.2.2 for more
358details.
359
360
3612. Configuring HAProxy
362----------------------
363
3642.1. Configuration file format
365------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200366
367HAProxy's configuration process involves 3 major sources of parameters :
368
369 - the arguments from the command-line, which always take precedence
370 - the "global" section, which sets process-wide parameters
371 - the proxies sections which can take form of "defaults", "listen",
372 "frontend" and "backend".
373
Willy Tarreau0ba27502007-12-24 16:55:16 +0100374The configuration file syntax consists in lines beginning with a keyword
375referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200376delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100377
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200378
William Lallemandf9873ba2015-05-05 17:37:14 +02003792.2. Quoting and escaping
380-------------------------
381
382HAProxy's configuration introduces a quoting and escaping system similar to
383many programming languages. The configuration file supports 3 types: escaping
384with a backslash, weak quoting with double quotes, and strong quoting with
385single quotes.
386
387If spaces have to be entered in strings, then they must be escaped by preceding
388them by a backslash ('\') or by quoting them. Backslashes also have to be
389escaped by doubling or strong quoting them.
390
391Escaping is achieved by preceding a special character by a backslash ('\'):
392
393 \ to mark a space and differentiate it from a delimiter
394 \# to mark a hash and differentiate it from a comment
395 \\ to use a backslash
396 \' to use a single quote and differentiate it from strong quoting
397 \" to use a double quote and differentiate it from weak quoting
398
399Weak quoting is achieved by using double quotes (""). Weak quoting prevents
400the interpretation of:
401
402 space as a parameter separator
403 ' single quote as a strong quoting delimiter
404 # hash as a comment start
405
William Lallemandb2f07452015-05-12 14:27:13 +0200406Weak quoting permits the interpretation of variables, if you want to use a non
407-interpreted dollar within a double quoted string, you should escape it with a
408backslash ("\$"), it does not work outside weak quoting.
409
410Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200411quoting.
412
413Strong quoting is achieved by using single quotes (''). Inside single quotes,
414nothing is interpreted, it's the efficient way to quote regexes.
415
416Quoted and escaped strings are replaced in memory by their interpreted
417equivalent, it allows you to perform concatenation.
418
419 Example:
420 # those are equivalents:
421 log-format %{+Q}o\ %t\ %s\ %{-Q}r
422 log-format "%{+Q}o %t %s %{-Q}r"
423 log-format '%{+Q}o %t %s %{-Q}r'
424 log-format "%{+Q}o %t"' %s %{-Q}r'
425 log-format "%{+Q}o %t"' %s'\ %{-Q}r
426
427 # those are equivalents:
428 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
429 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
430 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
431 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
432
433
William Lallemandb2f07452015-05-12 14:27:13 +02004342.3. Environment variables
435--------------------------
436
437HAProxy's configuration supports environment variables. Those variables are
438interpreted only within double quotes. Variables are expanded during the
439configuration parsing. Variable names must be preceded by a dollar ("$") and
440optionally enclosed with braces ("{}") similarly to what is done in Bourne
441shell. Variable names can contain alphanumerical characters or the character
442underscore ("_") but should not start with a digit.
443
444 Example:
445
446 bind "fd@${FD_APP1}"
447
448 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
449
450 user "$HAPROXY_USER"
451
452
4532.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200454----------------
455
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100456Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100457values are generally expressed in milliseconds (unless explicitly stated
458otherwise) but may be expressed in any other unit by suffixing the unit to the
459numeric value. It is important to consider this because it will not be repeated
460for every keyword. Supported units are :
461
462 - us : microseconds. 1 microsecond = 1/1000000 second
463 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
464 - s : seconds. 1s = 1000ms
465 - m : minutes. 1m = 60s = 60000ms
466 - h : hours. 1h = 60m = 3600s = 3600000ms
467 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
468
469
Lukas Tribusaa83a312017-03-21 09:25:09 +00004702.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200471-------------
472
473 # Simple configuration for an HTTP proxy listening on port 80 on all
474 # interfaces and forwarding requests to a single backend "servers" with a
475 # single server "server1" listening on 127.0.0.1:8000
476 global
477 daemon
478 maxconn 256
479
480 defaults
481 mode http
482 timeout connect 5000ms
483 timeout client 50000ms
484 timeout server 50000ms
485
486 frontend http-in
487 bind *:80
488 default_backend servers
489
490 backend servers
491 server server1 127.0.0.1:8000 maxconn 32
492
493
494 # The same configuration defined with a single listen block. Shorter but
495 # less expressive, especially in HTTP mode.
496 global
497 daemon
498 maxconn 256
499
500 defaults
501 mode http
502 timeout connect 5000ms
503 timeout client 50000ms
504 timeout server 50000ms
505
506 listen http-in
507 bind *:80
508 server server1 127.0.0.1:8000 maxconn 32
509
510
511Assuming haproxy is in $PATH, test these configurations in a shell with:
512
Willy Tarreauccb289d2010-12-11 20:19:38 +0100513 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200514
515
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005163. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200517--------------------
518
519Parameters in the "global" section are process-wide and often OS-specific. They
520are generally set once for all and do not need being changed once correct. Some
521of them have command-line equivalents.
522
523The following keywords are supported in the "global" section :
524
525 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200526 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200527 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200528 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200529 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200530 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200531 - description
532 - deviceatlas-json-file
533 - deviceatlas-log-level
534 - deviceatlas-separator
535 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900536 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200537 - gid
538 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100539 - hard-stop-after
Willy Tarreau6a06a402007-07-15 20:15:28 +0200540 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200541 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100542 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200543 - lua-load
Willy Tarreau6a06a402007-07-15 20:15:28 +0200544 - nbproc
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200545 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200546 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100547 - presetenv
548 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200549 - uid
550 - ulimit-n
551 - user
Willy Tarreau1d549722016-02-16 12:41:57 +0100552 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200553 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200554 - ssl-default-bind-ciphers
555 - ssl-default-bind-options
556 - ssl-default-server-ciphers
557 - ssl-default-server-options
558 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100559 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100560 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100561 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100562 - 51degrees-data-file
563 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200564 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200565 - 51degrees-cache-size
scientiamobiled0027ed2016-11-04 10:55:08 +0100566 - wurfl-data-file
567 - wurfl-information-list
568 - wurfl-information-list-separator
569 - wurfl-engine-mode
570 - wurfl-cache-size
571 - wurfl-useragent-priority
Willy Tarreaud72758d2010-01-12 10:42:19 +0100572
Willy Tarreau6a06a402007-07-15 20:15:28 +0200573 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200574 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200575 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200576 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100577 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100578 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100579 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200580 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200581 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200582 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200583 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200584 - noepoll
585 - nokqueue
586 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100587 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300588 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000589 - noreuseport
Willy Tarreaufe255b72007-10-14 23:09:26 +0200590 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200591 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200592 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000593 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000594 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200595 - tune.buffers.limit
596 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200597 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200598 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100599 - tune.comp.maxlevel
Willy Tarreau193b8c62012-11-22 00:17:38 +0100600 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200601 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200602 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100603 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100604 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100605 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100606 - tune.lua.session-timeout
607 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200608 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100609 - tune.maxaccept
610 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200611 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200612 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200613 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100614 - tune.rcvbuf.client
615 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100616 - tune.recv_enough
Willy Tarreaue803de22010-01-21 17:43:04 +0100617 - tune.sndbuf.client
618 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100619 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100620 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200621 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100622 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200623 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200624 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100625 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200626 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100627 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200628 - tune.vars.reqres-max-size
629 - tune.vars.sess-max-size
630 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100631 - tune.zlib.memlevel
632 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100633
Willy Tarreau6a06a402007-07-15 20:15:28 +0200634 * Debugging
635 - debug
636 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200637
638
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006393.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200640------------------------------------
641
Emeric Brunc8e8d122012-10-02 18:42:10 +0200642ca-base <dir>
643 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200644 relative path is used with "ca-file" or "crl-file" directives. Absolute
645 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200646
Willy Tarreau6a06a402007-07-15 20:15:28 +0200647chroot <jail dir>
648 Changes current directory to <jail dir> and performs a chroot() there before
649 dropping privileges. This increases the security level in case an unknown
650 vulnerability would be exploited, since it would make it very hard for the
651 attacker to exploit the system. This only works when the process is started
652 with superuser privileges. It is important to ensure that <jail_dir> is both
653 empty and unwritable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100654
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100655cpu-map <"all"|"odd"|"even"|process_num> <cpu-set>...
656 On Linux 2.6 and above, it is possible to bind a process to a specific CPU
657 set. This means that the process will never run on other CPUs. The "cpu-map"
658 directive specifies CPU sets for process sets. The first argument is the
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100659 process number to bind. This process must have a number between 1 and 32 or
660 64, depending on the machine's word size, and any process IDs above nbproc
661 are ignored. It is possible to specify all processes at once using "all",
662 only odd numbers using "odd" or even numbers using "even", just like with the
663 "bind-process" directive. The second and forthcoming arguments are CPU sets.
664 Each CPU set is either a unique number between 0 and 31 or 63 or a range with
665 two such numbers delimited by a dash ('-'). Multiple CPU numbers or ranges
666 may be specified, and the processes will be allowed to bind to all of them.
667 Obviously, multiple "cpu-map" directives may be specified. Each "cpu-map"
668 directive will replace the previous ones when they overlap.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100669
Emeric Brunc8e8d122012-10-02 18:42:10 +0200670crt-base <dir>
671 Assigns a default directory to fetch SSL certificates from when a relative
672 path is used with "crtfile" directives. Absolute locations specified after
673 "crtfile" prevail and ignore "crt-base".
674
Willy Tarreau6a06a402007-07-15 20:15:28 +0200675daemon
676 Makes the process fork into background. This is the recommended mode of
677 operation. It is equivalent to the command line "-D" argument. It can be
678 disabled by the command line "-db" argument.
679
David Carlier8167f302015-06-01 13:50:06 +0200680deviceatlas-json-file <path>
681 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
682 The path must be a valid JSON data file and accessible by Haproxy process.
683
684deviceatlas-log-level <value>
685 Sets the level of informations returned by the API. This directive is
686 optional and set to 0 by default if not set.
687
688deviceatlas-separator <char>
689 Sets the character separator for the API properties results. This directive
690 is optional and set to | by default if not set.
691
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100692deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200693 Sets the client cookie's name used for the detection if the DeviceAtlas
694 Client-side component was used during the request. This directive is optional
695 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100696
Simon Horman98637e52014-06-20 12:30:16 +0900697external-check
698 Allows the use of an external agent to perform health checks.
699 This is disabled by default as a security precaution.
700 See "option external-check".
701
Willy Tarreau6a06a402007-07-15 20:15:28 +0200702gid <number>
703 Changes the process' group ID to <number>. It is recommended that the group
704 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
705 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100706 Note that if haproxy is started from a user having supplementary groups, it
707 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200708 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100709
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100710hard-stop-after <time>
711 Defines the maximum time allowed to perform a clean soft-stop.
712
713 Arguments :
714 <time> is the maximum time (by default in milliseconds) for which the
715 instance will remain alive when a soft-stop is received via the
716 SIGUSR1 signal.
717
718 This may be used to ensure that the instance will quit even if connections
719 remain opened during a soft-stop (for example with long timeouts for a proxy
720 in tcp mode). It applies both in TCP and HTTP mode.
721
722 Example:
723 global
724 hard-stop-after 30s
725
Willy Tarreau6a06a402007-07-15 20:15:28 +0200726group <group name>
727 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
728 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100729
Dragan Dosen7ad31542015-09-28 17:16:47 +0200730log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Willy Tarreau6a06a402007-07-15 20:15:28 +0200731 Adds a global syslog server. Up to two global servers can be defined. They
732 will receive logs for startups and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100733 configured with "log global".
734
735 <address> can be one of:
736
Willy Tarreau2769aa02007-12-27 18:26:09 +0100737 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100738 no port is specified, 514 is used by default (the standard syslog
739 port).
740
David du Colombier24bb5f52011-03-17 10:40:23 +0100741 - An IPv6 address followed by a colon and optionally a UDP port. If
742 no port is specified, 514 is used by default (the standard syslog
743 port).
744
Robert Tsai81ae1952007-12-05 10:47:29 +0100745 - A filesystem path to a UNIX domain socket, keeping in mind
746 considerations for chroot (be sure the path is accessible inside
747 the chroot) and uid/gid (be sure the path is appropriately
748 writeable).
749
William Lallemandb2f07452015-05-12 14:27:13 +0200750 You may want to reference some environment variables in the address
751 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100752
Willy Tarreau18324f52014-06-27 18:10:07 +0200753 <length> is an optional maximum line length. Log lines larger than this value
754 will be truncated before being sent. The reason is that syslog
755 servers act differently on log line length. All servers support the
756 default value of 1024, but some servers simply drop larger lines
757 while others do log them. If a server supports long lines, it may
758 make sense to set this value here in order to avoid truncating long
759 lines. Similarly, if a server drops long lines, it is preferable to
760 truncate them before sending them. Accepted values are 80 to 65535
761 inclusive. The default value of 1024 is generally fine for all
762 standard usages. Some specific cases of long captures or
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200763 JSON-formated logs may require larger values. You may also need to
764 increase "tune.http.logurilen" if your request uris are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200765
Dragan Dosen7ad31542015-09-28 17:16:47 +0200766 <format> is the log format used when generating syslog messages. It may be
767 one of the following :
768
769 rfc3164 The RFC3164 syslog message format. This is the default.
770 (https://tools.ietf.org/html/rfc3164)
771
772 rfc5424 The RFC5424 syslog message format.
773 (https://tools.ietf.org/html/rfc5424)
774
Robert Tsai81ae1952007-12-05 10:47:29 +0100775 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200776
777 kern user mail daemon auth syslog lpr news
778 uucp cron auth2 ftp ntp audit alert cron2
779 local0 local1 local2 local3 local4 local5 local6 local7
780
781 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200782 all messages are sent. If a maximum level is specified, only messages with a
783 severity at least as important as this level will be sent. An optional minimum
784 level can be specified. If it is set, logs emitted with a more severe level
785 than this one will be capped to this level. This is used to avoid sending
786 "emerg" messages on all terminals on some default syslog configurations.
787 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200788
Cyril Bontédc4d9032012-04-08 21:57:39 +0200789 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200790
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100791log-send-hostname [<string>]
792 Sets the hostname field in the syslog header. If optional "string" parameter
793 is set the header is set to the string contents, otherwise uses the hostname
794 of the system. Generally used if one is not relaying logs through an
795 intermediate syslog server or for simply customizing the hostname printed in
796 the logs.
797
Kevinm48936af2010-12-22 16:08:21 +0000798log-tag <string>
799 Sets the tag field in the syslog header to this string. It defaults to the
800 program name as launched from the command line, which usually is "haproxy".
801 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100802 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000803
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100804lua-load <file>
805 This global directive loads and executes a Lua file. This directive can be
806 used multiple times.
807
William Lallemande202b1e2017-06-01 17:38:56 +0200808master-worker [exit-on-failure]
809 Master-worker mode. It is equivalent to the command line "-W" argument.
810 This mode will launch a "master" which will monitor the "workers". Using
811 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
812 the master. The master-worker mode is compatible either with the foreground
813 or daemon mode. It is recommended to use this mode with multiprocess and
814 systemd.
815 The "exit-on-failure" option allows the master to kill every workers and
816 exit when one of the current workers died. It is convenient to combine this
817 option with Restart=on-failure in a systemd unit file in order to relaunch
818 the whole process.
819
820 See alors "-W" in the management guide.
821
Willy Tarreau6a06a402007-07-15 20:15:28 +0200822nbproc <number>
823 Creates <number> processes when going daemon. This requires the "daemon"
824 mode. By default, only one process is created, which is the recommended mode
825 of operation. For systems limited to small sets of file descriptors per
826 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
827 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
828
829pidfile <pidfile>
830 Writes pids of all daemons into file <pidfile>. This option is equivalent to
831 the "-p" command line argument. The file must be accessible to the user
832 starting the process. See also "daemon".
833
Willy Tarreau1d549722016-02-16 12:41:57 +0100834presetenv <name> <value>
835 Sets environment variable <name> to value <value>. If the variable exists, it
836 is NOT overwritten. The changes immediately take effect so that the next line
837 in the configuration file sees the new value. See also "setenv", "resetenv",
838 and "unsetenv".
839
840resetenv [<name> ...]
841 Removes all environment variables except the ones specified in argument. It
842 allows to use a clean controlled environment before setting new values with
843 setenv or unsetenv. Please note that some internal functions may make use of
844 some environment variables, such as time manipulation functions, but also
845 OpenSSL or even external checks. This must be used with extreme care and only
846 after complete validation. The changes immediately take effect so that the
847 next line in the configuration file sees the new environment. See also
848 "setenv", "presetenv", and "unsetenv".
849
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100850stats bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200851 Limits the stats socket to a certain set of processes numbers. By default the
852 stats socket is bound to all processes, causing a warning to be emitted when
853 nbproc is greater than 1 because there is no way to select the target process
854 when connecting. However, by using this setting, it becomes possible to pin
855 the stats socket to a specific set of processes, typically the first one. The
856 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100857 the number of processes used. The maximum process ID depends on the machine's
Willy Tarreauae302532014-05-07 19:22:24 +0200858 word size (32 or 64). A better option consists in using the "process" setting
859 of the "stats socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +0200860
Baptiste Assmann5626f482015-08-23 10:00:10 +0200861server-state-base <directory>
862 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200863 file names which do not start with a '/'. See also "server-state-file",
864 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200865
866server-state-file <file>
867 Specifies the path to the file containing state of servers. If the path starts
868 with a slash ('/'), it is considered absolute, otherwise it is considered
869 relative to the directory specified using "server-state-base" (if set) or to
870 the current directory. Before reloading HAProxy, it is possible to save the
871 servers' current state using the stats command "show servers state". The
872 output of this command must be written in the file pointed by <file>. When
873 starting up, before handling traffic, HAProxy will read, load and apply state
874 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200875 configuration. See also "server-state-base" and "show servers state",
876 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +0200877
Willy Tarreau1d549722016-02-16 12:41:57 +0100878setenv <name> <value>
879 Sets environment variable <name> to value <value>. If the variable exists, it
880 is overwritten. The changes immediately take effect so that the next line in
881 the configuration file sees the new value. See also "presetenv", "resetenv",
882 and "unsetenv".
883
Willy Tarreau610f04b2014-02-13 11:36:41 +0100884ssl-default-bind-ciphers <ciphers>
885 This setting is only available when support for OpenSSL was built in. It sets
886 the default string describing the list of cipher algorithms ("cipher suite")
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300887 that are negotiated during the SSL/TLS handshake for all "bind" lines which
Willy Tarreau610f04b2014-02-13 11:36:41 +0100888 do not explicitly define theirs. The format of the string is defined in
889 "man 1 ciphers" from OpenSSL man pages, and can be for instance a string such
890 as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). Please check the
891 "bind" keyword for more information.
892
Emeric Brun2c86cbf2014-10-30 15:56:50 +0100893ssl-default-bind-options [<option>]...
894 This setting is only available when support for OpenSSL was built in. It sets
895 default ssl-options to force on all "bind" lines. Please check the "bind"
896 keyword to see available options.
897
898 Example:
899 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +0200900 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +0100901
Willy Tarreau610f04b2014-02-13 11:36:41 +0100902ssl-default-server-ciphers <ciphers>
903 This setting is only available when support for OpenSSL was built in. It
904 sets the default string describing the list of cipher algorithms that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300905 negotiated during the SSL/TLS handshake with the server, for all "server"
Willy Tarreau610f04b2014-02-13 11:36:41 +0100906 lines which do not explicitly define theirs. The format of the string is
907 defined in "man 1 ciphers". Please check the "server" keyword for more
908 information.
909
Emeric Brun2c86cbf2014-10-30 15:56:50 +0100910ssl-default-server-options [<option>]...
911 This setting is only available when support for OpenSSL was built in. It sets
912 default ssl-options to force on all "server" lines. Please check the "server"
913 keyword to see available options.
914
Remi Gacogne47783ef2015-05-29 15:53:22 +0200915ssl-dh-param-file <file>
916 This setting is only available when support for OpenSSL was built in. It sets
917 the default DH parameters that are used during the SSL/TLS handshake when
918 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
919 which do not explicitely define theirs. It will be overridden by custom DH
920 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200921 are not specified either by using ssl-dh-param-file or by setting them
922 directly in the certificate file, pre-generated DH parameters of the size
923 specified by tune.ssl.default-dh-param will be used. Custom parameters are
924 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +0200925 Custom DH parameters may be generated by using the OpenSSL command
926 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
927 parameters should not be considered secure anymore.
928
Emeric Brun850efd52014-01-29 12:24:34 +0100929ssl-server-verify [none|required]
930 The default behavior for SSL verify on servers side. If specified to 'none',
931 servers certificates are not verified. The default is 'required' except if
932 forced using cmdline option '-dV'.
933
Willy Tarreauabb175f2012-09-24 12:43:26 +0200934stats socket [<address:port>|<path>] [param*]
935 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
936 Connections to this socket will return various statistics outputs and even
937 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +0200938 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +0200939 details.
Willy Tarreau6162db22009-10-10 17:13:00 +0200940
Willy Tarreauabb175f2012-09-24 12:43:26 +0200941 All parameters supported by "bind" lines are supported, for instance to
942 restrict access to some users or their access rights. Please consult
943 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200944
945stats timeout <timeout, in milliseconds>
946 The default timeout on the stats socket is set to 10 seconds. It is possible
947 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +0100948 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +0200949
950stats maxconn <connections>
951 By default, the stats socket is limited to 10 concurrent connections. It is
952 possible to change this value with "stats maxconn".
953
Willy Tarreau6a06a402007-07-15 20:15:28 +0200954uid <number>
955 Changes the process' user ID to <number>. It is recommended that the user ID
956 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
957 be started with superuser privileges in order to be able to switch to another
958 one. See also "gid" and "user".
959
960ulimit-n <number>
961 Sets the maximum number of per-process file-descriptors to <number>. By
962 default, it is automatically computed, so it is recommended not to use this
963 option.
964
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100965unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
966 [ group <group> ] [ gid <gid> ]
967
968 Fixes common settings to UNIX listening sockets declared in "bind" statements.
969 This is mainly used to simplify declaration of those UNIX sockets and reduce
970 the risk of errors, since those settings are most commonly required but are
971 also process-specific. The <prefix> setting can be used to force all socket
972 path to be relative to that directory. This might be needed to access another
973 component's chroot. Note that those paths are resolved before haproxy chroots
974 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
975 all have the same meaning as their homonyms used by the "bind" statement. If
976 both are specified, the "bind" statement has priority, meaning that the
977 "unix-bind" settings may be seen as process-wide default settings.
978
Willy Tarreau1d549722016-02-16 12:41:57 +0100979unsetenv [<name> ...]
980 Removes environment variables specified in arguments. This can be useful to
981 hide some sensitive information that are occasionally inherited from the
982 user's environment during some operations. Variables which did not exist are
983 silently ignored so that after the operation, it is certain that none of
984 these variables remain. The changes immediately take effect so that the next
985 line in the configuration file will not see these variables. See also
986 "setenv", "presetenv", and "resetenv".
987
Willy Tarreau6a06a402007-07-15 20:15:28 +0200988user <user name>
989 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
990 See also "uid" and "group".
991
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +0200992node <name>
993 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
994
995 This statement is useful in HA configurations where two or more processes or
996 servers share the same IP address. By setting a different node-name on all
997 nodes, it becomes easy to immediately spot what server is handling the
998 traffic.
999
1000description <text>
1001 Add a text that describes the instance.
1002
1003 Please note that it is required to escape certain characters (# for example)
1004 and this text is inserted into a html page so you should avoid using
1005 "<" and ">" characters.
1006
Thomas Holmesdb04f192015-05-18 13:21:39 +0100100751degrees-data-file <file path>
1008 The path of the 51Degrees data file to provide device detection services. The
1009 file should be unzipped and accessible by HAProxy with relevavnt permissions.
1010
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001011 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001012 compiled with USE_51DEGREES.
1013
Ben Shillitof25e8e52016-12-02 14:25:37 +0000101451degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001015 A list of 51Degrees property names to be load from the dataset. A full list
1016 of names is available on the 51Degrees website:
1017 https://51degrees.com/resources/property-dictionary
1018
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001019 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001020 compiled with USE_51DEGREES.
1021
Dragan Dosen93b38d92015-06-29 16:43:25 +0200102251degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001023 A char that will be appended to every property value in a response header
1024 containing 51Degrees results. If not set that will be set as ','.
1025
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001026 Please note that this option is only available when haproxy has been
1027 compiled with USE_51DEGREES.
1028
102951degrees-cache-size <number>
1030 Sets the size of the 51Degrees converter cache to <number> entries. This
1031 is an LRU cache which reminds previous device detections and their results.
1032 By default, this cache is disabled.
1033
1034 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001035 compiled with USE_51DEGREES.
1036
scientiamobiled0027ed2016-11-04 10:55:08 +01001037wurfl-data-file <file path>
1038 The path of the WURFL data file to provide device detection services. The
1039 file should be accessible by HAProxy with relevant permissions.
1040
1041 Please note that this option is only available when haproxy has been compiled
1042 with USE_WURFL=1.
1043
1044wurfl-information-list [<capability>]*
1045 A space-delimited list of WURFL capabilities, virtual capabilities, property
1046 names we plan to use in injected headers. A full list of capability and
1047 virtual capability names is available on the Scientiamobile website :
1048
1049 https://www.scientiamobile.com/wurflCapability
1050
1051 Valid WURFL properties are:
1052 - wurfl_id Contains the device ID of the matched device.
1053
1054 - wurfl_root_id Contains the device root ID of the matched
1055 device.
1056
1057 - wurfl_isdevroot Tells if the matched device is a root device.
1058 Possible values are "TRUE" or "FALSE".
1059
1060 - wurfl_useragent The original useragent coming with this
1061 particular web request.
1062
1063 - wurfl_api_version Contains a string representing the currently
1064 used Libwurfl API version.
1065
1066 - wurfl_engine_target Contains a string representing the currently
1067 set WURFL Engine Target. Possible values are
1068 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1069
1070 - wurfl_info A string containing information on the parsed
1071 wurfl.xml and its full path.
1072
1073 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1074 WURFL has been loaded successfully.
1075
1076 - wurfl_normalized_useragent The normalized useragent.
1077
1078 - wurfl_useragent_priority The user agent priority used by WURFL.
1079
1080 Please note that this option is only available when haproxy has been compiled
1081 with USE_WURFL=1.
1082
1083wurfl-information-list-separator <char>
1084 A char that will be used to separate values in a response header containing
1085 WURFL results. If not set that a comma (',') will be used by default.
1086
1087 Please note that this option is only available when haproxy has been compiled
1088 with USE_WURFL=1.
1089
1090wurfl-patch-file [<file path>]
1091 A list of WURFL patch file paths. Note that patches are loaded during startup
1092 thus before the chroot.
1093
1094 Please note that this option is only available when haproxy has been compiled
1095 with USE_WURFL=1.
1096
1097wurfl-engine-mode { accuracy | performance }
1098 Sets the WURFL engine target. You can choose between 'accuracy' or
1099 'performance' targets. In performance mode, desktop web browser detection is
1100 done programmatically without referencing the WURFL data. As a result, most
1101 desktop web browsers are returned as generic_web_browser WURFL ID for
1102 performance. If either performance or accuracy are not defined, performance
1103 mode is enabled by default.
1104
1105 Please note that this option is only available when haproxy has been compiled
1106 with USE_WURFL=1.
1107
1108wurfl-cache-size <U>[,<D>]
1109 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1110 <D> is the internal device cache size. There are three possibilities here :
1111 - "0" : no cache is used.
1112 - <U> : the Single LRU cache is used, the size is expressed in elements.
1113 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1114 the highest performing option.
1115
1116 Please note that this option is only available when haproxy has been compiled
1117 with USE_WURFL=1.
1118
1119wurfl-useragent-priority { plain | sideloaded_browser }
1120 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1121 over the default sideloaded browser user agent ('sideloaded_browser').
1122
1123 Please note that this option is only available when haproxy has been compiled
1124 with USE_WURFL=1.
1125
Willy Tarreau6a06a402007-07-15 20:15:28 +02001126
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011273.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001128-----------------------
1129
Willy Tarreau1746eec2014-04-25 10:46:47 +02001130max-spread-checks <delay in milliseconds>
1131 By default, haproxy tries to spread the start of health checks across the
1132 smallest health check interval of all the servers in a farm. The principle is
1133 to avoid hammering services running on the same server. But when using large
1134 check intervals (10 seconds or more), the last servers in the farm take some
1135 time before starting to be tested, which can be a problem. This parameter is
1136 used to enforce an upper bound on delay between the first and the last check,
1137 even if the servers' check intervals are larger. When servers run with
1138 shorter intervals, their intervals will be respected though.
1139
Willy Tarreau6a06a402007-07-15 20:15:28 +02001140maxconn <number>
1141 Sets the maximum per-process number of concurrent connections to <number>. It
1142 is equivalent to the command-line argument "-n". Proxies will stop accepting
1143 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001144 automatically adjusted according to this value. See also "ulimit-n". Note:
1145 the "select" poller cannot reliably use more than 1024 file descriptors on
1146 some platforms. If your platform only supports select and reports "select
1147 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001148 below 500 in general). If this value is not set, it will default to the value
1149 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1150 limit is enforced, or will be computed based on the memory limit, the buffer
1151 size, memory allocated to compression, SSL cache size, and use or not of SSL
1152 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001153
Willy Tarreau81c25d02011-09-07 15:17:21 +02001154maxconnrate <number>
1155 Sets the maximum per-process number of connections per second to <number>.
1156 Proxies will stop accepting connections when this limit is reached. It can be
1157 used to limit the global capacity regardless of each frontend capacity. It is
1158 important to note that this can only be used as a service protection measure,
1159 as there will not necessarily be a fair share between frontends when the
1160 limit is reached, so it's a good idea to also limit each frontend to some
1161 value close to its expected share. Also, lowering tune.maxaccept can improve
1162 fairness.
1163
William Lallemandd85f9172012-11-09 17:05:39 +01001164maxcomprate <number>
1165 Sets the maximum per-process input compression rate to <number> kilobytes
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001166 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001167 level will be decreased during the session. If the maximum is reached at the
1168 beginning of a session, the session will not compress at all. If the maximum
1169 is not reached, the compression level will be increased up to
1170 tune.comp.maxlevel. A value of zero means there is no limit, this is the
1171 default value.
1172
William Lallemand072a2bf2012-11-20 17:01:01 +01001173maxcompcpuusage <number>
1174 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1175 for new requests or decreasing the compression level of current requests.
1176 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1177 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1178 case of multiple processes (nbproc > 1), each process manages its individual
1179 usage. A value of 100 disable the limit. The default value is 100. Setting
1180 a lower value will prevent the compression work from slowing the whole
1181 process down and from introducing high latencies.
1182
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001183maxpipes <number>
1184 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1185 are only used by kernel-based tcp splicing. Since a pipe contains two file
1186 descriptors, the "ulimit-n" value will be increased accordingly. The default
1187 value is maxconn/4, which seems to be more than enough for most heavy usages.
1188 The splice code dynamically allocates and releases pipes, and can fall back
1189 to standard copy, so setting this value too low may only impact performance.
1190
Willy Tarreau93e7c002013-10-07 18:51:07 +02001191maxsessrate <number>
1192 Sets the maximum per-process number of sessions per second to <number>.
1193 Proxies will stop accepting connections when this limit is reached. It can be
1194 used to limit the global capacity regardless of each frontend capacity. It is
1195 important to note that this can only be used as a service protection measure,
1196 as there will not necessarily be a fair share between frontends when the
1197 limit is reached, so it's a good idea to also limit each frontend to some
1198 value close to its expected share. Also, lowering tune.maxaccept can improve
1199 fairness.
1200
Willy Tarreau403edff2012-09-06 11:58:37 +02001201maxsslconn <number>
1202 Sets the maximum per-process number of concurrent SSL connections to
1203 <number>. By default there is no SSL-specific limit, which means that the
1204 global maxconn setting will apply to all connections. Setting this limit
1205 avoids having openssl use too much memory and crash when malloc returns NULL
1206 (since it unfortunately does not reliably check for such conditions). Note
1207 that the limit applies both to incoming and outgoing connections, so one
1208 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001209 If this value is not set, but a memory limit is enforced, this value will be
1210 automatically computed based on the memory limit, maxconn, the buffer size,
1211 memory allocated to compression, SSL cache size, and use of SSL in either
1212 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1213 when there is a memory limit, haproxy will automatically adjust these values
1214 so that 100% of the connections can be made over SSL with no risk, and will
1215 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001216
Willy Tarreaue43d5322013-10-07 20:01:52 +02001217maxsslrate <number>
1218 Sets the maximum per-process number of SSL sessions per second to <number>.
1219 SSL listeners will stop accepting connections when this limit is reached. It
1220 can be used to limit the global SSL CPU usage regardless of each frontend
1221 capacity. It is important to note that this can only be used as a service
1222 protection measure, as there will not necessarily be a fair share between
1223 frontends when the limit is reached, so it's a good idea to also limit each
1224 frontend to some value close to its expected share. It is also important to
1225 note that the sessions are accounted before they enter the SSL stack and not
1226 after, which also protects the stack against bad handshakes. Also, lowering
1227 tune.maxaccept can improve fairness.
1228
William Lallemand9d5f5482012-11-07 16:12:57 +01001229maxzlibmem <number>
1230 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1231 When the maximum amount is reached, future sessions will not compress as long
1232 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001233 The default value is 0. The value is available in bytes on the UNIX socket
1234 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1235 "ZlibMemUsage" in bytes.
1236
Willy Tarreau6a06a402007-07-15 20:15:28 +02001237noepoll
1238 Disables the use of the "epoll" event polling system on Linux. It is
1239 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001240 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001241
1242nokqueue
1243 Disables the use of the "kqueue" event polling system on BSD. It is
1244 equivalent to the command-line argument "-dk". The next polling system
1245 used will generally be "poll". See also "nopoll".
1246
1247nopoll
1248 Disables the use of the "poll" event polling system. It is equivalent to the
1249 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001250 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001251 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001252
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001253nosplice
1254 Disables the use of kernel tcp splicing between sockets on Linux. It is
1255 equivalent to the command line argument "-dS". Data will then be copied
1256 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001257 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001258 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1259 be used. This option makes it easier to globally disable kernel splicing in
1260 case of doubt. See also "option splice-auto", "option splice-request" and
1261 "option splice-response".
1262
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001263nogetaddrinfo
1264 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1265 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1266
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001267noreuseport
1268 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1269 command line argument "-dR".
1270
Willy Tarreaufe255b72007-10-14 23:09:26 +02001271spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001272 Sometimes it is desirable to avoid sending agent and health checks to
1273 servers at exact intervals, for instance when many logical servers are
1274 located on the same physical server. With the help of this parameter, it
1275 becomes possible to add some randomness in the check interval between 0
1276 and +/- 50%. A value between 2 and 5 seems to show good results. The
1277 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001278
Grant Zhang872f9c22017-01-21 01:10:18 +00001279ssl-engine <name> [algo <comma-seperated list of algorithms>]
1280 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
1281 obtained using the command "openssl engine". This statement may be used
1282 multiple times, it will simply enable multiple crypto engines. Referencing an
1283 unsupported engine will prevent haproxy from starting. Note that many engines
1284 will lead to lower HTTPS performance than pure software with recent
1285 processors. The optional command "algo" sets the default algorithms an ENGINE
1286 will supply using the OPENSSL function ENGINE_set_default_string(). A value
1287 of "ALL" uses the engine for all cryptographic operations. If no list of
1288 algo is specified then the value of "ALL" is used. A comma-seperated list
1289 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1290 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1291 openssl configuration file uses:
1292 https://www.openssl.org/docs/man1.0.2/apps/config.html
1293
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001294ssl-mode-async
1295 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001296 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001297 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1298 doesn't support moving read/write buffers and is not compliant with
1299 haproxy's buffer management. So the asynchronous mode is disabled on
1300 read/write operations (it is only enabled during initial and reneg
1301 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001302
Willy Tarreau33cb0652014-12-23 22:52:37 +01001303tune.buffers.limit <number>
1304 Sets a hard limit on the number of buffers which may be allocated per process.
1305 The default value is zero which means unlimited. The minimum non-zero value
1306 will always be greater than "tune.buffers.reserve" and should ideally always
1307 be about twice as large. Forcing this value can be particularly useful to
1308 limit the amount of memory a process may take, while retaining a sane
1309 behaviour. When this limit is reached, sessions which need a buffer wait for
1310 another one to be released by another session. Since buffers are dynamically
1311 allocated and released, the waiting time is very short and not perceptible
1312 provided that limits remain reasonable. In fact sometimes reducing the limit
1313 may even increase performance by increasing the CPU cache's efficiency. Tests
1314 have shown good results on average HTTP traffic with a limit to 1/10 of the
1315 expected global maxconn setting, which also significantly reduces memory
1316 usage. The memory savings come from the fact that a number of connections
1317 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1318 advised to do so by an haproxy core developer.
1319
Willy Tarreau1058ae72014-12-23 22:40:40 +01001320tune.buffers.reserve <number>
1321 Sets the number of buffers which are pre-allocated and reserved for use only
1322 during memory shortage conditions resulting in failed memory allocations. The
1323 minimum value is 2 and is also the default. There is no reason a user would
1324 want to change this value, it's mostly aimed at haproxy core developers.
1325
Willy Tarreau27a674e2009-08-17 07:23:33 +02001326tune.bufsize <number>
1327 Sets the buffer size to this size (in bytes). Lower values allow more
1328 sessions to coexist in the same amount of RAM, and higher values allow some
1329 applications with very large cookies to work. The default value is 16384 and
1330 can be changed at build time. It is strongly recommended not to change this
1331 from the default value, as very low values will break some services such as
1332 statistics, and values larger than default size will increase memory usage,
1333 possibly causing the system to run out of memory. At least the global maxconn
1334 parameter should be decreased by the same factor as this one is increased.
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001335 If HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
1336 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
1337 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +02001338
Willy Tarreau43961d52010-10-04 20:39:20 +02001339tune.chksize <number>
1340 Sets the check buffer size to this size (in bytes). Higher values may help
1341 find string or regex patterns in very large pages, though doing so may imply
1342 more memory and CPU usage. The default value is 16384 and can be changed at
1343 build time. It is not recommended to change this value, but to use better
1344 checks whenever possible.
1345
William Lallemandf3747832012-11-09 12:33:10 +01001346tune.comp.maxlevel <number>
1347 Sets the maximum compression level. The compression level affects CPU
1348 usage during compression. This value affects CPU usage during compression.
1349 Each session using compression initializes the compression algorithm with
1350 this value. The default value is 1.
1351
Willy Tarreau193b8c62012-11-22 00:17:38 +01001352tune.http.cookielen <number>
1353 Sets the maximum length of captured cookies. This is the maximum value that
1354 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1355 will automatically be truncated to this one. It is important not to set too
1356 high a value because all cookie captures still allocate this size whatever
1357 their configured value (they share a same pool). This value is per request
1358 per response, so the memory allocated is twice this value per connection.
1359 When not specified, the limit is set to 63 characters. It is recommended not
1360 to change this value.
1361
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001362tune.http.logurilen <number>
1363 Sets the maximum length of request uri in logs. This prevent to truncate long
1364 requests uris with valuable query strings in log lines. This is not related
1365 to syslog limits. If you increase this limit, you may also increase the
1366 'log ... len yyyy' parameter. Your syslog deamon may also need specific
1367 configuration directives too.
1368 The default value is 1024.
1369
Willy Tarreauac1932d2011-10-24 19:14:41 +02001370tune.http.maxhdr <number>
1371 Sets the maximum number of headers in a request. When a request comes with a
1372 number of headers greater than this value (including the first line), it is
1373 rejected with a "400 Bad Request" status code. Similarly, too large responses
1374 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1375 for all usages, considering that the widely deployed Apache server uses the
1376 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001377 a buggy application to work by the time it gets fixed. The accepted range is
1378 1..32767. Keep in mind that each new header consumes 32bits of memory for
1379 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001380
Willy Tarreau7e312732014-02-12 16:35:14 +01001381tune.idletimer <timeout>
1382 Sets the duration after which haproxy will consider that an empty buffer is
1383 probably associated with an idle stream. This is used to optimally adjust
1384 some packet sizes while forwarding large and small data alternatively. The
1385 decision to use splice() or to send large buffers in SSL is modulated by this
1386 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1387 means that haproxy will not try to detect idle streams. The default is 1000,
1388 which seems to correctly detect end user pauses (eg: read a page before
1389 clicking). There should be not reason for changing this value. Please check
1390 tune.ssl.maxrecord below.
1391
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001392tune.lua.forced-yield <number>
1393 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001394 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001395 HAProxy scheduler to process other tasks like accepting connections or
1396 forwarding traffic. The default value is 10000 instructions. If HAProxy often
1397 executes some Lua code but more reactivity is required, this value can be
1398 lowered. If the Lua code is quite long and its result is absolutely required
1399 to process the data, the <number> can be increased.
1400
Willy Tarreau32f61e22015-03-18 17:54:59 +01001401tune.lua.maxmem
1402 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1403 default it is zero which means unlimited. It is important to set a limit to
1404 ensure that a bug in a script will not result in the system running out of
1405 memory.
1406
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001407tune.lua.session-timeout <timeout>
1408 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001409 preventing infinite loops or spending too much time in Lua. This timeout
1410 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
1411 not taked in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001412
1413tune.lua.task-timeout <timeout>
1414 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1415 dedicated to the tasks. By default, this timeout isn't set because a task may
1416 remain alive during of the lifetime of HAProxy. For example, a task used to
1417 check servers.
1418
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001419tune.lua.service-timeout <timeout>
1420 This is the execution timeout for the Lua services. This is useful for
1421 preventing infinite loops or spending too much time in Lua. This timeout
1422 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
1423 not taked in account. The default timeout is 4s.
1424
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001425tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001426 Sets the maximum number of consecutive connections a process may accept in a
1427 row before switching to other work. In single process mode, higher numbers
1428 give better performance at high connection rates. However in multi-process
1429 modes, keeping a bit of fairness between processes generally is better to
1430 increase performance. This value applies individually to each listener, so
1431 that the number of processes a listener is bound to is taken into account.
1432 This value defaults to 64. In multi-process mode, it is divided by twice
1433 the number of processes the listener is bound to. Setting this value to -1
1434 completely disables the limitation. It should normally not be needed to tweak
1435 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001436
1437tune.maxpollevents <number>
1438 Sets the maximum amount of events that can be processed at once in a call to
1439 the polling system. The default value is adapted to the operating system. It
1440 has been noticed that reducing it below 200 tends to slightly decrease
1441 latency at the expense of network bandwidth, and increasing it above 200
1442 tends to trade latency for slightly increased bandwidth.
1443
Willy Tarreau27a674e2009-08-17 07:23:33 +02001444tune.maxrewrite <number>
1445 Sets the reserved buffer space to this size in bytes. The reserved space is
1446 used for header rewriting or appending. The first reads on sockets will never
1447 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1448 bufsize, though that does not make much sense since there are rarely large
1449 numbers of headers to add. Setting it too high prevents processing of large
1450 requests or responses. Setting it too low prevents addition of new headers
1451 to already large requests or to POST requests. It is generally wise to set it
1452 to about 1024. It is automatically readjusted to half of bufsize if it is
1453 larger than that. This means you don't have to worry about it when changing
1454 bufsize.
1455
Willy Tarreauf3045d22015-04-29 16:24:50 +02001456tune.pattern.cache-size <number>
1457 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1458 cache which reminds previous lookups and their results. It is used by ACLs
1459 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1460 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1461 strings. It applies to pattern expressions which means that it will be able
1462 to memorize the result of a lookup among all the patterns specified on a
1463 configuration line (including all those loaded from files). It automatically
1464 invalidates entries which are updated using HTTP actions or on the CLI. The
1465 default cache size is set to 10000 entries, which limits its footprint to
1466 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1467 risk of collision in this cache, which is in the order of the size of the
1468 cache divided by 2^64. Typically, at 10000 requests per second with the
1469 default cache size of 10000 entries, there's 1% chance that a brute force
1470 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1471 This is considered much lower than the risk of a memory corruption caused by
1472 aging components. If this is not acceptable, the cache can be disabled by
1473 setting this parameter to 0.
1474
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001475tune.pipesize <number>
1476 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1477 are the default size for the system. But sometimes when using TCP splicing,
1478 it can improve performance to increase pipe sizes, especially if it is
1479 suspected that pipes are not filled and that many calls to splice() are
1480 performed. This has an impact on the kernel's memory footprint, so this must
1481 not be changed if impacts are not understood.
1482
Willy Tarreaue803de22010-01-21 17:43:04 +01001483tune.rcvbuf.client <number>
1484tune.rcvbuf.server <number>
1485 Forces the kernel socket receive buffer size on the client or the server side
1486 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1487 and backends. It should normally never be set, and the default size (0) lets
1488 the kernel autotune this value depending on the amount of available memory.
1489 However it can sometimes help to set it to very low values (eg: 4096) in
1490 order to save kernel memory by preventing it from buffering too large amounts
1491 of received data. Lower values will significantly increase CPU usage though.
1492
Willy Tarreaub22fc302015-12-14 12:04:35 +01001493tune.recv_enough <number>
1494 Haproxy uses some hints to detect that a short read indicates the end of the
1495 socket buffers. One of them is that a read returns more than <recv_enough>
1496 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1497 may be changed by this setting to better deal with workloads involving lots
1498 of short messages such as telnet or SSH sessions.
1499
Willy Tarreaue803de22010-01-21 17:43:04 +01001500tune.sndbuf.client <number>
1501tune.sndbuf.server <number>
1502 Forces the kernel socket send buffer size on the client or the server side to
1503 the specified value in bytes. This value applies to all TCP/HTTP frontends
1504 and backends. It should normally never be set, and the default size (0) lets
1505 the kernel autotune this value depending on the amount of available memory.
1506 However it can sometimes help to set it to very low values (eg: 4096) in
1507 order to save kernel memory by preventing it from buffering too large amounts
1508 of received data. Lower values will significantly increase CPU usage though.
1509 Another use case is to prevent write timeouts with extremely slow clients due
1510 to the kernel waiting for a large part of the buffer to be read before
1511 notifying haproxy again.
1512
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001513tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001514 Sets the size of the global SSL session cache, in a number of blocks. A block
1515 is large enough to contain an encoded session without peer certificate.
1516 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001517 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001518 200 bytes of memory. The default value may be forced at build time, otherwise
1519 defaults to 20000. When the cache is full, the most idle entries are purged
1520 and reassigned. Higher values reduce the occurrence of such a purge, hence
1521 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1522 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001523 and are shared between all processes if "nbproc" is greater than 1. Setting
1524 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001525
Emeric Brun8dc60392014-05-09 13:52:00 +02001526tune.ssl.force-private-cache
1527 This boolean disables SSL session cache sharing between all processes. It
1528 should normally not be used since it will force many renegotiations due to
1529 clients hitting a random process. But it may be required on some operating
1530 systems where none of the SSL cache synchronization method may be used. In
1531 this case, adding a first layer of hash-based load balancing before the SSL
1532 layer might limit the impact of the lack of session sharing.
1533
Emeric Brun4f65bff2012-11-16 15:11:00 +01001534tune.ssl.lifetime <timeout>
1535 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001536 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001537 does not guarantee that sessions will last that long, because if the cache is
1538 full, the longest idle sessions will be purged despite their configured
1539 lifetime. The real usefulness of this setting is to prevent sessions from
1540 being used for too long.
1541
Willy Tarreaubfd59462013-02-21 07:46:09 +01001542tune.ssl.maxrecord <number>
1543 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1544 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1545 data only once it has received a full record. With large records, it means
1546 that clients might have to download up to 16kB of data before starting to
1547 process them. Limiting the value can improve page load times on browsers
1548 located over high latency or low bandwidth networks. It is suggested to find
1549 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1550 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1551 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1552 2859 gave good results during tests. Use "strace -e trace=write" to find the
Willy Tarreau7e312732014-02-12 16:35:14 +01001553 best value. Haproxy will automatically switch to this setting after an idle
1554 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001555
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001556tune.ssl.default-dh-param <number>
1557 Sets the maximum size of the Diffie-Hellman parameters used for generating
1558 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1559 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1560 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1561 this maximum value. Default value if 1024. Only 1024 or higher values are
1562 allowed. Higher values will increase the CPU load, and values greater than
1563 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001564 used if static Diffie-Hellman parameters are supplied either directly
1565 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001566
Christopher Faulet31af49d2015-06-09 17:29:50 +02001567tune.ssl.ssl-ctx-cache-size <number>
1568 Sets the size of the cache used to store generated certificates to <number>
1569 entries. This is a LRU cache. Because generating a SSL certificate
1570 dynamically is expensive, they are cached. The default cache size is set to
1571 1000 entries.
1572
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001573tune.ssl.capture-cipherlist-size <number>
1574 Sets the maximum size of the buffer used for capturing client-hello cipher
1575 list. If the value is 0 (default value) the capture is disabled, otherwise
1576 a buffer is allocated for each SSL/TLS connection.
1577
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001578tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001579tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001580tune.vars.reqres-max-size <size>
1581tune.vars.sess-max-size <size>
1582tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001583 These five tunes help to manage the maximum amount of memory used by the
1584 variables system. "global" limits the overall amount of memory available for
1585 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1586 memory for the session scope, "txn" for the transaction scope, and "reqres"
1587 limits the memory for each request or response processing.
1588 Memory accounting is hierarchical, meaning more coarse grained limits include
1589 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1590 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001591
Daniel Schneller0b547052016-03-21 20:46:57 +01001592 For example, when "tune.vars.sess-max-size" is limited to 100,
1593 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1594 100 either. If we create a variable "txn.var" that contains 100 bytes,
1595 all available space is consumed.
1596 Notice that exceeding the limits at runtime will not result in an error
1597 message, but values might be cut off or corrupted. So make sure to accurately
1598 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001599
William Lallemanda509e4c2012-11-07 16:54:34 +01001600tune.zlib.memlevel <number>
1601 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001602 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001603 state. A value of 1 uses minimum memory but is slow and reduces compression
1604 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
1605 between 1 and 9. The default value is 8.
1606
1607tune.zlib.windowsize <number>
1608 Sets the window size (the size of the history buffer) as a parameter of the
1609 zlib initialization for each session. Larger values of this parameter result
1610 in better compression at the expense of memory usage. Can be a value between
1611 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001612
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016133.3. Debugging
1614--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001615
1616debug
1617 Enables debug mode which dumps to stdout all exchanges, and disables forking
1618 into background. It is the equivalent of the command-line argument "-d". It
1619 should never be used in a production configuration since it may prevent full
1620 system startup.
1621
1622quiet
1623 Do not display any message during startup. It is equivalent to the command-
1624 line argument "-q".
1625
Emeric Brunf099e792010-09-27 12:05:28 +02001626
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016273.4. Userlists
1628--------------
1629It is possible to control access to frontend/backend/listen sections or to
1630http stats by allowing only authenticated and authorized users. To do this,
1631it is required to create at least one userlist and to define users.
1632
1633userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001634 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001635 used to store authentication & authorization data for independent customers.
1636
1637group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001638 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001639 attach users to this group by using a comma separated list of names
1640 proceeded by "users" keyword.
1641
Cyril Bontéf0c60612010-02-06 14:44:47 +01001642user <username> [password|insecure-password <password>]
1643 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001644 Adds user <username> to the current userlist. Both secure (encrypted) and
1645 insecure (unencrypted) passwords can be used. Encrypted passwords are
Cyril Bonté78caf842010-03-10 22:41:43 +01001646 evaluated using the crypt(3) function so depending of the system's
1647 capabilities, different algorithms are supported. For example modern Glibc
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001648 based Linux system supports MD5, SHA-256, SHA-512 and of course classic,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001649 DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001650
1651
1652 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001653 userlist L1
1654 group G1 users tiger,scott
1655 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001656
Cyril Bontéf0c60612010-02-06 14:44:47 +01001657 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1658 user scott insecure-password elgato
1659 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001660
Cyril Bontéf0c60612010-02-06 14:44:47 +01001661 userlist L2
1662 group G1
1663 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001664
Cyril Bontéf0c60612010-02-06 14:44:47 +01001665 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1666 user scott insecure-password elgato groups G1,G2
1667 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001668
1669 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001670
Emeric Brunf099e792010-09-27 12:05:28 +02001671
16723.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001673----------
Emeric Brun94900952015-06-11 18:25:54 +02001674It is possible to propagate entries of any data-types in stick-tables between
1675several haproxy instances over TCP connections in a multi-master fashion. Each
1676instance pushes its local updates and insertions to remote peers. The pushed
1677values overwrite remote ones without aggregation. Interrupted exchanges are
1678automatically detected and recovered from the last known point.
1679In addition, during a soft restart, the old process connects to the new one
1680using such a TCP connection to push all its entries before the new process
1681tries to connect to other peers. That ensures very fast replication during a
1682reload, it typically takes a fraction of a second even for large tables.
1683Note that Server IDs are used to identify servers remotely, so it is important
1684that configurations look similar or at least that the same IDs are forced on
1685each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001686
1687peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001688 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001689 which is referenced by one or more stick-tables.
1690
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001691disabled
1692 Disables a peers section. It disables both listening and any synchronization
1693 related to this section. This is provided to disable synchronization of stick
1694 tables without having to comment out all "peers" references.
1695
1696enable
1697 This re-enables a disabled peers section which was previously disabled.
1698
Emeric Brunf099e792010-09-27 12:05:28 +02001699peer <peername> <ip>:<port>
1700 Defines a peer inside a peers section.
1701 If <peername> is set to the local peer name (by default hostname, or forced
1702 using "-L" command line option), haproxy will listen for incoming remote peer
1703 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1704 to join the remote peer, and <peername> is used at the protocol level to
1705 identify and validate the remote peer on the server side.
1706
1707 During a soft restart, local peer <ip>:<port> is used by the old instance to
1708 connect the new one and initiate a complete replication (teaching process).
1709
1710 It is strongly recommended to have the exact same peers declaration on all
1711 peers and to only rely on the "-L" command line argument to change the local
1712 peer name. This makes it easier to maintain coherent configuration files
1713 across all peers.
1714
William Lallemandb2f07452015-05-12 14:27:13 +02001715 You may want to reference some environment variables in the address
1716 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001717
Cyril Bontédc4d9032012-04-08 21:57:39 +02001718 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001719 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001720 peer haproxy1 192.168.0.1:1024
1721 peer haproxy2 192.168.0.2:1024
1722 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001723
1724 backend mybackend
1725 mode tcp
1726 balance roundrobin
1727 stick-table type ip size 20k peers mypeers
1728 stick on src
1729
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001730 server srv1 192.168.0.30:80
1731 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001732
1733
Simon Horman51a1cf62015-02-03 13:00:44 +090017343.6. Mailers
1735------------
1736It is possible to send email alerts when the state of servers changes.
1737If configured email alerts are sent to each mailer that is configured
1738in a mailers section. Email is sent to mailers using SMTP.
1739
Pieter Baauw386a1272015-08-16 15:26:24 +02001740mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001741 Creates a new mailer list with the name <mailersect>. It is an
1742 independent section which is referenced by one or more proxies.
1743
1744mailer <mailername> <ip>:<port>
1745 Defines a mailer inside a mailers section.
1746
1747 Example:
1748 mailers mymailers
1749 mailer smtp1 192.168.0.1:587
1750 mailer smtp2 192.168.0.2:587
1751
1752 backend mybackend
1753 mode tcp
1754 balance roundrobin
1755
1756 email-alert mailers mymailers
1757 email-alert from test1@horms.org
1758 email-alert to test2@horms.org
1759
1760 server srv1 192.168.0.30:80
1761 server srv2 192.168.0.31:80
1762
Pieter Baauw235fcfc2016-02-13 15:33:40 +01001763timeout mail <time>
1764 Defines the time available for a mail/connection to be made and send to
1765 the mail-server. If not defined the default value is 10 seconds. To allow
1766 for at least two SYN-ACK packets to be send during initial TCP handshake it
1767 is advised to keep this value above 4 seconds.
1768
1769 Example:
1770 mailers mymailers
1771 timeout mail 20s
1772 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09001773
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017744. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001775----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001776
Willy Tarreau6a06a402007-07-15 20:15:28 +02001777Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02001778 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001779 - frontend <name>
1780 - backend <name>
1781 - listen <name>
1782
1783A "defaults" section sets default parameters for all other sections following
1784its declaration. Those default parameters are reset by the next "defaults"
1785section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001786section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001787
1788A "frontend" section describes a set of listening sockets accepting client
1789connections.
1790
1791A "backend" section describes a set of servers to which the proxy will connect
1792to forward incoming connections.
1793
1794A "listen" section defines a complete proxy with its frontend and backend
1795parts combined in one section. It is generally useful for TCP-only traffic.
1796
Willy Tarreau0ba27502007-12-24 16:55:16 +01001797All proxy names must be formed from upper and lower case letters, digits,
1798'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1799case-sensitive, which means that "www" and "WWW" are two different proxies.
1800
1801Historically, all proxy names could overlap, it just caused troubles in the
1802logs. Since the introduction of content switching, it is mandatory that two
1803proxies with overlapping capabilities (frontend/backend) have different names.
1804However, it is still permitted that a frontend and a backend share the same
1805name, as this configuration seems to be commonly encountered.
1806
1807Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1808and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001809bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001810protocol, and can interact with it by allowing, blocking, switching, adding,
1811modifying, or removing arbitrary contents in requests or responses, based on
1812arbitrary criteria.
1813
Willy Tarreau70dffda2014-01-30 03:07:23 +01001814In HTTP mode, the processing applied to requests and responses flowing over
1815a connection depends in the combination of the frontend's HTTP options and
1816the backend's. HAProxy supports 5 connection modes :
1817
1818 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1819 requests and responses are processed, and connections remain open but idle
1820 between responses and new requests.
1821
1822 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1823 1.0 to 1.5-dev21 : only the first request and response are processed, and
1824 everything else is forwarded with no analysis at all. This mode should not
1825 be used as it creates lots of trouble with logging and HTTP processing.
1826
1827 - PCL: passive close ("option httpclose") : exactly the same as tunnel mode,
1828 but with "Connection: close" appended in both directions to try to make
1829 both ends close after the first request/response exchange.
1830
1831 - SCL: server close ("option http-server-close") : the server-facing
1832 connection is closed after the end of the response is received, but the
1833 client-facing connection remains open.
1834
1835 - FCL: forced close ("option forceclose") : the connection is actively closed
1836 after the end of the response.
1837
1838The effective mode that will be applied to a connection passing through a
1839frontend and a backend can be determined by both proxy modes according to the
1840following matrix, but in short, the modes are symmetric, keep-alive is the
1841weakest option and force close is the strongest.
1842
1843 Backend mode
1844
1845 | KAL | TUN | PCL | SCL | FCL
1846 ----+-----+-----+-----+-----+----
1847 KAL | KAL | TUN | PCL | SCL | FCL
1848 ----+-----+-----+-----+-----+----
1849 TUN | TUN | TUN | PCL | SCL | FCL
1850 Frontend ----+-----+-----+-----+-----+----
1851 mode PCL | PCL | PCL | PCL | FCL | FCL
1852 ----+-----+-----+-----+-----+----
1853 SCL | SCL | SCL | FCL | SCL | FCL
1854 ----+-----+-----+-----+-----+----
1855 FCL | FCL | FCL | FCL | FCL | FCL
1856
Willy Tarreau0ba27502007-12-24 16:55:16 +01001857
Willy Tarreau70dffda2014-01-30 03:07:23 +01001858
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018594.1. Proxy keywords matrix
1860--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001861
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001862The following list of keywords is supported. Most of them may only be used in a
1863limited set of section types. Some of them are marked as "deprecated" because
1864they are inherited from an old syntax which may be confusing or functionally
1865limited, and there are new recommended keywords to replace them. Keywords
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001866marked with "(*)" can be optionally inverted using the "no" prefix, eg. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001867option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02001868and must be disabled for a specific instance. Such options may also be prefixed
1869with "default" in order to restore default settings regardless of what has been
1870specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01001871
Willy Tarreau6a06a402007-07-15 20:15:28 +02001872
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001873 keyword defaults frontend listen backend
1874------------------------------------+----------+----------+---------+---------
1875acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02001876appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001877backlog X X X -
1878balance X - X X
1879bind - X X -
1880bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02001881block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001882capture cookie - X X -
1883capture request header - X X -
1884capture response header - X X -
1885clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02001886compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001887contimeout (deprecated) X - X X
1888cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02001889declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001890default-server X - X X
1891default_backend X X X -
1892description - X X X
1893disabled X X X X
1894dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09001895email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09001896email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09001897email-alert mailers X X X X
1898email-alert myhostname X X X X
1899email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001900enabled X X X X
1901errorfile X X X X
1902errorloc X X X X
1903errorloc302 X X X X
1904-- keyword -------------------------- defaults - frontend - listen -- backend -
1905errorloc303 X X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001906force-persist - X X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001907filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001908fullconn X - X X
1909grace X X X X
1910hash-type X - X X
1911http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01001912http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02001913http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001914http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02001915http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02001916http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02001917http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001918id - X X X
Cyril Bonté0d4bf012010-04-25 23:21:46 +02001919ignore-persist - X X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001920load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02001921log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01001922log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02001923log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01001924log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02001925max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001926maxconn X X X -
1927mode X X X X
1928monitor fail - X X -
1929monitor-net X X X -
1930monitor-uri X X X -
1931option abortonclose (*) X - X X
1932option accept-invalid-http-request (*) X X X -
1933option accept-invalid-http-response (*) X - X X
1934option allbackups (*) X - X X
1935option checkcache (*) X - X X
1936option clitcpka (*) X X X -
1937option contstats (*) X X X -
1938option dontlog-normal (*) X X X -
1939option dontlognull (*) X X X -
1940option forceclose (*) X X X X
1941-- keyword -------------------------- defaults - frontend - listen -- backend -
1942option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02001943option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02001944option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01001945option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02001946option http-no-delay (*) X X X X
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02001947option http-pretend-keepalive (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001948option http-server-close (*) X X X X
Willy Tarreau02bce8b2014-01-30 00:15:28 +01001949option http-tunnel (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001950option http-use-proxy-header (*) X X X -
1951option httpchk X - X X
1952option httpclose (*) X X X X
1953option httplog X X X X
1954option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001955option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02001956option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09001957option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001958option log-health-checks (*) X - X X
1959option log-separate-errors (*) X X X -
1960option logasap (*) X X X -
1961option mysql-check X - X X
1962option nolinger (*) X X X X
1963option originalto X X X X
1964option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02001965option pgsql-check X - X X
1966option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001967option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02001968option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001969option smtpchk X - X X
1970option socket-stats (*) X X X -
1971option splice-auto (*) X X X X
1972option splice-request (*) X X X X
1973option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01001974option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001975option srvtcpka (*) X - X X
1976option ssl-hello-chk X - X X
1977-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01001978option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001979option tcp-smart-accept (*) X X X -
1980option tcp-smart-connect (*) X - X X
1981option tcpka X X X X
1982option tcplog X X X X
1983option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09001984external-check command X - X X
1985external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01001986persist rdp-cookie X - X X
1987rate-limit sessions X X X -
1988redirect - X X X
1989redisp (deprecated) X - X X
1990redispatch (deprecated) X - X X
1991reqadd - X X X
1992reqallow - X X X
1993reqdel - X X X
1994reqdeny - X X X
1995reqiallow - X X X
1996reqidel - X X X
1997reqideny - X X X
1998reqipass - X X X
1999reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002000reqitarpit - X X X
2001reqpass - X X X
2002reqrep - X X X
2003-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002004reqtarpit - X X X
2005retries X - X X
2006rspadd - X X X
2007rspdel - X X X
2008rspdeny - X X X
2009rspidel - X X X
2010rspideny - X X X
2011rspirep - X X X
2012rsprep - X X X
2013server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002014server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002015server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002016source X - X X
2017srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002018stats admin - X X X
2019stats auth X X X X
2020stats enable X X X X
2021stats hide-version X X X X
2022stats http-request - X X X
2023stats realm X X X X
2024stats refresh X X X X
2025stats scope X X X X
2026stats show-desc X X X X
2027stats show-legends X X X X
2028stats show-node X X X X
2029stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002030-- keyword -------------------------- defaults - frontend - listen -- backend -
2031stick match - - X X
2032stick on - - X X
2033stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002034stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002035stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002036tcp-check connect - - X X
2037tcp-check expect - - X X
2038tcp-check send - - X X
2039tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002040tcp-request connection - X X -
2041tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002042tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002043tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002044tcp-response content - - X X
2045tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002046timeout check X - X X
2047timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002048timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002049timeout clitimeout (deprecated) X X X -
2050timeout connect X - X X
2051timeout contimeout (deprecated) X - X X
2052timeout http-keep-alive X X X X
2053timeout http-request X X X X
2054timeout queue X - X X
2055timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002056timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002057timeout srvtimeout (deprecated) X - X X
2058timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002059timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002060transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002061unique-id-format X X X -
2062unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002063use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002064use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002065------------------------------------+----------+----------+---------+---------
2066 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002067
Willy Tarreau0ba27502007-12-24 16:55:16 +01002068
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020694.2. Alphabetically sorted keywords reference
2070---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002071
2072This section provides a description of each keyword and its usage.
2073
2074
2075acl <aclname> <criterion> [flags] [operator] <value> ...
2076 Declare or complete an access list.
2077 May be used in sections : defaults | frontend | listen | backend
2078 no | yes | yes | yes
2079 Example:
2080 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2081 acl invalid_src src_port 0:1023
2082 acl local_dst hdr(host) -i localhost
2083
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002084 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002085
2086
Cyril Bontéb21570a2009-11-29 20:04:48 +01002087appsession <cookie> len <length> timeout <holdtime>
2088 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002089 Define session stickiness on an existing application cookie.
2090 May be used in sections : defaults | frontend | listen | backend
2091 no | no | yes | yes
2092 Arguments :
2093 <cookie> this is the name of the cookie used by the application and which
2094 HAProxy will have to learn for each new session.
2095
Cyril Bontéb21570a2009-11-29 20:04:48 +01002096 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002097 checked in each cookie value.
2098
2099 <holdtime> this is the time after which the cookie will be removed from
2100 memory if unused. If no unit is specified, this time is in
2101 milliseconds.
2102
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002103 request-learn
2104 If this option is specified, then haproxy will be able to learn
2105 the cookie found in the request in case the server does not
2106 specify any in response. This is typically what happens with
2107 PHPSESSID cookies, or when haproxy's session expires before
2108 the application's session and the correct server is selected.
2109 It is recommended to specify this option to improve reliability.
2110
Cyril Bontéb21570a2009-11-29 20:04:48 +01002111 prefix When this option is specified, haproxy will match on the cookie
2112 prefix (or URL parameter prefix). The appsession value is the
2113 data following this prefix.
2114
2115 Example :
2116 appsession ASPSESSIONID len 64 timeout 3h prefix
2117
2118 This will match the cookie ASPSESSIONIDXXXX=XXXXX,
2119 the appsession value will be XXXX=XXXXX.
2120
2121 mode This option allows to change the URL parser mode.
2122 2 modes are currently supported :
2123 - path-parameters :
2124 The parser looks for the appsession in the path parameters
2125 part (each parameter is separated by a semi-colon), which is
2126 convenient for JSESSIONID for example.
2127 This is the default mode if the option is not set.
2128 - query-string :
2129 In this mode, the parser will look for the appsession in the
2130 query string.
2131
Willy Tarreau294d0f02015-08-10 19:40:12 +02002132 As of version 1.6, appsessions was removed. It is more flexible and more
2133 convenient to use stick-tables instead, and stick-tables support multi-master
2134 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002135
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002136 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2137 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002138
2139
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002140backlog <conns>
2141 Give hints to the system about the approximate listen backlog desired size
2142 May be used in sections : defaults | frontend | listen | backend
2143 yes | yes | yes | no
2144 Arguments :
2145 <conns> is the number of pending connections. Depending on the operating
2146 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002147 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002148
2149 In order to protect against SYN flood attacks, one solution is to increase
2150 the system's SYN backlog size. Depending on the system, sometimes it is just
2151 tunable via a system parameter, sometimes it is not adjustable at all, and
2152 sometimes the system relies on hints given by the application at the time of
2153 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2154 to the listen() syscall. On systems which can make use of this value, it can
2155 sometimes be useful to be able to specify a different value, hence this
2156 backlog parameter.
2157
2158 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2159 used as a hint and the system accepts up to the smallest greater power of
2160 two, and never more than some limits (usually 32768).
2161
2162 See also : "maxconn" and the target operating system's tuning guide.
2163
2164
Willy Tarreau0ba27502007-12-24 16:55:16 +01002165balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002166balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002167 Define the load balancing algorithm to be used in a backend.
2168 May be used in sections : defaults | frontend | listen | backend
2169 yes | no | yes | yes
2170 Arguments :
2171 <algorithm> is the algorithm used to select a server when doing load
2172 balancing. This only applies when no persistence information
2173 is available, or when a connection is redispatched to another
2174 server. <algorithm> may be one of the following :
2175
2176 roundrobin Each server is used in turns, according to their weights.
2177 This is the smoothest and fairest algorithm when the server's
2178 processing time remains equally distributed. This algorithm
2179 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002180 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002181 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002182 large farms, when a server becomes up after having been down
2183 for a very short time, it may sometimes take a few hundreds
2184 requests for it to be re-integrated into the farm and start
2185 receiving traffic. This is normal, though very rare. It is
2186 indicated here in case you would have the chance to observe
2187 it, so that you don't worry.
2188
2189 static-rr Each server is used in turns, according to their weights.
2190 This algorithm is as similar to roundrobin except that it is
2191 static, which means that changing a server's weight on the
2192 fly will have no effect. On the other hand, it has no design
2193 limitation on the number of servers, and when a server goes
2194 up, it is always immediately reintroduced into the farm, once
2195 the full map is recomputed. It also uses slightly less CPU to
2196 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002197
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002198 leastconn The server with the lowest number of connections receives the
2199 connection. Round-robin is performed within groups of servers
2200 of the same load to ensure that all servers will be used. Use
2201 of this algorithm is recommended where very long sessions are
2202 expected, such as LDAP, SQL, TSE, etc... but is not very well
2203 suited for protocols using short sessions such as HTTP. This
2204 algorithm is dynamic, which means that server weights may be
2205 adjusted on the fly for slow starts for instance.
2206
Willy Tarreauf09c6602012-02-13 17:12:08 +01002207 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002208 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002209 identifier to the highest (see server parameter "id"), which
2210 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002211 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002212 not make sense to use this algorithm without setting maxconn.
2213 The purpose of this algorithm is to always use the smallest
2214 number of servers so that extra servers can be powered off
2215 during non-intensive hours. This algorithm ignores the server
2216 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002217 or IMAP than HTTP, though it can be useful there too. In
2218 order to use this algorithm efficiently, it is recommended
2219 that a cloud controller regularly checks server usage to turn
2220 them off when unused, and regularly checks backend queue to
2221 turn new servers on when the queue inflates. Alternatively,
2222 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002223
Willy Tarreau0ba27502007-12-24 16:55:16 +01002224 source The source IP address is hashed and divided by the total
2225 weight of the running servers to designate which server will
2226 receive the request. This ensures that the same client IP
2227 address will always reach the same server as long as no
2228 server goes down or up. If the hash result changes due to the
2229 number of running servers changing, many clients will be
2230 directed to a different server. This algorithm is generally
2231 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002232 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002233 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002234 static by default, which means that changing a server's
2235 weight on the fly will have no effect, but this can be
2236 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002237
Oskar Stolc8dc41842012-05-19 10:19:54 +01002238 uri This algorithm hashes either the left part of the URI (before
2239 the question mark) or the whole URI (if the "whole" parameter
2240 is present) and divides the hash value by the total weight of
2241 the running servers. The result designates which server will
2242 receive the request. This ensures that the same URI will
2243 always be directed to the same server as long as no server
2244 goes up or down. This is used with proxy caches and
2245 anti-virus proxies in order to maximize the cache hit rate.
2246 Note that this algorithm may only be used in an HTTP backend.
2247 This algorithm is static by default, which means that
2248 changing a server's weight on the fly will have no effect,
2249 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002250
Oskar Stolc8dc41842012-05-19 10:19:54 +01002251 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002252 "depth", both followed by a positive integer number. These
2253 options may be helpful when it is needed to balance servers
2254 based on the beginning of the URI only. The "len" parameter
2255 indicates that the algorithm should only consider that many
2256 characters at the beginning of the URI to compute the hash.
2257 Note that having "len" set to 1 rarely makes sense since most
2258 URIs start with a leading "/".
2259
2260 The "depth" parameter indicates the maximum directory depth
2261 to be used to compute the hash. One level is counted for each
2262 slash in the request. If both parameters are specified, the
2263 evaluation stops when either is reached.
2264
Willy Tarreau0ba27502007-12-24 16:55:16 +01002265 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002266 the query string of each HTTP GET request.
2267
2268 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002269 request entity will be searched for the parameter argument,
2270 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002271 ('?') in the URL. The message body will only start to be
2272 analyzed once either the advertised amount of data has been
2273 received or the request buffer is full. In the unlikely event
2274 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002275 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002276 be randomly balanced if at all. This keyword used to support
2277 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002278
2279 If the parameter is found followed by an equal sign ('=') and
2280 a value, then the value is hashed and divided by the total
2281 weight of the running servers. The result designates which
2282 server will receive the request.
2283
2284 This is used to track user identifiers in requests and ensure
2285 that a same user ID will always be sent to the same server as
2286 long as no server goes up or down. If no value is found or if
2287 the parameter is not found, then a round robin algorithm is
2288 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002289 backend. This algorithm is static by default, which means
2290 that changing a server's weight on the fly will have no
2291 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002292
Cyril Bontédc4d9032012-04-08 21:57:39 +02002293 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2294 request. Just as with the equivalent ACL 'hdr()' function,
2295 the header name in parenthesis is not case sensitive. If the
2296 header is absent or if it does not contain any value, the
2297 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002298
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002299 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002300 reducing the hash algorithm to the main domain part with some
2301 specific headers such as 'Host'. For instance, in the Host
2302 value "haproxy.1wt.eu", only "1wt" will be considered.
2303
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002304 This algorithm is static by default, which means that
2305 changing a server's weight on the fly will have no effect,
2306 but this can be changed using "hash-type".
2307
Emeric Brun736aa232009-06-30 17:56:00 +02002308 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002309 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002310 The RDP cookie <name> (or "mstshash" if omitted) will be
2311 looked up and hashed for each incoming TCP request. Just as
2312 with the equivalent ACL 'req_rdp_cookie()' function, the name
2313 is not case-sensitive. This mechanism is useful as a degraded
2314 persistence mode, as it makes it possible to always send the
2315 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002316 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002317 used instead.
2318
2319 Note that for this to work, the frontend must ensure that an
2320 RDP cookie is already present in the request buffer. For this
2321 you must use 'tcp-request content accept' rule combined with
2322 a 'req_rdp_cookie_cnt' ACL.
2323
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002324 This algorithm is static by default, which means that
2325 changing a server's weight on the fly will have no effect,
2326 but this can be changed using "hash-type".
2327
Cyril Bontédc4d9032012-04-08 21:57:39 +02002328 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002329
Willy Tarreau0ba27502007-12-24 16:55:16 +01002330 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002331 algorithms. Right now, only "url_param" and "uri" support an
2332 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002333
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002334 The load balancing algorithm of a backend is set to roundrobin when no other
2335 algorithm, mode nor option have been set. The algorithm may only be set once
2336 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002337
2338 Examples :
2339 balance roundrobin
2340 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002341 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002342 balance hdr(User-Agent)
2343 balance hdr(host)
2344 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002345
2346 Note: the following caveats and limitations on using the "check_post"
2347 extension with "url_param" must be considered :
2348
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002349 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002350 to determine if the parameters will be found in the body or entity which
2351 may contain binary data. Therefore another method may be required to
2352 restrict consideration of POST requests that have no URL parameters in
2353 the body. (see acl reqideny http_end)
2354
2355 - using a <max_wait> value larger than the request buffer size does not
2356 make sense and is useless. The buffer size is set at build time, and
2357 defaults to 16 kB.
2358
2359 - Content-Encoding is not supported, the parameter search will probably
2360 fail; and load balancing will fall back to Round Robin.
2361
2362 - Expect: 100-continue is not supported, load balancing will fall back to
2363 Round Robin.
2364
Lukas Tribus23953682017-04-28 13:24:30 +00002365 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002366 If the entire parameter value is not present in the first chunk, the
2367 selection of server is undefined (actually, defined by how little
2368 actually appeared in the first chunk).
2369
2370 - This feature does not support generation of a 100, 411 or 501 response.
2371
2372 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002373 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002374 white space or control characters are found, indicating the end of what
2375 might be a URL parameter list. This is probably not a concern with SGML
2376 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002377
Willy Tarreau294d0f02015-08-10 19:40:12 +02002378 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002379
2380
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002381bind [<address>]:<port_range> [, ...] [param*]
2382bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002383 Define one or several listening addresses and/or ports in a frontend.
2384 May be used in sections : defaults | frontend | listen | backend
2385 no | yes | yes | no
2386 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002387 <address> is optional and can be a host name, an IPv4 address, an IPv6
2388 address, or '*'. It designates the address the frontend will
2389 listen on. If unset, all IPv4 addresses of the system will be
2390 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002391 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002392 Optionally, an address family prefix may be used before the
2393 address to force the family regardless of the address format,
2394 which can be useful to specify a path to a unix socket with
2395 no slash ('/'). Currently supported prefixes are :
2396 - 'ipv4@' -> address is always IPv4
2397 - 'ipv6@' -> address is always IPv6
2398 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002399 - 'abns@' -> address is in abstract namespace (Linux only).
2400 Note: since abstract sockets are not "rebindable", they
2401 do not cope well with multi-process mode during
2402 soft-restart, so it is better to avoid them if
2403 nbproc is greater than 1. The effect is that if the
2404 new process fails to start, only one of the old ones
2405 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002406 - 'fd@<n>' -> use file descriptor <n> inherited from the
2407 parent. The fd must be bound and may or may not already
2408 be listening.
William Lallemandb2f07452015-05-12 14:27:13 +02002409 You may want to reference some environment variables in the
2410 address parameter, see section 2.3 about environment
2411 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002412
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002413 <port_range> is either a unique TCP port, or a port range for which the
2414 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002415 above. The port is mandatory for TCP listeners. Note that in
2416 the case of an IPv6 address, the port is always the number
2417 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002418 - a numerical port (ex: '80')
2419 - a dash-delimited ports range explicitly stating the lower
2420 and upper bounds (ex: '2000-2100') which are included in
2421 the range.
2422
2423 Particular care must be taken against port ranges, because
2424 every <address:port> couple consumes one socket (= a file
2425 descriptor), so it's easy to consume lots of descriptors
2426 with a simple range, and to run out of sockets. Also, each
2427 <address:port> couple must be used only once among all
2428 instances running on a same system. Please note that binding
2429 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002430 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002431 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002432
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002433 <path> is a UNIX socket path beginning with a slash ('/'). This is
2434 alternative to the TCP listening port. Haproxy will then
2435 receive UNIX connections on the socket located at this place.
2436 The path must begin with a slash and by default is absolute.
2437 It can be relative to the prefix defined by "unix-bind" in
2438 the global section. Note that the total length of the prefix
2439 followed by the socket path cannot exceed some system limits
2440 for UNIX sockets, which commonly are set to 107 characters.
2441
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002442 <param*> is a list of parameters common to all sockets declared on the
2443 same line. These numerous parameters depend on OS and build
2444 options and have a complete section dedicated to them. Please
2445 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002446
Willy Tarreau0ba27502007-12-24 16:55:16 +01002447 It is possible to specify a list of address:port combinations delimited by
2448 commas. The frontend will then listen on all of these addresses. There is no
2449 fixed limit to the number of addresses and ports which can be listened on in
2450 a frontend, as well as there is no limit to the number of "bind" statements
2451 in a frontend.
2452
2453 Example :
2454 listen http_proxy
2455 bind :80,:443
2456 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002457 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002458
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002459 listen http_https_proxy
2460 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002461 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002462
Willy Tarreau24709282013-03-10 21:32:12 +01002463 listen http_https_proxy_explicit
2464 bind ipv6@:80
2465 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2466 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2467
Willy Tarreaudad36a32013-03-11 01:20:04 +01002468 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002469 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002470
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002471 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2472 sun_path length is used for the address length. Some other programs
2473 such as socat use the string length only by default. Pass the option
2474 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2475 make it compatible with HAProxy's.
2476
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002477 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002478 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002479
2480
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002481bind-process [ all | odd | even | <number 1-64>[-<number 1-64>] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002482 Limit visibility of an instance to a certain set of processes numbers.
2483 May be used in sections : defaults | frontend | listen | backend
2484 yes | yes | yes | yes
2485 Arguments :
2486 all All process will see this instance. This is the default. It
2487 may be used to override a default value.
2488
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002489 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002490 option may be combined with other numbers.
2491
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002492 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002493 option may be combined with other numbers. Do not use it
2494 with less than 2 processes otherwise some instances might be
2495 missing from all processes.
2496
Willy Tarreau110ecc12012-11-15 17:50:01 +01002497 number The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002498 whose values must all be between 1 and 32 or 64 depending on
Willy Tarreau102df612014-05-07 23:56:38 +02002499 the machine's word size. If a proxy is bound to process
2500 numbers greater than the configured global.nbproc, it will
2501 either be forced to process #1 if a single process was
2502 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002503
2504 This keyword limits binding of certain instances to certain processes. This
2505 is useful in order not to have too many processes listening to the same
2506 ports. For instance, on a dual-core machine, it might make sense to set
2507 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2508 and 'even' instances.
2509
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002510 At the moment, it is not possible to reference more than 32 or 64 processes
2511 using this keyword, but this should be more than enough for most setups.
2512 Please note that 'all' really means all processes regardless of the machine's
2513 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002514
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002515 Each "bind" line may further be limited to a subset of the proxy's processes,
2516 please consult the "process" bind keyword in section 5.1.
2517
Willy Tarreaub369a042014-09-16 13:21:03 +02002518 When a frontend has no explicit "bind-process" line, it tries to bind to all
2519 the processes referenced by its "bind" lines. That means that frontends can
2520 easily adapt to their listeners' processes.
2521
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002522 If some backends are referenced by frontends bound to other processes, the
2523 backend automatically inherits the frontend's processes.
2524
2525 Example :
2526 listen app_ip1
2527 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002528 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002529
2530 listen app_ip2
2531 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002532 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002533
2534 listen management
2535 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002536 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002537
Willy Tarreau110ecc12012-11-15 17:50:01 +01002538 listen management
2539 bind 10.0.0.4:80
2540 bind-process 1-4
2541
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002542 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002543
2544
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002545block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002546 Block a layer 7 request if/unless a condition is matched
2547 May be used in sections : defaults | frontend | listen | backend
2548 no | yes | yes | yes
2549
2550 The HTTP request will be blocked very early in the layer 7 processing
2551 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002552 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002553 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002554 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002555 "block" statements per instance. To block connections at layer 4 (without
2556 sending a 403 error) see "tcp-request connection reject" and
2557 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002558
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002559 This form is deprecated, do not use it in any new configuration, use the new
2560 "http-request deny" instead.
2561
Willy Tarreau0ba27502007-12-24 16:55:16 +01002562 Example:
2563 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2564 acl invalid_src src_port 0:1023
2565 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002566 # block is deprecated. Use http-request deny instead:
2567 #block if invalid_src || local_dst
2568 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002569
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002570 See also : section 7 about ACL usage, "http-request deny",
2571 "http-response deny", "tcp-request connection reject" and
2572 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002573
2574capture cookie <name> len <length>
2575 Capture and log a cookie in the request and in the response.
2576 May be used in sections : defaults | frontend | listen | backend
2577 no | yes | yes | no
2578 Arguments :
2579 <name> is the beginning of the name of the cookie to capture. In order
2580 to match the exact name, simply suffix the name with an equal
2581 sign ('='). The full name will appear in the logs, which is
2582 useful with application servers which adjust both the cookie name
2583 and value (eg: ASPSESSIONXXXXX).
2584
2585 <length> is the maximum number of characters to report in the logs, which
2586 include the cookie name, the equal sign and the value, all in the
2587 standard "name=value" form. The string will be truncated on the
2588 right if it exceeds <length>.
2589
2590 Only the first cookie is captured. Both the "cookie" request headers and the
2591 "set-cookie" response headers are monitored. This is particularly useful to
2592 check for application bugs causing session crossing or stealing between
2593 users, because generally the user's cookies can only change on a login page.
2594
2595 When the cookie was not presented by the client, the associated log column
2596 will report "-". When a request does not cause a cookie to be assigned by the
2597 server, a "-" is reported in the response column.
2598
2599 The capture is performed in the frontend only because it is necessary that
2600 the log format does not change for a given frontend depending on the
2601 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002602 "capture cookie" statement in a frontend. The maximum capture length is set
2603 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2604 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002605
2606 Example:
2607 capture cookie ASPSESSION len 32
2608
2609 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002610 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002611
2612
2613capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002614 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002615 May be used in sections : defaults | frontend | listen | backend
2616 no | yes | yes | no
2617 Arguments :
2618 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002619 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002620 appear in the requests, with the first letter of each word in
2621 upper case. The header name will not appear in the logs, only the
2622 value is reported, but the position in the logs is respected.
2623
2624 <length> is the maximum number of characters to extract from the value and
2625 report in the logs. The string will be truncated on the right if
2626 it exceeds <length>.
2627
Willy Tarreau4460d032012-11-21 23:37:37 +01002628 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002629 value will be added to the logs between braces ('{}'). If multiple headers
2630 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002631 in the same order they were declared in the configuration. Non-existent
2632 headers will be logged just as an empty string. Common uses for request
2633 header captures include the "Host" field in virtual hosting environments, the
2634 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002635 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002636 environments to find where the request came from.
2637
2638 Note that when capturing headers such as "User-agent", some spaces may be
2639 logged, making the log analysis more difficult. Thus be careful about what
2640 you log if you know your log parser is not smart enough to rely on the
2641 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002642
Willy Tarreau0900abb2012-11-22 00:21:46 +01002643 There is no limit to the number of captured request headers nor to their
2644 length, though it is wise to keep them low to limit memory usage per session.
2645 In order to keep log format consistent for a same frontend, header captures
2646 can only be declared in a frontend. It is not possible to specify a capture
2647 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002648
2649 Example:
2650 capture request header Host len 15
2651 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002652 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002653
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002654 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002655 about logging.
2656
2657
2658capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002659 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002660 May be used in sections : defaults | frontend | listen | backend
2661 no | yes | yes | no
2662 Arguments :
2663 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002664 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002665 appear in the response, with the first letter of each word in
2666 upper case. The header name will not appear in the logs, only the
2667 value is reported, but the position in the logs is respected.
2668
2669 <length> is the maximum number of characters to extract from the value and
2670 report in the logs. The string will be truncated on the right if
2671 it exceeds <length>.
2672
Willy Tarreau4460d032012-11-21 23:37:37 +01002673 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002674 result will be added to the logs between braces ('{}') after the captured
2675 request headers. If multiple headers are captured, they will be delimited by
2676 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002677 the configuration. Non-existent headers will be logged just as an empty
2678 string. Common uses for response header captures include the "Content-length"
2679 header which indicates how many bytes are expected to be returned, the
2680 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002681
Willy Tarreau0900abb2012-11-22 00:21:46 +01002682 There is no limit to the number of captured response headers nor to their
2683 length, though it is wise to keep them low to limit memory usage per session.
2684 In order to keep log format consistent for a same frontend, header captures
2685 can only be declared in a frontend. It is not possible to specify a capture
2686 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002687
2688 Example:
2689 capture response header Content-length len 9
2690 capture response header Location len 15
2691
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002692 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002693 about logging.
2694
2695
Cyril Bontéf0c60612010-02-06 14:44:47 +01002696clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002697 Set the maximum inactivity time on the client side.
2698 May be used in sections : defaults | frontend | listen | backend
2699 yes | yes | yes | no
2700 Arguments :
2701 <timeout> is the timeout value is specified in milliseconds by default, but
2702 can be in any other unit if the number is suffixed by the unit,
2703 as explained at the top of this document.
2704
2705 The inactivity timeout applies when the client is expected to acknowledge or
2706 send data. In HTTP mode, this timeout is particularly important to consider
2707 during the first phase, when the client sends the request, and during the
2708 response while it is reading data sent by the server. The value is specified
2709 in milliseconds by default, but can be in any other unit if the number is
2710 suffixed by the unit, as specified at the top of this document. In TCP mode
2711 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2712 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002713 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002714 losses by specifying timeouts that are slightly above multiples of 3 seconds
2715 (eg: 4 or 5 seconds).
2716
2717 This parameter is specific to frontends, but can be specified once for all in
2718 "defaults" sections. This is in fact one of the easiest solutions not to
2719 forget about it. An unspecified timeout results in an infinite timeout, which
2720 is not recommended. Such a usage is accepted and works but reports a warning
2721 during startup because it may results in accumulation of expired sessions in
2722 the system if the system's timeouts are not configured either.
2723
2724 This parameter is provided for compatibility but is currently deprecated.
2725 Please use "timeout client" instead.
2726
Willy Tarreau036fae02008-01-06 13:24:40 +01002727 See also : "timeout client", "timeout http-request", "timeout server", and
2728 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002729
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002730compression algo <algorithm> ...
2731compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002732compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002733 Enable HTTP compression.
2734 May be used in sections : defaults | frontend | listen | backend
2735 yes | yes | yes | yes
2736 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002737 algo is followed by the list of supported compression algorithms.
2738 type is followed by the list of MIME types that will be compressed.
2739 offload makes haproxy work as a compression offloader only (see notes).
2740
2741 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002742 identity this is mostly for debugging, and it was useful for developing
2743 the compression feature. Identity does not apply any change on
2744 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002745
Willy Tarreauc91840a2015-03-28 17:00:39 +01002746 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01002747 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01002748
2749 deflate same as "gzip", but with deflate algorithm and zlib format.
2750 Note that this algorithm has ambiguous support on many
2751 browsers and no support at all from recent ones. It is
2752 strongly recommended not to use it for anything else than
2753 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01002754 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002755
Willy Tarreauc91840a2015-03-28 17:00:39 +01002756 raw-deflate same as "deflate" without the zlib wrapper, and used as an
2757 alternative when the browser wants "deflate". All major
2758 browsers understand it and despite violating the standards,
2759 it is known to work better than "deflate", at least on MSIE
2760 and some versions of Safari. Do not use it in conjunction
2761 with "deflate", use either one or the other since both react
2762 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01002763 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002764
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002765 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002766 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002767 If backend servers support HTTP compression, these directives
2768 will be no-op: haproxy will see the compressed response and will not
2769 compress again. If backend servers do not support HTTP compression and
2770 there is Accept-Encoding header in request, haproxy will compress the
2771 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002772
2773 The "offload" setting makes haproxy remove the Accept-Encoding header to
2774 prevent backend servers from compressing responses. It is strongly
2775 recommended not to do this because this means that all the compression work
2776 will be done on the single point where haproxy is located. However in some
2777 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002778 with broken HTTP compression implementation which can't be turned off.
2779 In that case haproxy can be used to prevent that gateway from emitting
2780 invalid payloads. In this case, simply removing the header in the
2781 configuration does not work because it applies before the header is parsed,
2782 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02002783 then be used for such scenarios. Note: for now, the "offload" setting is
2784 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002785
William Lallemand05097442012-11-20 12:14:28 +01002786 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002787 * the request does not advertise a supported compression algorithm in the
2788 "Accept-Encoding" header
2789 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002790 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002791 * response header "Transfer-Encoding" contains "chunked" (Temporary
2792 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002793 * response contain neither a "Content-Length" header nor a
2794 "Transfer-Encoding" whose last value is "chunked"
2795 * response contains a "Content-Type" header whose first value starts with
2796 "multipart"
2797 * the response contains the "no-transform" value in the "Cache-control"
2798 header
2799 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2800 and later
2801 * The response contains a "Content-Encoding" header, indicating that the
2802 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002803
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002804 Note: The compression does not rewrite Etag headers, and does not emit the
2805 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002806
William Lallemand82fe75c2012-10-23 10:25:10 +02002807 Examples :
2808 compression algo gzip
2809 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002810
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002811
Cyril Bontéf0c60612010-02-06 14:44:47 +01002812contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002813 Set the maximum time to wait for a connection attempt to a server to succeed.
2814 May be used in sections : defaults | frontend | listen | backend
2815 yes | no | yes | yes
2816 Arguments :
2817 <timeout> is the timeout value is specified in milliseconds by default, but
2818 can be in any other unit if the number is suffixed by the unit,
2819 as explained at the top of this document.
2820
2821 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002822 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002823 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01002824 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
2825 connect timeout also presets the queue timeout to the same value if this one
2826 has not been specified. Historically, the contimeout was also used to set the
2827 tarpit timeout in a listen section, which is not possible in a pure frontend.
2828
2829 This parameter is specific to backends, but can be specified once for all in
2830 "defaults" sections. This is in fact one of the easiest solutions not to
2831 forget about it. An unspecified timeout results in an infinite timeout, which
2832 is not recommended. Such a usage is accepted and works but reports a warning
2833 during startup because it may results in accumulation of failed sessions in
2834 the system if the system's timeouts are not configured either.
2835
2836 This parameter is provided for backwards compatibility but is currently
2837 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2838 instead.
2839
2840 See also : "timeout connect", "timeout queue", "timeout tarpit",
2841 "timeout server", "contimeout".
2842
2843
Willy Tarreau55165fe2009-05-10 12:02:55 +02002844cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002845 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2846 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01002847 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002848 Enable cookie-based persistence in a backend.
2849 May be used in sections : defaults | frontend | listen | backend
2850 yes | no | yes | yes
2851 Arguments :
2852 <name> is the name of the cookie which will be monitored, modified or
2853 inserted in order to bring persistence. This cookie is sent to
2854 the client via a "Set-Cookie" header in the response, and is
2855 brought back by the client in a "Cookie" header in all requests.
2856 Special care should be taken to choose a name which does not
2857 conflict with any likely application cookie. Also, if the same
2858 backends are subject to be used by the same clients (eg:
2859 HTTP/HTTPS), care should be taken to use different cookie names
2860 between all backends if persistence between them is not desired.
2861
2862 rewrite This keyword indicates that the cookie will be provided by the
2863 server and that haproxy will have to modify its value to set the
2864 server's identifier in it. This mode is handy when the management
2865 of complex combinations of "Set-cookie" and "Cache-control"
2866 headers is left to the application. The application can then
2867 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01002868 cookie. Since all responses should be monitored, this mode
2869 doesn't work in HTTP tunnel mode. Unless the application
2870 behaviour is very complex and/or broken, it is advised not to
2871 start with this mode for new deployments. This keyword is
2872 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002873
2874 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02002875 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002876
Willy Tarreaua79094d2010-08-31 22:54:15 +02002877 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002878 server. When used without the "preserve" option, if the server
2879 emits a cookie with the same name, it will be remove before
2880 processing. For this reason, this mode can be used to upgrade
2881 existing configurations running in the "rewrite" mode. The cookie
2882 will only be a session cookie and will not be stored on the
2883 client's disk. By default, unless the "indirect" option is added,
2884 the server will see the cookies emitted by the client. Due to
2885 caching effects, it is generally wise to add the "nocache" or
2886 "postonly" keywords (see below). The "insert" keyword is not
2887 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002888
2889 prefix This keyword indicates that instead of relying on a dedicated
2890 cookie for the persistence, an existing one will be completed.
2891 This may be needed in some specific environments where the client
2892 does not support more than one single cookie and the application
2893 already needs it. In this case, whenever the server sets a cookie
2894 named <name>, it will be prefixed with the server's identifier
2895 and a delimiter. The prefix will be removed from all client
2896 requests so that the server still finds the cookie it emitted.
2897 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01002898 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02002899 not compatible with "rewrite" and "insert". Note: it is highly
2900 recommended not to use "indirect" with "prefix", otherwise server
2901 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002902
Willy Tarreaua79094d2010-08-31 22:54:15 +02002903 indirect When this option is specified, no cookie will be emitted to a
2904 client which already has a valid one for the server which has
2905 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002906 it will be removed, unless the "preserve" option is also set. In
2907 "insert" mode, this will additionally remove cookies from the
2908 requests transmitted to the server, making the persistence
2909 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02002910 Note: it is highly recommended not to use "indirect" with
2911 "prefix", otherwise server cookie updates would not be sent to
2912 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002913
2914 nocache This option is recommended in conjunction with the insert mode
2915 when there is a cache between the client and HAProxy, as it
2916 ensures that a cacheable response will be tagged non-cacheable if
2917 a cookie needs to be inserted. This is important because if all
2918 persistence cookies are added on a cacheable home page for
2919 instance, then all customers will then fetch the page from an
2920 outer cache and will all share the same persistence cookie,
2921 leading to one server receiving much more traffic than others.
2922 See also the "insert" and "postonly" options.
2923
2924 postonly This option ensures that cookie insertion will only be performed
2925 on responses to POST requests. It is an alternative to the
2926 "nocache" option, because POST responses are not cacheable, so
2927 this ensures that the persistence cookie will never get cached.
2928 Since most sites do not need any sort of persistence before the
2929 first POST which generally is a login request, this is a very
2930 efficient method to optimize caching without risking to find a
2931 persistence cookie in the cache.
2932 See also the "insert" and "nocache" options.
2933
Willy Tarreauba4c5be2010-10-23 12:46:42 +02002934 preserve This option may only be used with "insert" and/or "indirect". It
2935 allows the server to emit the persistence cookie itself. In this
2936 case, if a cookie is found in the response, haproxy will leave it
2937 untouched. This is useful in order to end persistence after a
2938 logout request for instance. For this, the server just has to
2939 emit a cookie with an invalid value (eg: empty) or with a date in
2940 the past. By combining this mechanism with the "disable-on-404"
2941 check option, it is possible to perform a completely graceful
2942 shutdown because users will definitely leave the server after
2943 they logout.
2944
Willy Tarreau4992dd22012-05-31 21:02:17 +02002945 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
2946 when a cookie is inserted. This attribute is used so that a
2947 user agent doesn't share the cookie with non-HTTP components.
2948 Please check RFC6265 for more information on this attribute.
2949
2950 secure This option tells haproxy to add a "Secure" cookie attribute when
2951 a cookie is inserted. This attribute is used so that a user agent
2952 never emits this cookie over non-secure channels, which means
2953 that a cookie learned with this flag will be presented only over
2954 SSL/TLS connections. Please check RFC6265 for more information on
2955 this attribute.
2956
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002957 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002958 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01002959 name. If the domain begins with a dot, the browser is allowed to
2960 use it for any host ending with that name. It is also possible to
2961 specify several domain names by invoking this option multiple
2962 times. Some browsers might have small limits on the number of
2963 domains, so be careful when doing that. For the record, sending
2964 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02002965
Willy Tarreau996a92c2010-10-13 19:30:47 +02002966 maxidle This option allows inserted cookies to be ignored after some idle
2967 time. It only works with insert-mode cookies. When a cookie is
2968 sent to the client, the date this cookie was emitted is sent too.
2969 Upon further presentations of this cookie, if the date is older
2970 than the delay indicated by the parameter (in seconds), it will
2971 be ignored. Otherwise, it will be refreshed if needed when the
2972 response is sent to the client. This is particularly useful to
2973 prevent users who never close their browsers from remaining for
2974 too long on the same server (eg: after a farm size change). When
2975 this option is set and a cookie has no date, it is always
2976 accepted, but gets refreshed in the response. This maintains the
2977 ability for admins to access their sites. Cookies that have a
2978 date in the future further than 24 hours are ignored. Doing so
2979 lets admins fix timezone issues without risking kicking users off
2980 the site.
2981
2982 maxlife This option allows inserted cookies to be ignored after some life
2983 time, whether they're in use or not. It only works with insert
2984 mode cookies. When a cookie is first sent to the client, the date
2985 this cookie was emitted is sent too. Upon further presentations
2986 of this cookie, if the date is older than the delay indicated by
2987 the parameter (in seconds), it will be ignored. If the cookie in
2988 the request has no date, it is accepted and a date will be set.
2989 Cookies that have a date in the future further than 24 hours are
2990 ignored. Doing so lets admins fix timezone issues without risking
2991 kicking users off the site. Contrary to maxidle, this value is
2992 not refreshed, only the first visit date counts. Both maxidle and
2993 maxlife may be used at the time. This is particularly useful to
2994 prevent users who never close their browsers from remaining for
2995 too long on the same server (eg: after a farm size change). This
2996 is stronger than the maxidle method in that it forces a
2997 redispatch after some absolute delay.
2998
Olivier Houchard4e694042017-03-14 20:01:29 +01002999 dynamic Activate dynamic cookies. When used, a session cookie is
3000 dynamically created for each server, based on the IP and port
3001 of the server, and a secret key, specified in the
3002 "dynamic-cookie-key" backend directive.
3003 The cookie will be regenerated each time the IP address change,
3004 and is only generated for IPv4/IPv6.
3005
Willy Tarreau0ba27502007-12-24 16:55:16 +01003006 There can be only one persistence cookie per HTTP backend, and it can be
3007 declared in a defaults section. The value of the cookie will be the value
3008 indicated after the "cookie" keyword in a "server" statement. If no cookie
3009 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003010
Willy Tarreau0ba27502007-12-24 16:55:16 +01003011 Examples :
3012 cookie JSESSIONID prefix
3013 cookie SRV insert indirect nocache
3014 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003015 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003016
Willy Tarreau294d0f02015-08-10 19:40:12 +02003017 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003018
Willy Tarreau983e01e2010-01-11 18:42:06 +01003019
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003020declare capture [ request | response ] len <length>
3021 Declares a capture slot.
3022 May be used in sections : defaults | frontend | listen | backend
3023 no | yes | yes | no
3024 Arguments:
3025 <length> is the length allowed for the capture.
3026
3027 This declaration is only available in the frontend or listen section, but the
3028 reserved slot can be used in the backends. The "request" keyword allocates a
3029 capture slot for use in the request, and "response" allocates a capture slot
3030 for use in the response.
3031
3032 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003033 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003034 "http-request capture" and "http-response capture".
3035
3036
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003037default-server [param*]
3038 Change default options for a server in a backend
3039 May be used in sections : defaults | frontend | listen | backend
3040 yes | no | yes | yes
3041 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003042 <param*> is a list of parameters for this server. The "default-server"
3043 keyword accepts an important number of options and has a complete
3044 section dedicated to it. Please refer to section 5 for more
3045 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003046
Willy Tarreau983e01e2010-01-11 18:42:06 +01003047 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003048 default-server inter 1000 weight 13
3049
3050 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003051
Willy Tarreau983e01e2010-01-11 18:42:06 +01003052
Willy Tarreau0ba27502007-12-24 16:55:16 +01003053default_backend <backend>
3054 Specify the backend to use when no "use_backend" rule has been matched.
3055 May be used in sections : defaults | frontend | listen | backend
3056 yes | yes | yes | no
3057 Arguments :
3058 <backend> is the name of the backend to use.
3059
3060 When doing content-switching between frontend and backends using the
3061 "use_backend" keyword, it is often useful to indicate which backend will be
3062 used when no rule has matched. It generally is the dynamic backend which
3063 will catch all undetermined requests.
3064
Willy Tarreau0ba27502007-12-24 16:55:16 +01003065 Example :
3066
3067 use_backend dynamic if url_dyn
3068 use_backend static if url_css url_img extension_img
3069 default_backend dynamic
3070
Willy Tarreau98d04852015-05-26 12:18:29 +02003071 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003072
Willy Tarreau0ba27502007-12-24 16:55:16 +01003073
Baptiste Assmann27f51342013-10-09 06:51:49 +02003074description <string>
3075 Describe a listen, frontend or backend.
3076 May be used in sections : defaults | frontend | listen | backend
3077 no | yes | yes | yes
3078 Arguments : string
3079
3080 Allows to add a sentence to describe the related object in the HAProxy HTML
3081 stats page. The description will be printed on the right of the object name
3082 it describes.
3083 No need to backslash spaces in the <string> arguments.
3084
3085
Willy Tarreau0ba27502007-12-24 16:55:16 +01003086disabled
3087 Disable a proxy, frontend or backend.
3088 May be used in sections : defaults | frontend | listen | backend
3089 yes | yes | yes | yes
3090 Arguments : none
3091
3092 The "disabled" keyword is used to disable an instance, mainly in order to
3093 liberate a listening port or to temporarily disable a service. The instance
3094 will still be created and its configuration will be checked, but it will be
3095 created in the "stopped" state and will appear as such in the statistics. It
3096 will not receive any traffic nor will it send any health-checks or logs. It
3097 is possible to disable many instances at once by adding the "disabled"
3098 keyword in a "defaults" section.
3099
3100 See also : "enabled"
3101
3102
Willy Tarreau5ce94572010-06-07 14:35:41 +02003103dispatch <address>:<port>
3104 Set a default server address
3105 May be used in sections : defaults | frontend | listen | backend
3106 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003107 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003108
3109 <address> is the IPv4 address of the default server. Alternatively, a
3110 resolvable hostname is supported, but this name will be resolved
3111 during start-up.
3112
3113 <ports> is a mandatory port specification. All connections will be sent
3114 to this port, and it is not permitted to use port offsets as is
3115 possible with normal servers.
3116
Willy Tarreau787aed52011-04-15 06:45:37 +02003117 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003118 server can take the connection. In the past it was used to forward non
3119 persistent connections to an auxiliary load balancer. Due to its simple
3120 syntax, it has also been used for simple TCP relays. It is recommended not to
3121 use it for more clarity, and to use the "server" directive instead.
3122
3123 See also : "server"
3124
Olivier Houchard4e694042017-03-14 20:01:29 +01003125
3126dynamic-cookie-key <string>
3127 Set the dynamic cookie secret key for a backend.
3128 May be used in sections : defaults | frontend | listen | backend
3129 yes | no | yes | yes
3130 Arguments : The secret key to be used.
3131
3132 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
3133 a dynamic cookie is created for each server (unless one is explicitely
3134 specified on the "server" line), using a hash of the IP address of the
3135 server, the TCP port, and the secret key.
3136 That way, we can ensure session persistence accross multiple load-balancers,
3137 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003138
Willy Tarreau0ba27502007-12-24 16:55:16 +01003139enabled
3140 Enable a proxy, frontend or backend.
3141 May be used in sections : defaults | frontend | listen | backend
3142 yes | yes | yes | yes
3143 Arguments : none
3144
3145 The "enabled" keyword is used to explicitly enable an instance, when the
3146 defaults has been set to "disabled". This is very rarely used.
3147
3148 See also : "disabled"
3149
3150
3151errorfile <code> <file>
3152 Return a file contents instead of errors generated by HAProxy
3153 May be used in sections : defaults | frontend | listen | backend
3154 yes | yes | yes | yes
3155 Arguments :
3156 <code> is the HTTP status code. Currently, HAProxy is capable of
CJ Ess108b1dd2015-04-07 12:03:37 -04003157 generating codes 200, 400, 403, 405, 408, 429, 500, 502, 503, and
3158 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003159
3160 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003161 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003162 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003163 error pages, and to use absolute paths, since files are read
3164 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003165
3166 It is important to understand that this keyword is not meant to rewrite
3167 errors returned by the server, but errors detected and returned by HAProxy.
3168 This is why the list of supported errors is limited to a small set.
3169
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003170 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3171
Willy Tarreau0ba27502007-12-24 16:55:16 +01003172 The files are returned verbatim on the TCP socket. This allows any trick such
3173 as redirections to another URL or site, as well as tricks to clean cookies,
3174 force enable or disable caching, etc... The package provides default error
3175 files returning the same contents as default errors.
3176
Willy Tarreau59140a22009-02-22 12:02:30 +01003177 The files should not exceed the configured buffer size (BUFSIZE), which
3178 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
3179 not to put any reference to local contents (eg: images) in order to avoid
3180 loops between the client and HAProxy when all servers are down, causing an
3181 error to be returned instead of an image. For better HTTP compliance, it is
3182 recommended that all header lines end with CR-LF and not LF alone.
3183
Willy Tarreau0ba27502007-12-24 16:55:16 +01003184 The files are read at the same time as the configuration and kept in memory.
3185 For this reason, the errors continue to be returned even when the process is
3186 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003187 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003188 403 status code and interrogating a blocked URL.
3189
3190 See also : "errorloc", "errorloc302", "errorloc303"
3191
Willy Tarreau59140a22009-02-22 12:02:30 +01003192 Example :
3193 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003194 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003195 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3196 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3197
Willy Tarreau2769aa02007-12-27 18:26:09 +01003198
3199errorloc <code> <url>
3200errorloc302 <code> <url>
3201 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3202 May be used in sections : defaults | frontend | listen | backend
3203 yes | yes | yes | yes
3204 Arguments :
3205 <code> is the HTTP status code. Currently, HAProxy is capable of
Jarno Huuskonen013a84f2017-04-22 11:26:50 +03003206 generating codes 200, 400, 403, 405, 408, 429, 500, 502, 503, and
3207 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003208
3209 <url> it is the exact contents of the "Location" header. It may contain
3210 either a relative URI to an error page hosted on the same site,
3211 or an absolute URI designating an error page on another site.
3212 Special care should be given to relative URIs to avoid redirect
3213 loops if the URI itself may generate the same error (eg: 500).
3214
3215 It is important to understand that this keyword is not meant to rewrite
3216 errors returned by the server, but errors detected and returned by HAProxy.
3217 This is why the list of supported errors is limited to a small set.
3218
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003219 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3220
Willy Tarreau2769aa02007-12-27 18:26:09 +01003221 Note that both keyword return the HTTP 302 status code, which tells the
3222 client to fetch the designated URL using the same HTTP method. This can be
3223 quite problematic in case of non-GET methods such as POST, because the URL
3224 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003225 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003226 status code, indicating to the client that the URL must be fetched with a GET
3227 request.
3228
3229 See also : "errorfile", "errorloc303"
3230
3231
3232errorloc303 <code> <url>
3233 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3234 May be used in sections : defaults | frontend | listen | backend
3235 yes | yes | yes | yes
3236 Arguments :
3237 <code> is the HTTP status code. Currently, HAProxy is capable of
Jarno Huuskonen013a84f2017-04-22 11:26:50 +03003238 generating codes 200, 400, 403, 405, 408, 429, 500, 502, 503, and
3239 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003240
3241 <url> it is the exact contents of the "Location" header. It may contain
3242 either a relative URI to an error page hosted on the same site,
3243 or an absolute URI designating an error page on another site.
3244 Special care should be given to relative URIs to avoid redirect
3245 loops if the URI itself may generate the same error (eg: 500).
3246
3247 It is important to understand that this keyword is not meant to rewrite
3248 errors returned by the server, but errors detected and returned by HAProxy.
3249 This is why the list of supported errors is limited to a small set.
3250
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003251 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3252
Willy Tarreau2769aa02007-12-27 18:26:09 +01003253 Note that both keyword return the HTTP 303 status code, which tells the
3254 client to fetch the designated URL using the same HTTP GET method. This
3255 solves the usual problems associated with "errorloc" and the 302 code. It is
3256 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003257 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003258
3259 See also : "errorfile", "errorloc", "errorloc302"
3260
3261
Simon Horman51a1cf62015-02-03 13:00:44 +09003262email-alert from <emailaddr>
3263 Declare the from email address to be used in both the envelope and header
3264 of email alerts. This is the address that email alerts are sent from.
3265 May be used in sections: defaults | frontend | listen | backend
3266 yes | yes | yes | yes
3267
3268 Arguments :
3269
3270 <emailaddr> is the from email address to use when sending email alerts
3271
3272 Also requires "email-alert mailers" and "email-alert to" to be set
3273 and if so sending email alerts is enabled for the proxy.
3274
Simon Horman64e34162015-02-06 11:11:57 +09003275 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003276 "email-alert myhostname", "email-alert to", section 3.6 about
3277 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003278
3279
3280email-alert level <level>
3281 Declare the maximum log level of messages for which email alerts will be
3282 sent. This acts as a filter on the sending of email alerts.
3283 May be used in sections: defaults | frontend | listen | backend
3284 yes | yes | yes | yes
3285
3286 Arguments :
3287
3288 <level> One of the 8 syslog levels:
3289 emerg alert crit err warning notice info debug
3290 The above syslog levels are ordered from lowest to highest.
3291
3292 By default level is alert
3293
3294 Also requires "email-alert from", "email-alert mailers" and
3295 "email-alert to" to be set and if so sending email alerts is enabled
3296 for the proxy.
3297
Simon Horman1421e212015-04-30 13:10:35 +09003298 Alerts are sent when :
3299
3300 * An un-paused server is marked as down and <level> is alert or lower
3301 * A paused server is marked as down and <level> is notice or lower
3302 * A server is marked as up or enters the drain state and <level>
3303 is notice or lower
3304 * "option log-health-checks" is enabled, <level> is info or lower,
3305 and a health check status update occurs
3306
Simon Horman64e34162015-02-06 11:11:57 +09003307 See also : "email-alert from", "email-alert mailers",
3308 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003309 section 3.6 about mailers.
3310
3311
3312email-alert mailers <mailersect>
3313 Declare the mailers to be used when sending email alerts
3314 May be used in sections: defaults | frontend | listen | backend
3315 yes | yes | yes | yes
3316
3317 Arguments :
3318
3319 <mailersect> is the name of the mailers section to send email alerts.
3320
3321 Also requires "email-alert from" and "email-alert to" to be set
3322 and if so sending email alerts is enabled for the proxy.
3323
Simon Horman64e34162015-02-06 11:11:57 +09003324 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3325 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003326
3327
3328email-alert myhostname <hostname>
3329 Declare the to hostname address to be used when communicating with
3330 mailers.
3331 May be used in sections: defaults | frontend | listen | backend
3332 yes | yes | yes | yes
3333
3334 Arguments :
3335
Baptiste Assmann738bad92015-12-21 15:27:53 +01003336 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003337
3338 By default the systems hostname is used.
3339
3340 Also requires "email-alert from", "email-alert mailers" and
3341 "email-alert to" to be set and if so sending email alerts is enabled
3342 for the proxy.
3343
Simon Horman64e34162015-02-06 11:11:57 +09003344 See also : "email-alert from", "email-alert level", "email-alert mailers",
3345 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003346
3347
3348email-alert to <emailaddr>
3349 Declare both the recipent address in the envelope and to address in the
3350 header of email alerts. This is the address that email alerts are sent to.
3351 May be used in sections: defaults | frontend | listen | backend
3352 yes | yes | yes | yes
3353
3354 Arguments :
3355
3356 <emailaddr> is the to email address to use when sending email alerts
3357
3358 Also requires "email-alert mailers" and "email-alert to" to be set
3359 and if so sending email alerts is enabled for the proxy.
3360
Simon Horman64e34162015-02-06 11:11:57 +09003361 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003362 "email-alert myhostname", section 3.6 about mailers.
3363
3364
Willy Tarreau4de91492010-01-22 19:10:05 +01003365force-persist { if | unless } <condition>
3366 Declare a condition to force persistence on down servers
3367 May be used in sections: defaults | frontend | listen | backend
3368 no | yes | yes | yes
3369
3370 By default, requests are not dispatched to down servers. It is possible to
3371 force this using "option persist", but it is unconditional and redispatches
3372 to a valid server if "option redispatch" is set. That leaves with very little
3373 possibilities to force some requests to reach a server which is artificially
3374 marked down for maintenance operations.
3375
3376 The "force-persist" statement allows one to declare various ACL-based
3377 conditions which, when met, will cause a request to ignore the down status of
3378 a server and still try to connect to it. That makes it possible to start a
3379 server, still replying an error to the health checks, and run a specially
3380 configured browser to test the service. Among the handy methods, one could
3381 use a specific source IP address, or a specific cookie. The cookie also has
3382 the advantage that it can easily be added/removed on the browser from a test
3383 page. Once the service is validated, it is then possible to open the service
3384 to the world by returning a valid response to health checks.
3385
3386 The forced persistence is enabled when an "if" condition is met, or unless an
3387 "unless" condition is met. The final redispatch is always disabled when this
3388 is used.
3389
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003390 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003391 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003392
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003393
3394filter <name> [param*]
3395 Add the filter <name> in the filter list attached to the proxy.
3396 May be used in sections : defaults | frontend | listen | backend
3397 no | yes | yes | yes
3398 Arguments :
3399 <name> is the name of the filter. Officially supported filters are
3400 referenced in section 9.
3401
Tim Düsterhus4896c442016-11-29 02:15:19 +01003402 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003403 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003404 filter. Please refer to the documentation of the corresponding
3405 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003406
3407 Multiple occurrences of the filter line can be used for the same proxy. The
3408 same filter can be referenced many times if needed.
3409
3410 Example:
3411 listen
3412 bind *:80
3413
3414 filter trace name BEFORE-HTTP-COMP
3415 filter compression
3416 filter trace name AFTER-HTTP-COMP
3417
3418 compression algo gzip
3419 compression offload
3420
3421 server srv1 192.168.0.1:80
3422
3423 See also : section 9.
3424
Willy Tarreau4de91492010-01-22 19:10:05 +01003425
Willy Tarreau2769aa02007-12-27 18:26:09 +01003426fullconn <conns>
3427 Specify at what backend load the servers will reach their maxconn
3428 May be used in sections : defaults | frontend | listen | backend
3429 yes | no | yes | yes
3430 Arguments :
3431 <conns> is the number of connections on the backend which will make the
3432 servers use the maximal number of connections.
3433
Willy Tarreau198a7442008-01-17 12:05:32 +01003434 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003435 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003436 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003437 load. The server will then always accept at least <minconn> connections,
3438 never more than <maxconn>, and the limit will be on the ramp between both
3439 values when the backend has less than <conns> concurrent connections. This
3440 makes it possible to limit the load on the servers during normal loads, but
3441 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003442 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003443
Willy Tarreaufbb78422011-06-05 15:38:35 +02003444 Since it's hard to get this value right, haproxy automatically sets it to
3445 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003446 backend (based on "use_backend" and "default_backend" rules). That way it's
3447 safe to leave it unset. However, "use_backend" involving dynamic names are
3448 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003449
Willy Tarreau2769aa02007-12-27 18:26:09 +01003450 Example :
3451 # The servers will accept between 100 and 1000 concurrent connections each
3452 # and the maximum of 1000 will be reached when the backend reaches 10000
3453 # connections.
3454 backend dynamic
3455 fullconn 10000
3456 server srv1 dyn1:80 minconn 100 maxconn 1000
3457 server srv2 dyn2:80 minconn 100 maxconn 1000
3458
3459 See also : "maxconn", "server"
3460
3461
3462grace <time>
3463 Maintain a proxy operational for some time after a soft stop
3464 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003465 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003466 Arguments :
3467 <time> is the time (by default in milliseconds) for which the instance
3468 will remain operational with the frontend sockets still listening
3469 when a soft-stop is received via the SIGUSR1 signal.
3470
3471 This may be used to ensure that the services disappear in a certain order.
3472 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003473 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003474 needed by the equipment to detect the failure.
3475
3476 Note that currently, there is very little benefit in using this parameter,
3477 and it may in fact complicate the soft-reconfiguration process more than
3478 simplify it.
3479
Willy Tarreau0ba27502007-12-24 16:55:16 +01003480
Andrew Rodland17be45e2016-10-25 17:04:12 -04003481hash-balance-factor <factor>
3482 Specify the balancing factor for bounded-load consistent hashing
3483 May be used in sections : defaults | frontend | listen | backend
3484 yes | no | no | yes
3485 Arguments :
3486 <factor> is the control for the maximum number of concurrent requests to
3487 send to a server, expressed as a percentage of the average number
3488 of concurrent requests across all of the active servers.
3489
3490 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3491 enables an algorithm that prevents any one server from getting too many
3492 requests at once, even if some hash buckets receive many more requests than
3493 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3494 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3495 then no server will be allowed to have a load more than 1.5 times the average.
3496 If server weights are used, they will be respected.
3497
3498 If the first-choice server is disqualified, the algorithm will choose another
3499 server based on the request hash, until a server with additional capacity is
3500 found. A higher <factor> allows more imbalance between the servers, while a
3501 lower <factor> means that more servers will be checked on average, affecting
3502 performance. Reasonable values are from 125 to 200.
3503
3504 See also : "balance" and "hash-type".
3505
3506
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003507hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003508 Specify a method to use for mapping hashes to servers
3509 May be used in sections : defaults | frontend | listen | backend
3510 yes | no | yes | yes
3511 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003512 <method> is the method used to select a server from the hash computed by
3513 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003514
Bhaskar98634f02013-10-29 23:30:51 -04003515 map-based the hash table is a static array containing all alive servers.
3516 The hashes will be very smooth, will consider weights, but
3517 will be static in that weight changes while a server is up
3518 will be ignored. This means that there will be no slow start.
3519 Also, since a server is selected by its position in the array,
3520 most mappings are changed when the server count changes. This
3521 means that when a server goes up or down, or when a server is
3522 added to a farm, most connections will be redistributed to
3523 different servers. This can be inconvenient with caches for
3524 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003525
Bhaskar98634f02013-10-29 23:30:51 -04003526 consistent the hash table is a tree filled with many occurrences of each
3527 server. The hash key is looked up in the tree and the closest
3528 server is chosen. This hash is dynamic, it supports changing
3529 weights while the servers are up, so it is compatible with the
3530 slow start feature. It has the advantage that when a server
3531 goes up or down, only its associations are moved. When a
3532 server is added to the farm, only a few part of the mappings
3533 are redistributed, making it an ideal method for caches.
3534 However, due to its principle, the distribution will never be
3535 very smooth and it may sometimes be necessary to adjust a
3536 server's weight or its ID to get a more balanced distribution.
3537 In order to get the same distribution on multiple load
3538 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003539 same IDs. Note: consistent hash uses sdbm and avalanche if no
3540 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003541
3542 <function> is the hash function to be used :
3543
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003544 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003545 reimplementation of ndbm) database library. It was found to do
3546 well in scrambling bits, causing better distribution of the keys
3547 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003548 function with good distribution, unless the total server weight
3549 is a multiple of 64, in which case applying the avalanche
3550 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003551
3552 djb2 this function was first proposed by Dan Bernstein many years ago
3553 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003554 function provides a better distribution than sdbm. It generally
3555 works well with text-based inputs though it can perform extremely
3556 poorly with numeric-only input or when the total server weight is
3557 a multiple of 33, unless the avalanche modifier is also used.
3558
Willy Tarreaua0f42712013-11-14 14:30:35 +01003559 wt6 this function was designed for haproxy while testing other
3560 functions in the past. It is not as smooth as the other ones, but
3561 is much less sensible to the input data set or to the number of
3562 servers. It can make sense as an alternative to sdbm+avalanche or
3563 djb2+avalanche for consistent hashing or when hashing on numeric
3564 data such as a source IP address or a visitor identifier in a URL
3565 parameter.
3566
Willy Tarreau324f07f2015-01-20 19:44:50 +01003567 crc32 this is the most common CRC32 implementation as used in Ethernet,
3568 gzip, PNG, etc. It is slower than the other ones but may provide
3569 a better distribution or less predictable results especially when
3570 used on strings.
3571
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003572 <modifier> indicates an optional method applied after hashing the key :
3573
3574 avalanche This directive indicates that the result from the hash
3575 function above should not be used in its raw form but that
3576 a 4-byte full avalanche hash must be applied first. The
3577 purpose of this step is to mix the resulting bits from the
3578 previous hash in order to avoid any undesired effect when
3579 the input contains some limited values or when the number of
3580 servers is a multiple of one of the hash's components (64
3581 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3582 result less predictable, but it's also not as smooth as when
3583 using the original function. Some testing might be needed
3584 with some workloads. This hash is one of the many proposed
3585 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003586
Bhaskar98634f02013-10-29 23:30:51 -04003587 The default hash type is "map-based" and is recommended for most usages. The
3588 default function is "sdbm", the selection of a function should be based on
3589 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003590
Andrew Rodland17be45e2016-10-25 17:04:12 -04003591 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003592
3593
Willy Tarreau0ba27502007-12-24 16:55:16 +01003594http-check disable-on-404
3595 Enable a maintenance mode upon HTTP/404 response to health-checks
3596 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003597 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003598 Arguments : none
3599
3600 When this option is set, a server which returns an HTTP code 404 will be
3601 excluded from further load-balancing, but will still receive persistent
3602 connections. This provides a very convenient method for Web administrators
3603 to perform a graceful shutdown of their servers. It is also important to note
3604 that a server which is detected as failed while it was in this mode will not
3605 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3606 will immediately be reinserted into the farm. The status on the stats page
3607 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003608 option only works in conjunction with the "httpchk" option. If this option
3609 is used with "http-check expect", then it has precedence over it so that 404
3610 responses will still be considered as soft-stop.
3611
3612 See also : "option httpchk", "http-check expect"
3613
3614
3615http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003616 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003617 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003618 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003619 Arguments :
3620 <match> is a keyword indicating how to look for a specific pattern in the
3621 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003622 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003623 exclamation mark ("!") to negate the match. Spaces are allowed
3624 between the exclamation mark and the keyword. See below for more
3625 details on the supported keywords.
3626
3627 <pattern> is the pattern to look for. It may be a string or a regular
3628 expression. If the pattern contains spaces, they must be escaped
3629 with the usual backslash ('\').
3630
3631 By default, "option httpchk" considers that response statuses 2xx and 3xx
3632 are valid, and that others are invalid. When "http-check expect" is used,
3633 it defines what is considered valid or invalid. Only one "http-check"
3634 statement is supported in a backend. If a server fails to respond or times
3635 out, the check obviously fails. The available matches are :
3636
3637 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003638 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003639 response's status code is exactly this string. If the
3640 "status" keyword is prefixed with "!", then the response
3641 will be considered invalid if the status code matches.
3642
3643 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003644 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003645 response's status code matches the expression. If the
3646 "rstatus" keyword is prefixed with "!", then the response
3647 will be considered invalid if the status code matches.
3648 This is mostly used to check for multiple codes.
3649
3650 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003651 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003652 response's body contains this exact string. If the
3653 "string" keyword is prefixed with "!", then the response
3654 will be considered invalid if the body contains this
3655 string. This can be used to look for a mandatory word at
3656 the end of a dynamic page, or to detect a failure when a
3657 specific error appears on the check page (eg: a stack
3658 trace).
3659
3660 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003661 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003662 response's body matches this expression. If the "rstring"
3663 keyword is prefixed with "!", then the response will be
3664 considered invalid if the body matches the expression.
3665 This can be used to look for a mandatory word at the end
3666 of a dynamic page, or to detect a failure when a specific
3667 error appears on the check page (eg: a stack trace).
3668
3669 It is important to note that the responses will be limited to a certain size
3670 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3671 Thus, too large responses may not contain the mandatory pattern when using
3672 "string" or "rstring". If a large response is absolutely required, it is
3673 possible to change the default max size by setting the global variable.
3674 However, it is worth keeping in mind that parsing very large responses can
3675 waste some CPU cycles, especially when regular expressions are used, and that
3676 it is always better to focus the checks on smaller resources.
3677
Cyril Bonté32602d22015-01-30 00:07:07 +01003678 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3679 will automatically append a "Connection: close" header, meaning that this
3680 header should not be present in the request provided by "option httpchk".
3681
Willy Tarreaubd741542010-03-16 18:46:54 +01003682 Last, if "http-check expect" is combined with "http-check disable-on-404",
3683 then this last one has precedence when the server responds with 404.
3684
3685 Examples :
3686 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003687 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003688
3689 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003690 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003691
3692 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003693 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003694
3695 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003696 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003697
Willy Tarreaubd741542010-03-16 18:46:54 +01003698 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003699
3700
Willy Tarreauef781042010-01-27 11:53:01 +01003701http-check send-state
3702 Enable emission of a state header with HTTP health checks
3703 May be used in sections : defaults | frontend | listen | backend
3704 yes | no | yes | yes
3705 Arguments : none
3706
3707 When this option is set, haproxy will systematically send a special header
3708 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3709 how they are seen by haproxy. This can be used for instance when a server is
3710 manipulated without access to haproxy and the operator needs to know whether
3711 haproxy still sees it up or not, or if the server is the last one in a farm.
3712
3713 The header is composed of fields delimited by semi-colons, the first of which
3714 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3715 checks on the total number before transition, just as appears in the stats
3716 interface. Next headers are in the form "<variable>=<value>", indicating in
3717 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003718 - a variable "address", containing the address of the backend server.
3719 This corresponds to the <address> field in the server declaration. For
3720 unix domain sockets, it will read "unix".
3721
3722 - a variable "port", containing the port of the backend server. This
3723 corresponds to the <port> field in the server declaration. For unix
3724 domain sockets, it will read "unix".
3725
Willy Tarreauef781042010-01-27 11:53:01 +01003726 - a variable "name", containing the name of the backend followed by a slash
3727 ("/") then the name of the server. This can be used when a server is
3728 checked in multiple backends.
3729
3730 - a variable "node" containing the name of the haproxy node, as set in the
3731 global "node" variable, otherwise the system's hostname if unspecified.
3732
3733 - a variable "weight" indicating the weight of the server, a slash ("/")
3734 and the total weight of the farm (just counting usable servers). This
3735 helps to know if other servers are available to handle the load when this
3736 one fails.
3737
3738 - a variable "scur" indicating the current number of concurrent connections
3739 on the server, followed by a slash ("/") then the total number of
3740 connections on all servers of the same backend.
3741
3742 - a variable "qcur" indicating the current number of requests in the
3743 server's queue.
3744
3745 Example of a header received by the application server :
3746 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
3747 scur=13/22; qcur=0
3748
3749 See also : "option httpchk", "http-check disable-on-404"
3750
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003751http-request { allow | auth [realm <realm>] | redirect <rule> |
3752 tarpit [deny_status <status>] | deny [deny_status <status>] |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003753 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003754 capture <sample> [ len <length> | id <id> ] |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003755 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003756 replace-header <name> <match-regex> <replace-fmt> |
3757 replace-value <name> <match-regex> <replace-fmt> |
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01003758 set-method <fmt> | set-path <fmt> | set-query <fmt> |
3759 set-uri <fmt> | set-tos <tos> | set-mark <mark> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003760 add-acl(<file name>) <key fmt> |
3761 del-acl(<file name>) <key fmt> |
3762 del-map(<file name>) <key fmt> |
Baptiste Assmannbb7e86a2014-09-03 18:29:47 +02003763 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02003764 set-var(<var name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01003765 unset-var(<var name>) |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01003766 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02003767 sc-inc-gpc0(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02003768 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02003769 silent-drop |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003770 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01003771 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003772 Access control for Layer 7 requests
3773
3774 May be used in sections: defaults | frontend | listen | backend
3775 no | yes | yes | yes
3776
Willy Tarreau20b0de52012-12-24 15:45:22 +01003777 The http-request statement defines a set of rules which apply to layer 7
3778 processing. The rules are evaluated in their declaration order when they are
3779 met in a frontend, listen or backend section. Any rule may optionally be
3780 followed by an ACL-based condition, in which case it will only be evaluated
3781 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003782
Willy Tarreau20b0de52012-12-24 15:45:22 +01003783 The first keyword is the rule's action. Currently supported actions include :
3784 - "allow" : this stops the evaluation of the rules and lets the request
3785 pass the check. No further "http-request" rules are evaluated.
3786
3787 - "deny" : this stops the evaluation of the rules and immediately rejects
Willy Tarreaube1d34d2016-06-26 19:37:59 +02003788 the request and emits an HTTP 403 error, or optionally the status code
3789 specified as an argument to "deny_status". The list of permitted status
3790 codes is limited to those that can be overridden by the "errorfile"
3791 directive. No further "http-request" rules are evaluated.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003792
Willy Tarreauccbcc372012-12-27 12:37:57 +01003793 - "tarpit" : this stops the evaluation of the rules and immediately blocks
3794 the request without responding for a delay specified by "timeout tarpit"
3795 or "timeout connect" if the former is not set. After that delay, if the
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003796 client is still connected, an HTTP error 500 (or optionally the status
3797 code specified as an argument to "deny_status") is returned so that the
Willy Tarreauccbcc372012-12-27 12:37:57 +01003798 client does not suspect it has been tarpitted. Logs will report the flags
3799 "PT". The goal of the tarpit rule is to slow down robots during an attack
3800 when they're limited on the number of concurrent requests. It can be very
3801 efficient against very dumb robots, and will significantly reduce the
3802 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003803 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreau2d392c22015-08-24 01:43:45 +02003804 front firewall to support insane number of concurrent connections. See
3805 also the "silent-drop" action below.
Willy Tarreauccbcc372012-12-27 12:37:57 +01003806
Willy Tarreau20b0de52012-12-24 15:45:22 +01003807 - "auth" : this stops the evaluation of the rules and immediately responds
3808 with an HTTP 401 or 407 error code to invite the user to present a valid
3809 user name and password. No further "http-request" rules are evaluated. An
3810 optional "realm" parameter is supported, it sets the authentication realm
3811 that is returned with the response (typically the application's name).
3812
Willy Tarreau81499eb2012-12-27 12:19:02 +01003813 - "redirect" : this performs an HTTP redirection based on a redirect rule.
3814 This is exactly the same as the "redirect" statement except that it
3815 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01003816 "http-request" rules and that these rules use the "log-format" strings.
3817 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01003818
Willy Tarreau20b0de52012-12-24 15:45:22 +01003819 - "add-header" appends an HTTP header field whose name is specified in
3820 <name> and whose value is defined by <fmt> which follows the log-format
3821 rules (see Custom Log Format in section 8.2.4). This is particularly
3822 useful to pass connection-specific information to the server (eg: the
3823 client's SSL certificate), or to combine several headers into one. This
3824 rule is not final, so it is possible to add other similar rules. Note
3825 that header addition is performed immediately, so one rule might reuse
3826 the resulting header from a previous rule.
3827
3828 - "set-header" does the same as "add-header" except that the header name
3829 is first removed if it existed. This is useful when passing security
3830 information to the server, where the header must not be manipulated by
Willy Tarreau85603282015-01-21 20:39:27 +01003831 external users. Note that the new value is computed before the removal so
3832 it is possible to concatenate a value to an existing header.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003833
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003834 - "del-header" removes all HTTP header fields whose name is specified in
3835 <name>.
3836
Sasha Pachev218f0642014-06-16 12:05:59 -06003837 - "replace-header" matches the regular expression in all occurrences of
3838 header field <name> according to <match-regex>, and replaces them with
3839 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3840 and work like in <fmt> arguments in "add-header". The match is only
3841 case-sensitive. It is important to understand that this action only
3842 considers whole header lines, regardless of the number of values they
3843 may contain. This usage is suited to headers naturally containing commas
3844 in their value, such as If-Modified-Since and so on.
3845
3846 Example:
3847
3848 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
3849
3850 applied to:
3851
3852 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3853
3854 outputs:
3855
3856 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3857
3858 assuming the backend IP is 192.168.1.20
3859
3860 - "replace-value" works like "replace-header" except that it matches the
3861 regex against every comma-delimited value of the header field <name>
3862 instead of the entire header. This is suited for all headers which are
3863 allowed to carry more than one value. An example could be the Accept
3864 header.
3865
3866 Example:
3867
3868 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
3869
3870 applied to:
3871
3872 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
3873
3874 outputs:
3875
3876 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
3877
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01003878 - "set-method" rewrites the request method with the result of the
3879 evaluation of format string <fmt>. There should be very few valid reasons
3880 for having to do so as this is more likely to break something than to fix
3881 it.
3882
3883 - "set-path" rewrites the request path with the result of the evaluation of
3884 format string <fmt>. The query string, if any, is left intact. If a
3885 scheme and authority is found before the path, they are left intact as
3886 well. If the request doesn't have a path ("*"), this one is replaced with
3887 the format. This can be used to prepend a directory component in front of
3888 a path for example. See also "set-query" and "set-uri".
3889
3890 Example :
3891 # prepend the host name before the path
3892 http-request set-path /%[hdr(host)]%[path]
3893
3894 - "set-query" rewrites the request's query string which appears after the
3895 first question mark ("?") with the result of the evaluation of format
3896 string <fmt>. The part prior to the question mark is left intact. If the
3897 request doesn't contain a question mark and the new value is not empty,
3898 then one is added at the end of the URI, followed by the new value. If
3899 a question mark was present, it will never be removed even if the value
3900 is empty. This can be used to add or remove parameters from the query
3901 string. See also "set-query" and "set-uri".
3902
3903 Example :
3904 # replace "%3D" with "=" in the query string
3905 http-request set-query %[query,regsub(%3D,=,g)]
3906
3907 - "set-uri" rewrites the request URI with the result of the evaluation of
3908 format string <fmt>. The scheme, authority, path and query string are all
3909 replaced at once. This can be used to rewrite hosts in front of proxies,
3910 or to perform complex modifications to the URI such as moving parts
3911 between the path and the query string. See also "set-path" and
3912 "set-query".
3913
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003914 - "set-nice" sets the "nice" factor of the current request being processed.
3915 It only has effect against the other requests being processed at the same
3916 time. The default value is 0, unless altered by the "nice" setting on the
3917 "bind" line. The accepted range is -1024..1024. The higher the value, the
3918 nicest the request will be. Lower values will make the request more
3919 important than other ones. This can be useful to improve the speed of
3920 some requests, or lower the priority of non-important requests. Using
3921 this setting without prior experimentation can cause some major slowdown.
3922
Willy Tarreau9a355ec2013-06-11 17:45:46 +02003923 - "set-log-level" is used to change the log level of the current request
3924 when a certain condition is met. Valid levels are the 8 syslog levels
3925 (see the "log" keyword) plus the special level "silent" which disables
3926 logging for this request. This rule is not final so the last matching
3927 rule wins. This rule can be useful to disable health checks coming from
3928 another equipment.
3929
Willy Tarreau42cf39e2013-06-11 18:51:32 +02003930 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
3931 the client to the value passed in <tos> on platforms which support this.
3932 This value represents the whole 8 bits of the IP TOS field, and can be
3933 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
3934 that only the 6 higher bits are used in DSCP or TOS, and the two lower
3935 bits are always 0. This can be used to adjust some routing behaviour on
3936 border routers based on some information from the request. See RFC 2474,
3937 2597, 3260 and 4594 for more information.
3938
Willy Tarreau51347ed2013-06-11 19:34:13 +02003939 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
3940 client to the value passed in <mark> on platforms which support it. This
3941 value is an unsigned 32 bit value which can be matched by netfilter and
3942 by the routing table. It can be expressed both in decimal or hexadecimal
3943 format (prefixed by "0x"). This can be useful to force certain packets to
3944 take a different route (for example a cheaper network path for bulk
3945 downloads). This works on Linux kernels 2.6.32 and above and requires
3946 admin privileges.
3947
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003948 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
3949 from a file (even a dummy empty file). The file name of the ACL to be
3950 updated is passed between parentheses. It takes one argument: <key fmt>,
3951 which follows log-format rules, to collect content of the new entry. It
3952 performs a lookup in the ACL before insertion, to avoid duplicated (or
3953 more) values. This lookup is done by a linear search and can be expensive
3954 with large lists! It is the equivalent of the "add acl" command from the
3955 stats socket, but can be triggered by an HTTP request.
3956
3957 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
3958 from a file (even a dummy empty file). The file name of the ACL to be
3959 updated is passed between parentheses. It takes one argument: <key fmt>,
3960 which follows log-format rules, to collect content of the entry to delete.
3961 It is the equivalent of the "del acl" command from the stats socket, but
3962 can be triggered by an HTTP request.
3963
3964 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
3965 from a file (even a dummy empty file). The file name of the MAP to be
3966 updated is passed between parentheses. It takes one argument: <key fmt>,
3967 which follows log-format rules, to collect content of the entry to delete.
3968 It takes one argument: "file name" It is the equivalent of the "del map"
3969 command from the stats socket, but can be triggered by an HTTP request.
3970
3971 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
3972 from a file (even a dummy empty file). The file name of the MAP to be
3973 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
3974 which follows log-format rules, used to collect MAP key, and <value fmt>,
3975 which follows log-format rules, used to collect content for the new entry.
3976 It performs a lookup in the MAP before insertion, to avoid duplicated (or
3977 more) values. This lookup is done by a linear search and can be expensive
3978 with large lists! It is the equivalent of the "set map" command from the
3979 stats socket, but can be triggered by an HTTP request.
3980
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003981 - capture <sample> [ len <length> | id <id> ] :
Willy Tarreaua9083d02015-05-08 15:27:59 +02003982 captures sample expression <sample> from the request buffer, and converts
3983 it to a string of at most <len> characters. The resulting string is
3984 stored into the next request "capture" slot, so it will possibly appear
3985 next to some captured HTTP headers. It will then automatically appear in
3986 the logs, and it will be possible to extract it using sample fetch rules
3987 to feed it into headers or anything. The length should be limited given
3988 that this size will be allocated for each capture during the whole
3989 session life. Please check section 7.3 (Fetching samples) and "capture
3990 request header" for more information.
3991
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003992 If the keyword "id" is used instead of "len", the action tries to store
3993 the captured string in a previously declared capture slot. This is useful
3994 to run captures in backends. The slot id can be declared by a previous
3995 directive "http-request capture" or with the "declare capture" keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01003996 If the slot <id> doesn't exist, then HAProxy fails parsing the
3997 configuration to prevent unexpected behavior at run time.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003998
Willy Tarreau09448f72014-06-25 18:12:15 +02003999 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4000 enables tracking of sticky counters from current request. These rules
4001 do not stop evaluation and do not change default action. Three sets of
4002 counters may be simultaneously tracked by the same connection. The first
4003 "track-sc0" rule executed enables tracking of the counters of the
4004 specified table as the first set. The first "track-sc1" rule executed
4005 enables tracking of the counters of the specified table as the second
4006 set. The first "track-sc2" rule executed enables tracking of the
4007 counters of the specified table as the third set. It is a recommended
4008 practice to use the first set of counters for the per-frontend counters
4009 and the second set for the per-backend ones. But this is just a
4010 guideline, all may be used everywhere.
4011
4012 These actions take one or two arguments :
4013 <key> is mandatory, and is a sample expression rule as described
4014 in section 7.3. It describes what elements of the incoming
4015 request or connection will be analysed, extracted, combined,
4016 and used to select which table entry to update the counters.
4017
4018 <table> is an optional table to be used instead of the default one,
4019 which is the stick-table declared in the current proxy. All
4020 the counters for the matches and updates for the key will
4021 then be performed in that table until the session ends.
4022
4023 Once a "track-sc*" rule is executed, the key is looked up in the table
4024 and if it is not found, an entry is allocated for it. Then a pointer to
4025 that entry is kept during all the session's life, and this entry's
4026 counters are updated as often as possible, every time the session's
4027 counters are updated, and also systematically when the session ends.
4028 Counters are only updated for events that happen after the tracking has
4029 been started. As an exception, connection counters and request counters
4030 are systematically updated so that they reflect useful information.
4031
4032 If the entry tracks concurrent connection counters, one connection is
4033 counted for as long as the entry is tracked, and the entry will not
4034 expire during that time. Tracking counters also provides a performance
4035 advantage over just checking the keys, because only one table lookup is
4036 performed for all ACL checks that make use of it.
4037
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004038 - sc-set-gpt0(<sc-id>) <int> :
4039 This action sets the GPT0 tag according to the sticky counter designated
4040 by <sc-id> and the value of <int>. The expected result is a boolean. If
4041 an error occurs, this action silently fails and the actions evaluation
4042 continues.
4043
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004044 - sc-inc-gpc0(<sc-id>):
4045 This action increments the GPC0 counter according with the sticky counter
4046 designated by <sc-id>. If an error occurs, this action silently fails and
4047 the actions evaluation continues.
4048
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004049 - set-var(<var-name>) <expr> :
4050 Is used to set the contents of a variable. The variable is declared
4051 inline.
4052
Daniel Schneller0b547052016-03-21 20:46:57 +01004053 <var-name> The name of the variable starts with an indication about
4054 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004055 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004056 "sess" : the variable is shared with the whole session
4057 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004058 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004059 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004060 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004061 "res" : the variable is shared only during response
4062 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004063 This prefix is followed by a name. The separator is a '.'.
Daniel Schneller0b547052016-03-21 20:46:57 +01004064 The name may only contain characters 'a-z', 'A-Z', '0-9'
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004065 and '_'.
4066
4067 <expr> Is a standard HAProxy expression formed by a sample-fetch
4068 followed by some converters.
4069
4070 Example:
4071
4072 http-request set-var(req.my_var) req.fhdr(user-agent),lower
4073
Christopher Faulet85d79c92016-11-09 16:54:56 +01004074 - unset-var(<var-name>) :
4075 Is used to unset a variable. See above for details about <var-name>.
4076
4077 Example:
4078
4079 http-request unset-var(req.my_var)
4080
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004081 - set-src <expr> :
4082 Is used to set the source IP address to the value of specified
4083 expression. Useful when a proxy in front of HAProxy rewrites source IP,
4084 but provides the correct IP in a HTTP header; or you want to mask
4085 source IP for privacy.
4086
4087 <expr> Is a standard HAProxy expression formed by a sample-fetch
4088 followed by some converters.
4089
4090 Example:
4091
4092 http-request set-src hdr(x-forwarded-for)
4093 http-request set-src src,ipmask(24)
4094
Willy Tarreau00005ce2016-10-21 15:07:45 +02004095 When possible, set-src preserves the original source port as long as the
4096 address family allows it, otherwise the source port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004097
William Lallemand44be6402016-05-25 01:51:35 +02004098 - set-src-port <expr> :
4099 Is used to set the source port address to the value of specified
4100 expression.
4101
4102 <expr> Is a standard HAProxy expression formed by a sample-fetch
4103 followed by some converters.
4104
4105 Example:
4106
4107 http-request set-src-port hdr(x-port)
4108 http-request set-src-port int(4000)
4109
Willy Tarreau00005ce2016-10-21 15:07:45 +02004110 When possible, set-src-port preserves the original source address as long
4111 as the address family supports a port, otherwise it forces the source
4112 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004113
William Lallemand13e9b0c2016-05-25 02:34:07 +02004114 - set-dst <expr> :
4115 Is used to set the destination IP address to the value of specified
4116 expression. Useful when a proxy in front of HAProxy rewrites destination
4117 IP, but provides the correct IP in a HTTP header; or you want to mask
4118 the IP for privacy. If you want to connect to the new address/port, use
4119 '0.0.0.0:0' as a server address in the backend.
4120
4121 <expr> Is a standard HAProxy expression formed by a sample-fetch
4122 followed by some converters.
4123
4124 Example:
4125
4126 http-request set-dst hdr(x-dst)
4127 http-request set-dst dst,ipmask(24)
4128
Willy Tarreau00005ce2016-10-21 15:07:45 +02004129 When possible, set-dst preserves the original destination port as long as
4130 the address family allows it, otherwise the destination port is set to 0.
4131
William Lallemand13e9b0c2016-05-25 02:34:07 +02004132 - set-dst-port <expr> :
4133 Is used to set the destination port address to the value of specified
4134 expression. If you want to connect to the new address/port, use
4135 '0.0.0.0:0' as a server address in the backend.
4136
4137 <expr> Is a standard HAProxy expression formed by a sample-fetch
4138 followed by some converters.
4139
4140 Example:
4141
4142 http-request set-dst-port hdr(x-port)
4143 http-request set-dst-port int(4000)
4144
Willy Tarreau00005ce2016-10-21 15:07:45 +02004145 When possible, set-dst-port preserves the original destination address as
4146 long as the address family supports a port, otherwise it forces the
4147 destination address to IPv4 "0.0.0.0" before rewriting the port.
4148
Willy Tarreau2d392c22015-08-24 01:43:45 +02004149 - "silent-drop" : this stops the evaluation of the rules and makes the
4150 client-facing connection suddenly disappear using a system-dependant way
4151 that tries to prevent the client from being notified. The effect it then
4152 that the client still sees an established connection while there's none
4153 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4154 except that it doesn't use any local resource at all on the machine
4155 running HAProxy. It can resist much higher loads than "tarpit", and slow
4156 down stronger attackers. It is important to undestand the impact of using
4157 this mechanism. All stateful equipments placed between the client and
4158 HAProxy (firewalls, proxies, load balancers) will also keep the
4159 established connection for a long time and may suffer from this action.
4160 On modern Linux systems running with enough privileges, the TCP_REPAIR
4161 socket option is used to block the emission of a TCP reset. On other
4162 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4163 pass the first router, though it's still delivered to local networks. Do
4164 not use it unless you fully understand how it works.
4165
Willy Tarreau20b0de52012-12-24 15:45:22 +01004166 There is no limit to the number of http-request statements per instance.
4167
4168 It is important to know that http-request rules are processed very early in
4169 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004170 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4171 visible by almost all further ACL rules.
4172
4173 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4174 in newer versions (>= 1.5). But if you need to use regular expression to
4175 delete headers, you can still use "reqdel". Also please use
4176 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004177
4178 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01004179 acl nagios src 192.168.129.3
4180 acl local_net src 192.168.0.0/16
4181 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004182
Cyril Bonté78caf842010-03-10 22:41:43 +01004183 http-request allow if nagios
4184 http-request allow if local_net auth_ok
4185 http-request auth realm Gimme if local_net auth_ok
4186 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004187
Cyril Bonté78caf842010-03-10 22:41:43 +01004188 Example:
4189 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01004190 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004191
Willy Tarreau20b0de52012-12-24 15:45:22 +01004192 Example:
4193 http-request set-header X-Haproxy-Current-Date %T
4194 http-request set-header X-SSL %[ssl_fc]
Willy Tarreaufca42612015-08-27 17:15:05 +02004195 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004196 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4197 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4198 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4199 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4200 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4201 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
4202
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004203 Example:
4204 acl key req.hdr(X-Add-Acl-Key) -m found
4205 acl add path /addacl
4206 acl del path /delacl
4207
4208 acl myhost hdr(Host) -f myhost.lst
4209
4210 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4211 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
4212
4213 Example:
4214 acl value req.hdr(X-Value) -m found
4215 acl setmap path /setmap
4216 acl delmap path /delmap
4217
4218 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4219
4220 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4221 http-request del-map(map.lst) %[src] if delmap
4222
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02004223 See also : "stats http-request", section 3.4 about userlists and section 7
4224 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01004225
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004226http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Willy Tarreau51d861a2015-05-22 17:30:48 +02004227 capture <sample> id <id> | redirect <rule> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004228 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06004229 replace-header <name> <regex-match> <replace-fmt> |
4230 replace-value <name> <regex-match> <replace-fmt> |
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004231 set-status <status> [reason <str>] |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004232 set-log-level <level> | set-mark <mark> | set-tos <tos> |
4233 add-acl(<file name>) <key fmt> |
4234 del-acl(<file name>) <key fmt> |
4235 del-map(<file name>) <key fmt> |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01004236 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004237 set-var(<var-name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01004238 unset-var(<var-name>) |
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004239 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004240 sc-inc-gpc0(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004241 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02004242 silent-drop |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004243 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02004244 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004245 Access control for Layer 7 responses
4246
4247 May be used in sections: defaults | frontend | listen | backend
4248 no | yes | yes | yes
4249
4250 The http-response statement defines a set of rules which apply to layer 7
4251 processing. The rules are evaluated in their declaration order when they are
4252 met in a frontend, listen or backend section. Any rule may optionally be
4253 followed by an ACL-based condition, in which case it will only be evaluated
4254 if the condition is true. Since these rules apply on responses, the backend
4255 rules are applied first, followed by the frontend's rules.
4256
4257 The first keyword is the rule's action. Currently supported actions include :
4258 - "allow" : this stops the evaluation of the rules and lets the response
4259 pass the check. No further "http-response" rules are evaluated for the
4260 current section.
4261
4262 - "deny" : this stops the evaluation of the rules and immediately rejects
4263 the response and emits an HTTP 502 error. No further "http-response"
4264 rules are evaluated.
4265
4266 - "add-header" appends an HTTP header field whose name is specified in
4267 <name> and whose value is defined by <fmt> which follows the log-format
4268 rules (see Custom Log Format in section 8.2.4). This may be used to send
4269 a cookie to a client for example, or to pass some internal information.
4270 This rule is not final, so it is possible to add other similar rules.
4271 Note that header addition is performed immediately, so one rule might
4272 reuse the resulting header from a previous rule.
4273
4274 - "set-header" does the same as "add-header" except that the header name
4275 is first removed if it existed. This is useful when passing security
4276 information to the server, where the header must not be manipulated by
4277 external users.
4278
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004279 - "del-header" removes all HTTP header fields whose name is specified in
4280 <name>.
4281
Sasha Pachev218f0642014-06-16 12:05:59 -06004282 - "replace-header" matches the regular expression in all occurrences of
4283 header field <name> according to <match-regex>, and replaces them with
4284 the <replace-fmt> argument. Format characters are allowed in replace-fmt
4285 and work like in <fmt> arguments in "add-header". The match is only
4286 case-sensitive. It is important to understand that this action only
4287 considers whole header lines, regardless of the number of values they
4288 may contain. This usage is suited to headers naturally containing commas
4289 in their value, such as Set-Cookie, Expires and so on.
4290
4291 Example:
4292
4293 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4294
4295 applied to:
4296
4297 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4298
4299 outputs:
4300
4301 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4302
4303 assuming the backend IP is 192.168.1.20.
4304
4305 - "replace-value" works like "replace-header" except that it matches the
4306 regex against every comma-delimited value of the header field <name>
4307 instead of the entire header. This is suited for all headers which are
4308 allowed to carry more than one value. An example could be the Accept
4309 header.
4310
4311 Example:
4312
4313 http-response replace-value Cache-control ^public$ private
4314
4315 applied to:
4316
4317 Cache-Control: max-age=3600, public
4318
4319 outputs:
4320
4321 Cache-Control: max-age=3600, private
4322
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004323 - "set-status" replaces the response status code with <status> which must
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004324 be an integer between 100 and 999. Optionally, a custom reason text can be
4325 provided defined by <str>, or the default reason for the specified code
4326 will be used as a fallback.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004327
4328 Example:
4329
4330 # return "431 Request Header Fields Too Large"
4331 http-response set-status 431
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004332 # return "503 Slow Down", custom reason
4333 http-response set-status 503 reason "Slow Down".
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004334
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004335 - "set-nice" sets the "nice" factor of the current request being processed.
4336 It only has effect against the other requests being processed at the same
4337 time. The default value is 0, unless altered by the "nice" setting on the
4338 "bind" line. The accepted range is -1024..1024. The higher the value, the
4339 nicest the request will be. Lower values will make the request more
4340 important than other ones. This can be useful to improve the speed of
4341 some requests, or lower the priority of non-important requests. Using
4342 this setting without prior experimentation can cause some major slowdown.
4343
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004344 - "set-log-level" is used to change the log level of the current request
4345 when a certain condition is met. Valid levels are the 8 syslog levels
4346 (see the "log" keyword) plus the special level "silent" which disables
4347 logging for this request. This rule is not final so the last matching
4348 rule wins. This rule can be useful to disable health checks coming from
4349 another equipment.
4350
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004351 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
4352 the client to the value passed in <tos> on platforms which support this.
4353 This value represents the whole 8 bits of the IP TOS field, and can be
4354 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
4355 that only the 6 higher bits are used in DSCP or TOS, and the two lower
4356 bits are always 0. This can be used to adjust some routing behaviour on
4357 border routers based on some information from the request. See RFC 2474,
4358 2597, 3260 and 4594 for more information.
4359
Willy Tarreau51347ed2013-06-11 19:34:13 +02004360 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
4361 client to the value passed in <mark> on platforms which support it. This
4362 value is an unsigned 32 bit value which can be matched by netfilter and
4363 by the routing table. It can be expressed both in decimal or hexadecimal
4364 format (prefixed by "0x"). This can be useful to force certain packets to
4365 take a different route (for example a cheaper network path for bulk
4366 downloads). This works on Linux kernels 2.6.32 and above and requires
4367 admin privileges.
4368
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004369 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
4370 from a file (even a dummy empty file). The file name of the ACL to be
4371 updated is passed between parentheses. It takes one argument: <key fmt>,
4372 which follows log-format rules, to collect content of the new entry. It
4373 performs a lookup in the ACL before insertion, to avoid duplicated (or
4374 more) values. This lookup is done by a linear search and can be expensive
4375 with large lists! It is the equivalent of the "add acl" command from the
4376 stats socket, but can be triggered by an HTTP response.
4377
4378 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
4379 from a file (even a dummy empty file). The file name of the ACL to be
4380 updated is passed between parentheses. It takes one argument: <key fmt>,
4381 which follows log-format rules, to collect content of the entry to delete.
4382 It is the equivalent of the "del acl" command from the stats socket, but
4383 can be triggered by an HTTP response.
4384
4385 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
4386 from a file (even a dummy empty file). The file name of the MAP to be
4387 updated is passed between parentheses. It takes one argument: <key fmt>,
4388 which follows log-format rules, to collect content of the entry to delete.
4389 It takes one argument: "file name" It is the equivalent of the "del map"
4390 command from the stats socket, but can be triggered by an HTTP response.
4391
4392 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
4393 from a file (even a dummy empty file). The file name of the MAP to be
4394 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
4395 which follows log-format rules, used to collect MAP key, and <value fmt>,
4396 which follows log-format rules, used to collect content for the new entry.
4397 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4398 more) values. This lookup is done by a linear search and can be expensive
4399 with large lists! It is the equivalent of the "set map" command from the
4400 stats socket, but can be triggered by an HTTP response.
4401
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004402 - capture <sample> id <id> :
4403 captures sample expression <sample> from the response buffer, and converts
4404 it to a string. The resulting string is stored into the next request
4405 "capture" slot, so it will possibly appear next to some captured HTTP
4406 headers. It will then automatically appear in the logs, and it will be
4407 possible to extract it using sample fetch rules to feed it into headers or
4408 anything. Please check section 7.3 (Fetching samples) and "capture
4409 response header" for more information.
4410
4411 The keyword "id" is the id of the capture slot which is used for storing
4412 the string. The capture slot must be defined in an associated frontend.
4413 This is useful to run captures in backends. The slot id can be declared by
4414 a previous directive "http-response capture" or with the "declare capture"
4415 keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01004416 If the slot <id> doesn't exist, then HAProxy fails parsing the
4417 configuration to prevent unexpected behavior at run time.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004418
Willy Tarreau51d861a2015-05-22 17:30:48 +02004419 - "redirect" : this performs an HTTP redirection based on a redirect rule.
4420 This supports a format string similarly to "http-request redirect" rules,
4421 with the exception that only the "location" type of redirect is possible
4422 on the response. See the "redirect" keyword for the rule's syntax. When
4423 a redirect rule is applied during a response, connections to the server
4424 are closed so that no data can be forwarded from the server to the client.
4425
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004426 - set-var(<var-name>) expr:
4427 Is used to set the contents of a variable. The variable is declared
4428 inline.
4429
Daniel Schneller0b547052016-03-21 20:46:57 +01004430 <var-name> The name of the variable starts with an indication about
4431 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004432 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004433 "sess" : the variable is shared with the whole session
4434 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004435 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004436 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004437 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004438 "res" : the variable is shared only during response
4439 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004440 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01004441 The name may only contain characters 'a-z', 'A-Z', '0-9',
4442 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004443
4444 <expr> Is a standard HAProxy expression formed by a sample-fetch
4445 followed by some converters.
4446
4447 Example:
4448
4449 http-response set-var(sess.last_redir) res.hdr(location)
4450
Christopher Faulet85d79c92016-11-09 16:54:56 +01004451 - unset-var(<var-name>) :
4452 Is used to unset a variable. See above for details about <var-name>.
4453
4454 Example:
4455
4456 http-response unset-var(sess.last_redir)
4457
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004458 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4459 enables tracking of sticky counters from current response. Please refer to
4460 "http-request track-sc" for a complete description. The only difference
4461 from "http-request track-sc" is the <key> sample expression can only make
4462 use of samples in response (eg. res.*, status etc.) and samples below
4463 Layer 6 (eg. ssl related samples, see section 7.3.4). If the sample is
4464 not supported, haproxy will fail and warn while parsing the config.
4465
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004466 - sc-set-gpt0(<sc-id>) <int> :
4467 This action sets the GPT0 tag according to the sticky counter designated
4468 by <sc-id> and the value of <int>. The expected result is a boolean. If
4469 an error occurs, this action silently fails and the actions evaluation
4470 continues.
4471
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004472 - sc-inc-gpc0(<sc-id>):
4473 This action increments the GPC0 counter according with the sticky counter
4474 designated by <sc-id>. If an error occurs, this action silently fails and
4475 the actions evaluation continues.
4476
Willy Tarreau2d392c22015-08-24 01:43:45 +02004477 - "silent-drop" : this stops the evaluation of the rules and makes the
4478 client-facing connection suddenly disappear using a system-dependant way
4479 that tries to prevent the client from being notified. The effect it then
4480 that the client still sees an established connection while there's none
4481 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4482 except that it doesn't use any local resource at all on the machine
4483 running HAProxy. It can resist much higher loads than "tarpit", and slow
4484 down stronger attackers. It is important to undestand the impact of using
4485 this mechanism. All stateful equipments placed between the client and
4486 HAProxy (firewalls, proxies, load balancers) will also keep the
4487 established connection for a long time and may suffer from this action.
4488 On modern Linux systems running with enough privileges, the TCP_REPAIR
4489 socket option is used to block the emission of a TCP reset. On other
4490 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4491 pass the first router, though it's still delivered to local networks. Do
4492 not use it unless you fully understand how it works.
4493
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004494 There is no limit to the number of http-response statements per instance.
4495
Godbach09250262013-07-02 01:19:15 +08004496 It is important to know that http-response rules are processed very early in
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004497 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4498 headers added by "add-header"/"set-header" are visible by almost all further ACL
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004499 rules.
4500
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004501 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4502 in newer versions (>= 1.5). But if you need to use regular expression to
4503 delete headers, you can still use "rspdel". Also please use
4504 "http-response deny" instead of "rspdeny".
4505
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004506 Example:
4507 acl key_acl res.hdr(X-Acl-Key) -m found
4508
4509 acl myhost hdr(Host) -f myhost.lst
4510
4511 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4512 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4513
4514 Example:
4515 acl value res.hdr(X-Value) -m found
4516
4517 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4518
4519 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4520 http-response del-map(map.lst) %[src] if ! value
4521
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004522 See also : "http-request", section 3.4 about userlists and section 7 about
4523 ACL usage.
4524
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004525
Willy Tarreau30631952015-08-06 15:05:24 +02004526http-reuse { never | safe | aggressive | always }
4527 Declare how idle HTTP connections may be shared between requests
4528
4529 May be used in sections: defaults | frontend | listen | backend
4530 yes | no | yes | yes
4531
4532 By default, a connection established between haproxy and the backend server
4533 belongs to the session that initiated it. The downside is that between the
4534 response and the next request, the connection remains idle and is not used.
4535 In many cases for performance reasons it is desirable to make it possible to
4536 reuse these idle connections to serve other requests from different sessions.
4537 This directive allows to tune this behaviour.
4538
4539 The argument indicates the desired connection reuse strategy :
4540
4541 - "never" : idle connections are never shared between sessions. This is
4542 the default choice. It may be enforced to cancel a different
4543 strategy inherited from a defaults section or for
4544 troubleshooting. For example, if an old bogus application
4545 considers that multiple requests over the same connection come
4546 from the same client and it is not possible to fix the
4547 application, it may be desirable to disable connection sharing
4548 in a single backend. An example of such an application could
4549 be an old haproxy using cookie insertion in tunnel mode and
4550 not checking any request past the first one.
4551
4552 - "safe" : this is the recommended strategy. The first request of a
4553 session is always sent over its own connection, and only
4554 subsequent requests may be dispatched over other existing
4555 connections. This ensures that in case the server closes the
4556 connection when the request is being sent, the browser can
4557 decide to silently retry it. Since it is exactly equivalent to
4558 regular keep-alive, there should be no side effects.
4559
4560 - "aggressive" : this mode may be useful in webservices environments where
4561 all servers are not necessarily known and where it would be
4562 appreciable to deliver most first requests over existing
4563 connections. In this case, first requests are only delivered
4564 over existing connections that have been reused at least once,
4565 proving that the server correctly supports connection reuse.
4566 It should only be used when it's sure that the client can
4567 retry a failed request once in a while and where the benefit
4568 of aggressive connection reuse significantly outweights the
4569 downsides of rare connection failures.
4570
4571 - "always" : this mode is only recommended when the path to the server is
4572 known for never breaking existing connections quickly after
4573 releasing them. It allows the first request of a session to be
4574 sent to an existing connection. This can provide a significant
4575 performance increase over the "safe" strategy when the backend
4576 is a cache farm, since such components tend to show a
4577 consistent behaviour and will benefit from the connection
4578 sharing. It is recommended that the "http-keep-alive" timeout
4579 remains low in this mode so that no dead connections remain
4580 usable. In most cases, this will lead to the same performance
4581 gains as "aggressive" but with more risks. It should only be
4582 used when it improves the situation over "aggressive".
4583
4584 When http connection sharing is enabled, a great care is taken to respect the
4585 connection properties and compatiblities. Specifically :
4586 - connections made with "usesrc" followed by a client-dependant value
4587 ("client", "clientip", "hdr_ip") are marked private and never shared ;
4588
4589 - connections sent to a server with a TLS SNI extension are marked private
4590 and are never shared ;
4591
4592 - connections receiving a status code 401 or 407 expect some authentication
4593 to be sent in return. Due to certain bogus authentication schemes (such
4594 as NTLM) relying on the connection, these connections are marked private
4595 and are never shared ;
4596
4597 No connection pool is involved, once a session dies, the last idle connection
4598 it was attached to is deleted at the same time. This ensures that connections
4599 may not last after all sessions are closed.
4600
4601 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4602 because almost no new connection will be established while idle connections
4603 remain available. This is particularly true with the "always" strategy.
4604
4605 See also : "option http-keep-alive", "server maxconn"
4606
4607
Mark Lamourinec2247f02012-01-04 13:02:01 -05004608http-send-name-header [<header>]
4609 Add the server name to a request. Use the header string given by <header>
4610
4611 May be used in sections: defaults | frontend | listen | backend
4612 yes | no | yes | yes
4613
4614 Arguments :
4615
4616 <header> The header string to use to send the server name
4617
4618 The "http-send-name-header" statement causes the name of the target
4619 server to be added to the headers of an HTTP request. The name
4620 is added with the header string proved.
4621
4622 See also : "server"
4623
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004624id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02004625 Set a persistent ID to a proxy.
4626 May be used in sections : defaults | frontend | listen | backend
4627 no | yes | yes | yes
4628 Arguments : none
4629
4630 Set a persistent ID for the proxy. This ID must be unique and positive.
4631 An unused ID will automatically be assigned if unset. The first assigned
4632 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004633
4634
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004635ignore-persist { if | unless } <condition>
4636 Declare a condition to ignore persistence
4637 May be used in sections: defaults | frontend | listen | backend
4638 no | yes | yes | yes
4639
4640 By default, when cookie persistence is enabled, every requests containing
4641 the cookie are unconditionally persistent (assuming the target server is up
4642 and running).
4643
4644 The "ignore-persist" statement allows one to declare various ACL-based
4645 conditions which, when met, will cause a request to ignore persistence.
4646 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004647 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004648 persistence for a specific User-Agent (for example, some web crawler bots).
4649
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004650 The persistence is ignored when an "if" condition is met, or unless an
4651 "unless" condition is met.
4652
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004653 Example:
4654 acl url_static path_beg /static /images /img /css
4655 acl url_static path_end .gif .png .jpg .css .js
4656 ignore-persist if url_static
4657
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004658 See also : "force-persist", "cookie", and section 7 about ACL usage.
4659
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004660load-server-state-from-file { global | local | none }
4661 Allow seamless reload of HAProxy
4662 May be used in sections: defaults | frontend | listen | backend
4663 yes | no | yes | yes
4664
4665 This directive points HAProxy to a file where server state from previous
4666 running process has been saved. That way, when starting up, before handling
4667 traffic, the new process can apply old states to servers exactly has if no
4668 reload occured. The purpose of the "load-server-state-from-file" directive is
4669 to tell haproxy which file to use. For now, only 2 arguments to either prevent
4670 loading state or load states from a file containing all backends and servers.
4671 The state file can be generated by running the command "show servers state"
4672 over the stats socket and redirect output.
4673
4674 The format of the file is versionned and is very specific. To understand it,
4675 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02004676 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004677
4678 Arguments:
4679 global load the content of the file pointed by the global directive
4680 named "server-state-file".
4681
4682 local load the content of the file pointed by the directive
4683 "server-state-file-name" if set. If not set, then the backend
4684 name is used as a file name.
4685
4686 none don't load any stat for this backend
4687
4688 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01004689 - server's IP address is preserved across reloads by default, but the
4690 order can be changed thanks to the server's "init-addr" setting. This
4691 means that an IP address change performed on the CLI at run time will
4692 be preserved, and that any change to the local resolver (eg: /etc/hosts)
4693 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004694
4695 - server's weight is applied from previous running process unless it has
4696 has changed between previous and new configuration files.
4697
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004698 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004699
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004700 global
4701 stats socket /tmp/socket
4702 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004703
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004704 defaults
4705 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004706
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004707 backend bk
4708 server s1 127.0.0.1:22 check weight 11
4709 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004710
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004711
4712 Then one can run :
4713
4714 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
4715
4716 Content of the file /tmp/server_state would be like this:
4717
4718 1
4719 # <field names skipped for the doc example>
4720 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
4721 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
4722
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004723 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004724
4725 global
4726 stats socket /tmp/socket
4727 server-state-base /etc/haproxy/states
4728
4729 defaults
4730 load-server-state-from-file local
4731
4732 backend bk
4733 server s1 127.0.0.1:22 check weight 11
4734 server s2 127.0.0.1:22 check weight 12
4735
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004736
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004737 Then one can run :
4738
4739 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
4740
4741 Content of the file /etc/haproxy/states/bk would be like this:
4742
4743 1
4744 # <field names skipped for the doc example>
4745 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
4746 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
4747
4748 See also: "server-state-file", "server-state-file-name", and
4749 "show servers state"
4750
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004751
Willy Tarreau2769aa02007-12-27 18:26:09 +01004752log global
Willy Tarreau18324f52014-06-27 18:10:07 +02004753log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02004754no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01004755 Enable per-instance logging of events and traffic.
4756 May be used in sections : defaults | frontend | listen | backend
4757 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02004758
4759 Prefix :
4760 no should be used when the logger list must be flushed. For example,
4761 if you don't want to inherit from the default logger list. This
4762 prefix does not allow arguments.
4763
Willy Tarreau2769aa02007-12-27 18:26:09 +01004764 Arguments :
4765 global should be used when the instance's logging parameters are the
4766 same as the global ones. This is the most common usage. "global"
4767 replaces <address>, <facility> and <level> with those of the log
4768 entries found in the "global" section. Only one "log global"
4769 statement may be used per instance, and this form takes no other
4770 parameter.
4771
4772 <address> indicates where to send the logs. It takes the same format as
4773 for the "global" section's logs, and can be one of :
4774
4775 - An IPv4 address optionally followed by a colon (':') and a UDP
4776 port. If no port is specified, 514 is used by default (the
4777 standard syslog port).
4778
David du Colombier24bb5f52011-03-17 10:40:23 +01004779 - An IPv6 address followed by a colon (':') and optionally a UDP
4780 port. If no port is specified, 514 is used by default (the
4781 standard syslog port).
4782
Willy Tarreau2769aa02007-12-27 18:26:09 +01004783 - A filesystem path to a UNIX domain socket, keeping in mind
4784 considerations for chroot (be sure the path is accessible
4785 inside the chroot) and uid/gid (be sure the path is
4786 appropriately writeable).
4787
William Lallemandb2f07452015-05-12 14:27:13 +02004788 You may want to reference some environment variables in the
4789 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01004790
Willy Tarreau18324f52014-06-27 18:10:07 +02004791 <length> is an optional maximum line length. Log lines larger than this
4792 value will be truncated before being sent. The reason is that
4793 syslog servers act differently on log line length. All servers
4794 support the default value of 1024, but some servers simply drop
4795 larger lines while others do log them. If a server supports long
4796 lines, it may make sense to set this value here in order to avoid
4797 truncating long lines. Similarly, if a server drops long lines,
4798 it is preferable to truncate them before sending them. Accepted
4799 values are 80 to 65535 inclusive. The default value of 1024 is
4800 generally fine for all standard usages. Some specific cases of
4801 long captures or JSON-formated logs may require larger values.
4802
Willy Tarreau2769aa02007-12-27 18:26:09 +01004803 <facility> must be one of the 24 standard syslog facilities :
4804
4805 kern user mail daemon auth syslog lpr news
4806 uucp cron auth2 ftp ntp audit alert cron2
4807 local0 local1 local2 local3 local4 local5 local6 local7
4808
4809 <level> is optional and can be specified to filter outgoing messages. By
4810 default, all messages are sent. If a level is specified, only
4811 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02004812 will be sent. An optional minimum level can be specified. If it
4813 is set, logs emitted with a more severe level than this one will
4814 be capped to this level. This is used to avoid sending "emerg"
4815 messages on all terminals on some default syslog configurations.
4816 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01004817
4818 emerg alert crit err warning notice info debug
4819
William Lallemand0f99e342011-10-12 17:50:54 +02004820 It is important to keep in mind that it is the frontend which decides what to
4821 log from a connection, and that in case of content switching, the log entries
4822 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004823
4824 However, backend log declaration define how and where servers status changes
4825 will be logged. Level "notice" will be used to indicate a server going up,
4826 "warning" will be used for termination signals and definitive service
4827 termination, and "alert" will be used for when a server goes down.
4828
4829 Note : According to RFC3164, messages are truncated to 1024 bytes before
4830 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004831
4832 Example :
4833 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02004834 log 127.0.0.1:514 local0 notice # only send important events
4835 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02004836 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01004837
Willy Tarreau2769aa02007-12-27 18:26:09 +01004838
William Lallemand48940402012-01-30 16:47:22 +01004839log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01004840 Specifies the log format string to use for traffic logs
4841 May be used in sections: defaults | frontend | listen | backend
4842 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01004843
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01004844 This directive specifies the log format string that will be used for all logs
4845 resulting from traffic passing through the frontend using this line. If the
4846 directive is used in a defaults section, all subsequent frontends will use
4847 the same log format. Please see section 8.2.4 which covers the log format
4848 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01004849
Guillaume de Lafond29f45602017-03-31 19:52:15 +02004850 "log-format" directive overrides previous "option tcplog", "log-format" and
4851 "option httplog" directives.
4852
Dragan Dosen7ad31542015-09-28 17:16:47 +02004853log-format-sd <string>
4854 Specifies the RFC5424 structured-data log format string
4855 May be used in sections: defaults | frontend | listen | backend
4856 yes | yes | yes | no
4857
4858 This directive specifies the RFC5424 structured-data log format string that
4859 will be used for all logs resulting from traffic passing through the frontend
4860 using this line. If the directive is used in a defaults section, all
4861 subsequent frontends will use the same log format. Please see section 8.2.4
4862 which covers the log format string in depth.
4863
4864 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
4865 about the RFC5424 structured-data part.
4866
4867 Note : This log format string will be used only for loggers that have set
4868 log format to "rfc5424".
4869
4870 Example :
4871 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
4872
4873
Willy Tarreau094af4e2015-01-07 15:03:42 +01004874log-tag <string>
4875 Specifies the log tag to use for all outgoing logs
4876 May be used in sections: defaults | frontend | listen | backend
4877 yes | yes | yes | yes
4878
4879 Sets the tag field in the syslog header to this string. It defaults to the
4880 log-tag set in the global section, otherwise the program name as launched
4881 from the command line, which usually is "haproxy". Sometimes it can be useful
4882 to differentiate between multiple processes running on the same host, or to
4883 differentiate customer instances running in the same process. In the backend,
4884 logs about servers up/down will use this tag. As a hint, it can be convenient
4885 to set a log-tag related to a hosted customer in a defaults section then put
4886 all the frontends and backends for that customer, then start another customer
4887 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004888
Willy Tarreauc35362a2014-04-25 13:58:37 +02004889max-keep-alive-queue <value>
4890 Set the maximum server queue size for maintaining keep-alive connections
4891 May be used in sections: defaults | frontend | listen | backend
4892 yes | no | yes | yes
4893
4894 HTTP keep-alive tries to reuse the same server connection whenever possible,
4895 but sometimes it can be counter-productive, for example if a server has a lot
4896 of connections while other ones are idle. This is especially true for static
4897 servers.
4898
4899 The purpose of this setting is to set a threshold on the number of queued
4900 connections at which haproxy stops trying to reuse the same server and prefers
4901 to find another one. The default value, -1, means there is no limit. A value
4902 of zero means that keep-alive requests will never be queued. For very close
4903 servers which can be reached with a low latency and which are not sensible to
4904 breaking keep-alive, a low value is recommended (eg: local static server can
4905 use a value of 10 or less). For remote servers suffering from a high latency,
4906 higher values might be needed to cover for the latency and/or the cost of
4907 picking a different server.
4908
4909 Note that this has no impact on responses which are maintained to the same
4910 server consecutively to a 401 response. They will still go to the same server
4911 even if they have to be queued.
4912
4913 See also : "option http-server-close", "option prefer-last-server", server
4914 "maxconn" and cookie persistence.
4915
4916
Willy Tarreau2769aa02007-12-27 18:26:09 +01004917maxconn <conns>
4918 Fix the maximum number of concurrent connections on a frontend
4919 May be used in sections : defaults | frontend | listen | backend
4920 yes | yes | yes | no
4921 Arguments :
4922 <conns> is the maximum number of concurrent connections the frontend will
4923 accept to serve. Excess connections will be queued by the system
4924 in the socket's listen queue and will be served once a connection
4925 closes.
4926
4927 If the system supports it, it can be useful on big sites to raise this limit
4928 very high so that haproxy manages connection queues, instead of leaving the
4929 clients with unanswered connection attempts. This value should not exceed the
4930 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01004931 of tune.bufsize (16kB by default) each, as well as some other data resulting
4932 in about 33 kB of RAM being consumed per established connection. That means
4933 that a medium system equipped with 1GB of RAM can withstand around
4934 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004935
4936 Also, when <conns> is set to large values, it is possible that the servers
4937 are not sized to accept such loads, and for this reason it is generally wise
4938 to assign them some reasonable connection limits.
4939
Vincent Bernat6341be52012-06-27 17:18:30 +02004940 By default, this value is set to 2000.
4941
Willy Tarreau2769aa02007-12-27 18:26:09 +01004942 See also : "server", global section's "maxconn", "fullconn"
4943
4944
4945mode { tcp|http|health }
4946 Set the running mode or protocol of the instance
4947 May be used in sections : defaults | frontend | listen | backend
4948 yes | yes | yes | yes
4949 Arguments :
4950 tcp The instance will work in pure TCP mode. A full-duplex connection
4951 will be established between clients and servers, and no layer 7
4952 examination will be performed. This is the default mode. It
4953 should be used for SSL, SSH, SMTP, ...
4954
4955 http The instance will work in HTTP mode. The client request will be
4956 analyzed in depth before connecting to any server. Any request
4957 which is not RFC-compliant will be rejected. Layer 7 filtering,
4958 processing and switching will be possible. This is the mode which
4959 brings HAProxy most of its value.
4960
4961 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02004962 to incoming connections and close the connection. Alternatively,
4963 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
4964 instead. Nothing will be logged in either case. This mode is used
4965 to reply to external components health checks. This mode is
4966 deprecated and should not be used anymore as it is possible to do
4967 the same and even better by combining TCP or HTTP modes with the
4968 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004969
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004970 When doing content switching, it is mandatory that the frontend and the
4971 backend are in the same mode (generally HTTP), otherwise the configuration
4972 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004973
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004974 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01004975 defaults http_instances
4976 mode http
4977
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004978 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004979
Willy Tarreau0ba27502007-12-24 16:55:16 +01004980
Cyril Bontéf0c60612010-02-06 14:44:47 +01004981monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01004982 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004983 May be used in sections : defaults | frontend | listen | backend
4984 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01004985 Arguments :
4986 if <cond> the monitor request will fail if the condition is satisfied,
4987 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004988 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01004989 are met, for instance a low number of servers both in a
4990 backend and its backup.
4991
4992 unless <cond> the monitor request will succeed only if the condition is
4993 satisfied, and will fail otherwise. Such a condition may be
4994 based on a test on the presence of a minimum number of active
4995 servers in a list of backends.
4996
4997 This statement adds a condition which can force the response to a monitor
4998 request to report a failure. By default, when an external component queries
4999 the URI dedicated to monitoring, a 200 response is returned. When one of the
5000 conditions above is met, haproxy will return 503 instead of 200. This is
5001 very useful to report a site failure to an external component which may base
5002 routing advertisements between multiple sites on the availability reported by
5003 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005004 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5005 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005006
5007 Example:
5008 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005009 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005010 acl site_dead nbsrv(dynamic) lt 2
5011 acl site_dead nbsrv(static) lt 2
5012 monitor-uri /site_alive
5013 monitor fail if site_dead
5014
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005015 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005016
5017
5018monitor-net <source>
5019 Declare a source network which is limited to monitor requests
5020 May be used in sections : defaults | frontend | listen | backend
5021 yes | yes | yes | no
5022 Arguments :
5023 <source> is the source IPv4 address or network which will only be able to
5024 get monitor responses to any request. It can be either an IPv4
5025 address, a host name, or an address followed by a slash ('/')
5026 followed by a mask.
5027
5028 In TCP mode, any connection coming from a source matching <source> will cause
5029 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005030 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005031 forwarding the connection to a remote server.
5032
5033 In HTTP mode, a connection coming from a source matching <source> will be
5034 accepted, the following response will be sent without waiting for a request,
5035 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5036 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005037 running without forwarding the request to a backend server. Note that this
5038 response is sent in raw format, without any transformation. This is important
5039 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005040
Willy Tarreau82569f92012-09-27 23:48:56 +02005041 Monitor requests are processed very early, just after tcp-request connection
5042 ACLs which are the only ones able to block them. These connections are short
5043 lived and never wait for any data from the client. They cannot be logged, and
5044 it is the intended purpose. They are only used to report HAProxy's health to
5045 an upper component, nothing more. Please note that "monitor fail" rules do
5046 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005047
Willy Tarreau95cd2832010-03-04 23:36:33 +01005048 Last, please note that only one "monitor-net" statement can be specified in
5049 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005050
Willy Tarreau2769aa02007-12-27 18:26:09 +01005051 Example :
5052 # addresses .252 and .253 are just probing us.
5053 frontend www
5054 monitor-net 192.168.0.252/31
5055
5056 See also : "monitor fail", "monitor-uri"
5057
5058
5059monitor-uri <uri>
5060 Intercept a URI used by external components' monitor requests
5061 May be used in sections : defaults | frontend | listen | backend
5062 yes | yes | yes | no
5063 Arguments :
5064 <uri> is the exact URI which we want to intercept to return HAProxy's
5065 health status instead of forwarding the request.
5066
5067 When an HTTP request referencing <uri> will be received on a frontend,
5068 HAProxy will not forward it nor log it, but instead will return either
5069 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5070 conditions defined with "monitor fail". This is normally enough for any
5071 front-end HTTP probe to detect that the service is UP and running without
5072 forwarding the request to a backend server. Note that the HTTP method, the
5073 version and all headers are ignored, but the request must at least be valid
5074 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5075
5076 Monitor requests are processed very early. It is not possible to block nor
5077 divert them using ACLs. They cannot be logged either, and it is the intended
5078 purpose. They are only used to report HAProxy's health to an upper component,
5079 nothing more. However, it is possible to add any number of conditions using
5080 "monitor fail" and ACLs so that the result can be adjusted to whatever check
5081 can be imagined (most often the number of available servers in a backend).
5082
5083 Example :
5084 # Use /haproxy_test to report haproxy's status
5085 frontend www
5086 mode http
5087 monitor-uri /haproxy_test
5088
5089 See also : "monitor fail", "monitor-net"
5090
Willy Tarreau0ba27502007-12-24 16:55:16 +01005091
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005092option abortonclose
5093no option abortonclose
5094 Enable or disable early dropping of aborted requests pending in queues.
5095 May be used in sections : defaults | frontend | listen | backend
5096 yes | no | yes | yes
5097 Arguments : none
5098
5099 In presence of very high loads, the servers will take some time to respond.
5100 The per-instance connection queue will inflate, and the response time will
5101 increase respective to the size of the queue times the average per-session
5102 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005103 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005104 the queue, and slowing down other users, and the servers as well, because the
5105 request will eventually be served, then aborted at the first error
5106 encountered while delivering the response.
5107
5108 As there is no way to distinguish between a full STOP and a simple output
5109 close on the client side, HTTP agents should be conservative and consider
5110 that the client might only have closed its output channel while waiting for
5111 the response. However, this introduces risks of congestion when lots of users
5112 do the same, and is completely useless nowadays because probably no client at
5113 all will close the session while waiting for the response. Some HTTP agents
Willy Tarreaud72758d2010-01-12 10:42:19 +01005114 support this behaviour (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005115 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005116 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005117 of being the single component to break rare but valid traffic is extremely
5118 low, which adds to the temptation to be able to abort a session early while
5119 still not served and not pollute the servers.
5120
5121 In HAProxy, the user can choose the desired behaviour using the option
5122 "abortonclose". By default (without the option) the behaviour is HTTP
5123 compliant and aborted requests will be served. But when the option is
5124 specified, a session with an incoming channel closed will be aborted while
5125 it is still possible, either pending in the queue for a connection slot, or
5126 during the connection establishment if the server has not yet acknowledged
5127 the connection request. This considerably reduces the queue size and the load
5128 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005129 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005130
5131 If this option has been enabled in a "defaults" section, it can be disabled
5132 in a specific instance by prepending the "no" keyword before it.
5133
5134 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5135
5136
Willy Tarreau4076a152009-04-02 15:18:36 +02005137option accept-invalid-http-request
5138no option accept-invalid-http-request
5139 Enable or disable relaxing of HTTP request parsing
5140 May be used in sections : defaults | frontend | listen | backend
5141 yes | yes | yes | no
5142 Arguments : none
5143
Willy Tarreau91852eb2015-05-01 13:26:00 +02005144 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005145 means that invalid characters in header names are not permitted and cause an
5146 error to be returned to the client. This is the desired behaviour as such
5147 forbidden characters are essentially used to build attacks exploiting server
5148 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5149 server will emit invalid header names for whatever reason (configuration,
5150 implementation) and the issue will not be immediately fixed. In such a case,
5151 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005152 even if that does not make sense, by specifying this option. Similarly, the
5153 list of characters allowed to appear in a URI is well defined by RFC3986, and
5154 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5155 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
5156 not allowed at all. Haproxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005157 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005158 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5159 to pass through (no version specified) and multiple digits for both the major
5160 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005161
5162 This option should never be enabled by default as it hides application bugs
5163 and open security breaches. It should only be deployed after a problem has
5164 been confirmed.
5165
5166 When this option is enabled, erroneous header names will still be accepted in
5167 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005168 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5169 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005170 also helps confirming that the issue has been solved.
5171
5172 If this option has been enabled in a "defaults" section, it can be disabled
5173 in a specific instance by prepending the "no" keyword before it.
5174
5175 See also : "option accept-invalid-http-response" and "show errors" on the
5176 stats socket.
5177
5178
5179option accept-invalid-http-response
5180no option accept-invalid-http-response
5181 Enable or disable relaxing of HTTP response parsing
5182 May be used in sections : defaults | frontend | listen | backend
5183 yes | no | yes | yes
5184 Arguments : none
5185
Willy Tarreau91852eb2015-05-01 13:26:00 +02005186 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005187 means that invalid characters in header names are not permitted and cause an
5188 error to be returned to the client. This is the desired behaviour as such
5189 forbidden characters are essentially used to build attacks exploiting server
5190 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5191 server will emit invalid header names for whatever reason (configuration,
5192 implementation) and the issue will not be immediately fixed. In such a case,
5193 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005194 even if that does not make sense, by specifying this option. This option also
5195 relaxes the test on the HTTP version format, it allows multiple digits for
5196 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005197
5198 This option should never be enabled by default as it hides application bugs
5199 and open security breaches. It should only be deployed after a problem has
5200 been confirmed.
5201
5202 When this option is enabled, erroneous header names will still be accepted in
5203 responses, but the complete response will be captured in order to permit
5204 later analysis using the "show errors" request on the UNIX stats socket.
5205 Doing this also helps confirming that the issue has been solved.
5206
5207 If this option has been enabled in a "defaults" section, it can be disabled
5208 in a specific instance by prepending the "no" keyword before it.
5209
5210 See also : "option accept-invalid-http-request" and "show errors" on the
5211 stats socket.
5212
5213
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005214option allbackups
5215no option allbackups
5216 Use either all backup servers at a time or only the first one
5217 May be used in sections : defaults | frontend | listen | backend
5218 yes | no | yes | yes
5219 Arguments : none
5220
5221 By default, the first operational backup server gets all traffic when normal
5222 servers are all down. Sometimes, it may be preferred to use multiple backups
5223 at once, because one will not be enough. When "option allbackups" is enabled,
5224 the load balancing will be performed among all backup servers when all normal
5225 ones are unavailable. The same load balancing algorithm will be used and the
5226 servers' weights will be respected. Thus, there will not be any priority
5227 order between the backup servers anymore.
5228
5229 This option is mostly used with static server farms dedicated to return a
5230 "sorry" page when an application is completely offline.
5231
5232 If this option has been enabled in a "defaults" section, it can be disabled
5233 in a specific instance by prepending the "no" keyword before it.
5234
5235
5236option checkcache
5237no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005238 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005239 May be used in sections : defaults | frontend | listen | backend
5240 yes | no | yes | yes
5241 Arguments : none
5242
5243 Some high-level frameworks set application cookies everywhere and do not
5244 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005245 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005246 high risk of session crossing or stealing between users traversing the same
5247 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005248 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005249
5250 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005251 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005252 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005253 response to check if there's a risk of caching a cookie on a client-side
5254 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005255 to the client are :
5256 - all those without "Set-Cookie" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005257 - all those with a return code other than 200, 203, 206, 300, 301, 410,
Willy Tarreau198a7442008-01-17 12:05:32 +01005258 provided that the server has not set a "Cache-control: public" header ;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005259 - all those that come from a POST request, provided that the server has not
5260 set a 'Cache-Control: public' header ;
5261 - those with a 'Pragma: no-cache' header
5262 - those with a 'Cache-control: private' header
5263 - those with a 'Cache-control: no-store' header
5264 - those with a 'Cache-control: max-age=0' header
5265 - those with a 'Cache-control: s-maxage=0' header
5266 - those with a 'Cache-control: no-cache' header
5267 - those with a 'Cache-control: no-cache="set-cookie"' header
5268 - those with a 'Cache-control: no-cache="set-cookie,' header
5269 (allowing other fields after set-cookie)
5270
5271 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005272 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005273 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005274 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005275 that admins are informed that there's something to be fixed.
5276
5277 Due to the high impact on the application, the application should be tested
5278 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005279 good practice to always activate it during tests, even if it is not used in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005280 production, as it will report potentially dangerous application behaviours.
5281
5282 If this option has been enabled in a "defaults" section, it can be disabled
5283 in a specific instance by prepending the "no" keyword before it.
5284
5285
5286option clitcpka
5287no option clitcpka
5288 Enable or disable the sending of TCP keepalive packets on the client side
5289 May be used in sections : defaults | frontend | listen | backend
5290 yes | yes | yes | no
5291 Arguments : none
5292
5293 When there is a firewall or any session-aware component between a client and
5294 a server, and when the protocol involves very long sessions with long idle
5295 periods (eg: remote desktops), there is a risk that one of the intermediate
5296 components decides to expire a session which has remained idle for too long.
5297
5298 Enabling socket-level TCP keep-alives makes the system regularly send packets
5299 to the other end of the connection, leaving it active. The delay between
5300 keep-alive probes is controlled by the system only and depends both on the
5301 operating system and its tuning parameters.
5302
5303 It is important to understand that keep-alive packets are neither emitted nor
5304 received at the application level. It is only the network stacks which sees
5305 them. For this reason, even if one side of the proxy already uses keep-alives
5306 to maintain its connection alive, those keep-alive packets will not be
5307 forwarded to the other side of the proxy.
5308
5309 Please note that this has nothing to do with HTTP keep-alive.
5310
5311 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5312 client side of a connection, which should help when session expirations are
5313 noticed between HAProxy and a client.
5314
5315 If this option has been enabled in a "defaults" section, it can be disabled
5316 in a specific instance by prepending the "no" keyword before it.
5317
5318 See also : "option srvtcpka", "option tcpka"
5319
5320
Willy Tarreau0ba27502007-12-24 16:55:16 +01005321option contstats
5322 Enable continuous traffic statistics updates
5323 May be used in sections : defaults | frontend | listen | backend
5324 yes | yes | yes | no
5325 Arguments : none
5326
5327 By default, counters used for statistics calculation are incremented
5328 only when a session finishes. It works quite well when serving small
5329 objects, but with big ones (for example large images or archives) or
5330 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005331 a hedgehog. With this option enabled counters get incremented frequently
5332 along the session, typically every 5 seconds, which is often enough to
5333 produce clean graphs. Recounting touches a hotpath directly so it is not
5334 not enabled by default, as it can cause a lot of wakeups for very large
5335 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005336
5337
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005338option dontlog-normal
5339no option dontlog-normal
5340 Enable or disable logging of normal, successful connections
5341 May be used in sections : defaults | frontend | listen | backend
5342 yes | yes | yes | no
5343 Arguments : none
5344
5345 There are large sites dealing with several thousand connections per second
5346 and for which logging is a major pain. Some of them are even forced to turn
5347 logs off and cannot debug production issues. Setting this option ensures that
5348 normal connections, those which experience no error, no timeout, no retry nor
5349 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5350 mode, the response status code is checked and return codes 5xx will still be
5351 logged.
5352
5353 It is strongly discouraged to use this option as most of the time, the key to
5354 complex issues is in the normal logs which will not be logged here. If you
5355 need to separate logs, see the "log-separate-errors" option instead.
5356
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005357 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005358 logging.
5359
5360
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005361option dontlognull
5362no option dontlognull
5363 Enable or disable logging of null connections
5364 May be used in sections : defaults | frontend | listen | backend
5365 yes | yes | yes | no
5366 Arguments : none
5367
5368 In certain environments, there are components which will regularly connect to
5369 various systems to ensure that they are still alive. It can be the case from
5370 another load balancer as well as from monitoring systems. By default, even a
5371 simple port probe or scan will produce a log. If those connections pollute
5372 the logs too much, it is possible to enable option "dontlognull" to indicate
5373 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005374 which typically corresponds to those probes. Note that errors will still be
5375 returned to the client and accounted for in the stats. If this is not what is
5376 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005377
5378 It is generally recommended not to use this option in uncontrolled
5379 environments (eg: internet), otherwise scans and other malicious activities
5380 would not be logged.
5381
5382 If this option has been enabled in a "defaults" section, it can be disabled
5383 in a specific instance by prepending the "no" keyword before it.
5384
Willy Tarreau0f228a02015-05-01 15:37:53 +02005385 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5386 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005387
5388
5389option forceclose
5390no option forceclose
5391 Enable or disable active connection closing after response is transferred.
5392 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01005393 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005394 Arguments : none
5395
5396 Some HTTP servers do not necessarily close the connections when they receive
5397 the "Connection: close" set by "option httpclose", and if the client does not
5398 close either, then the connection remains open till the timeout expires. This
5399 causes high number of simultaneous connections on the servers and shows high
5400 global session times in the logs.
5401
5402 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01005403 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01005404 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005405
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005406 This option may also be combined with "option http-pretend-keepalive", which
5407 will disable sending of the "Connection: close" header, but will still cause
5408 the connection to be closed once the whole response is received.
5409
Cyril Bonté653dcd62014-02-20 00:13:15 +01005410 This option disables and replaces any previous "option httpclose", "option
5411 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005412
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005413 If this option has been enabled in a "defaults" section, it can be disabled
5414 in a specific instance by prepending the "no" keyword before it.
5415
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005416 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005417
5418
Willy Tarreau87cf5142011-08-19 22:57:24 +02005419option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005420 Enable insertion of the X-Forwarded-For header to requests sent to servers
5421 May be used in sections : defaults | frontend | listen | backend
5422 yes | yes | yes | yes
5423 Arguments :
5424 <network> is an optional argument used to disable this option for sources
5425 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005426 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005427 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005428
5429 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5430 their client address. This is sometimes annoying when the client's IP address
5431 is expected in server logs. To solve this problem, the well-known HTTP header
5432 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5433 This header contains a value representing the client's IP address. Since this
5434 header is always appended at the end of the existing header list, the server
5435 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005436 the server's manual to find how to enable use of this standard header. Note
5437 that only the last occurrence of the header must be used, since it is really
5438 possible that the client has already brought one.
5439
Willy Tarreaud72758d2010-01-12 10:42:19 +01005440 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005441 the default "X-Forwarded-For". This can be useful where you might already
Willy Tarreaud72758d2010-01-12 10:42:19 +01005442 have a "X-Forwarded-For" header from a different application (eg: stunnel),
5443 and you need preserve it. Also if your backend server doesn't use the
Ross Westaf72a1d2008-08-03 10:51:45 +02005444 "X-Forwarded-For" header and requires different one (eg: Zeus Web Servers
5445 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005446
5447 Sometimes, a same HAProxy instance may be shared between a direct client
5448 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5449 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5450 header for a known source address or network by adding the "except" keyword
5451 followed by the network address. In this case, any source IP matching the
5452 network will not cause an addition of this header. Most common uses are with
5453 private networks or 127.0.0.1.
5454
Willy Tarreau87cf5142011-08-19 22:57:24 +02005455 Alternatively, the keyword "if-none" states that the header will only be
5456 added if it is not present. This should only be used in perfectly trusted
5457 environment, as this might cause a security issue if headers reaching haproxy
5458 are under the control of the end-user.
5459
Willy Tarreauc27debf2008-01-06 08:57:02 +01005460 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005461 least one of them uses it, the header will be added. Note that the backend's
5462 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005463 both are defined. In the case of the "if-none" argument, if at least one of
5464 the frontend or the backend does not specify it, it wants the addition to be
5465 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005466
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005467 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005468 # Public HTTP address also used by stunnel on the same machine
5469 frontend www
5470 mode http
5471 option forwardfor except 127.0.0.1 # stunnel already adds the header
5472
Ross Westaf72a1d2008-08-03 10:51:45 +02005473 # Those servers want the IP Address in X-Client
5474 backend www
5475 mode http
5476 option forwardfor header X-Client
5477
Willy Tarreau87cf5142011-08-19 22:57:24 +02005478 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01005479 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005480
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005481
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005482option http-buffer-request
5483no option http-buffer-request
5484 Enable or disable waiting for whole HTTP request body before proceeding
5485 May be used in sections : defaults | frontend | listen | backend
5486 yes | yes | yes | yes
5487 Arguments : none
5488
5489 It is sometimes desirable to wait for the body of an HTTP request before
5490 taking a decision. This is what is being done by "balance url_param" for
5491 example. The first use case is to buffer requests from slow clients before
5492 connecting to the server. Another use case consists in taking the routing
5493 decision based on the request body's contents. This option placed in a
5494 frontend or backend forces the HTTP processing to wait until either the whole
5495 body is received, or the request buffer is full, or the first chunk is
5496 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005497 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005498 the frontend and the backend, so this should definitely not be used by
5499 default.
5500
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005501 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005502
5503
Willy Tarreau0f228a02015-05-01 15:37:53 +02005504option http-ignore-probes
5505no option http-ignore-probes
5506 Enable or disable logging of null connections and request timeouts
5507 May be used in sections : defaults | frontend | listen | backend
5508 yes | yes | yes | no
5509 Arguments : none
5510
5511 Recently some browsers started to implement a "pre-connect" feature
5512 consisting in speculatively connecting to some recently visited web sites
5513 just in case the user would like to visit them. This results in many
5514 connections being established to web sites, which end up in 408 Request
5515 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5516 decides to close them first. These ones pollute the log and feed the error
5517 counters. There was already "option dontlognull" but it's insufficient in
5518 this case. Instead, this option does the following things :
5519 - prevent any 400/408 message from being sent to the client if nothing
5520 was received over a connection before it was closed ;
5521 - prevent any log from being emitted in this situation ;
5522 - prevent any error counter from being incremented
5523
5524 That way the empty connection is silently ignored. Note that it is better
5525 not to use this unless it is clear that it is needed, because it will hide
5526 real problems. The most common reason for not receiving a request and seeing
5527 a 408 is due to an MTU inconsistency between the client and an intermediary
5528 element such as a VPN, which blocks too large packets. These issues are
5529 generally seen with POST requests as well as GET with large cookies. The logs
5530 are often the only way to detect them.
5531
5532 If this option has been enabled in a "defaults" section, it can be disabled
5533 in a specific instance by prepending the "no" keyword before it.
5534
5535 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5536
5537
Willy Tarreau16bfb022010-01-16 19:48:41 +01005538option http-keep-alive
5539no option http-keep-alive
5540 Enable or disable HTTP keep-alive from client to server
5541 May be used in sections : defaults | frontend | listen | backend
5542 yes | yes | yes | yes
5543 Arguments : none
5544
Willy Tarreau70dffda2014-01-30 03:07:23 +01005545 By default HAProxy operates in keep-alive mode with regards to persistent
5546 connections: for each connection it processes each request and response, and
5547 leaves the connection idle on both sides between the end of a response and the
5548 start of a new request. This mode may be changed by several options such as
5549 "option http-server-close", "option forceclose", "option httpclose" or
5550 "option http-tunnel". This option allows to set back the keep-alive mode,
5551 which can be useful when another mode was used in a defaults section.
5552
5553 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5554 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005555 network) and the fastest session reuse on the server side at the expense
5556 of maintaining idle connections to the servers. In general, it is possible
5557 with this option to achieve approximately twice the request rate that the
5558 "http-server-close" option achieves on small objects. There are mainly two
5559 situations where this option may be useful :
5560
5561 - when the server is non-HTTP compliant and authenticates the connection
5562 instead of requests (eg: NTLM authentication)
5563
5564 - when the cost of establishing the connection to the server is significant
5565 compared to the cost of retrieving the associated object from the server.
5566
5567 This last case can happen when the server is a fast static server of cache.
5568 In this case, the server will need to be properly tuned to support high enough
5569 connection counts because connections will last until the client sends another
5570 request.
5571
5572 If the client request has to go to another backend or another server due to
5573 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01005574 immediately be closed and a new one re-opened. Option "prefer-last-server" is
5575 available to try optimize server selection so that if the server currently
5576 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01005577
5578 In general it is preferred to use "option http-server-close" with application
5579 servers, and some static servers might benefit from "option http-keep-alive".
5580
5581 At the moment, logs will not indicate whether requests came from the same
5582 session or not. The accept date reported in the logs corresponds to the end
5583 of the previous request, and the request time corresponds to the time spent
5584 waiting for a new request. The keep-alive request time is still bound to the
5585 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5586 not set.
5587
Cyril Bonté653dcd62014-02-20 00:13:15 +01005588 This option disables and replaces any previous "option httpclose", "option
5589 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01005590 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01005591 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005592
5593 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01005594 "option prefer-last-server", "option http-pretend-keepalive",
5595 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005596
5597
Willy Tarreau96e31212011-05-30 18:10:30 +02005598option http-no-delay
5599no option http-no-delay
5600 Instruct the system to favor low interactive delays over performance in HTTP
5601 May be used in sections : defaults | frontend | listen | backend
5602 yes | yes | yes | yes
5603 Arguments : none
5604
5605 In HTTP, each payload is unidirectional and has no notion of interactivity.
5606 Any agent is expected to queue data somewhat for a reasonably low delay.
5607 There are some very rare server-to-server applications that abuse the HTTP
5608 protocol and expect the payload phase to be highly interactive, with many
5609 interleaved data chunks in both directions within a single request. This is
5610 absolutely not supported by the HTTP specification and will not work across
5611 most proxies or servers. When such applications attempt to do this through
5612 haproxy, it works but they will experience high delays due to the network
5613 optimizations which favor performance by instructing the system to wait for
5614 enough data to be available in order to only send full packets. Typical
5615 delays are around 200 ms per round trip. Note that this only happens with
5616 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
5617 affected.
5618
5619 When "option http-no-delay" is present in either the frontend or the backend
5620 used by a connection, all such optimizations will be disabled in order to
5621 make the exchanges as fast as possible. Of course this offers no guarantee on
5622 the functionality, as it may break at any other place. But if it works via
5623 HAProxy, it will work as fast as possible. This option should never be used
5624 by default, and should never be used at all unless such a buggy application
5625 is discovered. The impact of using this option is an increase of bandwidth
5626 usage and CPU usage, which may significantly lower performance in high
5627 latency environments.
5628
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005629 See also : "option http-buffer-request"
5630
Willy Tarreau96e31212011-05-30 18:10:30 +02005631
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005632option http-pretend-keepalive
5633no option http-pretend-keepalive
5634 Define whether haproxy will announce keepalive to the server or not
5635 May be used in sections : defaults | frontend | listen | backend
5636 yes | yes | yes | yes
5637 Arguments : none
5638
5639 When running with "option http-server-close" or "option forceclose", haproxy
5640 adds a "Connection: close" header to the request forwarded to the server.
5641 Unfortunately, when some servers see this header, they automatically refrain
5642 from using the chunked encoding for responses of unknown length, while this
5643 is totally unrelated. The immediate effect is that this prevents haproxy from
5644 maintaining the client connection alive. A second effect is that a client or
5645 a cache could receive an incomplete response without being aware of it, and
5646 consider the response complete.
5647
5648 By setting "option http-pretend-keepalive", haproxy will make the server
5649 believe it will keep the connection alive. The server will then not fall back
5650 to the abnormal undesired above. When haproxy gets the whole response, it
5651 will close the connection with the server just as it would do with the
5652 "forceclose" option. That way the client gets a normal response and the
5653 connection is correctly closed on the server side.
5654
5655 It is recommended not to enable this option by default, because most servers
5656 will more efficiently close the connection themselves after the last packet,
5657 and release its buffers slightly earlier. Also, the added packet on the
5658 network could slightly reduce the overall peak performance. However it is
5659 worth noting that when this option is enabled, haproxy will have slightly
5660 less work to do. So if haproxy is the bottleneck on the whole architecture,
5661 enabling this option might save a few CPU cycles.
5662
5663 This option may be set both in a frontend and in a backend. It is enabled if
5664 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005665 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02005666 keepalive to be announced to the server and close to be announced to the
5667 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005668
5669 If this option has been enabled in a "defaults" section, it can be disabled
5670 in a specific instance by prepending the "no" keyword before it.
5671
Willy Tarreau16bfb022010-01-16 19:48:41 +01005672 See also : "option forceclose", "option http-server-close", and
5673 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005674
Willy Tarreauc27debf2008-01-06 08:57:02 +01005675
Willy Tarreaub608feb2010-01-02 22:47:18 +01005676option http-server-close
5677no option http-server-close
5678 Enable or disable HTTP connection closing on the server side
5679 May be used in sections : defaults | frontend | listen | backend
5680 yes | yes | yes | yes
5681 Arguments : none
5682
Willy Tarreau70dffda2014-01-30 03:07:23 +01005683 By default HAProxy operates in keep-alive mode with regards to persistent
5684 connections: for each connection it processes each request and response, and
5685 leaves the connection idle on both sides between the end of a response and
5686 the start of a new request. This mode may be changed by several options such
5687 as "option http-server-close", "option forceclose", "option httpclose" or
5688 "option http-tunnel". Setting "option http-server-close" enables HTTP
5689 connection-close mode on the server side while keeping the ability to support
5690 HTTP keep-alive and pipelining on the client side. This provides the lowest
5691 latency on the client side (slow network) and the fastest session reuse on
5692 the server side to save server resources, similarly to "option forceclose".
5693 It also permits non-keepalive capable servers to be served in keep-alive mode
Lukas Tribus23953682017-04-28 13:24:30 +00005694 to the clients if they conform to the requirements of RFC7230. Please note
Willy Tarreau70dffda2014-01-30 03:07:23 +01005695 that some servers do not always conform to those requirements when they see
5696 "Connection: close" in the request. The effect will be that keep-alive will
5697 never be used. A workaround consists in enabling "option
5698 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005699
5700 At the moment, logs will not indicate whether requests came from the same
5701 session or not. The accept date reported in the logs corresponds to the end
5702 of the previous request, and the request time corresponds to the time spent
5703 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01005704 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5705 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005706
5707 This option may be set both in a frontend and in a backend. It is enabled if
5708 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01005709 It disables and replaces any previous "option httpclose", "option forceclose",
5710 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01005711 ("Proxies") to see how this option combines with others when frontend and
5712 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005713
5714 If this option has been enabled in a "defaults" section, it can be disabled
5715 in a specific instance by prepending the "no" keyword before it.
5716
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02005717 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01005718 "option httpclose", "option http-keep-alive", and
5719 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005720
5721
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005722option http-tunnel
5723no option http-tunnel
5724 Disable or enable HTTP connection processing after first transaction
5725 May be used in sections : defaults | frontend | listen | backend
5726 yes | yes | yes | yes
5727 Arguments : none
5728
Willy Tarreau70dffda2014-01-30 03:07:23 +01005729 By default HAProxy operates in keep-alive mode with regards to persistent
5730 connections: for each connection it processes each request and response, and
5731 leaves the connection idle on both sides between the end of a response and
5732 the start of a new request. This mode may be changed by several options such
5733 as "option http-server-close", "option forceclose", "option httpclose" or
5734 "option http-tunnel".
5735
5736 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005737 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01005738 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
5739 is normally not needed anymore unless in very specific cases such as when
5740 using an in-house protocol that looks like HTTP but is not compatible, or
5741 just to log one request per client in order to reduce log size. Note that
5742 everything which works at the HTTP level, including header parsing/addition,
5743 cookie processing or content switching will only work for the first request
5744 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005745
5746 If this option has been enabled in a "defaults" section, it can be disabled
5747 in a specific instance by prepending the "no" keyword before it.
5748
5749 See also : "option forceclose", "option http-server-close",
5750 "option httpclose", "option http-keep-alive", and
5751 "1.1. The HTTP transaction model".
5752
5753
Willy Tarreau88d349d2010-01-25 12:15:43 +01005754option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01005755no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01005756 Make use of non-standard Proxy-Connection header instead of Connection
5757 May be used in sections : defaults | frontend | listen | backend
5758 yes | yes | yes | no
5759 Arguments : none
5760
Lukas Tribus23953682017-04-28 13:24:30 +00005761 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01005762 Connection header to indicate their wish of persistent or non-persistent
5763 connections, both browsers and proxies ignore this header for proxied
5764 connections and make use of the undocumented, non-standard Proxy-Connection
5765 header instead. The issue begins when trying to put a load balancer between
5766 browsers and such proxies, because there will be a difference between what
5767 haproxy understands and what the client and the proxy agree on.
5768
5769 By setting this option in a frontend, haproxy can automatically switch to use
5770 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01005771 defined here as one where the URI begins with neither a '/' nor a '*'. This
5772 is incompatible with the HTTP tunnel mode. Note that this option can only be
5773 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01005774
Willy Tarreau844a7e72010-01-31 21:46:18 +01005775 Also, when this option is set, a request which requires authentication will
5776 automatically switch to use proxy authentication headers if it is itself a
5777 proxied request. That makes it possible to check or enforce authentication in
5778 front of an existing proxy.
5779
Willy Tarreau88d349d2010-01-25 12:15:43 +01005780 This option should normally never be used, except in front of a proxy.
5781
5782 See also : "option httpclose", "option forceclose" and "option
5783 http-server-close".
5784
5785
Willy Tarreaud63335a2010-02-26 12:56:52 +01005786option httpchk
5787option httpchk <uri>
5788option httpchk <method> <uri>
5789option httpchk <method> <uri> <version>
5790 Enable HTTP protocol to check on the servers health
5791 May be used in sections : defaults | frontend | listen | backend
5792 yes | no | yes | yes
5793 Arguments :
5794 <method> is the optional HTTP method used with the requests. When not set,
5795 the "OPTIONS" method is used, as it generally requires low server
5796 processing and is easy to filter out from the logs. Any method
5797 may be used, though it is not recommended to invent non-standard
5798 ones.
5799
5800 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
5801 which is accessible by default on almost any server, but may be
5802 changed to any other URI. Query strings are permitted.
5803
5804 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
5805 but some servers might behave incorrectly in HTTP 1.0, so turning
5806 it to HTTP/1.1 may sometimes help. Note that the Host field is
5807 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
5808 after "\r\n" following the version string.
5809
5810 By default, server health checks only consist in trying to establish a TCP
5811 connection. When "option httpchk" is specified, a complete HTTP request is
5812 sent once the TCP connection is established, and responses 2xx and 3xx are
5813 considered valid, while all other ones indicate a server failure, including
5814 the lack of any response.
5815
5816 The port and interval are specified in the server configuration.
5817
5818 This option does not necessarily require an HTTP backend, it also works with
5819 plain TCP backends. This is particularly useful to check simple scripts bound
5820 to some dedicated ports using the inetd daemon.
5821
5822 Examples :
5823 # Relay HTTPS traffic to Apache instance and check service availability
5824 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
5825 backend https_relay
5826 mode tcp
5827 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
5828 server apache1 192.168.1.1:443 check port 80
5829
Simon Hormanafc47ee2013-11-25 10:46:35 +09005830 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
5831 "option pgsql-check", "http-check" and the "check", "port" and
5832 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01005833
5834
Willy Tarreauc27debf2008-01-06 08:57:02 +01005835option httpclose
5836no option httpclose
5837 Enable or disable passive HTTP connection closing
5838 May be used in sections : defaults | frontend | listen | backend
5839 yes | yes | yes | yes
5840 Arguments : none
5841
Willy Tarreau70dffda2014-01-30 03:07:23 +01005842 By default HAProxy operates in keep-alive mode with regards to persistent
5843 connections: for each connection it processes each request and response, and
5844 leaves the connection idle on both sides between the end of a response and
5845 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01005846 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01005847 "option http-tunnel".
5848
5849 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
5850 if a "Connection: close" header is already set in each direction, and will
5851 add one if missing. Each end should react to this by actively closing the TCP
5852 connection after each transfer, thus resulting in a switch to the HTTP close
5853 mode. Any "Connection" header different from "close" will also be removed.
5854 Note that this option is deprecated since what it does is very cheap but not
5855 reliable. Using "option http-server-close" or "option forceclose" is strongly
5856 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005857
5858 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005859 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01005860 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
5861 it is possible to use the "option forceclose" which actively closes the
5862 request connection once the server responds. Option "forceclose" also
5863 releases the server connection earlier because it does not have to wait for
5864 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005865
5866 This option may be set both in a frontend and in a backend. It is enabled if
5867 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01005868 It disables and replaces any previous "option http-server-close",
5869 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01005870 check section 4 ("Proxies") to see how this option combines with others when
5871 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005872
5873 If this option has been enabled in a "defaults" section, it can be disabled
5874 in a specific instance by prepending the "no" keyword before it.
5875
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02005876 See also : "option forceclose", "option http-server-close" and
5877 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01005878
5879
Emeric Brun3a058f32009-06-30 18:26:00 +02005880option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005881 Enable logging of HTTP request, session state and timers
5882 May be used in sections : defaults | frontend | listen | backend
5883 yes | yes | yes | yes
Emeric Brun3a058f32009-06-30 18:26:00 +02005884 Arguments :
5885 clf if the "clf" argument is added, then the output format will be
5886 the CLF format instead of HAProxy's default HTTP format. You can
5887 use this when you need to feed HAProxy's logs through a specific
5888 log analyser which only support the CLF format and which is not
5889 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005890
5891 By default, the log output format is very poor, as it only contains the
5892 source and destination addresses, and the instance name. By specifying
5893 "option httplog", each log line turns into a much richer format including,
5894 but not limited to, the HTTP request, the connection timers, the session
5895 status, the connections numbers, the captured headers and cookies, the
5896 frontend, backend and server name, and of course the source address and
5897 ports.
5898
5899 This option may be set either in the frontend or the backend.
5900
PiBa-NLbd556bf2014-12-11 21:31:54 +01005901 Specifying only "option httplog" will automatically clear the 'clf' mode
5902 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02005903
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005904 "option httplog" overrides any previous "log-format" directive.
5905
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005906 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005907
Willy Tarreau55165fe2009-05-10 12:02:55 +02005908
5909option http_proxy
5910no option http_proxy
5911 Enable or disable plain HTTP proxy mode
5912 May be used in sections : defaults | frontend | listen | backend
5913 yes | yes | yes | yes
5914 Arguments : none
5915
5916 It sometimes happens that people need a pure HTTP proxy which understands
5917 basic proxy requests without caching nor any fancy feature. In this case,
5918 it may be worth setting up an HAProxy instance with the "option http_proxy"
5919 set. In this mode, no server is declared, and the connection is forwarded to
5920 the IP address and port found in the URL after the "http://" scheme.
5921
5922 No host address resolution is performed, so this only works when pure IP
5923 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01005924 it will probably be used only by experts who know they need exactly it. This
5925 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02005926
5927 If this option has been enabled in a "defaults" section, it can be disabled
5928 in a specific instance by prepending the "no" keyword before it.
5929
5930 Example :
5931 # this backend understands HTTP proxy requests and forwards them directly.
5932 backend direct_forward
5933 option httpclose
5934 option http_proxy
5935
5936 See also : "option httpclose"
5937
Willy Tarreau211ad242009-10-03 21:45:07 +02005938
Jamie Gloudon801a0a32012-08-25 00:18:33 -04005939option independent-streams
5940no option independent-streams
5941 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02005942 May be used in sections : defaults | frontend | listen | backend
5943 yes | yes | yes | yes
5944 Arguments : none
5945
5946 By default, when data is sent over a socket, both the write timeout and the
5947 read timeout for that socket are refreshed, because we consider that there is
5948 activity on that socket, and we have no other means of guessing if we should
5949 receive data or not.
5950
5951 While this default behaviour is desirable for almost all applications, there
5952 exists a situation where it is desirable to disable it, and only refresh the
5953 read timeout if there are incoming data. This happens on sessions with large
5954 timeouts and low amounts of exchanged data such as telnet session. If the
5955 server suddenly disappears, the output data accumulates in the system's
5956 socket buffers, both timeouts are correctly refreshed, and there is no way
5957 to know the server does not receive them, so we don't timeout. However, when
5958 the underlying protocol always echoes sent data, it would be enough by itself
5959 to detect the issue using the read timeout. Note that this problem does not
5960 happen with more verbose protocols because data won't accumulate long in the
5961 socket buffers.
5962
5963 When this option is set on the frontend, it will disable read timeout updates
5964 on data sent to the client. There probably is little use of this case. When
5965 the option is set on the backend, it will disable read timeout updates on
5966 data sent to the server. Doing so will typically break large HTTP posts from
5967 slow lines, so use it with caution.
5968
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005969 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04005970 with a spelling mistake. This spelling is still supported but
5971 deprecated.
5972
Willy Tarreauce887fd2012-05-12 12:50:00 +02005973 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02005974
5975
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02005976option ldap-check
5977 Use LDAPv3 health checks for server testing
5978 May be used in sections : defaults | frontend | listen | backend
5979 yes | no | yes | yes
5980 Arguments : none
5981
5982 It is possible to test that the server correctly talks LDAPv3 instead of just
5983 testing that it accepts the TCP connection. When this option is set, an
5984 LDAPv3 anonymous simple bind message is sent to the server, and the response
5985 is analyzed to find an LDAPv3 bind response message.
5986
5987 The server is considered valid only when the LDAP response contains success
5988 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
5989
5990 Logging of bind requests is server dependent see your documentation how to
5991 configure it.
5992
5993 Example :
5994 option ldap-check
5995
5996 See also : "option httpchk"
5997
5998
Simon Horman98637e52014-06-20 12:30:16 +09005999option external-check
6000 Use external processes for server health checks
6001 May be used in sections : defaults | frontend | listen | backend
6002 yes | no | yes | yes
6003
6004 It is possible to test the health of a server using an external command.
6005 This is achieved by running the executable set using "external-check
6006 command".
6007
6008 Requires the "external-check" global to be set.
6009
6010 See also : "external-check", "external-check command", "external-check path"
6011
6012
Willy Tarreau211ad242009-10-03 21:45:07 +02006013option log-health-checks
6014no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006015 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006016 May be used in sections : defaults | frontend | listen | backend
6017 yes | no | yes | yes
6018 Arguments : none
6019
Willy Tarreaubef1b322014-05-13 21:01:39 +02006020 By default, failed health check are logged if server is UP and successful
6021 health checks are logged if server is DOWN, so the amount of additional
6022 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006023
Willy Tarreaubef1b322014-05-13 21:01:39 +02006024 When this option is enabled, any change of the health check status or to
6025 the server's health will be logged, so that it becomes possible to know
6026 that a server was failing occasional checks before crashing, or exactly when
6027 it failed to respond a valid HTTP status, then when the port started to
6028 reject connections, then when the server stopped responding at all.
6029
6030 Note that status changes not caused by health checks (eg: enable/disable on
6031 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006032
Willy Tarreaubef1b322014-05-13 21:01:39 +02006033 See also: "option httpchk", "option ldap-check", "option mysql-check",
6034 "option pgsql-check", "option redis-check", "option smtpchk",
6035 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006036
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006037
6038option log-separate-errors
6039no option log-separate-errors
6040 Change log level for non-completely successful connections
6041 May be used in sections : defaults | frontend | listen | backend
6042 yes | yes | yes | no
6043 Arguments : none
6044
6045 Sometimes looking for errors in logs is not easy. This option makes haproxy
6046 raise the level of logs containing potentially interesting information such
6047 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6048 level changes from "info" to "err". This makes it possible to log them
6049 separately to a different file with most syslog daemons. Be careful not to
6050 remove them from the original file, otherwise you would lose ordering which
6051 provides very important information.
6052
6053 Using this option, large sites dealing with several thousand connections per
6054 second may log normal traffic to a rotating buffer and only archive smaller
6055 error logs.
6056
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006057 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006058 logging.
6059
Willy Tarreauc27debf2008-01-06 08:57:02 +01006060
6061option logasap
6062no option logasap
6063 Enable or disable early logging of HTTP requests
6064 May be used in sections : defaults | frontend | listen | backend
6065 yes | yes | yes | no
6066 Arguments : none
6067
6068 By default, HTTP requests are logged upon termination so that the total
6069 transfer time and the number of bytes appear in the logs. When large objects
6070 are being transferred, it may take a while before the request appears in the
6071 logs. Using "option logasap", the request gets logged as soon as the server
6072 sends the complete headers. The only missing information in the logs will be
6073 the total number of bytes which will indicate everything except the amount
6074 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006075 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006076 "Content-Length" response header so that the logs at least indicate how many
6077 bytes are expected to be transferred.
6078
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006079 Examples :
6080 listen http_proxy 0.0.0.0:80
6081 mode http
6082 option httplog
6083 option logasap
6084 log 192.168.2.200 local3
6085
6086 >>> Feb 6 12:14:14 localhost \
6087 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6088 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6089 "GET /image.iso HTTP/1.0"
6090
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006091 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006092 logging.
6093
6094
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006095option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006096 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006097 May be used in sections : defaults | frontend | listen | backend
6098 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006099 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006100 <username> This is the username which will be used when connecting to MySQL
6101 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006102 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006103
6104 If you specify a username, the check consists of sending two MySQL packet,
6105 one Client Authentication packet, and one QUIT packet, to correctly close
6106 MySQL session. We then parse the MySQL Handshake Initialisation packet and/or
6107 Error packet. It is a basic but useful test which does not produce error nor
6108 aborted connect on the server. However, it requires adding an authorization
6109 in the MySQL table, like this :
6110
6111 USE mysql;
6112 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6113 FLUSH PRIVILEGES;
6114
6115 If you don't specify a username (it is deprecated and not recommended), the
6116 check only consists in parsing the Mysql Handshake Initialisation packet or
6117 Error packet, we don't send anything in this mode. It was reported that it
6118 can generate lockout if check is too frequent and/or if there is not enough
6119 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6120 value as if a connection is established successfully within fewer than MySQL
6121 "max_connect_errors" attempts after a previous connection was interrupted,
6122 the error count for the host is cleared to zero. If HAProxy's server get
6123 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6124
6125 Remember that this does not check database presence nor database consistency.
6126 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006127
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006128 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006129
6130 Most often, an incoming MySQL server needs to see the client's IP address for
6131 various purposes, including IP privilege matching and connection logging.
6132 When possible, it is often wise to masquerade the client's IP address when
6133 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006134 which requires the transparent proxy feature to be compiled in, and the MySQL
6135 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006136
6137 See also: "option httpchk"
6138
6139
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006140option nolinger
6141no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006142 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006143 May be used in sections: defaults | frontend | listen | backend
6144 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006145 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006146
6147 When clients or servers abort connections in a dirty way (eg: they are
6148 physically disconnected), the session timeouts triggers and the session is
6149 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6150 using some resources and possibly limiting the ability to establish newer
6151 connections.
6152
6153 When this happens, it is possible to activate "option nolinger" which forces
6154 the system to immediately remove any socket's pending data on close. Thus,
6155 the session is instantly purged from the system's tables. This usually has
6156 side effects such as increased number of TCP resets due to old retransmits
6157 getting immediately rejected. Some firewalls may sometimes complain about
6158 this too.
6159
6160 For this reason, it is not recommended to use this option when not absolutely
6161 needed. You know that you need it when you have thousands of FIN_WAIT1
6162 sessions on your system (TIME_WAIT ones do not count).
6163
6164 This option may be used both on frontends and backends, depending on the side
6165 where it is required. Use it on the frontend for clients, and on the backend
6166 for servers.
6167
6168 If this option has been enabled in a "defaults" section, it can be disabled
6169 in a specific instance by prepending the "no" keyword before it.
6170
6171
Willy Tarreau55165fe2009-05-10 12:02:55 +02006172option originalto [ except <network> ] [ header <name> ]
6173 Enable insertion of the X-Original-To header to requests sent to servers
6174 May be used in sections : defaults | frontend | listen | backend
6175 yes | yes | yes | yes
6176 Arguments :
6177 <network> is an optional argument used to disable this option for sources
6178 matching <network>
6179 <name> an optional argument to specify a different "X-Original-To"
6180 header name.
6181
6182 Since HAProxy can work in transparent mode, every request from a client can
6183 be redirected to the proxy and HAProxy itself can proxy every request to a
6184 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6185 be lost. This is annoying when you want access rules based on destination ip
6186 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6187 added by HAProxy to all requests sent to the server. This header contains a
6188 value representing the original destination IP address. Since this must be
6189 configured to always use the last occurrence of this header only. Note that
6190 only the last occurrence of the header must be used, since it is really
6191 possible that the client has already brought one.
6192
6193 The keyword "header" may be used to supply a different header name to replace
6194 the default "X-Original-To". This can be useful where you might already
6195 have a "X-Original-To" header from a different application, and you need
6196 preserve it. Also if your backend server doesn't use the "X-Original-To"
6197 header and requires different one.
6198
6199 Sometimes, a same HAProxy instance may be shared between a direct client
6200 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6201 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6202 header for a known source address or network by adding the "except" keyword
6203 followed by the network address. In this case, any source IP matching the
6204 network will not cause an addition of this header. Most common uses are with
6205 private networks or 127.0.0.1.
6206
6207 This option may be specified either in the frontend or in the backend. If at
6208 least one of them uses it, the header will be added. Note that the backend's
6209 setting of the header subargument takes precedence over the frontend's if
6210 both are defined.
6211
Willy Tarreau55165fe2009-05-10 12:02:55 +02006212 Examples :
6213 # Original Destination address
6214 frontend www
6215 mode http
6216 option originalto except 127.0.0.1
6217
6218 # Those servers want the IP Address in X-Client-Dst
6219 backend www
6220 mode http
6221 option originalto header X-Client-Dst
6222
Willy Tarreau87cf5142011-08-19 22:57:24 +02006223 See also : "option httpclose", "option http-server-close",
6224 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02006225
6226
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006227option persist
6228no option persist
6229 Enable or disable forced persistence on down servers
6230 May be used in sections: defaults | frontend | listen | backend
6231 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006232 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006233
6234 When an HTTP request reaches a backend with a cookie which references a dead
6235 server, by default it is redispatched to another server. It is possible to
6236 force the request to be sent to the dead server first using "option persist"
6237 if absolutely needed. A common use case is when servers are under extreme
6238 load and spend their time flapping. In this case, the users would still be
6239 directed to the server they opened the session on, in the hope they would be
6240 correctly served. It is recommended to use "option redispatch" in conjunction
6241 with this option so that in the event it would not be possible to connect to
6242 the server at all (server definitely dead), the client would finally be
6243 redirected to another valid server.
6244
6245 If this option has been enabled in a "defaults" section, it can be disabled
6246 in a specific instance by prepending the "no" keyword before it.
6247
Willy Tarreau4de91492010-01-22 19:10:05 +01006248 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006249
6250
Willy Tarreau0c122822013-12-15 18:49:01 +01006251option pgsql-check [ user <username> ]
6252 Use PostgreSQL health checks for server testing
6253 May be used in sections : defaults | frontend | listen | backend
6254 yes | no | yes | yes
6255 Arguments :
6256 <username> This is the username which will be used when connecting to
6257 PostgreSQL server.
6258
6259 The check sends a PostgreSQL StartupMessage and waits for either
6260 Authentication request or ErrorResponse message. It is a basic but useful
6261 test which does not produce error nor aborted connect on the server.
6262 This check is identical with the "mysql-check".
6263
6264 See also: "option httpchk"
6265
6266
Willy Tarreau9420b122013-12-15 18:58:25 +01006267option prefer-last-server
6268no option prefer-last-server
6269 Allow multiple load balanced requests to remain on the same server
6270 May be used in sections: defaults | frontend | listen | backend
6271 yes | no | yes | yes
6272 Arguments : none
6273
6274 When the load balancing algorithm in use is not deterministic, and a previous
6275 request was sent to a server to which haproxy still holds a connection, it is
6276 sometimes desirable that subsequent requests on a same session go to the same
6277 server as much as possible. Note that this is different from persistence, as
6278 we only indicate a preference which haproxy tries to apply without any form
6279 of warranty. The real use is for keep-alive connections sent to servers. When
6280 this option is used, haproxy will try to reuse the same connection that is
6281 attached to the server instead of rebalancing to another server, causing a
6282 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006283 not make much sense to use this in combination with hashing algorithms. Note,
6284 haproxy already automatically tries to stick to a server which sends a 401 or
6285 to a proxy which sends a 407 (authentication required). This is mandatory for
6286 use with the broken NTLM authentication challenge, and significantly helps in
6287 troubleshooting some faulty applications. Option prefer-last-server might be
6288 desirable in these environments as well, to avoid redistributing the traffic
6289 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006290
6291 If this option has been enabled in a "defaults" section, it can be disabled
6292 in a specific instance by prepending the "no" keyword before it.
6293
6294 See also: "option http-keep-alive"
6295
6296
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006297option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006298option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006299no option redispatch
6300 Enable or disable session redistribution in case of connection failure
6301 May be used in sections: defaults | frontend | listen | backend
6302 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006303 Arguments :
6304 <interval> The optional integer value that controls how often redispatches
6305 occur when retrying connections. Positive value P indicates a
6306 redispatch is desired on every Pth retry, and negative value
6307 N indicate a redispath is desired on the Nth retry prior to the
6308 last retry. For example, the default of -1 preserves the
6309 historical behaviour of redispatching on the last retry, a
6310 positive value of 1 would indicate a redispatch on every retry,
6311 and a positive value of 3 would indicate a redispatch on every
6312 third retry. You can disable redispatches with a value of 0.
6313
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006314
6315 In HTTP mode, if a server designated by a cookie is down, clients may
6316 definitely stick to it because they cannot flush the cookie, so they will not
6317 be able to access the service anymore.
6318
6319 Specifying "option redispatch" will allow the proxy to break their
6320 persistence and redistribute them to a working server.
6321
Joseph Lynch726ab712015-05-11 23:25:34 -07006322 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006323 connection failures. Of course, it requires having "retries" set to a nonzero
6324 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006325
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006326 This form is the preferred form, which replaces both the "redispatch" and
6327 "redisp" keywords.
6328
6329 If this option has been enabled in a "defaults" section, it can be disabled
6330 in a specific instance by prepending the "no" keyword before it.
6331
Willy Tarreau4de91492010-01-22 19:10:05 +01006332 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006333
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006334
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006335option redis-check
6336 Use redis health checks for server testing
6337 May be used in sections : defaults | frontend | listen | backend
6338 yes | no | yes | yes
6339 Arguments : none
6340
6341 It is possible to test that the server correctly talks REDIS protocol instead
6342 of just testing that it accepts the TCP connection. When this option is set,
6343 a PING redis command is sent to the server, and the response is analyzed to
6344 find the "+PONG" response message.
6345
6346 Example :
6347 option redis-check
6348
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006349 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006350
6351
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006352option smtpchk
6353option smtpchk <hello> <domain>
6354 Use SMTP health checks for server testing
6355 May be used in sections : defaults | frontend | listen | backend
6356 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006357 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006358 <hello> is an optional argument. It is the "hello" command to use. It can
6359 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
6360 values will be turned into the default command ("HELO").
6361
6362 <domain> is the domain name to present to the server. It may only be
6363 specified (and is mandatory) if the hello command has been
6364 specified. By default, "localhost" is used.
6365
6366 When "option smtpchk" is set, the health checks will consist in TCP
6367 connections followed by an SMTP command. By default, this command is
6368 "HELO localhost". The server's return code is analyzed and only return codes
6369 starting with a "2" will be considered as valid. All other responses,
6370 including a lack of response will constitute an error and will indicate a
6371 dead server.
6372
6373 This test is meant to be used with SMTP servers or relays. Depending on the
6374 request, it is possible that some servers do not log each connection attempt,
6375 so you may want to experiment to improve the behaviour. Using telnet on port
6376 25 is often easier than adjusting the configuration.
6377
6378 Most often, an incoming SMTP server needs to see the client's IP address for
6379 various purposes, including spam filtering, anti-spoofing and logging. When
6380 possible, it is often wise to masquerade the client's IP address when
6381 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006382 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006383
6384 Example :
6385 option smtpchk HELO mydomain.org
6386
6387 See also : "option httpchk", "source"
6388
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006389
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006390option socket-stats
6391no option socket-stats
6392
6393 Enable or disable collecting & providing separate statistics for each socket.
6394 May be used in sections : defaults | frontend | listen | backend
6395 yes | yes | yes | no
6396
6397 Arguments : none
6398
6399
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006400option splice-auto
6401no option splice-auto
6402 Enable or disable automatic kernel acceleration on sockets in both directions
6403 May be used in sections : defaults | frontend | listen | backend
6404 yes | yes | yes | yes
6405 Arguments : none
6406
6407 When this option is enabled either on a frontend or on a backend, haproxy
6408 will automatically evaluate the opportunity to use kernel tcp splicing to
6409 forward data between the client and the server, in either direction. Haproxy
6410 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006411 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006412 are not much aggressive in order to limit excessive use of splicing. This
6413 option requires splicing to be enabled at compile time, and may be globally
6414 disabled with the global option "nosplice". Since splice uses pipes, using it
6415 requires that there are enough spare pipes.
6416
6417 Important note: kernel-based TCP splicing is a Linux-specific feature which
6418 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6419 transfer data between sockets without copying these data to user-space, thus
6420 providing noticeable performance gains and CPU cycles savings. Since many
6421 early implementations are buggy, corrupt data and/or are inefficient, this
6422 feature is not enabled by default, and it should be used with extreme care.
6423 While it is not possible to detect the correctness of an implementation,
6424 2.6.29 is the first version offering a properly working implementation. In
6425 case of doubt, splicing may be globally disabled using the global "nosplice"
6426 keyword.
6427
6428 Example :
6429 option splice-auto
6430
6431 If this option has been enabled in a "defaults" section, it can be disabled
6432 in a specific instance by prepending the "no" keyword before it.
6433
6434 See also : "option splice-request", "option splice-response", and global
6435 options "nosplice" and "maxpipes"
6436
6437
6438option splice-request
6439no option splice-request
6440 Enable or disable automatic kernel acceleration on sockets for requests
6441 May be used in sections : defaults | frontend | listen | backend
6442 yes | yes | yes | yes
6443 Arguments : none
6444
6445 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006446 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006447 the client to the server. It might still use the recv/send scheme if there
6448 are no spare pipes left. This option requires splicing to be enabled at
6449 compile time, and may be globally disabled with the global option "nosplice".
6450 Since splice uses pipes, using it requires that there are enough spare pipes.
6451
6452 Important note: see "option splice-auto" for usage limitations.
6453
6454 Example :
6455 option splice-request
6456
6457 If this option has been enabled in a "defaults" section, it can be disabled
6458 in a specific instance by prepending the "no" keyword before it.
6459
6460 See also : "option splice-auto", "option splice-response", and global options
6461 "nosplice" and "maxpipes"
6462
6463
6464option splice-response
6465no option splice-response
6466 Enable or disable automatic kernel acceleration on sockets for responses
6467 May be used in sections : defaults | frontend | listen | backend
6468 yes | yes | yes | yes
6469 Arguments : none
6470
6471 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006472 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006473 the server to the client. It might still use the recv/send scheme if there
6474 are no spare pipes left. This option requires splicing to be enabled at
6475 compile time, and may be globally disabled with the global option "nosplice".
6476 Since splice uses pipes, using it requires that there are enough spare pipes.
6477
6478 Important note: see "option splice-auto" for usage limitations.
6479
6480 Example :
6481 option splice-response
6482
6483 If this option has been enabled in a "defaults" section, it can be disabled
6484 in a specific instance by prepending the "no" keyword before it.
6485
6486 See also : "option splice-auto", "option splice-request", and global options
6487 "nosplice" and "maxpipes"
6488
6489
Christopher Fauletba7bc162016-11-07 21:07:38 +01006490option spop-check
6491 Use SPOP health checks for server testing
6492 May be used in sections : defaults | frontend | listen | backend
6493 no | no | no | yes
6494 Arguments : none
6495
6496 It is possible to test that the server correctly talks SPOP protocol instead
6497 of just testing that it accepts the TCP connection. When this option is set,
6498 a HELLO handshake is performed between HAProxy and the server, and the
6499 response is analyzed to check no error is reported.
6500
6501 Example :
6502 option spop-check
6503
6504 See also : "option httpchk"
6505
6506
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006507option srvtcpka
6508no option srvtcpka
6509 Enable or disable the sending of TCP keepalive packets on the server side
6510 May be used in sections : defaults | frontend | listen | backend
6511 yes | no | yes | yes
6512 Arguments : none
6513
6514 When there is a firewall or any session-aware component between a client and
6515 a server, and when the protocol involves very long sessions with long idle
6516 periods (eg: remote desktops), there is a risk that one of the intermediate
6517 components decides to expire a session which has remained idle for too long.
6518
6519 Enabling socket-level TCP keep-alives makes the system regularly send packets
6520 to the other end of the connection, leaving it active. The delay between
6521 keep-alive probes is controlled by the system only and depends both on the
6522 operating system and its tuning parameters.
6523
6524 It is important to understand that keep-alive packets are neither emitted nor
6525 received at the application level. It is only the network stacks which sees
6526 them. For this reason, even if one side of the proxy already uses keep-alives
6527 to maintain its connection alive, those keep-alive packets will not be
6528 forwarded to the other side of the proxy.
6529
6530 Please note that this has nothing to do with HTTP keep-alive.
6531
6532 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6533 server side of a connection, which should help when session expirations are
6534 noticed between HAProxy and a server.
6535
6536 If this option has been enabled in a "defaults" section, it can be disabled
6537 in a specific instance by prepending the "no" keyword before it.
6538
6539 See also : "option clitcpka", "option tcpka"
6540
6541
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006542option ssl-hello-chk
6543 Use SSLv3 client hello health checks for server testing
6544 May be used in sections : defaults | frontend | listen | backend
6545 yes | no | yes | yes
6546 Arguments : none
6547
6548 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6549 possible to test that the server correctly talks SSL instead of just testing
6550 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6551 SSLv3 client hello messages are sent once the connection is established to
6552 the server, and the response is analyzed to find an SSL server hello message.
6553 The server is considered valid only when the response contains this server
6554 hello message.
6555
6556 All servers tested till there correctly reply to SSLv3 client hello messages,
6557 and most servers tested do not even log the requests containing only hello
6558 messages, which is appreciable.
6559
Willy Tarreau763a95b2012-10-04 23:15:39 +02006560 Note that this check works even when SSL support was not built into haproxy
6561 because it forges the SSL message. When SSL support is available, it is best
6562 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006563
Willy Tarreau763a95b2012-10-04 23:15:39 +02006564 See also: "option httpchk", "check-ssl"
6565
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006566
Willy Tarreaued179852013-12-16 01:07:00 +01006567option tcp-check
6568 Perform health checks using tcp-check send/expect sequences
6569 May be used in sections: defaults | frontend | listen | backend
6570 yes | no | yes | yes
6571
6572 This health check method is intended to be combined with "tcp-check" command
6573 lists in order to support send/expect types of health check sequences.
6574
6575 TCP checks currently support 4 modes of operations :
6576 - no "tcp-check" directive : the health check only consists in a connection
6577 attempt, which remains the default mode.
6578
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006579 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01006580 used to send a string along with a connection opening. With some
6581 protocols, it helps sending a "QUIT" message for example that prevents
6582 the server from logging a connection error for each health check. The
6583 check result will still be based on the ability to open the connection
6584 only.
6585
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006586 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01006587 The connection is opened and haproxy waits for the server to present some
6588 contents which must validate some rules. The check result will be based
6589 on the matching between the contents and the rules. This is suited for
6590 POP, IMAP, SMTP, FTP, SSH, TELNET.
6591
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006592 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01006593 used to test a hello-type protocol. Haproxy sends a message, the server
6594 responds and its response is analysed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006595 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01006596 suited for protocols which require a binding or a request/response model.
6597 LDAP, MySQL, Redis and SSL are example of such protocols, though they
6598 already all have their dedicated checks with a deeper understanding of
6599 the respective protocols.
6600 In this mode, many questions may be sent and many answers may be
6601 analysed.
6602
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006603 A fifth mode can be used to insert comments in different steps of the
6604 script.
6605
6606 For each tcp-check rule you create, you can add a "comment" directive,
6607 followed by a string. This string will be reported in the log and stderr
6608 in debug mode. It is useful to make user-friendly error reporting.
6609 The "comment" is of course optional.
6610
6611
Willy Tarreaued179852013-12-16 01:07:00 +01006612 Examples :
6613 # perform a POP check (analyse only server's banner)
6614 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006615 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006616
6617 # perform an IMAP check (analyse only server's banner)
6618 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006619 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006620
6621 # look for the redis master server after ensuring it speaks well
6622 # redis protocol, then it exits properly.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006623 # (send a command then analyse the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01006624 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006625 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006626 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02006627 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006628 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01006629 tcp-check send info\ replication\r\n
6630 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006631 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006632 tcp-check send QUIT\r\n
6633 tcp-check expect string +OK
6634
6635 forge a HTTP request, then analyse the response
6636 (send many headers before analyzing)
6637 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006638 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01006639 tcp-check send HEAD\ /\ HTTP/1.1\r\n
6640 tcp-check send Host:\ www.mydomain.com\r\n
6641 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
6642 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006643 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01006644
6645
6646 See also : "tcp-check expect", "tcp-check send"
6647
6648
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006649option tcp-smart-accept
6650no option tcp-smart-accept
6651 Enable or disable the saving of one ACK packet during the accept sequence
6652 May be used in sections : defaults | frontend | listen | backend
6653 yes | yes | yes | no
6654 Arguments : none
6655
6656 When an HTTP connection request comes in, the system acknowledges it on
6657 behalf of HAProxy, then the client immediately sends its request, and the
6658 system acknowledges it too while it is notifying HAProxy about the new
6659 connection. HAProxy then reads the request and responds. This means that we
6660 have one TCP ACK sent by the system for nothing, because the request could
6661 very well be acknowledged by HAProxy when it sends its response.
6662
6663 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
6664 sending this useless ACK on platforms which support it (currently at least
6665 Linux). It must not cause any problem, because the system will send it anyway
6666 after 40 ms if the response takes more time than expected to come.
6667
6668 During complex network debugging sessions, it may be desirable to disable
6669 this optimization because delayed ACKs can make troubleshooting more complex
6670 when trying to identify where packets are delayed. It is then possible to
6671 fall back to normal behaviour by specifying "no option tcp-smart-accept".
6672
6673 It is also possible to force it for non-HTTP proxies by simply specifying
6674 "option tcp-smart-accept". For instance, it can make sense with some services
6675 such as SMTP where the server speaks first.
6676
6677 It is recommended to avoid forcing this option in a defaults section. In case
6678 of doubt, consider setting it back to automatic values by prepending the
6679 "default" keyword before it, or disabling it using the "no" keyword.
6680
Willy Tarreaud88edf22009-06-14 15:48:17 +02006681 See also : "option tcp-smart-connect"
6682
6683
6684option tcp-smart-connect
6685no option tcp-smart-connect
6686 Enable or disable the saving of one ACK packet during the connect sequence
6687 May be used in sections : defaults | frontend | listen | backend
6688 yes | no | yes | yes
6689 Arguments : none
6690
6691 On certain systems (at least Linux), HAProxy can ask the kernel not to
6692 immediately send an empty ACK upon a connection request, but to directly
6693 send the buffer request instead. This saves one packet on the network and
6694 thus boosts performance. It can also be useful for some servers, because they
6695 immediately get the request along with the incoming connection.
6696
6697 This feature is enabled when "option tcp-smart-connect" is set in a backend.
6698 It is not enabled by default because it makes network troubleshooting more
6699 complex.
6700
6701 It only makes sense to enable it with protocols where the client speaks first
6702 such as HTTP. In other situations, if there is no data to send in place of
6703 the ACK, a normal ACK is sent.
6704
6705 If this option has been enabled in a "defaults" section, it can be disabled
6706 in a specific instance by prepending the "no" keyword before it.
6707
6708 See also : "option tcp-smart-accept"
6709
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006710
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006711option tcpka
6712 Enable or disable the sending of TCP keepalive packets on both sides
6713 May be used in sections : defaults | frontend | listen | backend
6714 yes | yes | yes | yes
6715 Arguments : none
6716
6717 When there is a firewall or any session-aware component between a client and
6718 a server, and when the protocol involves very long sessions with long idle
6719 periods (eg: remote desktops), there is a risk that one of the intermediate
6720 components decides to expire a session which has remained idle for too long.
6721
6722 Enabling socket-level TCP keep-alives makes the system regularly send packets
6723 to the other end of the connection, leaving it active. The delay between
6724 keep-alive probes is controlled by the system only and depends both on the
6725 operating system and its tuning parameters.
6726
6727 It is important to understand that keep-alive packets are neither emitted nor
6728 received at the application level. It is only the network stacks which sees
6729 them. For this reason, even if one side of the proxy already uses keep-alives
6730 to maintain its connection alive, those keep-alive packets will not be
6731 forwarded to the other side of the proxy.
6732
6733 Please note that this has nothing to do with HTTP keep-alive.
6734
6735 Using option "tcpka" enables the emission of TCP keep-alive probes on both
6736 the client and server sides of a connection. Note that this is meaningful
6737 only in "defaults" or "listen" sections. If this option is used in a
6738 frontend, only the client side will get keep-alives, and if this option is
6739 used in a backend, only the server side will get keep-alives. For this
6740 reason, it is strongly recommended to explicitly use "option clitcpka" and
6741 "option srvtcpka" when the configuration is split between frontends and
6742 backends.
6743
6744 See also : "option clitcpka", "option srvtcpka"
6745
Willy Tarreau844e3c52008-01-11 16:28:18 +01006746
6747option tcplog
6748 Enable advanced logging of TCP connections with session state and timers
6749 May be used in sections : defaults | frontend | listen | backend
6750 yes | yes | yes | yes
6751 Arguments : none
6752
6753 By default, the log output format is very poor, as it only contains the
6754 source and destination addresses, and the instance name. By specifying
6755 "option tcplog", each log line turns into a much richer format including, but
6756 not limited to, the connection timers, the session status, the connections
6757 numbers, the frontend, backend and server name, and of course the source
6758 address and ports. This option is useful for pure TCP proxies in order to
6759 find which of the client or server disconnects or times out. For normal HTTP
6760 proxies, it's better to use "option httplog" which is even more complete.
6761
6762 This option may be set either in the frontend or the backend.
6763
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006764 "option tcplog" overrides any previous "log-format" directive.
6765
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006766 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01006767
6768
Willy Tarreau844e3c52008-01-11 16:28:18 +01006769option transparent
6770no option transparent
6771 Enable client-side transparent proxying
6772 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01006773 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01006774 Arguments : none
6775
6776 This option was introduced in order to provide layer 7 persistence to layer 3
6777 load balancers. The idea is to use the OS's ability to redirect an incoming
6778 connection for a remote address to a local process (here HAProxy), and let
6779 this process know what address was initially requested. When this option is
6780 used, sessions without cookies will be forwarded to the original destination
6781 IP address of the incoming request (which should match that of another
6782 equipment), while requests with cookies will still be forwarded to the
6783 appropriate server.
6784
6785 Note that contrary to a common belief, this option does NOT make HAProxy
6786 present the client's IP to the server when establishing the connection.
6787
Willy Tarreaua1146052011-03-01 09:51:54 +01006788 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006789 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01006790
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006791
Simon Horman98637e52014-06-20 12:30:16 +09006792external-check command <command>
6793 Executable to run when performing an external-check
6794 May be used in sections : defaults | frontend | listen | backend
6795 yes | no | yes | yes
6796
6797 Arguments :
6798 <command> is the external command to run
6799
Simon Horman98637e52014-06-20 12:30:16 +09006800 The arguments passed to the to the command are:
6801
Cyril Bonté777be862014-12-02 21:21:35 +01006802 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09006803
Cyril Bonté777be862014-12-02 21:21:35 +01006804 The <proxy_address> and <proxy_port> are derived from the first listener
6805 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
6806 listener the proxy_address will be the path of the socket and the
6807 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
6808 possible to determine a listener, and both <proxy_address> and <proxy_port>
6809 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09006810
Cyril Bonté72cda2a2014-12-27 22:28:39 +01006811 Some values are also provided through environment variables.
6812
6813 Environment variables :
6814 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
6815 applicable, for example in a "backend" section).
6816
6817 HAPROXY_PROXY_ID The backend id.
6818
6819 HAPROXY_PROXY_NAME The backend name.
6820
6821 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
6822 applicable, for example in a "backend" section or
6823 for a UNIX socket).
6824
6825 HAPROXY_SERVER_ADDR The server address.
6826
6827 HAPROXY_SERVER_CURCONN The current number of connections on the server.
6828
6829 HAPROXY_SERVER_ID The server id.
6830
6831 HAPROXY_SERVER_MAXCONN The server max connections.
6832
6833 HAPROXY_SERVER_NAME The server name.
6834
6835 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
6836 socket).
6837
6838 PATH The PATH environment variable used when executing
6839 the command may be set using "external-check path".
6840
Simon Horman98637e52014-06-20 12:30:16 +09006841 If the command executed and exits with a zero status then the check is
6842 considered to have passed, otherwise the check is considered to have
6843 failed.
6844
6845 Example :
6846 external-check command /bin/true
6847
6848 See also : "external-check", "option external-check", "external-check path"
6849
6850
6851external-check path <path>
6852 The value of the PATH environment variable used when running an external-check
6853 May be used in sections : defaults | frontend | listen | backend
6854 yes | no | yes | yes
6855
6856 Arguments :
6857 <path> is the path used when executing external command to run
6858
6859 The default path is "".
6860
6861 Example :
6862 external-check path "/usr/bin:/bin"
6863
6864 See also : "external-check", "option external-check",
6865 "external-check command"
6866
6867
Emeric Brun647caf12009-06-30 17:57:00 +02006868persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02006869persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02006870 Enable RDP cookie-based persistence
6871 May be used in sections : defaults | frontend | listen | backend
6872 yes | no | yes | yes
6873 Arguments :
6874 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02006875 default cookie name "msts" will be used. There currently is no
6876 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02006877
6878 This statement enables persistence based on an RDP cookie. The RDP cookie
6879 contains all information required to find the server in the list of known
6880 servers. So when this option is set in the backend, the request is analysed
6881 and if an RDP cookie is found, it is decoded. If it matches a known server
6882 which is still UP (or if "option persist" is set), then the connection is
6883 forwarded to this server.
6884
6885 Note that this only makes sense in a TCP backend, but for this to work, the
6886 frontend must have waited long enough to ensure that an RDP cookie is present
6887 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006888 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02006889 a single "listen" section.
6890
Willy Tarreau61e28f22010-05-16 22:31:05 +02006891 Also, it is important to understand that the terminal server will emit this
6892 RDP cookie only if it is configured for "token redirection mode", which means
6893 that the "IP address redirection" option is disabled.
6894
Emeric Brun647caf12009-06-30 17:57:00 +02006895 Example :
6896 listen tse-farm
6897 bind :3389
6898 # wait up to 5s for an RDP cookie in the request
6899 tcp-request inspect-delay 5s
6900 tcp-request content accept if RDP_COOKIE
6901 # apply RDP cookie persistence
6902 persist rdp-cookie
6903 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02006904 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02006905 balance rdp-cookie
6906 server srv1 1.1.1.1:3389
6907 server srv2 1.1.1.2:3389
6908
Simon Hormanab814e02011-06-24 14:50:20 +09006909 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
6910 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02006911
6912
Willy Tarreau3a7d2072009-03-05 23:48:25 +01006913rate-limit sessions <rate>
6914 Set a limit on the number of new sessions accepted per second on a frontend
6915 May be used in sections : defaults | frontend | listen | backend
6916 yes | yes | yes | no
6917 Arguments :
6918 <rate> The <rate> parameter is an integer designating the maximum number
6919 of new sessions per second to accept on the frontend.
6920
6921 When the frontend reaches the specified number of new sessions per second, it
6922 stops accepting new connections until the rate drops below the limit again.
6923 During this time, the pending sessions will be kept in the socket's backlog
6924 (in system buffers) and haproxy will not even be aware that sessions are
6925 pending. When applying very low limit on a highly loaded service, it may make
6926 sense to increase the socket's backlog using the "backlog" keyword.
6927
6928 This feature is particularly efficient at blocking connection-based attacks
6929 or service abuse on fragile servers. Since the session rate is measured every
6930 millisecond, it is extremely accurate. Also, the limit applies immediately,
6931 no delay is needed at all to detect the threshold.
6932
6933 Example : limit the connection rate on SMTP to 10 per second max
6934 listen smtp
6935 mode tcp
6936 bind :25
6937 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02006938 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01006939
Willy Tarreaua17c2d92011-07-25 08:16:20 +02006940 Note : when the maximum rate is reached, the frontend's status is not changed
6941 but its sockets appear as "WAITING" in the statistics if the
6942 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01006943
6944 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
6945
6946
Willy Tarreau2e1dca82012-09-12 08:43:15 +02006947redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
6948redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
6949redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02006950 Return an HTTP redirection if/unless a condition is matched
6951 May be used in sections : defaults | frontend | listen | backend
6952 no | yes | yes | yes
6953
6954 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01006955 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02006956
Willy Tarreau0140f252008-11-19 21:07:09 +01006957 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02006958 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01006959 the HTTP "Location" header. When used in an "http-request" rule,
6960 <loc> value follows the log-format rules and can include some
6961 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02006962
6963 <pfx> With "redirect prefix", the "Location" header is built from the
6964 concatenation of <pfx> and the complete URI path, including the
6965 query string, unless the "drop-query" option is specified (see
6966 below). As a special case, if <pfx> equals exactly "/", then
6967 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01006968 redirect to the same URL (for instance, to insert a cookie). When
6969 used in an "http-request" rule, <pfx> value follows the log-format
6970 rules and can include some dynamic values (see Custom Log Format
6971 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02006972
6973 <sch> With "redirect scheme", then the "Location" header is built by
6974 concatenating <sch> with "://" then the first occurrence of the
6975 "Host" header, and then the URI path, including the query string
6976 unless the "drop-query" option is specified (see below). If no
6977 path is found or if the path is "*", then "/" is used instead. If
6978 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006979 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02006980 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01006981 HTTPS. When used in an "http-request" rule, <sch> value follows
6982 the log-format rules and can include some dynamic values (see
6983 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01006984
6985 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01006986 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
6987 with 302 used by default if no code is specified. 301 means
6988 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02006989 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01006990 cache the redirection. 303 is equivalent to 302 except that the
6991 browser will fetch the location with a GET method. 307 is just
6992 like 302 but makes it clear that the same method must be reused.
6993 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01006994
6995 <option> There are several options which can be specified to adjust the
6996 expected behaviour of a redirection :
6997
6998 - "drop-query"
6999 When this keyword is used in a prefix-based redirection, then the
7000 location will be set without any possible query-string, which is useful
7001 for directing users to a non-secure page for instance. It has no effect
7002 with a location-type redirect.
7003
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007004 - "append-slash"
7005 This keyword may be used in conjunction with "drop-query" to redirect
7006 users who use a URL not ending with a '/' to the same one with the '/'.
7007 It can be useful to ensure that search engines will only see one URL.
7008 For this, a return code 301 is preferred.
7009
Willy Tarreau0140f252008-11-19 21:07:09 +01007010 - "set-cookie NAME[=value]"
7011 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7012 to the response. This is sometimes used to indicate that a user has
7013 been seen, for instance to protect against some types of DoS. No other
7014 cookie option is added, so the cookie will be a session cookie. Note
7015 that for a browser, a sole cookie name without an equal sign is
7016 different from a cookie with an equal sign.
7017
7018 - "clear-cookie NAME[=]"
7019 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7020 with the "Max-Age" attribute set to zero. This will tell the browser to
7021 delete this cookie. It is useful for instance on logout pages. It is
7022 important to note that clearing the cookie "NAME" will not remove a
7023 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7024 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007025
7026 Example: move the login URL only to HTTPS.
7027 acl clear dst_port 80
7028 acl secure dst_port 8080
7029 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007030 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007031 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007032 acl cookie_set hdr_sub(cookie) SEEN=1
7033
7034 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007035 redirect prefix https://mysite.com if login_page !secure
7036 redirect prefix http://mysite.com drop-query if login_page !uid_given
7037 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007038 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007039
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007040 Example: send redirects for request for articles without a '/'.
7041 acl missing_slash path_reg ^/article/[^/]*$
7042 redirect code 301 prefix / drop-query append-slash if missing_slash
7043
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007044 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007045 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007046
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007047 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007048 http-request redirect code 301 location \
7049 http://www.%[hdr(host)]%[capture.req.uri] \
7050 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007051
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007052 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007053
7054
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007055redisp (deprecated)
7056redispatch (deprecated)
7057 Enable or disable session redistribution in case of connection failure
7058 May be used in sections: defaults | frontend | listen | backend
7059 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007060 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007061
7062 In HTTP mode, if a server designated by a cookie is down, clients may
7063 definitely stick to it because they cannot flush the cookie, so they will not
7064 be able to access the service anymore.
7065
7066 Specifying "redispatch" will allow the proxy to break their persistence and
7067 redistribute them to a working server.
7068
7069 It also allows to retry last connection to another server in case of multiple
7070 connection failures. Of course, it requires having "retries" set to a nonzero
7071 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007072
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007073 This form is deprecated, do not use it in any new configuration, use the new
7074 "option redispatch" instead.
7075
7076 See also : "option redispatch"
7077
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007078
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007079reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007080 Add a header at the end of the HTTP request
7081 May be used in sections : defaults | frontend | listen | backend
7082 no | yes | yes | yes
7083 Arguments :
7084 <string> is the complete line to be added. Any space or known delimiter
7085 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007086 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007087
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007088 <cond> is an optional matching condition built from ACLs. It makes it
7089 possible to ignore this rule when other conditions are not met.
7090
Willy Tarreau303c0352008-01-17 19:01:39 +01007091 A new line consisting in <string> followed by a line feed will be added after
7092 the last header of an HTTP request.
7093
7094 Header transformations only apply to traffic which passes through HAProxy,
7095 and not to traffic generated by HAProxy, such as health-checks or error
7096 responses.
7097
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007098 Example : add "X-Proto: SSL" to requests coming via port 81
7099 acl is-ssl dst_port 81
7100 reqadd X-Proto:\ SSL if is-ssl
7101
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007102 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7103 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007104
7105
Willy Tarreau5321c422010-01-28 20:35:13 +01007106reqallow <search> [{if | unless} <cond>]
7107reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007108 Definitely allow an HTTP request if a line matches a regular expression
7109 May be used in sections : defaults | frontend | listen | backend
7110 no | yes | yes | yes
7111 Arguments :
7112 <search> is the regular expression applied to HTTP headers and to the
7113 request line. This is an extended regular expression. Parenthesis
7114 grouping is supported and no preliminary backslash is required.
7115 Any space or known delimiter must be escaped using a backslash
7116 ('\'). The pattern applies to a full line at a time. The
7117 "reqallow" keyword strictly matches case while "reqiallow"
7118 ignores case.
7119
Willy Tarreau5321c422010-01-28 20:35:13 +01007120 <cond> is an optional matching condition built from ACLs. It makes it
7121 possible to ignore this rule when other conditions are not met.
7122
Willy Tarreau303c0352008-01-17 19:01:39 +01007123 A request containing any line which matches extended regular expression
7124 <search> will mark the request as allowed, even if any later test would
7125 result in a deny. The test applies both to the request line and to request
7126 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007127 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007128
7129 It is easier, faster and more powerful to use ACLs to write access policies.
7130 Reqdeny, reqallow and reqpass should be avoided in new designs.
7131
7132 Example :
7133 # allow www.* but refuse *.local
7134 reqiallow ^Host:\ www\.
7135 reqideny ^Host:\ .*\.local
7136
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007137 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7138 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007139
7140
Willy Tarreau5321c422010-01-28 20:35:13 +01007141reqdel <search> [{if | unless} <cond>]
7142reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007143 Delete all headers matching a regular expression in an HTTP request
7144 May be used in sections : defaults | frontend | listen | backend
7145 no | yes | yes | yes
7146 Arguments :
7147 <search> is the regular expression applied to HTTP headers and to the
7148 request line. This is an extended regular expression. Parenthesis
7149 grouping is supported and no preliminary backslash is required.
7150 Any space or known delimiter must be escaped using a backslash
7151 ('\'). The pattern applies to a full line at a time. The "reqdel"
7152 keyword strictly matches case while "reqidel" ignores case.
7153
Willy Tarreau5321c422010-01-28 20:35:13 +01007154 <cond> is an optional matching condition built from ACLs. It makes it
7155 possible to ignore this rule when other conditions are not met.
7156
Willy Tarreau303c0352008-01-17 19:01:39 +01007157 Any header line matching extended regular expression <search> in the request
7158 will be completely deleted. Most common use of this is to remove unwanted
7159 and/or dangerous headers or cookies from a request before passing it to the
7160 next servers.
7161
7162 Header transformations only apply to traffic which passes through HAProxy,
7163 and not to traffic generated by HAProxy, such as health-checks or error
7164 responses. Keep in mind that header names are not case-sensitive.
7165
7166 Example :
7167 # remove X-Forwarded-For header and SERVER cookie
7168 reqidel ^X-Forwarded-For:.*
7169 reqidel ^Cookie:.*SERVER=
7170
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007171 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7172 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007173
7174
Willy Tarreau5321c422010-01-28 20:35:13 +01007175reqdeny <search> [{if | unless} <cond>]
7176reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007177 Deny an HTTP request if a line matches a regular expression
7178 May be used in sections : defaults | frontend | listen | backend
7179 no | yes | yes | yes
7180 Arguments :
7181 <search> is the regular expression applied to HTTP headers and to the
7182 request line. This is an extended regular expression. Parenthesis
7183 grouping is supported and no preliminary backslash is required.
7184 Any space or known delimiter must be escaped using a backslash
7185 ('\'). The pattern applies to a full line at a time. The
7186 "reqdeny" keyword strictly matches case while "reqideny" ignores
7187 case.
7188
Willy Tarreau5321c422010-01-28 20:35:13 +01007189 <cond> is an optional matching condition built from ACLs. It makes it
7190 possible to ignore this rule when other conditions are not met.
7191
Willy Tarreau303c0352008-01-17 19:01:39 +01007192 A request containing any line which matches extended regular expression
7193 <search> will mark the request as denied, even if any later test would
7194 result in an allow. The test applies both to the request line and to request
7195 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007196 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007197
Willy Tarreauced27012008-01-17 20:35:34 +01007198 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007199 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007200 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007201
Willy Tarreau303c0352008-01-17 19:01:39 +01007202 It is easier, faster and more powerful to use ACLs to write access policies.
7203 Reqdeny, reqallow and reqpass should be avoided in new designs.
7204
7205 Example :
7206 # refuse *.local, then allow www.*
7207 reqideny ^Host:\ .*\.local
7208 reqiallow ^Host:\ www\.
7209
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007210 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7211 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007212
7213
Willy Tarreau5321c422010-01-28 20:35:13 +01007214reqpass <search> [{if | unless} <cond>]
7215reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007216 Ignore any HTTP request line matching a regular expression in next rules
7217 May be used in sections : defaults | frontend | listen | backend
7218 no | yes | yes | yes
7219 Arguments :
7220 <search> is the regular expression applied to HTTP headers and to the
7221 request line. This is an extended regular expression. Parenthesis
7222 grouping is supported and no preliminary backslash is required.
7223 Any space or known delimiter must be escaped using a backslash
7224 ('\'). The pattern applies to a full line at a time. The
7225 "reqpass" keyword strictly matches case while "reqipass" ignores
7226 case.
7227
Willy Tarreau5321c422010-01-28 20:35:13 +01007228 <cond> is an optional matching condition built from ACLs. It makes it
7229 possible to ignore this rule when other conditions are not met.
7230
Willy Tarreau303c0352008-01-17 19:01:39 +01007231 A request containing any line which matches extended regular expression
7232 <search> will skip next rules, without assigning any deny or allow verdict.
7233 The test applies both to the request line and to request headers. Keep in
7234 mind that URLs in request line are case-sensitive while header names are not.
7235
7236 It is easier, faster and more powerful to use ACLs to write access policies.
7237 Reqdeny, reqallow and reqpass should be avoided in new designs.
7238
7239 Example :
7240 # refuse *.local, then allow www.*, but ignore "www.private.local"
7241 reqipass ^Host:\ www.private\.local
7242 reqideny ^Host:\ .*\.local
7243 reqiallow ^Host:\ www\.
7244
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007245 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7246 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007247
7248
Willy Tarreau5321c422010-01-28 20:35:13 +01007249reqrep <search> <string> [{if | unless} <cond>]
7250reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007251 Replace a regular expression with a string in an HTTP request line
7252 May be used in sections : defaults | frontend | listen | backend
7253 no | yes | yes | yes
7254 Arguments :
7255 <search> is the regular expression applied to HTTP headers and to the
7256 request line. This is an extended regular expression. Parenthesis
7257 grouping is supported and no preliminary backslash is required.
7258 Any space or known delimiter must be escaped using a backslash
7259 ('\'). The pattern applies to a full line at a time. The "reqrep"
7260 keyword strictly matches case while "reqirep" ignores case.
7261
7262 <string> is the complete line to be added. Any space or known delimiter
7263 must be escaped using a backslash ('\'). References to matched
7264 pattern groups are possible using the common \N form, with N
7265 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007266 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007267
Willy Tarreau5321c422010-01-28 20:35:13 +01007268 <cond> is an optional matching condition built from ACLs. It makes it
7269 possible to ignore this rule when other conditions are not met.
7270
Willy Tarreau303c0352008-01-17 19:01:39 +01007271 Any line matching extended regular expression <search> in the request (both
7272 the request line and header lines) will be completely replaced with <string>.
7273 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7274
7275 Header transformations only apply to traffic which passes through HAProxy,
7276 and not to traffic generated by HAProxy, such as health-checks or error
7277 responses. Note that for increased readability, it is suggested to add enough
7278 spaces between the request and the response. Keep in mind that URLs in
7279 request line are case-sensitive while header names are not.
7280
7281 Example :
7282 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007283 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007284 # replace "www.mydomain.com" with "www" in the host name.
7285 reqirep ^Host:\ www.mydomain.com Host:\ www
7286
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007287 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7288 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007289
7290
Willy Tarreau5321c422010-01-28 20:35:13 +01007291reqtarpit <search> [{if | unless} <cond>]
7292reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007293 Tarpit an HTTP request containing a line matching a regular expression
7294 May be used in sections : defaults | frontend | listen | backend
7295 no | yes | yes | yes
7296 Arguments :
7297 <search> is the regular expression applied to HTTP headers and to the
7298 request line. This is an extended regular expression. Parenthesis
7299 grouping is supported and no preliminary backslash is required.
7300 Any space or known delimiter must be escaped using a backslash
7301 ('\'). The pattern applies to a full line at a time. The
7302 "reqtarpit" keyword strictly matches case while "reqitarpit"
7303 ignores case.
7304
Willy Tarreau5321c422010-01-28 20:35:13 +01007305 <cond> is an optional matching condition built from ACLs. It makes it
7306 possible to ignore this rule when other conditions are not met.
7307
Willy Tarreau303c0352008-01-17 19:01:39 +01007308 A request containing any line which matches extended regular expression
7309 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007310 be kept open for a pre-defined time, then will return an HTTP error 500 so
7311 that the attacker does not suspect it has been tarpitted. The status 500 will
7312 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007313 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7314 not set.
7315
7316 The goal of the tarpit is to slow down robots attacking servers with
7317 identifiable requests. Many robots limit their outgoing number of connections
7318 and stay connected waiting for a reply which can take several minutes to
7319 come. Depending on the environment and attack, it may be particularly
7320 efficient at reducing the load on the network and firewalls.
7321
Willy Tarreau5321c422010-01-28 20:35:13 +01007322 Examples :
Willy Tarreau303c0352008-01-17 19:01:39 +01007323 # ignore user-agents reporting any flavour of "Mozilla" or "MSIE", but
7324 # block all others.
7325 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7326 reqitarpit ^User-Agent:
7327
Willy Tarreau5321c422010-01-28 20:35:13 +01007328 # block bad guys
7329 acl badguys src 10.1.0.3 172.16.13.20/28
7330 reqitarpit . if badguys
7331
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007332 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7333 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007334
7335
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007336retries <value>
7337 Set the number of retries to perform on a server after a connection failure
7338 May be used in sections: defaults | frontend | listen | backend
7339 yes | no | yes | yes
7340 Arguments :
7341 <value> is the number of times a connection attempt should be retried on
7342 a server when a connection either is refused or times out. The
7343 default value is 3.
7344
7345 It is important to understand that this value applies to the number of
7346 connection attempts, not full requests. When a connection has effectively
7347 been established to a server, there will be no more retry.
7348
7349 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007350 a turn-around timer of min("timeout connect", one second) is applied before
7351 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007352
7353 When "option redispatch" is set, the last retry may be performed on another
7354 server even if a cookie references a different server.
7355
7356 See also : "option redispatch"
7357
7358
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007359rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007360 Add a header at the end of the HTTP response
7361 May be used in sections : defaults | frontend | listen | backend
7362 no | yes | yes | yes
7363 Arguments :
7364 <string> is the complete line to be added. Any space or known delimiter
7365 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007366 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007367
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007368 <cond> is an optional matching condition built from ACLs. It makes it
7369 possible to ignore this rule when other conditions are not met.
7370
Willy Tarreau303c0352008-01-17 19:01:39 +01007371 A new line consisting in <string> followed by a line feed will be added after
7372 the last header of an HTTP response.
7373
7374 Header transformations only apply to traffic which passes through HAProxy,
7375 and not to traffic generated by HAProxy, such as health-checks or error
7376 responses.
7377
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007378 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7379 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007380
7381
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007382rspdel <search> [{if | unless} <cond>]
7383rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007384 Delete all headers matching a regular expression in an HTTP response
7385 May be used in sections : defaults | frontend | listen | backend
7386 no | yes | yes | yes
7387 Arguments :
7388 <search> is the regular expression applied to HTTP headers and to the
7389 response line. This is an extended regular expression, so
7390 parenthesis grouping is supported and no preliminary backslash
7391 is required. Any space or known delimiter must be escaped using
7392 a backslash ('\'). The pattern applies to a full line at a time.
7393 The "rspdel" keyword strictly matches case while "rspidel"
7394 ignores case.
7395
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007396 <cond> is an optional matching condition built from ACLs. It makes it
7397 possible to ignore this rule when other conditions are not met.
7398
Willy Tarreau303c0352008-01-17 19:01:39 +01007399 Any header line matching extended regular expression <search> in the response
7400 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007401 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007402 client.
7403
7404 Header transformations only apply to traffic which passes through HAProxy,
7405 and not to traffic generated by HAProxy, such as health-checks or error
7406 responses. Keep in mind that header names are not case-sensitive.
7407
7408 Example :
7409 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007410 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007411
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007412 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7413 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007414
7415
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007416rspdeny <search> [{if | unless} <cond>]
7417rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007418 Block an HTTP response if a line matches a regular expression
7419 May be used in sections : defaults | frontend | listen | backend
7420 no | yes | yes | yes
7421 Arguments :
7422 <search> is the regular expression applied to HTTP headers and to the
7423 response line. This is an extended regular expression, so
7424 parenthesis grouping is supported and no preliminary backslash
7425 is required. Any space or known delimiter must be escaped using
7426 a backslash ('\'). The pattern applies to a full line at a time.
7427 The "rspdeny" keyword strictly matches case while "rspideny"
7428 ignores case.
7429
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007430 <cond> is an optional matching condition built from ACLs. It makes it
7431 possible to ignore this rule when other conditions are not met.
7432
Willy Tarreau303c0352008-01-17 19:01:39 +01007433 A response containing any line which matches extended regular expression
7434 <search> will mark the request as denied. The test applies both to the
7435 response line and to response headers. Keep in mind that header names are not
7436 case-sensitive.
7437
7438 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007439 block the response before it reaches the client. If a response is denied, it
7440 will be replaced with an HTTP 502 error so that the client never retrieves
7441 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007442
7443 It is easier, faster and more powerful to use ACLs to write access policies.
7444 Rspdeny should be avoided in new designs.
7445
7446 Example :
7447 # Ensure that no content type matching ms-word will leak
7448 rspideny ^Content-type:\.*/ms-word
7449
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007450 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7451 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007452
7453
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007454rsprep <search> <string> [{if | unless} <cond>]
7455rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007456 Replace a regular expression with a string in an HTTP response line
7457 May be used in sections : defaults | frontend | listen | backend
7458 no | yes | yes | yes
7459 Arguments :
7460 <search> is the regular expression applied to HTTP headers and to the
7461 response line. This is an extended regular expression, so
7462 parenthesis grouping is supported and no preliminary backslash
7463 is required. Any space or known delimiter must be escaped using
7464 a backslash ('\'). The pattern applies to a full line at a time.
7465 The "rsprep" keyword strictly matches case while "rspirep"
7466 ignores case.
7467
7468 <string> is the complete line to be added. Any space or known delimiter
7469 must be escaped using a backslash ('\'). References to matched
7470 pattern groups are possible using the common \N form, with N
7471 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007472 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007473
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007474 <cond> is an optional matching condition built from ACLs. It makes it
7475 possible to ignore this rule when other conditions are not met.
7476
Willy Tarreau303c0352008-01-17 19:01:39 +01007477 Any line matching extended regular expression <search> in the response (both
7478 the response line and header lines) will be completely replaced with
7479 <string>. Most common use of this is to rewrite Location headers.
7480
7481 Header transformations only apply to traffic which passes through HAProxy,
7482 and not to traffic generated by HAProxy, such as health-checks or error
7483 responses. Note that for increased readability, it is suggested to add enough
7484 spaces between the request and the response. Keep in mind that header names
7485 are not case-sensitive.
7486
7487 Example :
7488 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7489 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7490
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007491 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7492 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007493
7494
David du Colombier486df472011-03-17 10:40:26 +01007495server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007496 Declare a server in a backend
7497 May be used in sections : defaults | frontend | listen | backend
7498 no | no | yes | yes
7499 Arguments :
7500 <name> is the internal name assigned to this server. This name will
Cyril Bonté941a0c62012-10-15 19:44:24 +02007501 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007502 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007503
David du Colombier486df472011-03-17 10:40:26 +01007504 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7505 resolvable hostname is supported, but this name will be resolved
7506 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7507 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007508 address as the one from the client connection. This is useful in
7509 transparent proxy architectures where the client's connection is
7510 intercepted and haproxy must forward to the original destination
7511 address. This is more or less what the "transparent" keyword does
7512 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007513 to report statistics. Optionally, an address family prefix may be
7514 used before the address to force the family regardless of the
7515 address format, which can be useful to specify a path to a unix
7516 socket with no slash ('/'). Currently supported prefixes are :
7517 - 'ipv4@' -> address is always IPv4
7518 - 'ipv6@' -> address is always IPv6
7519 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007520 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemandb2f07452015-05-12 14:27:13 +02007521 You may want to reference some environment variables in the
7522 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007523 variables. The "init-addr" setting can be used to modify the way
7524 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007525
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007526 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007527 be sent to this port. If unset, the same port the client
7528 connected to will be used. The port may also be prefixed by a "+"
7529 or a "-". In this case, the server's port will be determined by
7530 adding this value to the client's port.
7531
7532 <param*> is a list of parameters for this server. The "server" keywords
7533 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007534 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007535
7536 Examples :
7537 server first 10.1.1.1:1080 cookie first check inter 1000
7538 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007539 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007540 server backup "${SRV_BACKUP}:1080" backup
7541 server www1_dc1 "${LAN_DC1}.101:80"
7542 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007543
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007544 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7545 sun_path length is used for the address length. Some other programs
7546 such as socat use the string length only by default. Pass the option
7547 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7548 make it compatible with HAProxy's.
7549
Mark Lamourinec2247f02012-01-04 13:02:01 -05007550 See also: "default-server", "http-send-name-header" and section 5 about
7551 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007552
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007553server-state-file-name [<file>]
7554 Set the server state file to read, load and apply to servers available in
7555 this backend. It only applies when the directive "load-server-state-from-file"
7556 is set to "local". When <file> is not provided or if this directive is not
7557 set, then backend name is used. If <file> starts with a slash '/', then it is
7558 considered as an absolute path. Otherwise, <file> is concatenated to the
7559 global directive "server-state-file-base".
7560
7561 Example: the minimal configuration below would make HAProxy look for the
7562 state server file '/etc/haproxy/states/bk':
7563
7564 global
7565 server-state-file-base /etc/haproxy/states
7566
7567 backend bk
7568 load-server-state-from-file
7569
7570 See also: "server-state-file-base", "load-server-state-from-file", and
7571 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007572
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007573server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7574 Set a template to initialize servers with shared parameters.
7575 The names of these servers are built from <prefix> and <num | range> parameters.
7576 May be used in sections : defaults | frontend | listen | backend
7577 no | no | yes | yes
7578
7579 Arguments:
7580 <prefix> A prefix for the server names to be built.
7581
7582 <num | range>
7583 If <num> is provided, this template initializes <num> servers
7584 with 1 up to <num> as server name suffixes. A range of numbers
7585 <num_low>-<num_high> may also be used to use <num_low> up to
7586 <num_high> as server name suffixes.
7587
7588 <fqdn> A FQDN for all the servers this template initializes.
7589
7590 <port> Same meaning as "server" <port> argument (see "server" keyword).
7591
7592 <params*>
7593 Remaining server parameters among all those supported by "server"
7594 keyword.
7595
7596 Examples:
7597 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7598 # google.com as FQDN, and health-check enabled.
7599 server-template srv 1-3 google.com:80 check
7600
7601 # or
7602 server-template srv 3 google.com:80 check
7603
7604 # would be equivalent to:
7605 server srv1 google.com:80 check
7606 server srv2 google.com:80 check
7607 server srv3 google.com:80 check
7608
7609
7610
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007611source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007612source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007613source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007614 Set the source address for outgoing connections
7615 May be used in sections : defaults | frontend | listen | backend
7616 yes | no | yes | yes
7617 Arguments :
7618 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7619 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007620
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007621 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007622 the most appropriate address to reach its destination. Optionally
7623 an address family prefix may be used before the address to force
7624 the family regardless of the address format, which can be useful
7625 to specify a path to a unix socket with no slash ('/'). Currently
7626 supported prefixes are :
7627 - 'ipv4@' -> address is always IPv4
7628 - 'ipv6@' -> address is always IPv6
7629 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007630 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007631 You may want to reference some environment variables in the
7632 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007633
7634 <port> is an optional port. It is normally not needed but may be useful
7635 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007636 the system will select a free port. Note that port ranges are not
7637 supported in the backend. If you want to force port ranges, you
7638 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007639
7640 <addr2> is the IP address to present to the server when connections are
7641 forwarded in full transparent proxy mode. This is currently only
7642 supported on some patched Linux kernels. When this address is
7643 specified, clients connecting to the server will be presented
7644 with this address, while health checks will still use the address
7645 <addr>.
7646
7647 <port2> is the optional port to present to the server when connections
7648 are forwarded in full transparent proxy mode (see <addr2> above).
7649 The default value of zero means the system will select a free
7650 port.
7651
Willy Tarreaubce70882009-09-07 11:51:47 +02007652 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7653 This is the name of a comma-separated header list which can
7654 contain multiple IP addresses. By default, the last occurrence is
7655 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01007656 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02007657 by previous proxy, typically Stunnel. In order to use another
7658 occurrence from the last one, please see the <occ> parameter
7659 below. When the header (or occurrence) is not found, no binding
7660 is performed so that the proxy's default IP address is used. Also
7661 keep in mind that the header name is case insensitive, as for any
7662 HTTP header.
7663
7664 <occ> is the occurrence number of a value to be used in a multi-value
7665 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007666 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02007667 address. Positive values indicate a position from the first
7668 occurrence, 1 being the first one. Negative values indicate
7669 positions relative to the last one, -1 being the last one. This
7670 is helpful for situations where an X-Forwarded-For header is set
7671 at the entry point of an infrastructure and must be used several
7672 proxy layers away. When this value is not specified, -1 is
7673 assumed. Passing a zero here disables the feature.
7674
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007675 <name> is an optional interface name to which to bind to for outgoing
7676 traffic. On systems supporting this features (currently, only
7677 Linux), this allows one to bind all traffic to the server to
7678 this interface even if it is not the one the system would select
7679 based on routing tables. This should be used with extreme care.
7680 Note that using this option requires root privileges.
7681
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007682 The "source" keyword is useful in complex environments where a specific
7683 address only is allowed to connect to the servers. It may be needed when a
7684 private address must be used through a public gateway for instance, and it is
7685 known that the system cannot determine the adequate source address by itself.
7686
7687 An extension which is available on certain patched Linux kernels may be used
7688 through the "usesrc" optional keyword. It makes it possible to connect to the
7689 servers with an IP address which does not belong to the system itself. This
7690 is called "full transparent proxy mode". For this to work, the destination
7691 servers have to route their traffic back to this address through the machine
7692 running HAProxy, and IP forwarding must generally be enabled on this machine.
7693
7694 In this "full transparent proxy" mode, it is possible to force a specific IP
7695 address to be presented to the servers. This is not much used in fact. A more
7696 common use is to tell HAProxy to present the client's IP address. For this,
7697 there are two methods :
7698
7699 - present the client's IP and port addresses. This is the most transparent
7700 mode, but it can cause problems when IP connection tracking is enabled on
7701 the machine, because a same connection may be seen twice with different
7702 states. However, this solution presents the huge advantage of not
7703 limiting the system to the 64k outgoing address+port couples, because all
7704 of the client ranges may be used.
7705
7706 - present only the client's IP address and select a spare port. This
7707 solution is still quite elegant but slightly less transparent (downstream
7708 firewalls logs will not match upstream's). It also presents the downside
7709 of limiting the number of concurrent connections to the usual 64k ports.
7710 However, since the upstream and downstream ports are different, local IP
7711 connection tracking on the machine will not be upset by the reuse of the
7712 same session.
7713
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007714 This option sets the default source for all servers in the backend. It may
7715 also be specified in a "defaults" section. Finer source address specification
7716 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007717 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007718
Baptiste Assmann91bd3372015-07-17 21:59:42 +02007719 In order to work, "usesrc" requires root privileges.
7720
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007721 Examples :
7722 backend private
7723 # Connect to the servers using our 192.168.1.200 source address
7724 source 192.168.1.200
7725
7726 backend transparent_ssl1
7727 # Connect to the SSL farm from the client's source address
7728 source 192.168.1.200 usesrc clientip
7729
7730 backend transparent_ssl2
7731 # Connect to the SSL farm from the client's source address and port
7732 # not recommended if IP conntrack is present on the local machine.
7733 source 192.168.1.200 usesrc client
7734
7735 backend transparent_ssl3
7736 # Connect to the SSL farm from the client's source address. It
7737 # is more conntrack-friendly.
7738 source 192.168.1.200 usesrc clientip
7739
7740 backend transparent_smtp
7741 # Connect to the SMTP farm from the client's source address/port
7742 # with Tproxy version 4.
7743 source 0.0.0.0 usesrc clientip
7744
Willy Tarreaubce70882009-09-07 11:51:47 +02007745 backend transparent_http
7746 # Connect to the servers using the client's IP as seen by previous
7747 # proxy.
7748 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
7749
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007750 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007751 the Linux kernel on www.balabit.com, the "bind" keyword.
7752
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007753
Willy Tarreau844e3c52008-01-11 16:28:18 +01007754srvtimeout <timeout> (deprecated)
7755 Set the maximum inactivity time on the server side.
7756 May be used in sections : defaults | frontend | listen | backend
7757 yes | no | yes | yes
7758 Arguments :
7759 <timeout> is the timeout value specified in milliseconds by default, but
7760 can be in any other unit if the number is suffixed by the unit,
7761 as explained at the top of this document.
7762
7763 The inactivity timeout applies when the server is expected to acknowledge or
7764 send data. In HTTP mode, this timeout is particularly important to consider
7765 during the first phase of the server's response, when it has to send the
7766 headers, as it directly represents the server's processing time for the
7767 request. To find out what value to put there, it's often good to start with
7768 what would be considered as unacceptable response times, then check the logs
7769 to observe the response time distribution, and adjust the value accordingly.
7770
7771 The value is specified in milliseconds by default, but can be in any other
7772 unit if the number is suffixed by the unit, as specified at the top of this
7773 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
7774 recommended that the client timeout remains equal to the server timeout in
7775 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007776 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01007777 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreaud72758d2010-01-12 10:42:19 +01007778 seconds (eg: 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01007779
7780 This parameter is specific to backends, but can be specified once for all in
7781 "defaults" sections. This is in fact one of the easiest solutions not to
7782 forget about it. An unspecified timeout results in an infinite timeout, which
7783 is not recommended. Such a usage is accepted and works but reports a warning
7784 during startup because it may results in accumulation of expired sessions in
7785 the system if the system's timeouts are not configured either.
7786
7787 This parameter is provided for compatibility but is currently deprecated.
7788 Please use "timeout server" instead.
7789
Willy Tarreauce887fd2012-05-12 12:50:00 +02007790 See also : "timeout server", "timeout tunnel", "timeout client" and
7791 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01007792
7793
Cyril Bonté66c327d2010-10-12 00:14:37 +02007794stats admin { if | unless } <cond>
7795 Enable statistics admin level if/unless a condition is matched
7796 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007797 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02007798
7799 This statement enables the statistics admin level if/unless a condition is
7800 matched.
7801
7802 The admin level allows to enable/disable servers from the web interface. By
7803 default, statistics page is read-only for security reasons.
7804
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01007805 Note : Consider not using this feature in multi-process mode (nbproc > 1)
7806 unless you know what you do : memory is not shared between the
7807 processes, which can result in random behaviours.
7808
Cyril Bonté23b39d92011-02-10 22:54:44 +01007809 Currently, the POST request is limited to the buffer size minus the reserved
7810 buffer space, which means that if the list of servers is too long, the
7811 request won't be processed. It is recommended to alter few servers at a
7812 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02007813
7814 Example :
7815 # statistics admin level only for localhost
7816 backend stats_localhost
7817 stats enable
7818 stats admin if LOCALHOST
7819
7820 Example :
7821 # statistics admin level always enabled because of the authentication
7822 backend stats_auth
7823 stats enable
7824 stats auth admin:AdMiN123
7825 stats admin if TRUE
7826
7827 Example :
7828 # statistics admin level depends on the authenticated user
7829 userlist stats-auth
7830 group admin users admin
7831 user admin insecure-password AdMiN123
7832 group readonly users haproxy
7833 user haproxy insecure-password haproxy
7834
7835 backend stats_auth
7836 stats enable
7837 acl AUTH http_auth(stats-auth)
7838 acl AUTH_ADMIN http_auth_group(stats-auth) admin
7839 stats http-request auth unless AUTH
7840 stats admin if AUTH_ADMIN
7841
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01007842 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
7843 "bind-process", section 3.4 about userlists and section 7 about
7844 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02007845
7846
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007847stats auth <user>:<passwd>
7848 Enable statistics with authentication and grant access to an account
7849 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007850 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007851 Arguments :
7852 <user> is a user name to grant access to
7853
7854 <passwd> is the cleartext password associated to this user
7855
7856 This statement enables statistics with default settings, and restricts access
7857 to declared users only. It may be repeated as many times as necessary to
7858 allow as many users as desired. When a user tries to access the statistics
7859 without a valid account, a "401 Forbidden" response will be returned so that
7860 the browser asks the user to provide a valid user and password. The real
7861 which will be returned to the browser is configurable using "stats realm".
7862
7863 Since the authentication method is HTTP Basic Authentication, the passwords
7864 circulate in cleartext on the network. Thus, it was decided that the
7865 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007866 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007867
7868 It is also possible to reduce the scope of the proxies which appear in the
7869 report using "stats scope".
7870
7871 Though this statement alone is enough to enable statistics reporting, it is
7872 recommended to set all other settings in order to avoid relying on default
7873 unobvious parameters.
7874
7875 Example :
7876 # public access (limited to this backend only)
7877 backend public_www
7878 server srv1 192.168.0.1:80
7879 stats enable
7880 stats hide-version
7881 stats scope .
7882 stats uri /admin?stats
7883 stats realm Haproxy\ Statistics
7884 stats auth admin1:AdMiN123
7885 stats auth admin2:AdMiN321
7886
7887 # internal monitoring access (unlimited)
7888 backend private_monitoring
7889 stats enable
7890 stats uri /admin?stats
7891 stats refresh 5s
7892
7893 See also : "stats enable", "stats realm", "stats scope", "stats uri"
7894
7895
7896stats enable
7897 Enable statistics reporting with default settings
7898 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007899 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007900 Arguments : none
7901
7902 This statement enables statistics reporting with default settings defined
7903 at build time. Unless stated otherwise, these settings are used :
7904 - stats uri : /haproxy?stats
7905 - stats realm : "HAProxy Statistics"
7906 - stats auth : no authentication
7907 - stats scope : no restriction
7908
7909 Though this statement alone is enough to enable statistics reporting, it is
7910 recommended to set all other settings in order to avoid relying on default
7911 unobvious parameters.
7912
7913 Example :
7914 # public access (limited to this backend only)
7915 backend public_www
7916 server srv1 192.168.0.1:80
7917 stats enable
7918 stats hide-version
7919 stats scope .
7920 stats uri /admin?stats
7921 stats realm Haproxy\ Statistics
7922 stats auth admin1:AdMiN123
7923 stats auth admin2:AdMiN321
7924
7925 # internal monitoring access (unlimited)
7926 backend private_monitoring
7927 stats enable
7928 stats uri /admin?stats
7929 stats refresh 5s
7930
7931 See also : "stats auth", "stats realm", "stats uri"
7932
7933
Willy Tarreaud63335a2010-02-26 12:56:52 +01007934stats hide-version
7935 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007936 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007937 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01007938 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007939
Willy Tarreaud63335a2010-02-26 12:56:52 +01007940 By default, the stats page reports some useful status information along with
7941 the statistics. Among them is HAProxy's version. However, it is generally
7942 considered dangerous to report precise version to anyone, as it can help them
7943 target known weaknesses with specific attacks. The "stats hide-version"
7944 statement removes the version from the statistics report. This is recommended
7945 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007946
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02007947 Though this statement alone is enough to enable statistics reporting, it is
7948 recommended to set all other settings in order to avoid relying on default
7949 unobvious parameters.
7950
Willy Tarreaud63335a2010-02-26 12:56:52 +01007951 Example :
7952 # public access (limited to this backend only)
7953 backend public_www
7954 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02007955 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01007956 stats hide-version
7957 stats scope .
7958 stats uri /admin?stats
7959 stats realm Haproxy\ Statistics
7960 stats auth admin1:AdMiN123
7961 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007962
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007963 # internal monitoring access (unlimited)
7964 backend private_monitoring
7965 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01007966 stats uri /admin?stats
7967 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01007968
Willy Tarreaud63335a2010-02-26 12:56:52 +01007969 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02007970
Willy Tarreau983e01e2010-01-11 18:42:06 +01007971
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02007972stats http-request { allow | deny | auth [realm <realm>] }
7973 [ { if | unless } <condition> ]
7974 Access control for statistics
7975
7976 May be used in sections: defaults | frontend | listen | backend
7977 no | no | yes | yes
7978
7979 As "http-request", these set of options allow to fine control access to
7980 statistics. Each option may be followed by if/unless and acl.
7981 First option with matched condition (or option without condition) is final.
7982 For "deny" a 403 error will be returned, for "allow" normal processing is
7983 performed, for "auth" a 401/407 error code is returned so the client
7984 should be asked to enter a username and password.
7985
7986 There is no fixed limit to the number of http-request statements per
7987 instance.
7988
7989 See also : "http-request", section 3.4 about userlists and section 7
7990 about ACL usage.
7991
7992
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007993stats realm <realm>
7994 Enable statistics and set authentication realm
7995 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007996 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007997 Arguments :
7998 <realm> is the name of the HTTP Basic Authentication realm reported to
7999 the browser. The browser uses it to display it in the pop-up
8000 inviting the user to enter a valid username and password.
8001
8002 The realm is read as a single word, so any spaces in it should be escaped
8003 using a backslash ('\').
8004
8005 This statement is useful only in conjunction with "stats auth" since it is
8006 only related to authentication.
8007
8008 Though this statement alone is enough to enable statistics reporting, it is
8009 recommended to set all other settings in order to avoid relying on default
8010 unobvious parameters.
8011
8012 Example :
8013 # public access (limited to this backend only)
8014 backend public_www
8015 server srv1 192.168.0.1:80
8016 stats enable
8017 stats hide-version
8018 stats scope .
8019 stats uri /admin?stats
8020 stats realm Haproxy\ Statistics
8021 stats auth admin1:AdMiN123
8022 stats auth admin2:AdMiN321
8023
8024 # internal monitoring access (unlimited)
8025 backend private_monitoring
8026 stats enable
8027 stats uri /admin?stats
8028 stats refresh 5s
8029
8030 See also : "stats auth", "stats enable", "stats uri"
8031
8032
8033stats refresh <delay>
8034 Enable statistics with automatic refresh
8035 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008036 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008037 Arguments :
8038 <delay> is the suggested refresh delay, specified in seconds, which will
8039 be returned to the browser consulting the report page. While the
8040 browser is free to apply any delay, it will generally respect it
8041 and refresh the page this every seconds. The refresh interval may
8042 be specified in any other non-default time unit, by suffixing the
8043 unit after the value, as explained at the top of this document.
8044
8045 This statement is useful on monitoring displays with a permanent page
8046 reporting the load balancer's activity. When set, the HTML report page will
8047 include a link "refresh"/"stop refresh" so that the user can select whether
8048 he wants automatic refresh of the page or not.
8049
8050 Though this statement alone is enough to enable statistics reporting, it is
8051 recommended to set all other settings in order to avoid relying on default
8052 unobvious parameters.
8053
8054 Example :
8055 # public access (limited to this backend only)
8056 backend public_www
8057 server srv1 192.168.0.1:80
8058 stats enable
8059 stats hide-version
8060 stats scope .
8061 stats uri /admin?stats
8062 stats realm Haproxy\ Statistics
8063 stats auth admin1:AdMiN123
8064 stats auth admin2:AdMiN321
8065
8066 # internal monitoring access (unlimited)
8067 backend private_monitoring
8068 stats enable
8069 stats uri /admin?stats
8070 stats refresh 5s
8071
8072 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8073
8074
8075stats scope { <name> | "." }
8076 Enable statistics and limit access scope
8077 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008078 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008079 Arguments :
8080 <name> is the name of a listen, frontend or backend section to be
8081 reported. The special name "." (a single dot) designates the
8082 section in which the statement appears.
8083
8084 When this statement is specified, only the sections enumerated with this
8085 statement will appear in the report. All other ones will be hidden. This
8086 statement may appear as many times as needed if multiple sections need to be
8087 reported. Please note that the name checking is performed as simple string
8088 comparisons, and that it is never checked that a give section name really
8089 exists.
8090
8091 Though this statement alone is enough to enable statistics reporting, it is
8092 recommended to set all other settings in order to avoid relying on default
8093 unobvious parameters.
8094
8095 Example :
8096 # public access (limited to this backend only)
8097 backend public_www
8098 server srv1 192.168.0.1:80
8099 stats enable
8100 stats hide-version
8101 stats scope .
8102 stats uri /admin?stats
8103 stats realm Haproxy\ Statistics
8104 stats auth admin1:AdMiN123
8105 stats auth admin2:AdMiN321
8106
8107 # internal monitoring access (unlimited)
8108 backend private_monitoring
8109 stats enable
8110 stats uri /admin?stats
8111 stats refresh 5s
8112
8113 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8114
Willy Tarreaud63335a2010-02-26 12:56:52 +01008115
Willy Tarreauc9705a12010-07-27 20:05:50 +02008116stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008117 Enable reporting of a description on the statistics page.
8118 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008119 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008120
Willy Tarreauc9705a12010-07-27 20:05:50 +02008121 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008122 description from global section is automatically used instead.
8123
8124 This statement is useful for users that offer shared services to their
8125 customers, where node or description should be different for each customer.
8126
8127 Though this statement alone is enough to enable statistics reporting, it is
8128 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04008129 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008130
8131 Example :
8132 # internal monitoring access (unlimited)
8133 backend private_monitoring
8134 stats enable
8135 stats show-desc Master node for Europe, Asia, Africa
8136 stats uri /admin?stats
8137 stats refresh 5s
8138
8139 See also: "show-node", "stats enable", "stats uri" and "description" in
8140 global section.
8141
8142
8143stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008144 Enable reporting additional information on the statistics page
8145 May be used in sections : defaults | frontend | listen | backend
8146 yes | yes | yes | yes
8147 Arguments : none
8148
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008149 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008150 - cap: capabilities (proxy)
8151 - mode: one of tcp, http or health (proxy)
8152 - id: SNMP ID (proxy, socket, server)
8153 - IP (socket, server)
8154 - cookie (backend, server)
8155
8156 Though this statement alone is enough to enable statistics reporting, it is
8157 recommended to set all other settings in order to avoid relying on default
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04008158 unobvious parameters. Default behaviour is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008159
8160 See also: "stats enable", "stats uri".
8161
8162
8163stats show-node [ <name> ]
8164 Enable reporting of a host name on the statistics page.
8165 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008166 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008167 Arguments:
8168 <name> is an optional name to be reported. If unspecified, the
8169 node name from global section is automatically used instead.
8170
8171 This statement is useful for users that offer shared services to their
8172 customers, where node or description might be different on a stats page
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04008173 provided for each customer. Default behaviour is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008174
8175 Though this statement alone is enough to enable statistics reporting, it is
8176 recommended to set all other settings in order to avoid relying on default
8177 unobvious parameters.
8178
8179 Example:
8180 # internal monitoring access (unlimited)
8181 backend private_monitoring
8182 stats enable
8183 stats show-node Europe-1
8184 stats uri /admin?stats
8185 stats refresh 5s
8186
8187 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8188 section.
8189
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008190
8191stats uri <prefix>
8192 Enable statistics and define the URI prefix to access them
8193 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008194 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008195 Arguments :
8196 <prefix> is the prefix of any URI which will be redirected to stats. This
8197 prefix may contain a question mark ('?') to indicate part of a
8198 query string.
8199
8200 The statistics URI is intercepted on the relayed traffic, so it appears as a
8201 page within the normal application. It is strongly advised to ensure that the
8202 selected URI will never appear in the application, otherwise it will never be
8203 possible to reach it in the application.
8204
8205 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008206 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008207 It is generally a good idea to include a question mark in the URI so that
8208 intermediate proxies refrain from caching the results. Also, since any string
8209 beginning with the prefix will be accepted as a stats request, the question
8210 mark helps ensuring that no valid URI will begin with the same words.
8211
8212 It is sometimes very convenient to use "/" as the URI prefix, and put that
8213 statement in a "listen" instance of its own. That makes it easy to dedicate
8214 an address or a port to statistics only.
8215
8216 Though this statement alone is enough to enable statistics reporting, it is
8217 recommended to set all other settings in order to avoid relying on default
8218 unobvious parameters.
8219
8220 Example :
8221 # public access (limited to this backend only)
8222 backend public_www
8223 server srv1 192.168.0.1:80
8224 stats enable
8225 stats hide-version
8226 stats scope .
8227 stats uri /admin?stats
8228 stats realm Haproxy\ Statistics
8229 stats auth admin1:AdMiN123
8230 stats auth admin2:AdMiN321
8231
8232 # internal monitoring access (unlimited)
8233 backend private_monitoring
8234 stats enable
8235 stats uri /admin?stats
8236 stats refresh 5s
8237
8238 See also : "stats auth", "stats enable", "stats realm"
8239
8240
Willy Tarreaud63335a2010-02-26 12:56:52 +01008241stick match <pattern> [table <table>] [{if | unless} <cond>]
8242 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008243 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008244 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008245
8246 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008247 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008248 describes what elements of the incoming request or connection
8249 will be analysed in the hope to find a matching entry in a
8250 stickiness table. This rule is mandatory.
8251
8252 <table> is an optional stickiness table name. If unspecified, the same
8253 backend's table is used. A stickiness table is declared using
8254 the "stick-table" statement.
8255
8256 <cond> is an optional matching condition. It makes it possible to match
8257 on a certain criterion only when other conditions are met (or
8258 not met). For instance, it could be used to match on a source IP
8259 address except when a request passes through a known proxy, in
8260 which case we'd match on a header containing that IP address.
8261
8262 Some protocols or applications require complex stickiness rules and cannot
8263 always simply rely on cookies nor hashing. The "stick match" statement
8264 describes a rule to extract the stickiness criterion from an incoming request
8265 or connection. See section 7 for a complete list of possible patterns and
8266 transformation rules.
8267
8268 The table has to be declared using the "stick-table" statement. It must be of
8269 a type compatible with the pattern. By default it is the one which is present
8270 in the same backend. It is possible to share a table with other backends by
8271 referencing it using the "table" keyword. If another table is referenced,
8272 the server's ID inside the backends are used. By default, all server IDs
8273 start at 1 in each backend, so the server ordering is enough. But in case of
8274 doubt, it is highly recommended to force server IDs using their "id" setting.
8275
8276 It is possible to restrict the conditions where a "stick match" statement
8277 will apply, using "if" or "unless" followed by a condition. See section 7 for
8278 ACL based conditions.
8279
8280 There is no limit on the number of "stick match" statements. The first that
8281 applies and matches will cause the request to be directed to the same server
8282 as was used for the request which created the entry. That way, multiple
8283 matches can be used as fallbacks.
8284
8285 The stick rules are checked after the persistence cookies, so they will not
8286 affect stickiness if a cookie has already been used to select a server. That
8287 way, it becomes very easy to insert cookies and match on IP addresses in
8288 order to maintain stickiness between HTTP and HTTPS.
8289
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008290 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8291 unless you know what you do : memory is not shared between the
8292 processes, which can result in random behaviours.
8293
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008294 Example :
8295 # forward SMTP users to the same server they just used for POP in the
8296 # last 30 minutes
8297 backend pop
8298 mode tcp
8299 balance roundrobin
8300 stick store-request src
8301 stick-table type ip size 200k expire 30m
8302 server s1 192.168.1.1:110
8303 server s2 192.168.1.1:110
8304
8305 backend smtp
8306 mode tcp
8307 balance roundrobin
8308 stick match src table pop
8309 server s1 192.168.1.1:25
8310 server s2 192.168.1.1:25
8311
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008312 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008313 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008314
8315
8316stick on <pattern> [table <table>] [{if | unless} <condition>]
8317 Define a request pattern to associate a user to a server
8318 May be used in sections : defaults | frontend | listen | backend
8319 no | no | yes | yes
8320
8321 Note : This form is exactly equivalent to "stick match" followed by
8322 "stick store-request", all with the same arguments. Please refer
8323 to both keywords for details. It is only provided as a convenience
8324 for writing more maintainable configurations.
8325
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008326 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8327 unless you know what you do : memory is not shared between the
8328 processes, which can result in random behaviours.
8329
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008330 Examples :
8331 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008332 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008333
8334 # ...is strictly equivalent to this one :
8335 stick match src table pop if !localhost
8336 stick store-request src table pop if !localhost
8337
8338
8339 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8340 # well as HTTP without cookie. Share the same table between both accesses.
8341 backend http
8342 mode http
8343 balance roundrobin
8344 stick on src table https
8345 cookie SRV insert indirect nocache
8346 server s1 192.168.1.1:80 cookie s1
8347 server s2 192.168.1.1:80 cookie s2
8348
8349 backend https
8350 mode tcp
8351 balance roundrobin
8352 stick-table type ip size 200k expire 30m
8353 stick on src
8354 server s1 192.168.1.1:443
8355 server s2 192.168.1.1:443
8356
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008357 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008358
8359
8360stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8361 Define a request pattern used to create an entry in a stickiness table
8362 May be used in sections : defaults | frontend | listen | backend
8363 no | no | yes | yes
8364
8365 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008366 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008367 describes what elements of the incoming request or connection
8368 will be analysed, extracted and stored in the table once a
8369 server is selected.
8370
8371 <table> is an optional stickiness table name. If unspecified, the same
8372 backend's table is used. A stickiness table is declared using
8373 the "stick-table" statement.
8374
8375 <cond> is an optional storage condition. It makes it possible to store
8376 certain criteria only when some conditions are met (or not met).
8377 For instance, it could be used to store the source IP address
8378 except when the request passes through a known proxy, in which
8379 case we'd store a converted form of a header containing that IP
8380 address.
8381
8382 Some protocols or applications require complex stickiness rules and cannot
8383 always simply rely on cookies nor hashing. The "stick store-request" statement
8384 describes a rule to decide what to extract from the request and when to do
8385 it, in order to store it into a stickiness table for further requests to
8386 match it using the "stick match" statement. Obviously the extracted part must
8387 make sense and have a chance to be matched in a further request. Storing a
8388 client's IP address for instance often makes sense. Storing an ID found in a
8389 URL parameter also makes sense. Storing a source port will almost never make
8390 any sense because it will be randomly matched. See section 7 for a complete
8391 list of possible patterns and transformation rules.
8392
8393 The table has to be declared using the "stick-table" statement. It must be of
8394 a type compatible with the pattern. By default it is the one which is present
8395 in the same backend. It is possible to share a table with other backends by
8396 referencing it using the "table" keyword. If another table is referenced,
8397 the server's ID inside the backends are used. By default, all server IDs
8398 start at 1 in each backend, so the server ordering is enough. But in case of
8399 doubt, it is highly recommended to force server IDs using their "id" setting.
8400
8401 It is possible to restrict the conditions where a "stick store-request"
8402 statement will apply, using "if" or "unless" followed by a condition. This
8403 condition will be evaluated while parsing the request, so any criteria can be
8404 used. See section 7 for ACL based conditions.
8405
8406 There is no limit on the number of "stick store-request" statements, but
8407 there is a limit of 8 simultaneous stores per request or response. This
8408 makes it possible to store up to 8 criteria, all extracted from either the
8409 request or the response, regardless of the number of rules. Only the 8 first
8410 ones which match will be kept. Using this, it is possible to feed multiple
8411 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008412 another protocol or access method. Using multiple store-request rules with
8413 the same table is possible and may be used to find the best criterion to rely
8414 on, by arranging the rules by decreasing preference order. Only the first
8415 extracted criterion for a given table will be stored. All subsequent store-
8416 request rules referencing the same table will be skipped and their ACLs will
8417 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008418
8419 The "store-request" rules are evaluated once the server connection has been
8420 established, so that the table will contain the real server that processed
8421 the request.
8422
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008423 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8424 unless you know what you do : memory is not shared between the
8425 processes, which can result in random behaviours.
8426
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008427 Example :
8428 # forward SMTP users to the same server they just used for POP in the
8429 # last 30 minutes
8430 backend pop
8431 mode tcp
8432 balance roundrobin
8433 stick store-request src
8434 stick-table type ip size 200k expire 30m
8435 server s1 192.168.1.1:110
8436 server s2 192.168.1.1:110
8437
8438 backend smtp
8439 mode tcp
8440 balance roundrobin
8441 stick match src table pop
8442 server s1 192.168.1.1:25
8443 server s2 192.168.1.1:25
8444
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008445 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008446 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008447
8448
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008449stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008450 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8451 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008452 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008453 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008454 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008455
8456 Arguments :
8457 ip a table declared with "type ip" will only store IPv4 addresses.
8458 This form is very compact (about 50 bytes per entry) and allows
8459 very fast entry lookup and stores with almost no overhead. This
8460 is mainly used to store client source IP addresses.
8461
David du Colombier9a6d3c92011-03-17 10:40:24 +01008462 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8463 This form is very compact (about 60 bytes per entry) and allows
8464 very fast entry lookup and stores with almost no overhead. This
8465 is mainly used to store client source IP addresses.
8466
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008467 integer a table declared with "type integer" will store 32bit integers
8468 which can represent a client identifier found in a request for
8469 instance.
8470
8471 string a table declared with "type string" will store substrings of up
8472 to <len> characters. If the string provided by the pattern
8473 extractor is larger than <len>, it will be truncated before
8474 being stored. During matching, at most <len> characters will be
8475 compared between the string in the table and the extracted
8476 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008477 to 32 characters.
8478
8479 binary a table declared with "type binary" will store binary blocks
8480 of <len> bytes. If the block provided by the pattern
8481 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008482 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008483 is shorter than <len>, it will be padded by 0. When not
8484 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008485
8486 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008487 "string" type table (See type "string" above). Or the number
8488 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008489 changing this parameter as memory usage will proportionally
8490 increase.
8491
8492 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008493 value directly impacts memory usage. Count approximately
8494 50 bytes per entry, plus the size of a string if any. The size
8495 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008496
8497 [nopurge] indicates that we refuse to purge older entries when the table
8498 is full. When not specified and the table is full when haproxy
8499 wants to store an entry in it, it will flush a few of the oldest
8500 entries in order to release some space for the new ones. This is
8501 most often the desired behaviour. In some specific cases, it
8502 be desirable to refuse new entries instead of purging the older
8503 ones. That may be the case when the amount of data to store is
8504 far above the hardware limits and we prefer not to offer access
8505 to new clients than to reject the ones already connected. When
8506 using this parameter, be sure to properly set the "expire"
8507 parameter (see below).
8508
Emeric Brunf099e792010-09-27 12:05:28 +02008509 <peersect> is the name of the peers section to use for replication. Entries
8510 which associate keys to server IDs are kept synchronized with
8511 the remote peers declared in this section. All entries are also
8512 automatically learned from the local peer (old process) during a
8513 soft restart.
8514
Willy Tarreau1abc6732015-05-01 19:21:02 +02008515 NOTE : each peers section may be referenced only by tables
8516 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008517
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008518 <expire> defines the maximum duration of an entry in the table since it
8519 was last created, refreshed or matched. The expiration delay is
8520 defined using the standard time format, similarly as the various
8521 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008522 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008523 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008524 be removed once full. Be sure not to use the "nopurge" parameter
8525 if not expiration delay is specified.
8526
Willy Tarreau08d5f982010-06-06 13:34:54 +02008527 <data_type> is used to store additional information in the stick-table. This
8528 may be used by ACLs in order to control various criteria related
8529 to the activity of the client matching the stick-table. For each
8530 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008531 that the additional data can fit. Several data types may be
8532 stored with an entry. Multiple data types may be specified after
8533 the "store" keyword, as a comma-separated list. Alternatively,
8534 it is possible to repeat the "store" keyword followed by one or
8535 several data types. Except for the "server_id" type which is
8536 automatically detected and enabled, all data types must be
8537 explicitly declared to be stored. If an ACL references a data
8538 type which is not stored, the ACL will simply not match. Some
8539 data types require an argument which must be passed just after
8540 the type between parenthesis. See below for the supported data
8541 types and their arguments.
8542
8543 The data types that can be stored with an entry are the following :
8544 - server_id : this is an integer which holds the numeric ID of the server a
8545 request was assigned to. It is used by the "stick match", "stick store",
8546 and "stick on" rules. It is automatically enabled when referenced.
8547
8548 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8549 integer which may be used for anything. Most of the time it will be used
8550 to put a special tag on some entries, for instance to note that a
8551 specific behaviour was detected and must be known for future matches.
8552
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008553 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8554 over a period. It is a positive 32-bit integer integer which may be used
8555 for anything. Just like <gpc0>, it counts events, but instead of keeping
8556 a cumulative count, it maintains the rate at which the counter is
8557 incremented. Most of the time it will be used to measure the frequency of
8558 occurrence of certain events (eg: requests to a specific URL).
8559
Willy Tarreauc9705a12010-07-27 20:05:50 +02008560 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8561 the absolute number of connections received from clients which matched
8562 this entry. It does not mean the connections were accepted, just that
8563 they were received.
8564
8565 - conn_cur : Current Connections. It is a positive 32-bit integer which
8566 stores the concurrent connection counts for the entry. It is incremented
8567 once an incoming connection matches the entry, and decremented once the
8568 connection leaves. That way it is possible to know at any time the exact
8569 number of concurrent connections for an entry.
8570
8571 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8572 integer parameter <period> which indicates in milliseconds the length
8573 of the period over which the average is measured. It reports the average
8574 incoming connection rate over that period, in connections per period. The
8575 result is an integer which can be matched using ACLs.
8576
8577 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8578 the absolute number of sessions received from clients which matched this
8579 entry. A session is a connection that was accepted by the layer 4 rules.
8580
8581 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8582 integer parameter <period> which indicates in milliseconds the length
8583 of the period over which the average is measured. It reports the average
8584 incoming session rate over that period, in sessions per period. The
8585 result is an integer which can be matched using ACLs.
8586
8587 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8588 counts the absolute number of HTTP requests received from clients which
8589 matched this entry. It does not matter whether they are valid requests or
8590 not. Note that this is different from sessions when keep-alive is used on
8591 the client side.
8592
8593 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8594 integer parameter <period> which indicates in milliseconds the length
8595 of the period over which the average is measured. It reports the average
8596 HTTP request rate over that period, in requests per period. The result is
8597 an integer which can be matched using ACLs. It does not matter whether
8598 they are valid requests or not. Note that this is different from sessions
8599 when keep-alive is used on the client side.
8600
8601 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8602 counts the absolute number of HTTP requests errors induced by clients
8603 which matched this entry. Errors are counted on invalid and truncated
8604 requests, as well as on denied or tarpitted requests, and on failed
8605 authentications. If the server responds with 4xx, then the request is
8606 also counted as an error since it's an error triggered by the client
8607 (eg: vulnerability scan).
8608
8609 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8610 integer parameter <period> which indicates in milliseconds the length
8611 of the period over which the average is measured. It reports the average
8612 HTTP request error rate over that period, in requests per period (see
8613 http_err_cnt above for what is accounted as an error). The result is an
8614 integer which can be matched using ACLs.
8615
8616 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
8617 integer which counts the cumulated amount of bytes received from clients
8618 which matched this entry. Headers are included in the count. This may be
8619 used to limit abuse of upload features on photo or video servers.
8620
8621 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8622 integer parameter <period> which indicates in milliseconds the length
8623 of the period over which the average is measured. It reports the average
8624 incoming bytes rate over that period, in bytes per period. It may be used
8625 to detect users which upload too much and too fast. Warning: with large
8626 uploads, it is possible that the amount of uploaded data will be counted
8627 once upon termination, thus causing spikes in the average transfer speed
8628 instead of having a smooth one. This may partially be smoothed with
8629 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8630 recommended for better fairness.
8631
8632 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
8633 integer which counts the cumulated amount of bytes sent to clients which
8634 matched this entry. Headers are included in the count. This may be used
8635 to limit abuse of bots sucking the whole site.
8636
8637 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8638 an integer parameter <period> which indicates in milliseconds the length
8639 of the period over which the average is measured. It reports the average
8640 outgoing bytes rate over that period, in bytes per period. It may be used
8641 to detect users which download too much and too fast. Warning: with large
8642 transfers, it is possible that the amount of transferred data will be
8643 counted once upon termination, thus causing spikes in the average
8644 transfer speed instead of having a smooth one. This may partially be
8645 smoothed with "option contstats" though this is not perfect yet. Use of
8646 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008647
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008648 There is only one stick-table per proxy. At the moment of writing this doc,
8649 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008650 to be required, simply create a dummy backend with a stick-table in it and
8651 reference it.
8652
8653 It is important to understand that stickiness based on learning information
8654 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008655 lost upon restart unless peers are properly configured to transfer such
8656 information upon restart (recommended). In general it can be good as a
8657 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008658
Willy Tarreauc9705a12010-07-27 20:05:50 +02008659 Last, memory requirements may be important when storing many data types.
8660 Indeed, storing all indicators above at once in each entry requires 116 bytes
8661 per entry, or 116 MB for a 1-million entries table. This is definitely not
8662 something that can be ignored.
8663
8664 Example:
8665 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8666 # and store a general purpose counter and the average connection rate
8667 # computed over a sliding window of 30 seconds.
8668 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8669
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008670 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008671 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008672
8673
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008674stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008675 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008676 May be used in sections : defaults | frontend | listen | backend
8677 no | no | yes | yes
8678
8679 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008680 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008681 describes what elements of the response or connection will
8682 be analysed, extracted and stored in the table once a
8683 server is selected.
8684
8685 <table> is an optional stickiness table name. If unspecified, the same
8686 backend's table is used. A stickiness table is declared using
8687 the "stick-table" statement.
8688
8689 <cond> is an optional storage condition. It makes it possible to store
8690 certain criteria only when some conditions are met (or not met).
8691 For instance, it could be used to store the SSL session ID only
8692 when the response is a SSL server hello.
8693
8694 Some protocols or applications require complex stickiness rules and cannot
8695 always simply rely on cookies nor hashing. The "stick store-response"
8696 statement describes a rule to decide what to extract from the response and
8697 when to do it, in order to store it into a stickiness table for further
8698 requests to match it using the "stick match" statement. Obviously the
8699 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008700 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008701 See section 7 for a complete list of possible patterns and transformation
8702 rules.
8703
8704 The table has to be declared using the "stick-table" statement. It must be of
8705 a type compatible with the pattern. By default it is the one which is present
8706 in the same backend. It is possible to share a table with other backends by
8707 referencing it using the "table" keyword. If another table is referenced,
8708 the server's ID inside the backends are used. By default, all server IDs
8709 start at 1 in each backend, so the server ordering is enough. But in case of
8710 doubt, it is highly recommended to force server IDs using their "id" setting.
8711
8712 It is possible to restrict the conditions where a "stick store-response"
8713 statement will apply, using "if" or "unless" followed by a condition. This
8714 condition will be evaluated while parsing the response, so any criteria can
8715 be used. See section 7 for ACL based conditions.
8716
8717 There is no limit on the number of "stick store-response" statements, but
8718 there is a limit of 8 simultaneous stores per request or response. This
8719 makes it possible to store up to 8 criteria, all extracted from either the
8720 request or the response, regardless of the number of rules. Only the 8 first
8721 ones which match will be kept. Using this, it is possible to feed multiple
8722 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008723 another protocol or access method. Using multiple store-response rules with
8724 the same table is possible and may be used to find the best criterion to rely
8725 on, by arranging the rules by decreasing preference order. Only the first
8726 extracted criterion for a given table will be stored. All subsequent store-
8727 response rules referencing the same table will be skipped and their ACLs will
8728 not be evaluated. However, even if a store-request rule references a table, a
8729 store-response rule may also use the same table. This means that each table
8730 may learn exactly one element from the request and one element from the
8731 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008732
8733 The table will contain the real server that processed the request.
8734
8735 Example :
8736 # Learn SSL session ID from both request and response and create affinity.
8737 backend https
8738 mode tcp
8739 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02008740 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008741 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008742
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008743 acl clienthello req_ssl_hello_type 1
8744 acl serverhello rep_ssl_hello_type 2
8745
8746 # use tcp content accepts to detects ssl client and server hello.
8747 tcp-request inspect-delay 5s
8748 tcp-request content accept if clienthello
8749
8750 # no timeout on response inspect delay by default.
8751 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008752
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008753 # SSL session ID (SSLID) may be present on a client or server hello.
8754 # Its length is coded on 1 byte at offset 43 and its value starts
8755 # at offset 44.
8756
8757 # Match and learn on request if client hello.
8758 stick on payload_lv(43,1) if clienthello
8759
8760 # Learn on response if server hello.
8761 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02008762
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008763 server s1 192.168.1.1:443
8764 server s2 192.168.1.1:443
8765
8766 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
8767 extraction.
8768
8769
Willy Tarreau938c7fe2014-04-25 14:21:39 +02008770tcp-check connect [params*]
8771 Opens a new connection
8772 May be used in sections: defaults | frontend | listen | backend
8773 no | no | yes | yes
8774
8775 When an application lies on more than a single TCP port or when HAProxy
8776 load-balance many services in a single backend, it makes sense to probe all
8777 the services individually before considering a server as operational.
8778
8779 When there are no TCP port configured on the server line neither server port
8780 directive, then the 'tcp-check connect port <port>' must be the first step
8781 of the sequence.
8782
8783 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
8784 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
8785 do.
8786
8787 Parameters :
8788 They are optional and can be used to describe how HAProxy should open and
8789 use the TCP connection.
8790
8791 port if not set, check port or server port is used.
8792 It tells HAProxy where to open the connection to.
8793 <port> must be a valid TCP port source integer, from 1 to 65535.
8794
8795 send-proxy send a PROXY protocol string
8796
8797 ssl opens a ciphered connection
8798
8799 Examples:
8800 # check HTTP and HTTPs services on a server.
8801 # first open port 80 thanks to server line port directive, then
8802 # tcp-check opens port 443, ciphered and run a request on it:
8803 option tcp-check
8804 tcp-check connect
8805 tcp-check send GET\ /\ HTTP/1.0\r\n
8806 tcp-check send Host:\ haproxy.1wt.eu\r\n
8807 tcp-check send \r\n
8808 tcp-check expect rstring (2..|3..)
8809 tcp-check connect port 443 ssl
8810 tcp-check send GET\ /\ HTTP/1.0\r\n
8811 tcp-check send Host:\ haproxy.1wt.eu\r\n
8812 tcp-check send \r\n
8813 tcp-check expect rstring (2..|3..)
8814 server www 10.0.0.1 check port 80
8815
8816 # check both POP and IMAP from a single server:
8817 option tcp-check
8818 tcp-check connect port 110
8819 tcp-check expect string +OK\ POP3\ ready
8820 tcp-check connect port 143
8821 tcp-check expect string *\ OK\ IMAP4\ ready
8822 server mail 10.0.0.1 check
8823
8824 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
8825
8826
8827tcp-check expect [!] <match> <pattern>
8828 Specify data to be collected and analysed during a generic health check
8829 May be used in sections: defaults | frontend | listen | backend
8830 no | no | yes | yes
8831
8832 Arguments :
8833 <match> is a keyword indicating how to look for a specific pattern in the
8834 response. The keyword may be one of "string", "rstring" or
8835 binary.
8836 The keyword may be preceded by an exclamation mark ("!") to negate
8837 the match. Spaces are allowed between the exclamation mark and the
8838 keyword. See below for more details on the supported keywords.
8839
8840 <pattern> is the pattern to look for. It may be a string or a regular
8841 expression. If the pattern contains spaces, they must be escaped
8842 with the usual backslash ('\').
8843 If the match is set to binary, then the pattern must be passed as
8844 a serie of hexadecimal digits in an even number. Each sequence of
8845 two digits will represent a byte. The hexadecimal digits may be
8846 used upper or lower case.
8847
8848
8849 The available matches are intentionally similar to their http-check cousins :
8850
8851 string <string> : test the exact string matches in the response buffer.
8852 A health check response will be considered valid if the
8853 response's buffer contains this exact string. If the
8854 "string" keyword is prefixed with "!", then the response
8855 will be considered invalid if the body contains this
8856 string. This can be used to look for a mandatory pattern
8857 in a protocol response, or to detect a failure when a
8858 specific error appears in a protocol banner.
8859
8860 rstring <regex> : test a regular expression on the response buffer.
8861 A health check response will be considered valid if the
8862 response's buffer matches this expression. If the
8863 "rstring" keyword is prefixed with "!", then the response
8864 will be considered invalid if the body matches the
8865 expression.
8866
8867 binary <hexstring> : test the exact string in its hexadecimal form matches
8868 in the response buffer. A health check response will
8869 be considered valid if the response's buffer contains
8870 this exact hexadecimal string.
8871 Purpose is to match data on binary protocols.
8872
8873 It is important to note that the responses will be limited to a certain size
8874 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
8875 Thus, too large responses may not contain the mandatory pattern when using
8876 "string", "rstring" or binary. If a large response is absolutely required, it
8877 is possible to change the default max size by setting the global variable.
8878 However, it is worth keeping in mind that parsing very large responses can
8879 waste some CPU cycles, especially when regular expressions are used, and that
8880 it is always better to focus the checks on smaller resources. Also, in its
8881 current state, the check will not find any string nor regex past a null
8882 character in the response. Similarly it is not possible to request matching
8883 the null character.
8884
8885 Examples :
8886 # perform a POP check
8887 option tcp-check
8888 tcp-check expect string +OK\ POP3\ ready
8889
8890 # perform an IMAP check
8891 option tcp-check
8892 tcp-check expect string *\ OK\ IMAP4\ ready
8893
8894 # look for the redis master server
8895 option tcp-check
8896 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02008897 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02008898 tcp-check send info\ replication\r\n
8899 tcp-check expect string role:master
8900 tcp-check send QUIT\r\n
8901 tcp-check expect string +OK
8902
8903
8904 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
8905 "tcp-check send-binary", "http-check expect", tune.chksize
8906
8907
8908tcp-check send <data>
8909 Specify a string to be sent as a question during a generic health check
8910 May be used in sections: defaults | frontend | listen | backend
8911 no | no | yes | yes
8912
8913 <data> : the data to be sent as a question during a generic health check
8914 session. For now, <data> must be a string.
8915
8916 Examples :
8917 # look for the redis master server
8918 option tcp-check
8919 tcp-check send info\ replication\r\n
8920 tcp-check expect string role:master
8921
8922 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
8923 "tcp-check send-binary", tune.chksize
8924
8925
8926tcp-check send-binary <hexastring>
8927 Specify an hexa digits string to be sent as a binary question during a raw
8928 tcp health check
8929 May be used in sections: defaults | frontend | listen | backend
8930 no | no | yes | yes
8931
8932 <data> : the data to be sent as a question during a generic health check
8933 session. For now, <data> must be a string.
8934 <hexastring> : test the exact string in its hexadecimal form matches in the
8935 response buffer. A health check response will be considered
8936 valid if the response's buffer contains this exact
8937 hexadecimal string.
8938 Purpose is to send binary data to ask on binary protocols.
8939
8940 Examples :
8941 # redis check in binary
8942 option tcp-check
8943 tcp-check send-binary 50494e470d0a # PING\r\n
8944 tcp-check expect binary 2b504F4e47 # +PONG
8945
8946
8947 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
8948 "tcp-check send", tune.chksize
8949
8950
Willy Tarreaue9656522010-08-17 15:40:09 +02008951tcp-request connection <action> [{if | unless} <condition>]
8952 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02008953 May be used in sections : defaults | frontend | listen | backend
8954 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02008955 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02008956 <action> defines the action to perform if the condition applies. See
8957 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02008958
Willy Tarreaue9656522010-08-17 15:40:09 +02008959 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02008960
8961 Immediately after acceptance of a new incoming connection, it is possible to
8962 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02008963 or dropped or have its counters tracked. Those conditions cannot make use of
8964 any data contents because the connection has not been read from yet, and the
8965 buffers are not yet allocated. This is used to selectively and very quickly
8966 accept or drop connections from various sources with a very low overhead. If
8967 some contents need to be inspected in order to take the decision, the
8968 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02008969
Willy Tarreaue9656522010-08-17 15:40:09 +02008970 The "tcp-request connection" rules are evaluated in their exact declaration
8971 order. If no rule matches or if there is no rule, the default action is to
8972 accept the incoming connection. There is no specific limit to the number of
8973 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02008974
Willy Tarreaua9083d02015-05-08 15:27:59 +02008975 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02008976 - accept :
8977 accepts the connection if the condition is true (when used with "if")
8978 or false (when used with "unless"). The first such rule executed ends
8979 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02008980
Willy Tarreaue9656522010-08-17 15:40:09 +02008981 - reject :
8982 rejects the connection if the condition is true (when used with "if")
8983 or false (when used with "unless"). The first such rule executed ends
8984 the rules evaluation. Rejected connections do not even become a
8985 session, which is why they are accounted separately for in the stats,
8986 as "denied connections". They are not considered for the session
8987 rate-limit and are not logged either. The reason is that these rules
8988 should only be used to filter extremely high connection rates such as
8989 the ones encountered during a massive DDoS attack. Under these extreme
8990 conditions, the simple action of logging each event would make the
8991 system collapse and would considerably lower the filtering capacity. If
8992 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02008993 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02008994
Willy Tarreau4f0d9192013-06-11 20:40:55 +02008995 - expect-proxy layer4 :
8996 configures the client-facing connection to receive a PROXY protocol
8997 header before any byte is read from the socket. This is equivalent to
8998 having the "accept-proxy" keyword on the "bind" line, except that using
8999 the TCP rule allows the PROXY protocol to be accepted only for certain
9000 IP address ranges using an ACL. This is convenient when multiple layers
9001 of load balancers are passed through by traffic coming from public
9002 hosts.
9003
Bertrand Jacquin90759682016-06-06 15:35:39 +01009004 - expect-netscaler-cip layer4 :
9005 configures the client-facing connection to receive a NetScaler Client
9006 IP insertion protocol header before any byte is read from the socket.
9007 This is equivalent to having the "accept-netscaler-cip" keyword on the
9008 "bind" line, except that using the TCP rule allows the PROXY protocol
9009 to be accepted only for certain IP address ranges using an ACL. This
9010 is convenient when multiple layers of load balancers are passed
9011 through by traffic coming from public hosts.
9012
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009013 - capture <sample> len <length> :
9014 This only applies to "tcp-request content" rules. It captures sample
9015 expression <sample> from the request buffer, and converts it to a
9016 string of at most <len> characters. The resulting string is stored into
9017 the next request "capture" slot, so it will possibly appear next to
9018 some captured HTTP headers. It will then automatically appear in the
9019 logs, and it will be possible to extract it using sample fetch rules to
9020 feed it into headers or anything. The length should be limited given
9021 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009022 session life. Please check section 7.3 (Fetching samples) and "capture
9023 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009024
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009025 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009026 enables tracking of sticky counters from current connection. These
Willy Tarreau09448f72014-06-25 18:12:15 +02009027 rules do not stop evaluation and do not change default action. 3 sets
Willy Tarreaue9656522010-08-17 15:40:09 +02009028 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009029 first "track-sc0" rule executed enables tracking of the counters of the
9030 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02009031 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009032 set. The first "track-sc2" rule executed enables tracking of the
9033 counters of the specified table as the third set. It is a recommended
9034 practice to use the first set of counters for the per-frontend counters
9035 and the second set for the per-backend ones. But this is just a
9036 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009037
Willy Tarreaue9656522010-08-17 15:40:09 +02009038 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009039 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009040 in section 7.3. It describes what elements of the incoming
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009041 request or connection will be analysed, extracted, combined,
9042 and used to select which table entry to update the counters.
9043 Note that "tcp-request connection" cannot use content-based
9044 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009045
Willy Tarreaue9656522010-08-17 15:40:09 +02009046 <table> is an optional table to be used instead of the default one,
9047 which is the stick-table declared in the current proxy. All
9048 the counters for the matches and updates for the key will
9049 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009050
Willy Tarreaue9656522010-08-17 15:40:09 +02009051 Once a "track-sc*" rule is executed, the key is looked up in the table
9052 and if it is not found, an entry is allocated for it. Then a pointer to
9053 that entry is kept during all the session's life, and this entry's
9054 counters are updated as often as possible, every time the session's
9055 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009056 Counters are only updated for events that happen after the tracking has
9057 been started. For example, connection counters will not be updated when
9058 tracking layer 7 information, since the connection event happens before
9059 layer7 information is extracted.
9060
Willy Tarreaue9656522010-08-17 15:40:09 +02009061 If the entry tracks concurrent connection counters, one connection is
9062 counted for as long as the entry is tracked, and the entry will not
9063 expire during that time. Tracking counters also provides a performance
9064 advantage over just checking the keys, because only one table lookup is
9065 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009066
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009067 - sc-inc-gpc0(<sc-id>):
9068 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9069 counter designated by <sc-id>. If an error occurs, this action silently
9070 fails and the actions evaluation continues.
9071
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009072 - sc-set-gpt0(<sc-id>) <int>:
9073 This action sets the GPT0 tag according to the sticky counter designated
9074 by <sc-id> and the value of <int>. The expected result is a boolean. If
9075 an error occurs, this action silently fails and the actions evaluation
9076 continues.
9077
William Lallemand2e785f22016-05-25 01:48:42 +02009078 - set-src <expr> :
9079 Is used to set the source IP address to the value of specified
9080 expression. Useful if you want to mask source IP for privacy.
9081 If you want to provide an IP from a HTTP header use "http-request
9082 set-src"
9083
9084 <expr> Is a standard HAProxy expression formed by a sample-fetch
9085 followed by some converters.
9086
9087 Example:
9088
9089 tcp-request connection set-src src,ipmask(24)
9090
Willy Tarreau0c630532016-10-21 17:52:58 +02009091 When possible, set-src preserves the original source port as long as the
9092 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009093
William Lallemand44be6402016-05-25 01:51:35 +02009094 - set-src-port <expr> :
9095 Is used to set the source port address to the value of specified
9096 expression.
9097
9098 <expr> Is a standard HAProxy expression formed by a sample-fetch
9099 followed by some converters.
9100
9101 Example:
9102
9103 tcp-request connection set-src-port int(4000)
9104
Willy Tarreau0c630532016-10-21 17:52:58 +02009105 When possible, set-src-port preserves the original source address as long
9106 as the address family supports a port, otherwise it forces the source
9107 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009108
William Lallemand13e9b0c2016-05-25 02:34:07 +02009109 - set-dst <expr> :
9110 Is used to set the destination IP address to the value of specified
9111 expression. Useful if you want to mask IP for privacy in log.
9112 If you want to provide an IP from a HTTP header use "http-request
9113 set-dst". If you want to connect to the new address/port, use
9114 '0.0.0.0:0' as a server address in the backend.
9115
9116 <expr> Is a standard HAProxy expression formed by a sample-fetch
9117 followed by some converters.
9118
9119 Example:
9120
9121 tcp-request connection set-dst dst,ipmask(24)
9122 tcp-request connection set-dst ipv4(10.0.0.1)
9123
Willy Tarreau0c630532016-10-21 17:52:58 +02009124 When possible, set-dst preserves the original destination port as long as
9125 the address family allows it, otherwise the destination port is set to 0.
9126
William Lallemand13e9b0c2016-05-25 02:34:07 +02009127 - set-dst-port <expr> :
9128 Is used to set the destination port address to the value of specified
9129 expression. If you want to connect to the new address/port, use
9130 '0.0.0.0:0' as a server address in the backend.
9131
9132
9133 <expr> Is a standard HAProxy expression formed by a sample-fetch
9134 followed by some converters.
9135
9136 Example:
9137
9138 tcp-request connection set-dst-port int(4000)
9139
Willy Tarreau0c630532016-10-21 17:52:58 +02009140 When possible, set-dst-port preserves the original destination address as
9141 long as the address family supports a port, otherwise it forces the
9142 destination address to IPv4 "0.0.0.0" before rewriting the port.
9143
Willy Tarreau2d392c22015-08-24 01:43:45 +02009144 - "silent-drop" :
9145 This stops the evaluation of the rules and makes the client-facing
9146 connection suddenly disappear using a system-dependant way that tries
9147 to prevent the client from being notified. The effect it then that the
9148 client still sees an established connection while there's none on
9149 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9150 except that it doesn't use any local resource at all on the machine
9151 running HAProxy. It can resist much higher loads than "tarpit", and
9152 slow down stronger attackers. It is important to undestand the impact
9153 of using this mechanism. All stateful equipments placed between the
9154 client and HAProxy (firewalls, proxies, load balancers) will also keep
9155 the established connection for a long time and may suffer from this
9156 action. On modern Linux systems running with enough privileges, the
9157 TCP_REPAIR socket option is used to block the emission of a TCP
9158 reset. On other systems, the socket's TTL is reduced to 1 so that the
9159 TCP reset doesn't pass the first router, though it's still delivered to
9160 local networks. Do not use it unless you fully understand how it works.
9161
Willy Tarreaue9656522010-08-17 15:40:09 +02009162 Note that the "if/unless" condition is optional. If no condition is set on
9163 the action, it is simply performed unconditionally. That can be useful for
9164 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009165
Willy Tarreaue9656522010-08-17 15:40:09 +02009166 Example: accept all connections from white-listed hosts, reject too fast
9167 connection without counting them, and track accepted connections.
9168 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009169
Willy Tarreaue9656522010-08-17 15:40:09 +02009170 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009171 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009172 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009173
Willy Tarreaue9656522010-08-17 15:40:09 +02009174 Example: accept all connections from white-listed hosts, count all other
9175 connections and reject too fast ones. This results in abusive ones
9176 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009177
Willy Tarreaue9656522010-08-17 15:40:09 +02009178 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009179 tcp-request connection track-sc0 src
9180 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009181
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009182 Example: enable the PROXY protocol for traffic coming from all known proxies.
9183
9184 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9185
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009186 See section 7 about ACL usage.
9187
Willy Tarreau4f614292016-10-21 17:49:36 +02009188 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009189
9190
Willy Tarreaue9656522010-08-17 15:40:09 +02009191tcp-request content <action> [{if | unless} <condition>]
9192 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009193 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009194 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009195 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009196 <action> defines the action to perform if the condition applies. See
9197 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009198
Willy Tarreaue9656522010-08-17 15:40:09 +02009199 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009200
Willy Tarreaue9656522010-08-17 15:40:09 +02009201 A request's contents can be analysed at an early stage of request processing
9202 called "TCP content inspection". During this stage, ACL-based rules are
9203 evaluated every time the request contents are updated, until either an
9204 "accept" or a "reject" rule matches, or the TCP request inspection delay
9205 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009206
Willy Tarreaue9656522010-08-17 15:40:09 +02009207 The first difference between these rules and "tcp-request connection" rules
9208 is that "tcp-request content" rules can make use of contents to take a
9209 decision. Most often, these decisions will consider a protocol recognition or
9210 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009211 both frontends and backends. In case of HTTP keep-alive with the client, all
9212 tcp-request content rules are evaluated again, so haproxy keeps a record of
9213 what sticky counters were assigned by a "tcp-request connection" versus a
9214 "tcp-request content" rule, and flushes all the content-related ones after
9215 processing an HTTP request, so that they may be evaluated again by the rules
9216 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009217 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009218 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009219
Willy Tarreaue9656522010-08-17 15:40:09 +02009220 Content-based rules are evaluated in their exact declaration order. If no
9221 rule matches or if there is no rule, the default action is to accept the
9222 contents. There is no specific limit to the number of rules which may be
9223 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009224
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009225 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009226 - accept : the request is accepted
9227 - reject : the request is rejected and the connection is closed
9228 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009229 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009230 - sc-inc-gpc0(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009231 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009232 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009233 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009234 - silent-drop
Willy Tarreau62644772008-07-16 18:36:06 +02009235
Willy Tarreaue9656522010-08-17 15:40:09 +02009236 They have the same meaning as their counter-parts in "tcp-request connection"
9237 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009238
Willy Tarreauf3338342014-01-28 21:40:28 +01009239 While there is nothing mandatory about it, it is recommended to use the
9240 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9241 content" rules in the frontend, and track-sc2 for "tcp-request content"
9242 rules in the backend, because that makes the configuration more readable
9243 and easier to troubleshoot, but this is just a guideline and all counters
9244 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009245
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009246 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009247 the action, it is simply performed unconditionally. That can be useful for
9248 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009249
Willy Tarreaue9656522010-08-17 15:40:09 +02009250 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009251 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9252 contents of a buffer before extracting the required data. If the buffered
9253 contents do not parse as a valid HTTP message, then the ACL does not match.
9254 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009255 processing, so there is no risk of parsing something differently. In an HTTP
9256 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9257 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009258
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009259 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009260 are present when the rule is processed. The rule processing engine is able to
9261 wait until the inspect delay expires when the data to be tracked is not yet
9262 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009263
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009264 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009265 declared inline. For "tcp-request session" rules, only session-level
9266 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009267
Daniel Schneller0b547052016-03-21 20:46:57 +01009268 <var-name> The name of the variable starts with an indication about
9269 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009270 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009271 "sess" : the variable is shared with the whole session
9272 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009273 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009274 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009275 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009276 "res" : the variable is shared only during response
9277 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009278 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009279 The name may only contain characters 'a-z', 'A-Z', '0-9',
9280 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009281
9282 <expr> Is a standard HAProxy expression formed by a sample-fetch
9283 followed by some converters.
9284
Christopher Faulet85d79c92016-11-09 16:54:56 +01009285 The "unset-var" is used to unset a variable. See above for details about
9286 <var-name>.
9287
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009288 Example:
9289
9290 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009291 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009292
Willy Tarreau62644772008-07-16 18:36:06 +02009293 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009294 # Accept HTTP requests containing a Host header saying "example.com"
9295 # and reject everything else.
9296 acl is_host_com hdr(Host) -i example.com
9297 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009298 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009299 tcp-request content reject
9300
9301 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009302 # reject SMTP connection if client speaks first
9303 tcp-request inspect-delay 30s
9304 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009305 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009306
9307 # Forward HTTPS connection only if client speaks
9308 tcp-request inspect-delay 30s
9309 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009310 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009311 tcp-request content reject
9312
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009313 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009314 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009315 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009316 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009317 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9318 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009319
9320 Example:
9321 # track request counts per "base" (concatenation of Host+URL)
9322 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009323 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009324
Willy Tarreaue9656522010-08-17 15:40:09 +02009325 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009326 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009327
9328 frontend http
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009329 # Use General Purpose Couter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009330 # protecting all our sites
9331 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009332 tcp-request connection track-sc0 src
9333 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009334 ...
9335 use_backend http_dynamic if { path_end .php }
9336
9337 backend http_dynamic
9338 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009339 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009340 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009341 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009342 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009343 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009344 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009345
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009346 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009347
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009348 See also : "tcp-request connection", "tcp-request session",
9349 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009350
9351
9352tcp-request inspect-delay <timeout>
9353 Set the maximum allowed time to wait for data during content inspection
9354 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009355 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009356 Arguments :
9357 <timeout> is the timeout value specified in milliseconds by default, but
9358 can be in any other unit if the number is suffixed by the unit,
9359 as explained at the top of this document.
9360
9361 People using haproxy primarily as a TCP relay are often worried about the
9362 risk of passing any type of protocol to a server without any analysis. In
9363 order to be able to analyze the request contents, we must first withhold
9364 the data then analyze them. This statement simply enables withholding of
9365 data for at most the specified amount of time.
9366
Willy Tarreaufb356202010-08-03 14:02:05 +02009367 TCP content inspection applies very early when a connection reaches a
9368 frontend, then very early when the connection is forwarded to a backend. This
9369 means that a connection may experience a first delay in the frontend and a
9370 second delay in the backend if both have tcp-request rules.
9371
Willy Tarreau62644772008-07-16 18:36:06 +02009372 Note that when performing content inspection, haproxy will evaluate the whole
9373 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009374 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009375 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009376 contents are definitive. If no delay is set, haproxy will not wait at all
9377 and will immediately apply a verdict based on the available information.
9378 Obviously this is unlikely to be very useful and might even be racy, so such
9379 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009380
9381 As soon as a rule matches, the request is released and continues as usual. If
9382 the timeout is reached and no rule matches, the default policy will be to let
9383 it pass through unaffected.
9384
9385 For most protocols, it is enough to set it to a few seconds, as most clients
9386 send the full request immediately upon connection. Add 3 or more seconds to
9387 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009388 to use large values, for instance to ensure that the client never talks
Willy Tarreau62644772008-07-16 18:36:06 +02009389 before the server (eg: SMTP), or to wait for a client to talk before passing
9390 data to the server (eg: SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009391 least the inspection delay, otherwise it will expire first. If the client
9392 closes the connection or if the buffer is full, the delay immediately expires
9393 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009394
Willy Tarreau55165fe2009-05-10 12:02:55 +02009395 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009396 "timeout client".
9397
9398
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009399tcp-response content <action> [{if | unless} <condition>]
9400 Perform an action on a session response depending on a layer 4-7 condition
9401 May be used in sections : defaults | frontend | listen | backend
9402 no | no | yes | yes
9403 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009404 <action> defines the action to perform if the condition applies. See
9405 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009406
9407 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9408
9409 Response contents can be analysed at an early stage of response processing
9410 called "TCP content inspection". During this stage, ACL-based rules are
9411 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009412 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9413 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009414
9415 Most often, these decisions will consider a protocol recognition or validity.
9416
9417 Content-based rules are evaluated in their exact declaration order. If no
9418 rule matches or if there is no rule, the default action is to accept the
9419 contents. There is no specific limit to the number of rules which may be
9420 inserted.
9421
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009422 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009423 - accept :
9424 accepts the response if the condition is true (when used with "if")
9425 or false (when used with "unless"). The first such rule executed ends
9426 the rules evaluation.
9427
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009428 - close :
9429 immediately closes the connection with the server if the condition is
9430 true (when used with "if"), or false (when used with "unless"). The
9431 first such rule executed ends the rules evaluation. The main purpose of
9432 this action is to force a connection to be finished between a client
9433 and a server after an exchange when the application protocol expects
9434 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009435 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009436 protocols.
9437
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009438 - reject :
9439 rejects the response if the condition is true (when used with "if")
9440 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009441 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009442
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009443 - set-var(<var-name>) <expr>
9444 Sets a variable.
9445
Christopher Faulet85d79c92016-11-09 16:54:56 +01009446 - unset-var(<var-name>)
9447 Unsets a variable.
9448
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009449 - sc-inc-gpc0(<sc-id>):
9450 This action increments the GPC0 counter according to the sticky
9451 counter designated by <sc-id>. If an error occurs, this action fails
9452 silently and the actions evaluation continues.
9453
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009454 - sc-set-gpt0(<sc-id>) <int> :
9455 This action sets the GPT0 tag according to the sticky counter designated
9456 by <sc-id> and the value of <int>. The expected result is a boolean. If
9457 an error occurs, this action silently fails and the actions evaluation
9458 continues.
9459
Willy Tarreau2d392c22015-08-24 01:43:45 +02009460 - "silent-drop" :
9461 This stops the evaluation of the rules and makes the client-facing
9462 connection suddenly disappear using a system-dependant way that tries
9463 to prevent the client from being notified. The effect it then that the
9464 client still sees an established connection while there's none on
9465 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9466 except that it doesn't use any local resource at all on the machine
9467 running HAProxy. It can resist much higher loads than "tarpit", and
9468 slow down stronger attackers. It is important to undestand the impact
9469 of using this mechanism. All stateful equipments placed between the
9470 client and HAProxy (firewalls, proxies, load balancers) will also keep
9471 the established connection for a long time and may suffer from this
9472 action. On modern Linux systems running with enough privileges, the
9473 TCP_REPAIR socket option is used to block the emission of a TCP
9474 reset. On other systems, the socket's TTL is reduced to 1 so that the
9475 TCP reset doesn't pass the first router, though it's still delivered to
9476 local networks. Do not use it unless you fully understand how it works.
9477
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009478 Note that the "if/unless" condition is optional. If no condition is set on
9479 the action, it is simply performed unconditionally. That can be useful for
9480 for changing the default action to a reject.
9481
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009482 It is perfectly possible to match layer 7 contents with "tcp-response
9483 content" rules, but then it is important to ensure that a full response has
9484 been buffered, otherwise no contents will match. In order to achieve this,
9485 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009486 period.
9487
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009488 The "set-var" is used to set the content of a variable. The variable is
9489 declared inline.
9490
Daniel Schneller0b547052016-03-21 20:46:57 +01009491 <var-name> The name of the variable starts with an indication about
9492 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009493 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009494 "sess" : the variable is shared with the whole session
9495 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009496 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009497 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009498 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009499 "res" : the variable is shared only during response
9500 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009501 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009502 The name may only contain characters 'a-z', 'A-Z', '0-9',
9503 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009504
9505 <expr> Is a standard HAProxy expression formed by a sample-fetch
9506 followed by some converters.
9507
9508 Example:
9509
9510 tcp-request content set-var(sess.my_var) src
9511
Christopher Faulet85d79c92016-11-09 16:54:56 +01009512 The "unset-var" is used to unset a variable. See above for details about
9513 <var-name>.
9514
9515 Example:
9516
9517 tcp-request content unset-var(sess.my_var)
9518
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009519 See section 7 about ACL usage.
9520
9521 See also : "tcp-request content", "tcp-response inspect-delay"
9522
9523
Willy Tarreau4f614292016-10-21 17:49:36 +02009524tcp-request session <action> [{if | unless} <condition>]
9525 Perform an action on a validated session depending on a layer 5 condition
9526 May be used in sections : defaults | frontend | listen | backend
9527 no | yes | yes | no
9528 Arguments :
9529 <action> defines the action to perform if the condition applies. See
9530 below.
9531
9532 <condition> is a standard layer5-only ACL-based condition (see section 7).
9533
9534 Once a session is validated, (ie. after all handshakes have been completed),
9535 it is possible to evaluate some conditions to decide whether this session
9536 must be accepted or dropped or have its counters tracked. Those conditions
9537 cannot make use of any data contents because no buffers are allocated yet and
9538 the processing cannot wait at this stage. The main use case it to copy some
9539 early information into variables (since variables are accessible in the
9540 session), or to keep track of some information collected after the handshake,
9541 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
9542 from the PROXY protocol header (eg: track a source forwarded this way). The
9543 extracted information can thus be copied to a variable or tracked using
9544 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9545 with other rulesets. Most operations performed here could also be performed
9546 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9547 for each new request, and that might not always be acceptable. For example a
9548 rule might increment a counter on each evaluation. It would also be possible
9549 that a country is resolved by geolocation from the source IP address,
9550 assigned to a session-wide variable, then the source address rewritten from
9551 an HTTP header for all requests. If some contents need to be inspected in
9552 order to take the decision, the "tcp-request content" statements must be used
9553 instead.
9554
9555 The "tcp-request session" rules are evaluated in their exact declaration
9556 order. If no rule matches or if there is no rule, the default action is to
9557 accept the incoming session. There is no specific limit to the number of
9558 rules which may be inserted.
9559
9560 Several types of actions are supported :
9561 - accept : the request is accepted
9562 - reject : the request is rejected and the connection is closed
9563 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9564 - sc-inc-gpc0(<sc-id>)
9565 - sc-set-gpt0(<sc-id>) <int>
9566 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009567 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009568 - silent-drop
9569
9570 These actions have the same meaning as their respective counter-parts in
9571 "tcp-request connection" and "tcp-request content", so please refer to these
9572 sections for a complete description.
9573
9574 Note that the "if/unless" condition is optional. If no condition is set on
9575 the action, it is simply performed unconditionally. That can be useful for
9576 "track-sc*" actions as well as for changing the default action to a reject.
9577
9578 Example: track the original source address by default, or the one advertised
9579 in the PROXY protocol header for connection coming from the local
9580 proxies. The first connection-level rule enables receipt of the
9581 PROXY protocol for these ones, the second rule tracks whatever
9582 address we decide to keep after optional decoding.
9583
9584 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9585 tcp-request session track-sc0 src
9586
9587 Example: accept all sessions from white-listed hosts, reject too fast
9588 sessions without counting them, and track accepted sessions.
9589 This results in session rate being capped from abusive sources.
9590
9591 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9592 tcp-request session reject if { src_sess_rate gt 10 }
9593 tcp-request session track-sc0 src
9594
9595 Example: accept all sessions from white-listed hosts, count all other
9596 sessions and reject too fast ones. This results in abusive ones
9597 being blocked as long as they don't slow down.
9598
9599 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9600 tcp-request session track-sc0 src
9601 tcp-request session reject if { sc0_sess_rate gt 10 }
9602
9603 See section 7 about ACL usage.
9604
9605 See also : "tcp-request connection", "tcp-request content", "stick-table"
9606
9607
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009608tcp-response inspect-delay <timeout>
9609 Set the maximum allowed time to wait for a response during content inspection
9610 May be used in sections : defaults | frontend | listen | backend
9611 no | no | yes | yes
9612 Arguments :
9613 <timeout> is the timeout value specified in milliseconds by default, but
9614 can be in any other unit if the number is suffixed by the unit,
9615 as explained at the top of this document.
9616
9617 See also : "tcp-response content", "tcp-request inspect-delay".
9618
9619
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009620timeout check <timeout>
9621 Set additional check timeout, but only after a connection has been already
9622 established.
9623
9624 May be used in sections: defaults | frontend | listen | backend
9625 yes | no | yes | yes
9626 Arguments:
9627 <timeout> is the timeout value specified in milliseconds by default, but
9628 can be in any other unit if the number is suffixed by the unit,
9629 as explained at the top of this document.
9630
9631 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
9632 for check and "timeout check" as an additional read timeout. The "min" is
9633 used so that people running with *very* long "timeout connect" (eg. those
9634 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01009635 (Please also note that there is no valid reason to have such long connect
9636 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
9637 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009638
9639 If "timeout check" is not set haproxy uses "inter" for complete check
9640 timeout (connect + read) exactly like all <1.3.15 version.
9641
9642 In most cases check request is much simpler and faster to handle than normal
9643 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01009644 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009645
9646 This parameter is specific to backends, but can be specified once for all in
9647 "defaults" sections. This is in fact one of the easiest solutions not to
9648 forget about it.
9649
Willy Tarreau41a340d2008-01-22 12:25:31 +01009650 See also: "timeout connect", "timeout queue", "timeout server",
9651 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009652
9653
Willy Tarreau0ba27502007-12-24 16:55:16 +01009654timeout client <timeout>
9655timeout clitimeout <timeout> (deprecated)
9656 Set the maximum inactivity time on the client side.
9657 May be used in sections : defaults | frontend | listen | backend
9658 yes | yes | yes | no
9659 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009660 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009661 can be in any other unit if the number is suffixed by the unit,
9662 as explained at the top of this document.
9663
9664 The inactivity timeout applies when the client is expected to acknowledge or
9665 send data. In HTTP mode, this timeout is particularly important to consider
9666 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +01009667 response while it is reading data sent by the server. That said, for the
9668 first phase, it is preferable to set the "timeout http-request" to better
9669 protect HAProxy from Slowloris like attacks. The value is specified in
9670 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +01009671 suffixed by the unit, as specified at the top of this document. In TCP mode
9672 (and to a lesser extent, in HTTP mode), it is highly recommended that the
9673 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009674 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01009675 losses by specifying timeouts that are slightly above multiples of 3 seconds
Willy Tarreauce887fd2012-05-12 12:50:00 +02009676 (eg: 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
9677 sessions (eg: WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02009678 which overrides "timeout client" and "timeout server" for tunnels, as well as
9679 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009680
9681 This parameter is specific to frontends, but can be specified once for all in
9682 "defaults" sections. This is in fact one of the easiest solutions not to
9683 forget about it. An unspecified timeout results in an infinite timeout, which
9684 is not recommended. Such a usage is accepted and works but reports a warning
9685 during startup because it may results in accumulation of expired sessions in
9686 the system if the system's timeouts are not configured either.
9687
9688 This parameter replaces the old, deprecated "clitimeout". It is recommended
9689 to use it to write new configurations. The form "timeout clitimeout" is
9690 provided only by backwards compatibility but its use is strongly discouraged.
9691
Baptiste Assmann2e1941e2016-03-06 23:24:12 +01009692 See also : "clitimeout", "timeout server", "timeout tunnel",
9693 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009694
9695
Willy Tarreau05cdd962014-05-10 14:30:07 +02009696timeout client-fin <timeout>
9697 Set the inactivity timeout on the client side for half-closed connections.
9698 May be used in sections : defaults | frontend | listen | backend
9699 yes | yes | yes | no
9700 Arguments :
9701 <timeout> is the timeout value specified in milliseconds by default, but
9702 can be in any other unit if the number is suffixed by the unit,
9703 as explained at the top of this document.
9704
9705 The inactivity timeout applies when the client is expected to acknowledge or
9706 send data while one direction is already shut down. This timeout is different
9707 from "timeout client" in that it only applies to connections which are closed
9708 in one direction. This is particularly useful to avoid keeping connections in
9709 FIN_WAIT state for too long when clients do not disconnect cleanly. This
9710 problem is particularly common long connections such as RDP or WebSocket.
9711 Note that this timeout can override "timeout tunnel" when a connection shuts
9712 down in one direction.
9713
9714 This parameter is specific to frontends, but can be specified once for all in
9715 "defaults" sections. By default it is not set, so half-closed connections
9716 will use the other timeouts (timeout.client or timeout.tunnel).
9717
9718 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
9719
9720
Willy Tarreau0ba27502007-12-24 16:55:16 +01009721timeout connect <timeout>
9722timeout contimeout <timeout> (deprecated)
9723 Set the maximum time to wait for a connection attempt to a server to succeed.
9724 May be used in sections : defaults | frontend | listen | backend
9725 yes | no | yes | yes
9726 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009727 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009728 can be in any other unit if the number is suffixed by the unit,
9729 as explained at the top of this document.
9730
9731 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009732 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01009733 cover one or several TCP packet losses by specifying timeouts that are
Willy Tarreau0ba27502007-12-24 16:55:16 +01009734 slightly above multiples of 3 seconds (eg: 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009735 connect timeout also presets both queue and tarpit timeouts to the same value
9736 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009737
9738 This parameter is specific to backends, but can be specified once for all in
9739 "defaults" sections. This is in fact one of the easiest solutions not to
9740 forget about it. An unspecified timeout results in an infinite timeout, which
9741 is not recommended. Such a usage is accepted and works but reports a warning
9742 during startup because it may results in accumulation of failed sessions in
9743 the system if the system's timeouts are not configured either.
9744
9745 This parameter replaces the old, deprecated "contimeout". It is recommended
9746 to use it to write new configurations. The form "timeout contimeout" is
9747 provided only by backwards compatibility but its use is strongly discouraged.
9748
Willy Tarreau41a340d2008-01-22 12:25:31 +01009749 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
9750 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009751
9752
Willy Tarreaub16a5742010-01-10 14:46:16 +01009753timeout http-keep-alive <timeout>
9754 Set the maximum allowed time to wait for a new HTTP request to appear
9755 May be used in sections : defaults | frontend | listen | backend
9756 yes | yes | yes | yes
9757 Arguments :
9758 <timeout> is the timeout value specified in milliseconds by default, but
9759 can be in any other unit if the number is suffixed by the unit,
9760 as explained at the top of this document.
9761
9762 By default, the time to wait for a new request in case of keep-alive is set
9763 by "timeout http-request". However this is not always convenient because some
9764 people want very short keep-alive timeouts in order to release connections
9765 faster, and others prefer to have larger ones but still have short timeouts
9766 once the request has started to present itself.
9767
9768 The "http-keep-alive" timeout covers these needs. It will define how long to
9769 wait for a new HTTP request to start coming after a response was sent. Once
9770 the first byte of request has been seen, the "http-request" timeout is used
9771 to wait for the complete request to come. Note that empty lines prior to a
9772 new request do not refresh the timeout and are not counted as a new request.
9773
9774 There is also another difference between the two timeouts : when a connection
9775 expires during timeout http-keep-alive, no error is returned, the connection
9776 just closes. If the connection expires in "http-request" while waiting for a
9777 connection to complete, a HTTP 408 error is returned.
9778
9779 In general it is optimal to set this value to a few tens to hundreds of
9780 milliseconds, to allow users to fetch all objects of a page at once but
9781 without waiting for further clicks. Also, if set to a very small value (eg:
9782 1 millisecond) it will probably only accept pipelined requests but not the
9783 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +02009784 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +01009785
9786 If this parameter is not set, the "http-request" timeout applies, and if both
9787 are not set, "timeout client" still applies at the lower level. It should be
9788 set in the frontend to take effect, unless the frontend is in TCP mode, in
9789 which case the HTTP backend's timeout will be used.
9790
9791 See also : "timeout http-request", "timeout client".
9792
9793
Willy Tarreau036fae02008-01-06 13:24:40 +01009794timeout http-request <timeout>
9795 Set the maximum allowed time to wait for a complete HTTP request
9796 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +02009797 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +01009798 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009799 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +01009800 can be in any other unit if the number is suffixed by the unit,
9801 as explained at the top of this document.
9802
9803 In order to offer DoS protection, it may be required to lower the maximum
9804 accepted time to receive a complete HTTP request without affecting the client
9805 timeout. This helps protecting against established connections on which
9806 nothing is sent. The client timeout cannot offer a good protection against
9807 this abuse because it is an inactivity timeout, which means that if the
9808 attacker sends one character every now and then, the timeout will not
9809 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +02009810 types, the request will be aborted if it does not complete in time. When the
9811 timeout expires, an HTTP 408 response is sent to the client to inform it
9812 about the problem, and the connection is closed. The logs will report
9813 termination codes "cR". Some recent browsers are having problems with this
9814 standard, well-documented behaviour, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +02009815 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
9816 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +01009817
Baptiste Assmanneccdf432015-10-28 13:49:01 +01009818 By default, this timeout only applies to the header part of the request,
9819 and not to any data. As soon as the empty line is received, this timeout is
9820 not used anymore. When combined with "option http-buffer-request", this
9821 timeout also applies to the body of the request..
9822 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +01009823 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +01009824
9825 Generally it is enough to set it to a few seconds, as most clients send the
9826 full request immediately upon connection. Add 3 or more seconds to cover TCP
9827 retransmits but that's all. Setting it to very low values (eg: 50 ms) will
9828 generally work on local networks as long as there are no packet losses. This
9829 will prevent people from sending bare HTTP requests using telnet.
9830
9831 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +02009832 chunk of the incoming request. It should be set in the frontend to take
9833 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
9834 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +01009835
Willy Tarreau0f228a02015-05-01 15:37:53 +02009836 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +01009837 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +01009838
Willy Tarreau844e3c52008-01-11 16:28:18 +01009839
9840timeout queue <timeout>
9841 Set the maximum time to wait in the queue for a connection slot to be free
9842 May be used in sections : defaults | frontend | listen | backend
9843 yes | no | yes | yes
9844 Arguments :
9845 <timeout> is the timeout value specified in milliseconds by default, but
9846 can be in any other unit if the number is suffixed by the unit,
9847 as explained at the top of this document.
9848
9849 When a server's maxconn is reached, connections are left pending in a queue
9850 which may be server-specific or global to the backend. In order not to wait
9851 indefinitely, a timeout is applied to requests pending in the queue. If the
9852 timeout is reached, it is considered that the request will almost never be
9853 served, so it is dropped and a 503 error is returned to the client.
9854
9855 The "timeout queue" statement allows to fix the maximum time for a request to
9856 be left pending in a queue. If unspecified, the same value as the backend's
9857 connection timeout ("timeout connect") is used, for backwards compatibility
9858 with older versions with no "timeout queue" parameter.
9859
9860 See also : "timeout connect", "contimeout".
9861
9862
9863timeout server <timeout>
9864timeout srvtimeout <timeout> (deprecated)
9865 Set the maximum inactivity time on the server side.
9866 May be used in sections : defaults | frontend | listen | backend
9867 yes | no | yes | yes
9868 Arguments :
9869 <timeout> is the timeout value specified in milliseconds by default, but
9870 can be in any other unit if the number is suffixed by the unit,
9871 as explained at the top of this document.
9872
9873 The inactivity timeout applies when the server is expected to acknowledge or
9874 send data. In HTTP mode, this timeout is particularly important to consider
9875 during the first phase of the server's response, when it has to send the
9876 headers, as it directly represents the server's processing time for the
9877 request. To find out what value to put there, it's often good to start with
9878 what would be considered as unacceptable response times, then check the logs
9879 to observe the response time distribution, and adjust the value accordingly.
9880
9881 The value is specified in milliseconds by default, but can be in any other
9882 unit if the number is suffixed by the unit, as specified at the top of this
9883 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
9884 recommended that the client timeout remains equal to the server timeout in
9885 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009886 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01009887 packet losses by specifying timeouts that are slightly above multiples of 3
Willy Tarreauce887fd2012-05-12 12:50:00 +02009888 seconds (eg: 4 or 5 seconds minimum). If some long-lived sessions are mixed
9889 with short-lived sessions (eg: WebSocket and HTTP), it's worth considering
9890 "timeout tunnel", which overrides "timeout client" and "timeout server" for
9891 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009892
9893 This parameter is specific to backends, but can be specified once for all in
9894 "defaults" sections. This is in fact one of the easiest solutions not to
9895 forget about it. An unspecified timeout results in an infinite timeout, which
9896 is not recommended. Such a usage is accepted and works but reports a warning
9897 during startup because it may results in accumulation of expired sessions in
9898 the system if the system's timeouts are not configured either.
9899
9900 This parameter replaces the old, deprecated "srvtimeout". It is recommended
9901 to use it to write new configurations. The form "timeout srvtimeout" is
9902 provided only by backwards compatibility but its use is strongly discouraged.
9903
Willy Tarreauce887fd2012-05-12 12:50:00 +02009904 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +01009905
Willy Tarreau05cdd962014-05-10 14:30:07 +02009906
9907timeout server-fin <timeout>
9908 Set the inactivity timeout on the server side for half-closed connections.
9909 May be used in sections : defaults | frontend | listen | backend
9910 yes | no | yes | yes
9911 Arguments :
9912 <timeout> is the timeout value specified in milliseconds by default, but
9913 can be in any other unit if the number is suffixed by the unit,
9914 as explained at the top of this document.
9915
9916 The inactivity timeout applies when the server is expected to acknowledge or
9917 send data while one direction is already shut down. This timeout is different
9918 from "timeout server" in that it only applies to connections which are closed
9919 in one direction. This is particularly useful to avoid keeping connections in
9920 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
9921 This problem is particularly common long connections such as RDP or WebSocket.
9922 Note that this timeout can override "timeout tunnel" when a connection shuts
9923 down in one direction. This setting was provided for completeness, but in most
9924 situations, it should not be needed.
9925
9926 This parameter is specific to backends, but can be specified once for all in
9927 "defaults" sections. By default it is not set, so half-closed connections
9928 will use the other timeouts (timeout.server or timeout.tunnel).
9929
9930 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
9931
Willy Tarreau844e3c52008-01-11 16:28:18 +01009932
9933timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +01009934 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +01009935 May be used in sections : defaults | frontend | listen | backend
9936 yes | yes | yes | yes
9937 Arguments :
9938 <timeout> is the tarpit duration specified in milliseconds by default, but
9939 can be in any other unit if the number is suffixed by the unit,
9940 as explained at the top of this document.
9941
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009942 When a connection is tarpitted using "http-request tarpit" or
9943 "reqtarpit", it is maintained open with no activity for a certain
9944 amount of time, then closed. "timeout tarpit" defines how long it will
9945 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009946
9947 The value is specified in milliseconds by default, but can be in any other
9948 unit if the number is suffixed by the unit, as specified at the top of this
9949 document. If unspecified, the same value as the backend's connection timeout
9950 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +01009951 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009952
9953 See also : "timeout connect", "contimeout".
9954
9955
Willy Tarreauce887fd2012-05-12 12:50:00 +02009956timeout tunnel <timeout>
9957 Set the maximum inactivity time on the client and server side for tunnels.
9958 May be used in sections : defaults | frontend | listen | backend
9959 yes | no | yes | yes
9960 Arguments :
9961 <timeout> is the timeout value specified in milliseconds by default, but
9962 can be in any other unit if the number is suffixed by the unit,
9963 as explained at the top of this document.
9964
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009965 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +02009966 between a client and a server, and the connection remains inactive in both
9967 directions. This timeout supersedes both the client and server timeouts once
9968 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
9969 analyser remains attached to either connection (eg: tcp content rules are
9970 accepted). In HTTP, this timeout is used when a connection is upgraded (eg:
9971 when switching to the WebSocket protocol, or forwarding a CONNECT request
9972 to a proxy), or after the first response when no keepalive/close option is
9973 specified.
9974
Willy Tarreau05cdd962014-05-10 14:30:07 +02009975 Since this timeout is usually used in conjunction with long-lived connections,
9976 it usually is a good idea to also set "timeout client-fin" to handle the
9977 situation where a client suddenly disappears from the net and does not
9978 acknowledge a close, or sends a shutdown and does not acknowledge pending
9979 data anymore. This can happen in lossy networks where firewalls are present,
9980 and is detected by the presence of large amounts of sessions in a FIN_WAIT
9981 state.
9982
Willy Tarreauce887fd2012-05-12 12:50:00 +02009983 The value is specified in milliseconds by default, but can be in any other
9984 unit if the number is suffixed by the unit, as specified at the top of this
9985 document. Whatever the expected normal idle time, it is a good practice to
9986 cover at least one or several TCP packet losses by specifying timeouts that
9987 are slightly above multiples of 3 seconds (eg: 4 or 5 seconds minimum).
9988
9989 This parameter is specific to backends, but can be specified once for all in
9990 "defaults" sections. This is in fact one of the easiest solutions not to
9991 forget about it.
9992
9993 Example :
9994 defaults http
9995 option http-server-close
9996 timeout connect 5s
9997 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +02009998 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +02009999 timeout server 30s
10000 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10001
Willy Tarreau05cdd962014-05-10 14:30:07 +020010002 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010003
10004
Willy Tarreau844e3c52008-01-11 16:28:18 +010010005transparent (deprecated)
10006 Enable client-side transparent proxying
10007 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010008 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010009 Arguments : none
10010
10011 This keyword was introduced in order to provide layer 7 persistence to layer
10012 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10013 connection for a remote address to a local process (here HAProxy), and let
10014 this process know what address was initially requested. When this option is
10015 used, sessions without cookies will be forwarded to the original destination
10016 IP address of the incoming request (which should match that of another
10017 equipment), while requests with cookies will still be forwarded to the
10018 appropriate server.
10019
10020 The "transparent" keyword is deprecated, use "option transparent" instead.
10021
10022 Note that contrary to a common belief, this option does NOT make HAProxy
10023 present the client's IP to the server when establishing the connection.
10024
Willy Tarreau844e3c52008-01-11 16:28:18 +010010025 See also: "option transparent"
10026
William Lallemanda73203e2012-03-12 12:48:57 +010010027unique-id-format <string>
10028 Generate a unique ID for each request.
10029 May be used in sections : defaults | frontend | listen | backend
10030 yes | yes | yes | no
10031 Arguments :
10032 <string> is a log-format string.
10033
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010034 This keyword creates a ID for each request using the custom log format. A
10035 unique ID is useful to trace a request passing through many components of
10036 a complex infrastructure. The newly created ID may also be logged using the
10037 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010038
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010039 The format should be composed from elements that are guaranteed to be
10040 unique when combined together. For instance, if multiple haproxy instances
10041 are involved, it might be important to include the node name. It is often
10042 needed to log the incoming connection's source and destination addresses
10043 and ports. Note that since multiple requests may be performed over the same
10044 connection, including a request counter may help differentiate them.
10045 Similarly, a timestamp may protect against a rollover of the counter.
10046 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010047
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010048 It is recommended to use hexadecimal notation for many fields since it
10049 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010050
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010051 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010052
Julien Vehentf21be322014-03-07 08:27:34 -050010053 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010054
10055 will generate:
10056
10057 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10058
10059 See also: "unique-id-header"
10060
10061unique-id-header <name>
10062 Add a unique ID header in the HTTP request.
10063 May be used in sections : defaults | frontend | listen | backend
10064 yes | yes | yes | no
10065 Arguments :
10066 <name> is the name of the header.
10067
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010068 Add a unique-id header in the HTTP request sent to the server, using the
10069 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010070
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010071 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010072
Julien Vehentf21be322014-03-07 08:27:34 -050010073 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010074 unique-id-header X-Unique-ID
10075
10076 will generate:
10077
10078 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10079
10080 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010081
Willy Tarreauf51658d2014-04-23 01:21:56 +020010082use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010083 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010084 May be used in sections : defaults | frontend | listen | backend
10085 no | yes | yes | no
10086 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010087 <backend> is the name of a valid backend or "listen" section, or a
10088 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010089
Willy Tarreauf51658d2014-04-23 01:21:56 +020010090 <condition> is a condition composed of ACLs, as described in section 7. If
10091 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010092
10093 When doing content-switching, connections arrive on a frontend and are then
10094 dispatched to various backends depending on a number of conditions. The
10095 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010096 "use_backend" keyword. While it is normally used with HTTP processing, it can
10097 also be used in pure TCP, either without content using stateless ACLs (eg:
10098 source address validation) or combined with a "tcp-request" rule to wait for
10099 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010100
10101 There may be as many "use_backend" rules as desired. All of these rules are
10102 evaluated in their declaration order, and the first one which matches will
10103 assign the backend.
10104
10105 In the first form, the backend will be used if the condition is met. In the
10106 second form, the backend will be used if the condition is not met. If no
10107 condition is valid, the backend defined with "default_backend" will be used.
10108 If no default backend is defined, either the servers in the same section are
10109 used (in case of a "listen" section) or, in case of a frontend, no server is
10110 used and a 503 service unavailable response is returned.
10111
Willy Tarreau51aecc72009-07-12 09:47:04 +020010112 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010113 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010114 and backend processing will immediately follow, or the backend will wait for
10115 a complete HTTP request to get in. This feature is useful when a frontend
10116 must decode several protocols on a unique port, one of them being HTTP.
10117
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010118 When <backend> is a simple name, it is resolved at configuration time, and an
10119 error is reported if the specified backend does not exist. If <backend> is
10120 a log-format string instead, no check may be done at configuration time, so
10121 the backend name is resolved dynamically at run time. If the resulting
10122 backend name does not correspond to any valid backend, no other rule is
10123 evaluated, and the default_backend directive is applied instead. Note that
10124 when using dynamic backend names, it is highly recommended to use a prefix
10125 that no other backend uses in order to ensure that an unauthorized backend
10126 cannot be forced from the request.
10127
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010128 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010129 used to detect the association between frontends and backends to compute the
10130 backend's "fullconn" setting. This cannot be done for dynamic names.
10131
10132 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10133 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010134
Willy Tarreau036fae02008-01-06 13:24:40 +010010135
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010136use-server <server> if <condition>
10137use-server <server> unless <condition>
10138 Only use a specific server if/unless an ACL-based condition is matched.
10139 May be used in sections : defaults | frontend | listen | backend
10140 no | no | yes | yes
10141 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010142 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010143
10144 <condition> is a condition composed of ACLs, as described in section 7.
10145
10146 By default, connections which arrive to a backend are load-balanced across
10147 the available servers according to the configured algorithm, unless a
10148 persistence mechanism such as a cookie is used and found in the request.
10149
10150 Sometimes it is desirable to forward a particular request to a specific
10151 server without having to declare a dedicated backend for this server. This
10152 can be achieved using the "use-server" rules. These rules are evaluated after
10153 the "redirect" rules and before evaluating cookies, and they have precedence
10154 on them. There may be as many "use-server" rules as desired. All of these
10155 rules are evaluated in their declaration order, and the first one which
10156 matches will assign the server.
10157
10158 If a rule designates a server which is down, and "option persist" is not used
10159 and no force-persist rule was validated, it is ignored and evaluation goes on
10160 with the next rules until one matches.
10161
10162 In the first form, the server will be used if the condition is met. In the
10163 second form, the server will be used if the condition is not met. If no
10164 condition is valid, the processing continues and the server will be assigned
10165 according to other persistence mechanisms.
10166
10167 Note that even if a rule is matched, cookie processing is still performed but
10168 does not assign the server. This allows prefixed cookies to have their prefix
10169 stripped.
10170
10171 The "use-server" statement works both in HTTP and TCP mode. This makes it
10172 suitable for use with content-based inspection. For instance, a server could
10173 be selected in a farm according to the TLS SNI field. And if these servers
10174 have their weight set to zero, they will not be used for other traffic.
10175
10176 Example :
10177 # intercept incoming TLS requests based on the SNI field
10178 use-server www if { req_ssl_sni -i www.example.com }
10179 server www 192.168.0.1:443 weight 0
10180 use-server mail if { req_ssl_sni -i mail.example.com }
10181 server mail 192.168.0.1:587 weight 0
10182 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010183 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010184 # all the rest is forwarded to this server
10185 server default 192.168.0.2:443 check
10186
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010187 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010188
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010189
101905. Bind and Server options
10191--------------------------
10192
10193The "bind", "server" and "default-server" keywords support a number of settings
10194depending on some build options and on the system HAProxy was built on. These
10195settings generally each consist in one word sometimes followed by a value,
10196written on the same line as the "bind" or "server" line. All these options are
10197described in this section.
10198
10199
102005.1. Bind options
10201-----------------
10202
10203The "bind" keyword supports a certain number of settings which are all passed
10204as arguments on the same line. The order in which those arguments appear makes
10205no importance, provided that they appear after the bind address. All of these
10206parameters are optional. Some of them consist in a single words (booleans),
10207while other ones expect a value after them. In this case, the value must be
10208provided immediately after the setting name.
10209
10210The currently supported settings are the following ones.
10211
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010212accept-netscaler-cip <magic number>
10213 Enforces the use of the NetScaler Client IP insertion protocol over any
10214 connection accepted by any of the TCP sockets declared on the same line. The
10215 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10216 the incoming connection to be used everywhere an address is used, with the
10217 only exception of "tcp-request connection" rules which will only see the
10218 real connection address. Logs will reflect the addresses indicated in the
10219 protocol, unless it is violated, in which case the real address will still
10220 be used. This keyword combined with support from external components can be
10221 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010222 mechanism which is not always reliable and not even always usable. See also
10223 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10224 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010225
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010226accept-proxy
10227 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010228 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10229 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010230 3/4 addresses of the incoming connection to be used everywhere an address is
10231 used, with the only exception of "tcp-request connection" rules which will
10232 only see the real connection address. Logs will reflect the addresses
10233 indicated in the protocol, unless it is violated, in which case the real
10234 address will still be used. This keyword combined with support from external
10235 components can be used as an efficient and reliable alternative to the
10236 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010237 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10238 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010239
Willy Tarreauab861d32013-04-02 02:30:41 +020010240alpn <protocols>
10241 This enables the TLS ALPN extension and advertises the specified protocol
10242 list as supported on top of ALPN. The protocol list consists in a comma-
10243 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10244 quotes). This requires that the SSL library is build with support for TLS
10245 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
10246 initial NPN extension.
10247
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010248backlog <backlog>
10249 Sets the socket's backlog to this value. If unspecified, the frontend's
10250 backlog is used instead, which generally defaults to the maxconn value.
10251
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010252curves <curves>
10253 This setting is only available when support for OpenSSL was built in. It sets
10254 the string describing the list of elliptic curves algorithms ("curve suite")
10255 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10256 string is a colon-delimited list of curve name.
10257 Example: "X25519:P-256" (without quote)
10258 When "curves" is set, "ecdhe" parameter is ignored.
10259
Emeric Brun7fb34422012-09-28 15:26:15 +020010260ecdhe <named curve>
10261 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010262 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10263 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010264
Emeric Brunfd33a262012-10-11 16:28:27 +020010265ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010266 This setting is only available when support for OpenSSL was built in. It
10267 designates a PEM file from which to load CA certificates used to verify
10268 client's certificate.
10269
Emeric Brunb6dc9342012-09-28 17:55:37 +020010270ca-ignore-err [all|<errorID>,...]
10271 This setting is only available when support for OpenSSL was built in.
10272 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10273 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10274 error is ignored.
10275
Christopher Faulet31af49d2015-06-09 17:29:50 +020010276ca-sign-file <cafile>
10277 This setting is only available when support for OpenSSL was built in. It
10278 designates a PEM file containing both the CA certificate and the CA private
10279 key used to create and sign server's certificates. This is a mandatory
10280 setting when the dynamic generation of certificates is enabled. See
10281 'generate-certificates' for details.
10282
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010283ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010284 This setting is only available when support for OpenSSL was built in. It is
10285 the CA private key passphrase. This setting is optional and used only when
10286 the dynamic generation of certificates is enabled. See
10287 'generate-certificates' for details.
10288
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010289ciphers <ciphers>
10290 This setting is only available when support for OpenSSL was built in. It sets
10291 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010292 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010293 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
10294 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
Daniel Schneller87e43022017-09-01 19:29:57 +020010295 Depending on the compatiblity and security requirements, the list of suitable
10296 ciphers depends on a variety of variables. For background information and
10297 recommendations see e. g. (https://wiki.mozilla.org/Security/Server_Side_TLS)
10298 and (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010299
Emeric Brunfd33a262012-10-11 16:28:27 +020010300crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010301 This setting is only available when support for OpenSSL was built in. It
10302 designates a PEM file from which to load certificate revocation list used
10303 to verify client's certificate.
10304
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010305crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010306 This setting is only available when support for OpenSSL was built in. It
10307 designates a PEM file containing both the required certificates and any
10308 associated private keys. This file can be built by concatenating multiple
10309 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10310 requires an intermediate certificate, this can also be concatenated into this
10311 file.
10312
10313 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10314 are loaded.
10315
10316 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010317 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010318 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10319 specified multiple times in order to load certificates from multiple files or
10320 directories. The certificates will be presented to clients who provide a
10321 valid TLS Server Name Indication field matching one of their CN or alt
10322 subjects. Wildcards are supported, where a wildcard character '*' is used
10323 instead of the first hostname component (eg: *.example.org matches
10324 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010325
10326 If no SNI is provided by the client or if the SSL library does not support
10327 TLS extensions, or if the client provides an SNI hostname which does not
10328 match any certificate, then the first loaded certificate will be presented.
10329 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010330 recommended to load the default one first as a file or to ensure that it will
10331 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010332
Emeric Brune032bfa2012-09-28 13:01:45 +020010333 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010334
Alex Davies0fbf0162013-03-02 16:04:50 +000010335 Some CAs (such as Godaddy) offer a drop down list of server types that do not
10336 include HAProxy when obtaining a certificate. If this happens be sure to
Godbach8bf60a12014-04-21 21:42:41 +080010337 choose a webserver that the CA believes requires an intermediate CA (for
Alex Davies0fbf0162013-03-02 16:04:50 +000010338 Godaddy, selection Apache Tomcat will get the correct bundle, but many
10339 others, e.g. nginx, result in a wrong bundle that will not work for some
10340 clients).
10341
Emeric Brun4147b2e2014-06-16 18:36:30 +020010342 For each PEM file, haproxy checks for the presence of file at the same path
10343 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10344 Status Request extension (also known as "OCSP stapling") is automatically
10345 enabled. The content of this file is optional. If not empty, it must contain
10346 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10347 must comply with the following rules: it has to indicate a good status,
10348 it has to be a single response for the certificate of the PEM file, and it
10349 has to be valid at the moment of addition. If these rules are not respected
10350 the OCSP Response is ignored and a warning is emitted. In order to identify
10351 which certificate an OCSP Response applies to, the issuer's certificate is
10352 necessary. If the issuer's certificate is not found in the PEM file, it will
10353 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10354 if it exists otherwise it will fail with an error.
10355
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010356 For each PEM file, haproxy also checks for the presence of file at the same
10357 path suffixed by ".sctl". If such file is found, support for Certificate
10358 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10359 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10360 to check basic syntax, but no signatures are verified.
10361
yanbzhu6c25e9e2016-01-05 12:52:02 -050010362 There are cases where it is desirable to support multiple key types, e.g. RSA
10363 and ECDSA in the cipher suites offered to the clients. This allows clients
10364 that support EC certificates to be able to use EC ciphers, while
10365 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010366
10367 In order to provide this functionality, multiple PEM files, each with a
10368 different key type, are required. To associate these PEM files into a
10369 "cert bundle" that is recognized by haproxy, they must be named in the
10370 following way: All PEM files that are to be bundled must have the same base
10371 name, with a suffix indicating the key type. Currently, three suffixes are
10372 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10373 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10374 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10375 suffix matters. To load this bundle into haproxy, specify the base name only:
10376
10377 Example : bind :8443 ssl crt example.pem
10378
yanbzhu6c25e9e2016-01-05 12:52:02 -050010379 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010380 a cert bundle.
10381
10382 Haproxy will load all PEM files in the bundle at the same time to try to
10383 support multiple key types. PEM files are combined based on Common Name
10384 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10385 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10386 entries in the certificates in that bundle, haproxy will not be able to
10387 provide multi-cert support.
10388
10389 Assuming bundle in the example above contained the following:
10390
10391 Filename | CN | SAN
10392 -------------------+-----------------+-------------------
10393 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010394 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010395 example.pem.ecdsa | www.example.com | ecdsa.example.com
10396 -------------------+-----------------+-------------------
10397
10398 Users connecting with an SNI of "www.example.com" will be able
10399 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10400 "rsa.example.com" will only be able to use RSA cipher suites, and users
10401 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010402 suites. With BoringSSL multi-cert is natively supported, no need to bundle
10403 certificates. ECDSA certificate will be preferred if client support it.
yanbzhud19630c2015-12-14 15:10:25 -050010404
10405 If a directory name is given as the <cert> argument, haproxy will
10406 automatically search and load bundled files in that directory.
10407
10408 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10409 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10410 time, sctl is not supported in multi-certificate bundling.
10411
Emeric Brunb6dc9342012-09-28 17:55:37 +020010412crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010413 This setting is only available when support for OpenSSL was built in. Sets a
10414 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010415 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010416 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010417
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010418crt-list <file>
10419 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010420 designates a list of PEM file with an optional ssl configuration and a SNI
10421 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010422
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010423 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10424
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010425 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10426 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
10427 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010428 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010429
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010430 Wildcards are supported in the SNI filter. Negative filter are also supported,
10431 only useful in combination with a wildcard filter to exclude a particular SNI.
10432 The certificates will be presented to clients who provide a valid TLS Server
10433 Name Indication field matching one of the SNI filters. If no SNI filter is
10434 specified, the CN and alt subjects are used. This directive may be specified
10435 multiple times. See the "crt" option for more information. The default
10436 certificate is still needed to meet OpenSSL expectations. If it is not used,
10437 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010438
yanbzhu6c25e9e2016-01-05 12:52:02 -050010439 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010440 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010441 all bundled certificates. With BoringSSL multi-cert is natively supported,
10442 avoid multi-cert bundling. RSA and ECDSA certificates can be declared in a
10443 row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010444
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010445 crt-list file example:
10446 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010447 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010448 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010449 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010450
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010451defer-accept
10452 Is an optional keyword which is supported only on certain Linux kernels. It
10453 states that a connection will only be accepted once some data arrive on it,
10454 or at worst after the first retransmit. This should be used only on protocols
10455 for which the client talks first (eg: HTTP). It can slightly improve
10456 performance by ensuring that most of the request is already available when
10457 the connection is accepted. On the other hand, it will not be able to detect
10458 connections which don't talk. It is important to note that this option is
10459 broken in all kernels up to 2.6.31, as the connection is never accepted until
10460 the client talks. This can cause issues with front firewalls which would see
10461 an established connection while the proxy will only see it in SYN_RECV. This
10462 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10463
William Lallemandf6975e92017-05-26 17:42:10 +020010464expose-fd listeners
10465 This option is only usable with the stats socket. It gives your stats socket
10466 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010467 During a reload with the master-worker mode, the process is automatically
10468 reexecuted adding -x and one of the stats socket with this option.
William Lallemandf6975e92017-05-26 17:42:10 +020010469 See alors "-x" in the management guide.
10470
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010471force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010472 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010473 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010474 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010475 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010476
10477force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010478 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010479 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010480 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010481
10482force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010483 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010484 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010485 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010486
10487force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010488 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010489 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010490 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010491
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010492force-tlsv13
10493 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10494 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010495 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010496
Christopher Faulet31af49d2015-06-09 17:29:50 +020010497generate-certificates
10498 This setting is only available when support for OpenSSL was built in. It
10499 enables the dynamic SSL certificates generation. A CA certificate and its
10500 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10501 a transparent forward proxy, SSL requests generate errors because of a common
10502 name mismatch on the certificate presented to the client. With this option
10503 enabled, HAProxy will try to forge a certificate using the SNI hostname
10504 indicated by the client. This is done only if no certificate matches the SNI
10505 hostname (see 'crt-list'). If an error occurs, the default certificate is
10506 used, else the 'strict-sni' option is set.
10507 It can also be used when HAProxy is configured as a reverse proxy to ease the
10508 deployment of an architecture with many backends.
10509
10510 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10511 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
10512 increases the HAProxy's memroy footprint to reduce latency when the same
10513 certificate is used many times.
10514
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010515gid <gid>
10516 Sets the group of the UNIX sockets to the designated system gid. It can also
10517 be set by default in the global section's "unix-bind" statement. Note that
10518 some platforms simply ignore this. This setting is equivalent to the "group"
10519 setting except that the group ID is used instead of its name. This setting is
10520 ignored by non UNIX sockets.
10521
10522group <group>
10523 Sets the group of the UNIX sockets to the designated system group. It can
10524 also be set by default in the global section's "unix-bind" statement. Note
10525 that some platforms simply ignore this. This setting is equivalent to the
10526 "gid" setting except that the group name is used instead of its gid. This
10527 setting is ignored by non UNIX sockets.
10528
10529id <id>
10530 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10531 sometimes it is more convenient to fix them to ease monitoring. This value
10532 must be strictly positive and unique within the listener/frontend. This
10533 option can only be used when defining only a single socket.
10534
10535interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010536 Restricts the socket to a specific interface. When specified, only packets
10537 received from that particular interface are processed by the socket. This is
10538 currently only supported on Linux. The interface must be a primary system
10539 interface, not an aliased interface. It is also possible to bind multiple
10540 frontends to the same address if they are bound to different interfaces. Note
10541 that binding to a network interface requires root privileges. This parameter
10542 is only compatible with TCPv4/TCPv6 sockets.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010543
Willy Tarreauabb175f2012-09-24 12:43:26 +020010544level <level>
10545 This setting is used with the stats sockets only to restrict the nature of
10546 the commands that can be issued on the socket. It is ignored by other
10547 sockets. <level> can be one of :
10548 - "user" is the least privileged level ; only non-sensitive stats can be
10549 read, and no change is allowed. It would make sense on systems where it
10550 is not easy to restrict access to the socket.
10551 - "operator" is the default level and fits most common uses. All data can
10552 be read, and only non-sensitive changes are permitted (eg: clear max
10553 counters).
10554 - "admin" should be used with care, as everything is permitted (eg: clear
10555 all counters).
10556
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010557severity-output <format>
10558 This setting is used with the stats sockets only to configure severity
10559 level output prepended to informational feedback messages. Severity
10560 level of messages can range between 0 and 7, conforming to syslog
10561 rfc5424. Valid and successful socket commands requesting data
10562 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10563 prepended. It is ignored by other sockets. <format> can be one of :
10564 - "none" (default) no severity level is prepended to feedback messages.
10565 - "number" severity level is prepended as a number.
10566 - "string" severity level is prepended as a string following the
10567 rfc5424 convention.
10568
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010569maxconn <maxconn>
10570 Limits the sockets to this number of concurrent connections. Extraneous
10571 connections will remain in the system's backlog until a connection is
10572 released. If unspecified, the limit will be the same as the frontend's
10573 maxconn. Note that in case of port ranges or multiple addresses, the same
10574 value will be applied to each socket. This setting enables different
10575 limitations on expensive sockets, for instance SSL entries which may easily
10576 eat all memory.
10577
10578mode <mode>
10579 Sets the octal mode used to define access permissions on the UNIX socket. It
10580 can also be set by default in the global section's "unix-bind" statement.
10581 Note that some platforms simply ignore this. This setting is ignored by non
10582 UNIX sockets.
10583
10584mss <maxseg>
10585 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10586 connections. This can be used to force a lower MSS for certain specific
10587 ports, for instance for connections passing through a VPN. Note that this
10588 relies on a kernel feature which is theoretically supported under Linux but
10589 was buggy in all versions prior to 2.6.28. It may or may not work on other
10590 operating systems. It may also not change the advertised value but change the
10591 effective size of outgoing segments. The commonly advertised value for TCPv4
10592 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
10593 positive, it will be used as the advertised MSS. If it is negative, it will
10594 indicate by how much to reduce the incoming connection's advertised MSS for
10595 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
10596
10597name <name>
10598 Sets an optional name for these sockets, which will be reported on the stats
10599 page.
10600
Willy Tarreaud72f0f32015-10-13 14:50:22 +020010601namespace <name>
10602 On Linux, it is possible to specify which network namespace a socket will
10603 belong to. This directive makes it possible to explicitly bind a listener to
10604 a namespace different from the default one. Please refer to your operating
10605 system's documentation to find more details about network namespaces.
10606
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010607nice <nice>
10608 Sets the 'niceness' of connections initiated from the socket. Value must be
10609 in the range -1024..1024 inclusive, and defaults to zero. Positive values
10610 means that such connections are more friendly to others and easily offer
10611 their place in the scheduler. On the opposite, negative values mean that
10612 connections want to run with a higher priority than others. The difference
10613 only happens under high loads when the system is close to saturation.
10614 Negative values are appropriate for low-latency or administration services,
10615 and high values are generally recommended for CPU intensive tasks such as SSL
10616 processing or bulk transfers which are less sensible to latency. For example,
10617 it may make sense to use a positive value for an SMTP socket and a negative
10618 one for an RDP socket.
10619
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010620no-ca-names
10621 This setting is only available when support for OpenSSL was built in. It
10622 prevents from send CA names in server hello message when ca-file is used.
10623
Emeric Brun9b3009b2012-10-05 11:55:06 +020010624no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010625 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010626 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010627 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010628 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010629 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
10630 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010631
Emeric Brun90ad8722012-10-02 14:00:59 +020010632no-tls-tickets
10633 This setting is only available when support for OpenSSL was built in. It
10634 disables the stateless session resumption (RFC 5077 TLS Ticket
10635 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010636 session resumption is more expensive in CPU usage. This option is also
10637 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020010638
Emeric Brun9b3009b2012-10-05 11:55:06 +020010639no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010640 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010641 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010642 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010643 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010644 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10645 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010646
Emeric Brun9b3009b2012-10-05 11:55:06 +020010647no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020010648 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010649 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010650 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010651 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010652 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10653 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020010654
Emeric Brun9b3009b2012-10-05 11:55:06 +020010655no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020010656 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010657 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010658 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010659 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010660 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10661 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020010662
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010663no-tlsv13
10664 This setting is only available when support for OpenSSL was built in. It
10665 disables support for TLSv1.3 on any sockets instantiated from the listener
10666 when SSL is supported. Note that SSLv2 is forced disabled in the code and
10667 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010668 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10669 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010670
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020010671npn <protocols>
10672 This enables the NPN TLS extension and advertises the specified protocol list
10673 as supported on top of NPN. The protocol list consists in a comma-delimited
10674 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
10675 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020010676 enabled (check with haproxy -vv). Note that the NPN extension has been
10677 replaced with the ALPN extension (see the "alpn" keyword).
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020010678
Lukas Tribus53ae85c2017-05-04 15:45:40 +000010679prefer-client-ciphers
10680 Use the client's preference when selecting the cipher suite, by default
10681 the server's preference is enforced. This option is also available on
10682 global statement "ssl-default-bind-options".
10683
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020010684process [ all | odd | even | <number 1-64>[-<number 1-64>] ]
10685 This restricts the list of processes on which this listener is allowed to
10686 run. It does not enforce any process but eliminates those which do not match.
10687 If the frontend uses a "bind-process" setting, the intersection between the
10688 two is applied. If in the end the listener is not allowed to run on any
10689 remaining process, a warning is emitted, and the listener will either run on
10690 the first process of the listener if a single process was specified, or on
10691 all of its processes if multiple processes were specified. For the unlikely
Willy Tarreauae302532014-05-07 19:22:24 +020010692 case where several ranges are needed, this directive may be repeated. The
10693 main purpose of this directive is to be used with the stats sockets and have
10694 one different socket per process. The second purpose is to have multiple bind
10695 lines sharing the same IP:port but not the same process in a listener, so
10696 that the system can distribute the incoming connections into multiple queues
10697 and allow a smoother inter-process load balancing. Currently Linux 3.9 and
10698 above is known for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020010699
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010700ssl
10701 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010702 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010703 certificate is necessary (see "crt" above). All contents in the buffers will
10704 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020010705 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
10706 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010707
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010708ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
10709 This option enforces use of <version> or lower on SSL connections instantiated
10710 from this listener. This option is also available on global statement
10711 "ssl-default-bind-options". See also "ssl-min-ver".
10712
10713ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
10714 This option enforces use of <version> or upper on SSL connections instantiated
10715 from this listener. This option is also available on global statement
10716 "ssl-default-bind-options". See also "ssl-max-ver".
10717
Emmanuel Hocdet65623372013-01-24 17:17:15 +010010718strict-sni
10719 This setting is only available when support for OpenSSL was built in. The
10720 SSL/TLS negotiation is allow only if the client provided an SNI which match
10721 a certificate. The default certificate is not used.
10722 See the "crt" option for more information.
10723
Willy Tarreau2af207a2015-02-04 00:45:58 +010010724tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010010725 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010010726 listening socket. This option is available on Linux since version 2.6.37. It
10727 allows haproxy to configure a timeout for sockets which contain data not
Tim Düsterhus4896c442016-11-29 02:15:19 +010010728 receiving an acknowledgement for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010010729 useful on long-lived connections experiencing long idle periods such as
10730 remote terminals or database connection pools, where the client and server
10731 timeouts must remain high to allow a long period of idle, but where it is
10732 important to detect that the client has disappeared in order to release all
10733 resources associated with its connection (and the server's session). The
10734 argument is a delay expressed in milliseconds by default. This only works
10735 for regular TCP connections, and is ignored for other protocols.
10736
Willy Tarreau1c862c52012-10-05 16:21:00 +020010737tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010010738 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020010739 enables TCP Fast Open on the listening socket, which means that clients which
10740 support this feature will be able to send a request and receive a response
10741 during the 3-way handshake starting from second connection, thus saving one
10742 round-trip after the first connection. This only makes sense with protocols
10743 that use high connection rates and where each round trip matters. This can
10744 possibly cause issues with many firewalls which do not accept data on SYN
10745 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020010746 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
10747 need to build HAProxy with USE_TFO=1 if your libc doesn't define
10748 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020010749
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010010750tls-ticket-keys <keyfile>
10751 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
10752 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
10753 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
10754 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
10755 used for decryption and the penultimate one for encryption. This enables easy
10756 key rotation by just appending new key to the file and reloading the process.
10757 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
10758 is compromised. It is also a good idea to keep the keys off any permanent
10759 storage such as hard drives (hint: use tmpfs and don't swap those files).
10760 Lifetime hint can be changed using tune.ssl.timeout.
10761
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010762transparent
10763 Is an optional keyword which is supported only on certain Linux kernels. It
10764 indicates that the addresses will be bound even if they do not belong to the
10765 local machine, and that packets targeting any of these addresses will be
10766 intercepted just as if the addresses were locally configured. This normally
10767 requires that IP forwarding is enabled. Caution! do not use this with the
10768 default address '*', as it would redirect any traffic for the specified port.
10769 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
10770 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
10771 kernel version. Some distribution kernels include backports of the feature,
10772 so check for support with your vendor.
10773
Willy Tarreau77e3af92012-11-24 15:07:23 +010010774v4v6
10775 Is an optional keyword which is supported only on most recent systems
10776 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
10777 and IPv6 when it uses the default address. Doing so is sometimes necessary
10778 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010779 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010010780
Willy Tarreau9b6700f2012-11-24 11:55:28 +010010781v6only
10782 Is an optional keyword which is supported only on most recent systems
10783 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
10784 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010010785 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
10786 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010010787
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010788uid <uid>
10789 Sets the owner of the UNIX sockets to the designated system uid. It can also
10790 be set by default in the global section's "unix-bind" statement. Note that
10791 some platforms simply ignore this. This setting is equivalent to the "user"
10792 setting except that the user numeric ID is used instead of its name. This
10793 setting is ignored by non UNIX sockets.
10794
10795user <user>
10796 Sets the owner of the UNIX sockets to the designated system user. It can also
10797 be set by default in the global section's "unix-bind" statement. Note that
10798 some platforms simply ignore this. This setting is equivalent to the "uid"
10799 setting except that the user name is used instead of its uid. This setting is
10800 ignored by non UNIX sockets.
10801
Emeric Brun1a073b42012-09-28 17:07:34 +020010802verify [none|optional|required]
10803 This setting is only available when support for OpenSSL was built in. If set
10804 to 'none', client certificate is not requested. This is the default. In other
10805 cases, a client certificate is requested. If the client does not provide a
10806 certificate after the request and if 'verify' is set to 'required', then the
10807 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020010808 certificate provided by the client is always verified using CAs from
10809 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
10810 is aborted, regardless of the 'verify' option, unless the error code exactly
10811 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010812
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200108135.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010010814------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020010815
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010010816The "server" and "default-server" keywords support a certain number of settings
10817which are all passed as arguments on the server line. The order in which those
10818arguments appear does not count, and they are all optional. Some of those
10819settings are single words (booleans) while others expect one or several values
10820after them. In this case, the values must immediately follow the setting name.
10821Except default-server, all those settings must be specified after the server's
10822address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020010823
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010824 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010010825 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020010826
Frédéric Lécailled2376272017-03-21 18:52:12 +010010827Note that all these settings are supported both by "server" and "default-server"
10828keywords, except "id" which is only supported by "server".
10829
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010010830The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010831
Willy Tarreauceb4ac92012-04-28 00:41:46 +020010832addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010833 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010010834 to send health-checks or to probe the agent-check. On some servers, it may be
10835 desirable to dedicate an IP address to specific component able to perform
10836 complex tests which are more suitable to health-checks than the application.
10837 This parameter is ignored if the "check" parameter is not set. See also the
10838 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020010839
Simon Hormand60d6912013-11-25 10:46:36 +090010840agent-check
10841 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010010842 health check. An agent health check is performed by making a TCP connection
10843 to the port set by the "agent-port" parameter and reading an ASCII string.
10844 The string is made of a series of words delimited by spaces, tabs or commas
10845 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090010846
Willy Tarreau81f5d942013-12-09 20:51:51 +010010847 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090010848 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020010849 weight of a server as configured when haproxy starts. Note that a zero
10850 weight is reported on the stats page as "DRAIN" since it has the same
10851 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090010852
Nenad Merdanovic174dd372016-04-24 23:10:06 +020010853 - The string "maxconn:" followed by an integer (no space between). Values in
10854 this format will set the maxconn of a server. The maximum number of
10855 connections advertised needs to be multipled by the number of load balancers
10856 and different backends that use this health check to get the total number
10857 of connections the server might receive. Example: maxconn:30
10858
Willy Tarreau81f5d942013-12-09 20:51:51 +010010859 - The word "ready". This will turn the server's administrative state to the
10860 READY mode, thus cancelling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090010861
Willy Tarreau81f5d942013-12-09 20:51:51 +010010862 - The word "drain". This will turn the server's administrative state to the
10863 DRAIN mode, thus it will not accept any new connections other than those
10864 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090010865
Willy Tarreau81f5d942013-12-09 20:51:51 +010010866 - The word "maint". This will turn the server's administrative state to the
10867 MAINT mode, thus it will not accept any new connections at all, and health
10868 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090010869
Willy Tarreau81f5d942013-12-09 20:51:51 +010010870 - The words "down", "failed", or "stopped", optionally followed by a
10871 description string after a sharp ('#'). All of these mark the server's
10872 operating state as DOWN, but since the word itself is reported on the stats
10873 page, the difference allows an administrator to know if the situation was
10874 expected or not : the service may intentionally be stopped, may appear up
10875 but fail some validity tests, or may be seen as down (eg: missing process,
10876 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090010877
Willy Tarreau81f5d942013-12-09 20:51:51 +010010878 - The word "up" sets back the server's operating state as UP if health checks
10879 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090010880
Willy Tarreau81f5d942013-12-09 20:51:51 +010010881 Parameters which are not advertised by the agent are not changed. For
10882 example, an agent might be designed to monitor CPU usage and only report a
10883 relative weight and never interact with the operating status. Similarly, an
10884 agent could be designed as an end-user interface with 3 radio buttons
10885 allowing an administrator to change only the administrative state. However,
10886 it is important to consider that only the agent may revert its own actions,
10887 so if a server is set to DRAIN mode or to DOWN state using the agent, the
10888 agent must implement the other equivalent actions to bring the service into
10889 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090010890
Simon Horman2f1f9552013-11-25 10:46:37 +090010891 Failure to connect to the agent is not considered an error as connectivity
10892 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010010893 parameter. Warning though, it is not a good idea to stop an agent after it
10894 reports "down", since only an agent reporting "up" will be able to turn the
10895 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010010896 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090010897
Willy Tarreau81f5d942013-12-09 20:51:51 +010010898 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010010899 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090010900
James Brown55f9ff12015-10-21 18:19:05 -070010901agent-send <string>
10902 If this option is specified, haproxy will send the given string (verbatim)
10903 to the agent server upon connection. You could, for example, encode
10904 the backend name into this string, which would enable your agent to send
10905 different responses based on the backend. Make sure to include a '\n' if
10906 you want to terminate your request with a newline.
10907
Simon Hormand60d6912013-11-25 10:46:36 +090010908agent-inter <delay>
10909 The "agent-inter" parameter sets the interval between two agent checks
10910 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
10911
10912 Just as with every other time-based parameter, it may be entered in any
10913 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
10914 parameter also serves as a timeout for agent checks "timeout check" is
10915 not set. In order to reduce "resonance" effects when multiple servers are
10916 hosted on the same hardware, the agent and health checks of all servers
10917 are started with a small time offset between them. It is also possible to
10918 add some random noise in the agent and health checks interval using the
10919 global "spread-checks" keyword. This makes sense for instance when a lot
10920 of backends use the same servers.
10921
10922 See also the "agent-check" and "agent-port" parameters.
10923
Misiek768d8602017-01-09 09:52:43 +010010924agent-addr <addr>
10925 The "agent-addr" parameter sets address for agent check.
10926
10927 You can offload agent-check to another target, so you can make single place
10928 managing status and weights of servers defined in haproxy in case you can't
10929 make self-aware and self-managing services. You can specify both IP or
10930 hostname, it will be resolved.
10931
Simon Hormand60d6912013-11-25 10:46:36 +090010932agent-port <port>
10933 The "agent-port" parameter sets the TCP port used for agent checks.
10934
10935 See also the "agent-check" and "agent-inter" parameters.
10936
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010937backup
10938 When "backup" is present on a server line, the server is only used in load
10939 balancing when all other non-backup servers are unavailable. Requests coming
10940 with a persistence cookie referencing the server will always be served
10941 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010010942 the "allbackups" option is set in the backend. See also the "no-backup" and
10943 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010010944
Emeric Brunef42d922012-10-11 16:11:36 +020010945ca-file <cafile>
10946 This setting is only available when support for OpenSSL was built in. It
10947 designates a PEM file from which to load CA certificates used to verify
10948 server's certificate.
10949
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010950check
10951 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010010952 always considered available. If "check" is set, the server is available when
10953 accepting periodic TCP connections, to ensure that it is really able to serve
10954 requests. The default address and port to send the tests to are those of the
10955 server, and the default source is the same as the one defined in the
10956 backend. It is possible to change the address using the "addr" parameter, the
10957 port using the "port" parameter, the source address using the "source"
10958 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090010959 parameters. The request method is define in the backend using the "httpchk",
10960 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010010961 refer to those options and parameters for more information. See also
10962 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010010963
Willy Tarreau6c16adc2012-10-05 00:04:16 +020010964check-send-proxy
10965 This option forces emission of a PROXY protocol line with outgoing health
10966 checks, regardless of whether the server uses send-proxy or not for the
10967 normal traffic. By default, the PROXY protocol is enabled for health checks
10968 if it is already enabled for normal traffic and if no "port" nor "addr"
10969 directive is present. However, if such a directive is present, the
10970 "check-send-proxy" option needs to be used to force the use of the
10971 protocol. See also the "send-proxy" option for more information.
10972
Willy Tarreau763a95b2012-10-04 23:15:39 +020010973check-ssl
10974 This option forces encryption of all health checks over SSL, regardless of
10975 whether the server uses SSL or not for the normal traffic. This is generally
10976 used when an explicit "port" or "addr" directive is specified and SSL health
10977 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010978 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020010979 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
10980 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
10981 All SSL settings are common to health checks and traffic (eg: ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010010982 See the "ssl" option for more information and "no-check-ssl" to disable
10983 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020010984
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020010985ciphers <ciphers>
10986 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010987 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020010988 string is defined in "man 1 ciphers". When SSL is used to communicate with
10989 servers on the local network, it is common to see a weaker set of algorithms
10990 than what is used over the internet. Doing so reduces CPU usage on both the
10991 server and haproxy while still keeping it compatible with deployed software.
10992 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
10993 is needed and just connectivity, using DES can be appropriate.
10994
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010995cookie <value>
10996 The "cookie" parameter sets the cookie value assigned to the server to
10997 <value>. This value will be checked in incoming requests, and the first
10998 operational server possessing the same value will be selected. In return, in
10999 cookie insertion or rewrite modes, this value will be assigned to the cookie
11000 sent to the client. There is nothing wrong in having several servers sharing
11001 the same cookie value, and it is in fact somewhat common between normal and
11002 backup servers. See also the "cookie" keyword in backend section.
11003
Emeric Brunef42d922012-10-11 16:11:36 +020011004crl-file <crlfile>
11005 This setting is only available when support for OpenSSL was built in. It
11006 designates a PEM file from which to load certificate revocation list used
11007 to verify server's certificate.
11008
Emeric Bruna7aa3092012-10-26 12:58:00 +020011009crt <cert>
11010 This setting is only available when support for OpenSSL was built in.
11011 It designates a PEM file from which to load both a certificate and the
11012 associated private key. This file can be built by concatenating both PEM
11013 files into one. This certificate will be sent if the server send a client
11014 certificate request.
11015
Willy Tarreau96839092010-03-29 10:02:24 +020011016disabled
11017 The "disabled" keyword starts the server in the "disabled" state. That means
11018 that it is marked down in maintenance mode, and no connection other than the
11019 ones allowed by persist mode will reach it. It is very well suited to setup
11020 new servers, because normal traffic will never reach them, while it is still
11021 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011022 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011023
Frédéric Lécailled2376272017-03-21 18:52:12 +010011024enabled
11025 This option may be used as 'server' setting to reset any 'disabled'
11026 setting which would have been inherited from 'default-server' directive as
11027 default value.
11028 It may also be used as 'default-server' setting to reset any previous
11029 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011030
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011031error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011032 If health observing is enabled, the "error-limit" parameter specifies the
11033 number of consecutive errors that triggers event selected by the "on-error"
11034 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011035
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011036 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011037
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011038fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011039 The "fall" parameter states that a server will be considered as dead after
11040 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11041 unspecified. See also the "check", "inter" and "rise" parameters.
11042
Emeric Brun8694b9a2012-10-05 14:39:07 +020011043force-sslv3
11044 This option enforces use of SSLv3 only when SSL is used to communicate with
11045 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011046 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011047 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011048
11049force-tlsv10
11050 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011051 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011052 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011053
11054force-tlsv11
11055 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011056 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011057 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011058
11059force-tlsv12
11060 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011061 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011062 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011063
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011064force-tlsv13
11065 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11066 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011067 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011068
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011069id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011070 Set a persistent ID for the server. This ID must be positive and unique for
11071 the proxy. An unused ID will automatically be assigned if unset. The first
11072 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011073
Willy Tarreau6a031d12016-11-07 19:42:35 +010011074init-addr {last | libc | none | <ip>},[...]*
11075 Indicate in what order the server's address should be resolved upon startup
11076 if it uses an FQDN. Attempts are made to resolve the address by applying in
11077 turn each of the methods mentionned in the comma-delimited list. The first
11078 method which succeeds is used. If the end of the list is reached without
11079 finding a working method, an error is thrown. Method "last" suggests to pick
11080 the address which appears in the state file (see "server-state-file"). Method
11081 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11082 depending on the operating system and build options). Method "none"
11083 specifically indicates that the server should start without any valid IP
11084 address in a down state. It can be useful to ignore some DNS issues upon
11085 startup, waiting for the situation to get fixed later. Finally, an IP address
11086 (IPv4 or IPv6) may be provided. It can be the currently known address of the
11087 server (eg: filled by a configuration generator), or the address of a dummy
11088 server used to catch old sessions and present them with a decent error
11089 message for example. When the "first" load balancing algorithm is used, this
11090 IP address could point to a fake server used to trigger the creation of new
11091 instances on the fly. This option defaults to "last,libc" indicating that the
11092 previous address found in the state file (if any) is used first, otherwise
11093 the libc's resolver is used. This ensures continued compatibility with the
11094 historic behaviour.
11095
11096 Example:
11097 defaults
11098 # never fail on address resolution
11099 default-server init-addr last,libc,none
11100
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011101inter <delay>
11102fastinter <delay>
11103downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011104 The "inter" parameter sets the interval between two consecutive health checks
11105 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11106 It is also possible to use "fastinter" and "downinter" to optimize delays
11107 between checks depending on the server state :
11108
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011109 Server state | Interval used
11110 ----------------------------------------+----------------------------------
11111 UP 100% (non-transitional) | "inter"
11112 ----------------------------------------+----------------------------------
11113 Transitionally UP (going down "fall"), | "fastinter" if set,
11114 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11115 or yet unchecked. |
11116 ----------------------------------------+----------------------------------
11117 DOWN 100% (non-transitional) | "downinter" if set,
11118 | "inter" otherwise.
11119 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011120
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011121 Just as with every other time-based parameter, they can be entered in any
11122 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11123 serves as a timeout for health checks sent to servers if "timeout check" is
11124 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011125 hosted on the same hardware, the agent and health checks of all servers
11126 are started with a small time offset between them. It is also possible to
11127 add some random noise in the agent and health checks interval using the
11128 global "spread-checks" keyword. This makes sense for instance when a lot
11129 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011130
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011131maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011132 The "maxconn" parameter specifies the maximal number of concurrent
11133 connections that will be sent to this server. If the number of incoming
11134 concurrent requests goes higher than this value, they will be queued, waiting
11135 for a connection to be released. This parameter is very important as it can
11136 save fragile servers from going down under extreme loads. If a "minconn"
11137 parameter is specified, the limit becomes dynamic. The default value is "0"
11138 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11139 the backend's "fullconn" keyword.
11140
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011141maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011142 The "maxqueue" parameter specifies the maximal number of connections which
11143 will wait in the queue for this server. If this limit is reached, next
11144 requests will be redispatched to other servers instead of indefinitely
11145 waiting to be served. This will break persistence but may allow people to
11146 quickly re-log in when the server they try to connect to is dying. The
11147 default value is "0" which means the queue is unlimited. See also the
11148 "maxconn" and "minconn" parameters.
11149
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011150minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011151 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11152 limit following the backend's load. The server will always accept at least
11153 <minconn> connections, never more than <maxconn>, and the limit will be on
11154 the ramp between both values when the backend has less than <fullconn>
11155 concurrent connections. This makes it possible to limit the load on the
11156 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011157 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011158 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011159
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011160namespace <name>
11161 On Linux, it is possible to specify which network namespace a socket will
11162 belong to. This directive makes it possible to explicitly bind a server to
11163 a namespace different from the default one. Please refer to your operating
11164 system's documentation to find more details about network namespaces.
11165
Frédéric Lécailled2376272017-03-21 18:52:12 +010011166no-agent-check
11167 This option may be used as "server" setting to reset any "agent-check"
11168 setting which would have been inherited from "default-server" directive as
11169 default value.
11170 It may also be used as "default-server" setting to reset any previous
11171 "default-server" "agent-check" setting.
11172
11173no-backup
11174 This option may be used as "server" setting to reset any "backup"
11175 setting which would have been inherited from "default-server" directive as
11176 default value.
11177 It may also be used as "default-server" setting to reset any previous
11178 "default-server" "backup" setting.
11179
11180no-check
11181 This option may be used as "server" setting to reset any "check"
11182 setting which would have been inherited from "default-server" directive as
11183 default value.
11184 It may also be used as "default-server" setting to reset any previous
11185 "default-server" "check" setting.
11186
11187no-check-ssl
11188 This option may be used as "server" setting to reset any "check-ssl"
11189 setting which would have been inherited from "default-server" directive as
11190 default value.
11191 It may also be used as "default-server" setting to reset any previous
11192 "default-server" "check-ssl" setting.
11193
Frédéric Lécailled2376272017-03-21 18:52:12 +010011194no-send-proxy
11195 This option may be used as "server" setting to reset any "send-proxy"
11196 setting which would have been inherited from "default-server" directive as
11197 default value.
11198 It may also be used as "default-server" setting to reset any previous
11199 "default-server" "send-proxy" setting.
11200
11201no-send-proxy-v2
11202 This option may be used as "server" setting to reset any "send-proxy-v2"
11203 setting which would have been inherited from "default-server" directive as
11204 default value.
11205 It may also be used as "default-server" setting to reset any previous
11206 "default-server" "send-proxy-v2" setting.
11207
11208no-send-proxy-v2-ssl
11209 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11210 setting which would have been inherited from "default-server" directive as
11211 default value.
11212 It may also be used as "default-server" setting to reset any previous
11213 "default-server" "send-proxy-v2-ssl" setting.
11214
11215no-send-proxy-v2-ssl-cn
11216 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11217 setting which would have been inherited from "default-server" directive as
11218 default value.
11219 It may also be used as "default-server" setting to reset any previous
11220 "default-server" "send-proxy-v2-ssl-cn" setting.
11221
11222no-ssl
11223 This option may be used as "server" setting to reset any "ssl"
11224 setting which would have been inherited from "default-server" directive as
11225 default value.
11226 It may also be used as "default-server" setting to reset any previous
11227 "default-server" "ssl" setting.
11228
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011229no-ssl-reuse
11230 This option disables SSL session reuse when SSL is used to communicate with
11231 the server. It will force the server to perform a full handshake for every
11232 new connection. It's probably only useful for benchmarking, troubleshooting,
11233 and for paranoid users.
11234
Emeric Brun9b3009b2012-10-05 11:55:06 +020011235no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011236 This option disables support for SSLv3 when SSL is used to communicate with
11237 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011238 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011239
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011240 Supported in default-server: No
11241
Emeric Brunf9c5c472012-10-11 15:28:34 +020011242no-tls-tickets
11243 This setting is only available when support for OpenSSL was built in. It
11244 disables the stateless session resumption (RFC 5077 TLS Ticket
11245 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011246 session resumption is more expensive in CPU usage for servers. This option
11247 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011248 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011249
Emeric Brun9b3009b2012-10-05 11:55:06 +020011250no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011251 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011252 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11253 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011254 often makes sense to disable it when communicating with local servers. This
11255 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011256 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011257
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011258 Supported in default-server: No
11259
Emeric Brun9b3009b2012-10-05 11:55:06 +020011260no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011261 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011262 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11263 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011264 often makes sense to disable it when communicating with local servers. This
11265 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011266 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011267
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011268 Supported in default-server: No
11269
Emeric Brun9b3009b2012-10-05 11:55:06 +020011270no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011271 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011272 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11273 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011274 often makes sense to disable it when communicating with local servers. This
11275 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011276 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011277
11278 Supported in default-server: No
11279
11280no-tlsv13
11281 This option disables support for TLSv1.3 when SSL is used to communicate with
11282 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11283 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11284 often makes sense to disable it when communicating with local servers. This
11285 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011286 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011287
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011288 Supported in default-server: No
11289
Frédéric Lécailled2376272017-03-21 18:52:12 +010011290no-verifyhost
11291 This option may be used as "server" setting to reset any "verifyhost"
11292 setting which would have been inherited from "default-server" directive as
11293 default value.
11294 It may also be used as "default-server" setting to reset any previous
11295 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011296
Simon Hormanfa461682011-06-25 09:39:49 +090011297non-stick
11298 Never add connections allocated to this sever to a stick-table.
11299 This may be used in conjunction with backup to ensure that
11300 stick-table persistence is disabled for backup servers.
11301
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011302observe <mode>
11303 This option enables health adjusting based on observing communication with
11304 the server. By default this functionality is disabled and enabling it also
11305 requires to enable health checks. There are two supported modes: "layer4" and
11306 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11307 significant. In layer7, which is only allowed for http proxies, responses
11308 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011309 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011310
11311 See also the "check", "on-error" and "error-limit".
11312
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011313on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011314 Select what should happen when enough consecutive errors are detected.
11315 Currently, four modes are available:
11316 - fastinter: force fastinter
11317 - fail-check: simulate a failed check, also forces fastinter (default)
11318 - sudden-death: simulate a pre-fatal failed health check, one more failed
11319 check will mark a server down, forces fastinter
11320 - mark-down: mark the server immediately down and force fastinter
11321
11322 See also the "check", "observe" and "error-limit".
11323
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011324on-marked-down <action>
11325 Modify what occurs when a server is marked down.
11326 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011327 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11328 all connections to the server are immediately terminated when the server
11329 goes down. It might be used if the health check detects more complex cases
11330 than a simple connection status, and long timeouts would cause the service
11331 to remain unresponsive for too long a time. For instance, a health check
11332 might detect that a database is stuck and that there's no chance to reuse
11333 existing connections anymore. Connections killed this way are logged with
11334 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011335
11336 Actions are disabled by default
11337
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011338on-marked-up <action>
11339 Modify what occurs when a server is marked up.
11340 Currently one action is available:
11341 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11342 done only if the server is not in backup state and if it is not disabled
11343 (it must have an effective weight > 0). This can be used sometimes to force
11344 an active server to take all the traffic back after recovery when dealing
11345 with long sessions (eg: LDAP, SQL, ...). Doing this can cause more trouble
11346 than it tries to solve (eg: incomplete transactions), so use this feature
11347 with extreme care. Sessions killed because a server comes up are logged
11348 with an 'U' termination code (for "Up").
11349
11350 Actions are disabled by default
11351
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011352port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011353 Using the "port" parameter, it becomes possible to use a different port to
11354 send health-checks. On some servers, it may be desirable to dedicate a port
11355 to a specific component able to perform complex tests which are more suitable
11356 to health-checks than the application. It is common to run a simple script in
11357 inetd for instance. This parameter is ignored if the "check" parameter is not
11358 set. See also the "addr" parameter.
11359
11360redir <prefix>
11361 The "redir" parameter enables the redirection mode for all GET and HEAD
11362 requests addressing this server. This means that instead of having HAProxy
11363 forward the request to the server, it will send an "HTTP 302" response with
11364 the "Location" header composed of this prefix immediately followed by the
11365 requested URI beginning at the leading '/' of the path component. That means
11366 that no trailing slash should be used after <prefix>. All invalid requests
11367 will be rejected, and all non-GET or HEAD requests will be normally served by
11368 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011369 mangling nor cookie insertion is possible in the response. However, cookies in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011370 requests are still analysed, making this solution completely usable to direct
11371 users to a remote location in case of local disaster. Main use consists in
11372 increasing bandwidth for static servers by having the clients directly
11373 connect to them. Note: never use a relative location here, it would cause a
11374 loop between the client and HAProxy!
11375
11376 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11377
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011378rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011379 The "rise" parameter states that a server will be considered as operational
11380 after <count> consecutive successful health checks. This value defaults to 2
11381 if unspecified. See also the "check", "inter" and "fall" parameters.
11382
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011383resolve-prefer <family>
11384 When DNS resolution is enabled for a server and multiple IP addresses from
11385 different families are returned, HAProxy will prefer using an IP address
11386 from the family mentioned in the "resolve-prefer" parameter.
11387 Available families: "ipv4" and "ipv6"
11388
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011389 Default value: ipv6
11390
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011391 Example:
11392
11393 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011394
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011395resolve-net <network>[,<network[,...]]
11396 This options prioritize th choice of an ip address matching a network. This is
11397 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011398 availability service can be announced with many ip addresses on many
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011399 differents datacenters. The latency between datacenter is not negligible, so
11400 this patch permitsto prefers a local datacenter. If none address matchs the
11401 configured network, another address is selected.
11402
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011403 Example:
11404
11405 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011406
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011407resolvers <id>
11408 Points to an existing "resolvers" section to resolve current server's
11409 hostname.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011410 In order to be operational, DNS resolution requires that health check is
11411 enabled on the server. Actually, health checks triggers the DNS resolution.
11412 You must precise one 'resolvers' parameter on each server line where DNS
11413 resolution is required.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011414
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011415 Example:
11416
11417 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011418
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011419 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011420
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011421send-proxy
11422 The "send-proxy" parameter enforces use of the PROXY protocol over any
11423 connection established to this server. The PROXY protocol informs the other
11424 end about the layer 3/4 addresses of the incoming connection, so that it can
11425 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011426 upper layer protocol. For connections accepted by an "accept-proxy" or
11427 "accept-netscaler-cip" listener, the advertised address will be used. Only
11428 TCPv4 and TCPv6 address families are supported. Other families such as
11429 Unix sockets, will report an UNKNOWN family. Servers using this option can
11430 fully be chained to another instance of haproxy listening with an
11431 "accept-proxy" setting. This setting must not be used if the server isn't
11432 aware of the protocol. When health checks are sent to the server, the PROXY
11433 protocol is automatically used when this option is set, unless there is an
11434 explicit "port" or "addr" directive, in which case an explicit
11435 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011436 See also the "no-send-proxy" option of this section and "accept-proxy" and
11437 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011438
David Safb76832014-05-08 23:42:08 -040011439send-proxy-v2
11440 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
11441 over any connection established to this server. The PROXY protocol informs
11442 the other end about the layer 3/4 addresses of the incoming connection, so
11443 that it can know the client's address or the public address it accessed to,
11444 whatever the upper layer protocol. This setting must not be used if the
Frédéric Lécailled2376272017-03-21 18:52:12 +010011445 server isn't aware of this version of the protocol. See also the
11446 "no-send-proxy-v2" option of this section and send-proxy" option of the
11447 "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011448
11449send-proxy-v2-ssl
11450 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11451 2 over any connection established to this server. The PROXY protocol informs
11452 the other end about the layer 3/4 addresses of the incoming connection, so
11453 that it can know the client's address or the public address it accessed to,
11454 whatever the upper layer protocol. In addition, the SSL information extension
11455 of the PROXY protocol is added to the PROXY protocol header. This setting
11456 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011457 See also the "no-send-proxy-v2-ssl" option of this section and the
11458 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011459
11460send-proxy-v2-ssl-cn
11461 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11462 2 over any connection established to this server. The PROXY protocol informs
11463 the other end about the layer 3/4 addresses of the incoming connection, so
11464 that it can know the client's address or the public address it accessed to,
11465 whatever the upper layer protocol. In addition, the SSL information extension
11466 of the PROXY protocol, along along with the Common Name from the subject of
11467 the client certificate (if any), is added to the PROXY protocol header. This
11468 setting must not be used if the server isn't aware of this version of the
Frédéric Lécailled2376272017-03-21 18:52:12 +010011469 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and the
11470 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011471
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011472slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011473 The "slowstart" parameter for a server accepts a value in milliseconds which
11474 indicates after how long a server which has just come back up will run at
11475 full speed. Just as with every other time-based parameter, it can be entered
11476 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
11477 linearly from 0 to 100% during this time. The limitation applies to two
11478 parameters :
11479
11480 - maxconn: the number of connections accepted by the server will grow from 1
11481 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
11482
11483 - weight: when the backend uses a dynamic weighted algorithm, the weight
11484 grows linearly from 1 to 100%. In this case, the weight is updated at every
11485 health-check. For this reason, it is important that the "inter" parameter
11486 is smaller than the "slowstart", in order to maximize the number of steps.
11487
11488 The slowstart never applies when haproxy starts, otherwise it would cause
11489 trouble to running servers. It only applies when a server has been previously
11490 seen as failed.
11491
Willy Tarreau732eac42015-07-09 11:40:25 +020011492sni <expression>
11493 The "sni" parameter evaluates the sample fetch expression, converts it to a
11494 string and uses the result as the host name sent in the SNI TLS extension to
11495 the server. A typical use case is to send the SNI received from the client in
11496 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020011497 expression, though alternatives such as req.hdr(host) can also make sense. If
11498 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011499 name will also be matched against the server certificate's names. See the
11500 "verify" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020011501
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011502source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020011503source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011504source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011505 The "source" parameter sets the source address which will be used when
11506 connecting to the server. It follows the exact same parameters and principle
11507 as the backend "source" keyword, except that it only applies to the server
11508 referencing it. Please consult the "source" keyword for details.
11509
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011510 Additionally, the "source" statement on a server line allows one to specify a
11511 source port range by indicating the lower and higher bounds delimited by a
11512 dash ('-'). Some operating systems might require a valid IP address when a
11513 source port range is specified. It is permitted to have the same IP/range for
11514 several servers. Doing so makes it possible to bypass the maximum of 64k
11515 total concurrent connections. The limit will then reach 64k connections per
11516 server.
11517
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000011518 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
11519 specifying the source address without port(s).
11520
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011521ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020011522 This option enables SSL ciphering on outgoing connections to the server. It
11523 is critical to verify server certificates using "verify" when using SSL to
11524 connect to servers, otherwise the communication is prone to trivial man in
11525 the-middle attacks rendering SSL useless. When this option is used, health
11526 checks are automatically sent in SSL too unless there is a "port" or an
11527 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011528 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
11529 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011530
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011531ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11532 This option enforces use of <version> or lower when SSL is used to communicate
11533 with the server. This option is also available on global statement
11534 "ssl-default-server-options". See also "ssl-min-ver".
11535
11536ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11537 This option enforces use of <version> or upper when SSL is used to communicate
11538 with the server. This option is also available on global statement
11539 "ssl-default-server-options". See also "ssl-max-ver".
11540
Frédéric Lécailled2376272017-03-21 18:52:12 +010011541ssl-reuse
11542 This option may be used as "server" setting to reset any "no-ssl-reuse"
11543 setting which would have been inherited from "default-server" directive as
11544 default value.
11545 It may also be used as "default-server" setting to reset any previous
11546 "default-server" "no-ssl-reuse" setting.
11547
11548stick
11549 This option may be used as "server" setting to reset any "non-stick"
11550 setting which would have been inherited from "default-server" directive as
11551 default value.
11552 It may also be used as "default-server" setting to reset any previous
11553 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011554
Willy Tarreau163d4622015-10-13 16:16:41 +020011555tcp-ut <delay>
11556 Sets the TCP User Timeout for all outgoing connections to this server. This
11557 option is available on Linux since version 2.6.37. It allows haproxy to
11558 configure a timeout for sockets which contain data not receiving an
Tim Düsterhus4896c442016-11-29 02:15:19 +010011559 acknowledgement for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020011560 long-lived connections experiencing long idle periods such as remote
11561 terminals or database connection pools, where the client and server timeouts
11562 must remain high to allow a long period of idle, but where it is important to
11563 detect that the server has disappeared in order to release all resources
11564 associated with its connection (and the client's session). One typical use
11565 case is also to force dead server connections to die when health checks are
11566 too slow or during a soft reload since health checks are then disabled. The
11567 argument is a delay expressed in milliseconds by default. This only works for
11568 regular TCP connections, and is ignored for other protocols.
11569
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011570track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020011571 This option enables ability to set the current state of the server by tracking
11572 another one. It is possible to track a server which itself tracks another
11573 server, provided that at the end of the chain, a server has health checks
11574 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011575 used, it has to be enabled on both proxies.
11576
Frédéric Lécailled2376272017-03-21 18:52:12 +010011577tls-tickets
11578 This option may be used as "server" setting to reset any "no-tls-tickets"
11579 setting which would have been inherited from "default-server" directive as
11580 default value.
11581 It may also be used as "default-server" setting to reset any previous
11582 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011583
Emeric Brunef42d922012-10-11 16:11:36 +020011584verify [none|required]
11585 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010011586 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011587 certificate provided by the server is verified using CAs from 'ca-file' and
11588 optional CRLs from 'crl-file' after having checked that the names provided in
11589 the certificate's subject and subjectAlternateNames attributs match either
11590 the name passed using the "sni" directive, or if not provided, the static
11591 host name passed using the "verifyhost" directive. When no name is found, the
11592 certificate's names are ignored. For this reason, without SNI it's important
11593 to use "verifyhost". On verification failure the handshake is aborted. It is
11594 critically important to verify server certificates when using SSL to connect
11595 to servers, otherwise the communication is prone to trivial man-in-the-middle
11596 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
11597 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020011598
Evan Broderbe554312013-06-27 00:05:25 -070011599verifyhost <hostname>
11600 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011601 only takes effect if 'verify required' is also specified. This directive sets
11602 a default static hostname to check the server's certificate against when no
11603 SNI was used to connect to the server. If SNI is not used, this is the only
11604 way to enable hostname verification. This static hostname, when set, will
11605 also be used for health checks (which cannot provide an SNI value). If none
11606 of the hostnames in the certificate match the specified hostname, the
11607 handshake is aborted. The hostnames in the server-provided certificate may
11608 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070011609
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011610weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011611 The "weight" parameter is used to adjust the server's weight relative to
11612 other servers. All servers will receive a load proportional to their weight
11613 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020011614 load. The default weight is 1, and the maximal value is 256. A value of 0
11615 means the server will not participate in load-balancing but will still accept
11616 persistent connections. If this parameter is used to distribute the load
11617 according to server's capacity, it is recommended to start with values which
11618 can both grow and shrink, for instance between 10 and 100 to leave enough
11619 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011620
11621
Cyril Bonté46175dd2015-07-02 22:45:32 +0200116225.3. Server IP address resolution using DNS
11623-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011624
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011625HAProxy allows using a host name on the server line to retrieve its IP address
11626using name servers. By default, HAProxy resolves the name when parsing the
11627configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011628This is not sufficient in some cases, such as in Amazon where a server's IP
11629can change after a reboot or an ELB Virtual IP can change based on current
11630workload.
11631This chapter describes how HAProxy can be configured to process server's name
11632resolution at run time.
11633Whether run time server name resolution has been enable or not, HAProxy will
11634carry on doing the first resolution when parsing the configuration.
11635
11636
Cyril Bonté46175dd2015-07-02 22:45:32 +0200116375.3.1. Global overview
11638----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011639
11640As we've seen in introduction, name resolution in HAProxy occurs at two
11641different steps of the process life:
11642
11643 1. when starting up, HAProxy parses the server line definition and matches a
11644 host name. It uses libc functions to get the host name resolved. This
11645 resolution relies on /etc/resolv.conf file.
11646
11647 2. at run time, when HAProxy gets prepared to run a health check on a server,
11648 it verifies if the current name resolution is still considered as valid.
11649 If not, it processes a new resolution, in parallel of the health check.
11650
11651A few other events can trigger a name resolution at run time:
11652 - when a server's health check ends up in a connection timeout: this may be
11653 because the server has a new IP address. So we need to trigger a name
11654 resolution to know this new IP.
11655
Olivier Houchardecfa18d2017-08-07 17:30:03 +020011656When using resolvers, the server name can either be a hostname, or s SRV label.
11657HAProxy considers anything that starts with an underscore a SRV label.
11658If a SRV label is specified, then the corresponding SRV records will be
11659retrieved from the DNS server, and the provided hostnames will be used. The
11660SRV label will be checked periodically, and if any server are added or removed,
11661haproxy will automatically do the same.
11662
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011663A few things important to notice:
11664 - all the name servers are queried in the mean time. HAProxy will process the
11665 first valid response.
11666
11667 - a resolution is considered as invalid (NX, timeout, refused), when all the
11668 servers return an error.
11669
11670
Cyril Bonté46175dd2015-07-02 22:45:32 +0200116715.3.2. The resolvers section
11672----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011673
11674This section is dedicated to host information related to name resolution in
11675HAProxy.
11676There can be as many as resolvers section as needed. Each section can contain
11677many name servers.
11678
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011679When multiple name servers are configured in a resolvers section, then HAProxy
11680uses the first valid response. In case of invalid responses, only the last one
11681is treated. Purpose is to give the chance to a slow server to deliver a valid
11682answer after a fast faulty or outdated server.
11683
11684When each server returns a different error type, then only the last error is
11685used by HAProxy to decide what type of behavior to apply.
11686
11687Two types of behavior can be applied:
11688 1. stop DNS resolution
11689 2. replay the DNS query with a new query type
11690 In such case, the following types are applied in this exact order:
11691 1. ANY query type
11692 2. query type corresponding to family pointed by resolve-prefer
11693 server's parameter
11694 3. remaining family type
11695
11696HAProxy stops DNS resolution when the following errors occur:
11697 - invalid DNS response packet
11698 - wrong name in the query section of the response
11699 - NX domain
11700 - Query refused by server
11701 - CNAME not pointing to an IP address
11702
11703HAProxy tries a new query type when the following errors occur:
11704 - no Answer records in the response
11705 - DNS response truncated
11706 - Error in DNS response
11707 - No expected DNS records found in the response
11708 - name server timeout
11709
11710For example, with 2 name servers configured in a resolvers section:
11711 - first response is valid and is applied directly, second response is ignored
11712 - first response is invalid and second one is valid, then second response is
11713 applied;
11714 - first response is a NX domain and second one a truncated response, then
11715 HAProxy replays the query with a new type;
11716 - first response is truncated and second one is a NX Domain, then HAProxy
11717 stops resolution.
11718
Olivier Houcharda8c6db82017-07-06 18:46:47 +020011719As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
11720a cache of previous answers, an answer will be considered obsolete after
11721"hold obsolete" seconds without the IP returned.
11722
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011723
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011724resolvers <resolvers id>
11725 Creates a new name server list labelled <resolvers id>
11726
11727A resolvers section accept the following parameters:
11728
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020011729accepted_payload_size <nb>
11730 Defines the maxium payload size accepted by HAProxy and announced to all the
11731 naeservers configured in this resolvers section.
11732 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
11733 by RFC 6891)
11734
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020011735 Note: to get bigger responses but still be sure that responses won't be
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020011736 dropped on the wire, one can choose a value between 1280 and 1410.
11737
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020011738 Note: the maximum allowed value is 8192.
11739
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011740nameserver <id> <ip>:<port>
11741 DNS server description:
11742 <id> : label of the server, should be unique
11743 <ip> : IP address of the server
11744 <port> : port where the DNS service actually runs
11745
11746hold <status> <period>
11747 Defines <period> during which the last name resolution should be kept based
11748 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010011749 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020011750 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011751 <period> : interval between two successive name resolution when the last
11752 answer was in <status>. It follows the HAProxy time format.
11753 <period> is in milliseconds by default.
11754
Baptiste Assmann686408b2017-08-18 10:15:42 +020011755 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011756
11757 Note: since the name resolution is triggered by the health checks, a new
11758 resolution is triggered after <period> modulo the <inter> parameter of
11759 the healch check.
11760
Baptiste Assmann201c07f2017-05-22 15:17:15 +020011761resolution_pool_size <nb>
11762 Defines the number of resolutions available in the pool for this resolvers.
11763 If not defines, it defaults to 64. If your configuration requires more than
11764 <nb>, then HAProxy will return an error when parsing the configuration.
11765
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011766resolve_retries <nb>
11767 Defines the number <nb> of queries to send to resolve a server name before
11768 giving up.
11769 Default value: 3
11770
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011771 A retry occurs on name server timeout or when the full sequence of DNS query
11772 type failover is over and we need to start up from the default ANY query
11773 type.
11774
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011775timeout <event> <time>
11776 Defines timeouts related to name resolution
11777 <event> : the event on which the <time> timeout period applies to.
11778 events available are:
11779 - retry: time between two DNS queries, when no response have
11780 been received.
11781 Default value: 1s
11782 <time> : time related to the event. It follows the HAProxy time format.
11783 <time> is expressed in milliseconds.
11784
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011785 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011786
11787 resolvers mydns
11788 nameserver dns1 10.0.0.1:53
11789 nameserver dns2 10.0.0.2:53
11790 resolve_retries 3
11791 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010011792 hold other 30s
11793 hold refused 30s
11794 hold nx 30s
11795 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011796 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020011797 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011798
11799
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118006. HTTP header manipulation
11801---------------------------
11802
11803In HTTP mode, it is possible to rewrite, add or delete some of the request and
11804response headers based on regular expressions. It is also possible to block a
11805request or a response if a particular header matches a regular expression,
11806which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010011807against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011808
Willy Tarreau70dffda2014-01-30 03:07:23 +010011809If HAProxy encounters an "Informational Response" (status code 1xx), it is able
11810to process all rsp* rules which can allow, deny, rewrite or delete a header,
11811but it will refuse to add a header to any such messages as this is not
11812HTTP-compliant. The reason for still processing headers in such responses is to
11813stop and/or fix any possible information leak which may happen, for instance
11814because another downstream equipment would unconditionally add a header, or if
11815a server name appears there. When such messages are seen, normal processing
11816still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020011817
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011818This section covers common usage of the following keywords, described in detail
11819in section 4.2 :
11820
11821 - reqadd <string>
11822 - reqallow <search>
11823 - reqiallow <search>
11824 - reqdel <search>
11825 - reqidel <search>
11826 - reqdeny <search>
11827 - reqideny <search>
11828 - reqpass <search>
11829 - reqipass <search>
11830 - reqrep <search> <replace>
11831 - reqirep <search> <replace>
11832 - reqtarpit <search>
11833 - reqitarpit <search>
11834 - rspadd <string>
11835 - rspdel <search>
11836 - rspidel <search>
11837 - rspdeny <search>
11838 - rspideny <search>
11839 - rsprep <search> <replace>
11840 - rspirep <search> <replace>
11841
11842With all these keywords, the same conventions are used. The <search> parameter
11843is a POSIX extended regular expression (regex) which supports grouping through
11844parenthesis (without the backslash). Spaces and other delimiters must be
11845prefixed with a backslash ('\') to avoid confusion with a field delimiter.
11846Other characters may be prefixed with a backslash to change their meaning :
11847
11848 \t for a tab
11849 \r for a carriage return (CR)
11850 \n for a new line (LF)
11851 \ to mark a space and differentiate it from a delimiter
11852 \# to mark a sharp and differentiate it from a comment
11853 \\ to use a backslash in a regex
11854 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
11855 \xXX to write the ASCII hex code XX as in the C language
11856
11857The <replace> parameter contains the string to be used to replace the largest
11858portion of text matching the regex. It can make use of the special characters
11859above, and can reference a substring which is delimited by parenthesis in the
11860regex, by writing a backslash ('\') immediately followed by one digit from 0 to
118619 indicating the group position (0 designating the entire line). This practice
11862is very common to users of the "sed" program.
11863
11864The <string> parameter represents the string which will systematically be added
11865after the last header line. It can also use special character sequences above.
11866
11867Notes related to these keywords :
11868---------------------------------
11869 - these keywords are not always convenient to allow/deny based on header
11870 contents. It is strongly recommended to use ACLs with the "block" keyword
11871 instead, resulting in far more flexible and manageable rules.
11872
11873 - lines are always considered as a whole. It is not possible to reference
11874 a header name only or a value only. This is important because of the way
11875 headers are written (notably the number of spaces after the colon).
11876
11877 - the first line is always considered as a header, which makes it possible to
11878 rewrite or filter HTTP requests URIs or response codes, but in turn makes
11879 it harder to distinguish between headers and request line. The regex prefix
11880 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
11881 ^[^ \t:]*: matches any header name followed by a colon.
11882
11883 - for performances reasons, the number of characters added to a request or to
11884 a response is limited at build time to values between 1 and 4 kB. This
11885 should normally be far more than enough for most usages. If it is too short
11886 on occasional usages, it is possible to gain some space by removing some
11887 useless headers before adding new ones.
11888
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011889 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011890 without the 'i' letter except that they ignore case when matching patterns.
11891
11892 - when a request passes through a frontend then a backend, all req* rules
11893 from the frontend will be evaluated, then all req* rules from the backend
11894 will be evaluated. The reverse path is applied to responses.
11895
11896 - req* statements are applied after "block" statements, so that "block" is
11897 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010011898 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011899
11900
Willy Tarreau74ca5042013-06-11 23:12:07 +0200119017. Using ACLs and fetching samples
11902----------------------------------
11903
11904Haproxy is capable of extracting data from request or response streams, from
11905client or server information, from tables, environmental information etc...
11906The action of extracting such data is called fetching a sample. Once retrieved,
11907these samples may be used for various purposes such as a key to a stick-table,
11908but most common usages consist in matching them against predefined constant
11909data called patterns.
11910
11911
119127.1. ACL basics
11913---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011914
11915The use of Access Control Lists (ACL) provides a flexible solution to perform
11916content switching and generally to take decisions based on content extracted
11917from the request, the response or any environmental status. The principle is
11918simple :
11919
Willy Tarreau74ca5042013-06-11 23:12:07 +020011920 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010011921 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020011922 - apply one or multiple pattern matching methods on this sample
11923 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011924
Willy Tarreau74ca5042013-06-11 23:12:07 +020011925The actions generally consist in blocking a request, selecting a backend, or
11926adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011927
11928In order to define a test, the "acl" keyword is used. The syntax is :
11929
Willy Tarreau74ca5042013-06-11 23:12:07 +020011930 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011931
11932This creates a new ACL <aclname> or completes an existing one with new tests.
11933Those tests apply to the portion of request/response specified in <criterion>
11934and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010011935an operator which may be specified before the set of values. Optionally some
11936conversion operators may be applied to the sample, and they will be specified
11937as a comma-delimited list of keywords just after the first keyword. The values
11938are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011939
11940ACL names must be formed from upper and lower case letters, digits, '-' (dash),
11941'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
11942which means that "my_acl" and "My_Acl" are two different ACLs.
11943
11944There is no enforced limit to the number of ACLs. The unused ones do not affect
11945performance, they just consume a small amount of memory.
11946
Willy Tarreau74ca5042013-06-11 23:12:07 +020011947The criterion generally is the name of a sample fetch method, or one of its ACL
11948specific declinations. The default test method is implied by the output type of
11949this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010011950methods of a same sample fetch method. The sample fetch methods are the only
11951ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020011952
11953Sample fetch methods return data which can be of the following types :
11954 - boolean
11955 - integer (signed or unsigned)
11956 - IPv4 or IPv6 address
11957 - string
11958 - data block
11959
Willy Tarreaue6b11e42013-11-26 19:02:32 +010011960Converters transform any of these data into any of these. For example, some
11961converters might convert a string to a lower-case string while other ones
11962would turn a string to an IPv4 address, or apply a netmask to an IP address.
11963The resulting sample is of the type of the last converter applied to the list,
11964which defaults to the type of the sample fetch method.
11965
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020011966Each sample or converter returns data of a specific type, specified with its
11967keyword in this documentation. When an ACL is declared using a standard sample
11968fetch method, certain types automatically involved a default matching method
11969which are summarized in the table below :
11970
11971 +---------------------+-----------------+
11972 | Sample or converter | Default |
11973 | output type | matching method |
11974 +---------------------+-----------------+
11975 | boolean | bool |
11976 +---------------------+-----------------+
11977 | integer | int |
11978 +---------------------+-----------------+
11979 | ip | ip |
11980 +---------------------+-----------------+
11981 | string | str |
11982 +---------------------+-----------------+
11983 | binary | none, use "-m" |
11984 +---------------------+-----------------+
11985
11986Note that in order to match a binary samples, it is mandatory to specify a
11987matching method, see below.
11988
Willy Tarreau74ca5042013-06-11 23:12:07 +020011989The ACL engine can match these types against patterns of the following types :
11990 - boolean
11991 - integer or integer range
11992 - IP address / network
11993 - string (exact, substring, suffix, prefix, subdir, domain)
11994 - regular expression
11995 - hex block
11996
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011997The following ACL flags are currently supported :
11998
Willy Tarreau2b5285d2010-05-09 23:45:24 +020011999 -i : ignore case during matching of all subsequent patterns.
12000 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012001 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012002 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012003 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012004 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012005 -- : force end of flags. Useful when a string looks like one of the flags.
12006
Willy Tarreau74ca5042013-06-11 23:12:07 +020012007The "-f" flag is followed by the name of a file from which all lines will be
12008read as individual values. It is even possible to pass multiple "-f" arguments
12009if the patterns are to be loaded from multiple files. Empty lines as well as
12010lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12011will be stripped. If it is absolutely necessary to insert a valid pattern
12012beginning with a sharp, just prefix it with a space so that it is not taken for
12013a comment. Depending on the data type and match method, haproxy may load the
12014lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12015exact string matching. In this case, duplicates will automatically be removed.
12016
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012017The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12018parsed as two column file. The first column contains the patterns used by the
12019ACL, and the second column contain the samples. The sample can be used later by
12020a map. This can be useful in some rare cases where an ACL would just be used to
12021check for the existence of a pattern in a map before a mapping is applied.
12022
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012023The "-u" flag forces the unique id of the ACL. This unique id is used with the
12024socket interface to identify ACL and dynamically change its values. Note that a
12025file is always identified by its name even if an id is set.
12026
Willy Tarreau74ca5042013-06-11 23:12:07 +020012027Also, note that the "-i" flag applies to subsequent entries and not to entries
12028loaded from files preceding it. For instance :
12029
12030 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12031
12032In this example, each line of "exact-ua.lst" will be exactly matched against
12033the "user-agent" header of the request. Then each line of "generic-ua" will be
12034case-insensitively matched. Then the word "test" will be insensitively matched
12035as well.
12036
12037The "-m" flag is used to select a specific pattern matching method on the input
12038sample. All ACL-specific criteria imply a pattern matching method and generally
12039do not need this flag. However, this flag is useful with generic sample fetch
12040methods to describe how they're going to be matched against the patterns. This
12041is required for sample fetches which return data type for which there is no
12042obvious matching method (eg: string or binary). When "-m" is specified and
12043followed by a pattern matching method name, this method is used instead of the
12044default one for the criterion. This makes it possible to match contents in ways
12045that were not initially planned, or with sample fetch methods which return a
12046string. The matching method also affects the way the patterns are parsed.
12047
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012048The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12049By default, if the parser cannot parse ip address it considers that the parsed
12050string is maybe a domain name and try dns resolution. The flag "-n" disable this
12051resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12052server is not reachable, the haproxy configuration parsing may last many minutes
12053waiting fir the timeout. During this time no error messages are displayed. The
12054flag "-n" disable this behavior. Note also that during the runtime, this
12055function is disabled for the dynamic acl modifications.
12056
Willy Tarreau74ca5042013-06-11 23:12:07 +020012057There are some restrictions however. Not all methods can be used with all
12058sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12059be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012060
12061 - "found" : only check if the requested sample could be found in the stream,
12062 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012063 to pass any pattern to avoid confusion. This matching method is
12064 particularly useful to detect presence of certain contents such
12065 as headers, cookies, etc... even if they are empty and without
12066 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012067
12068 - "bool" : check the value as a boolean. It can only be applied to fetches
12069 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012070 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012071
12072 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012073 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012074
12075 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012076 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012077
12078 - "bin" : match the contents against an hexadecimal string representing a
12079 binary sequence. This may be used with binary or string samples.
12080
12081 - "len" : match the sample's length as an integer. This may be used with
12082 binary or string samples.
12083
Willy Tarreau74ca5042013-06-11 23:12:07 +020012084 - "str" : exact match : match the contents against a string. This may be
12085 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012086
Willy Tarreau74ca5042013-06-11 23:12:07 +020012087 - "sub" : substring match : check that the contents contain at least one of
12088 the provided string patterns. This may be used with binary or
12089 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012090
Willy Tarreau74ca5042013-06-11 23:12:07 +020012091 - "reg" : regex match : match the contents against a list of regular
12092 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012093
Willy Tarreau74ca5042013-06-11 23:12:07 +020012094 - "beg" : prefix match : check that the contents begin like the provided
12095 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012096
Willy Tarreau74ca5042013-06-11 23:12:07 +020012097 - "end" : suffix match : check that the contents end like the provided
12098 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012099
Willy Tarreau74ca5042013-06-11 23:12:07 +020012100 - "dir" : subdir match : check that a slash-delimited portion of the
12101 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012102 This may be used with binary or string samples.
12103
Willy Tarreau74ca5042013-06-11 23:12:07 +020012104 - "dom" : domain match : check that a dot-delimited portion of the contents
12105 exactly match one of the provided string patterns. This may be
12106 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012107
12108For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12109request, it is possible to do :
12110
12111 acl jsess_present cook(JSESSIONID) -m found
12112
12113In order to apply a regular expression on the 500 first bytes of data in the
12114buffer, one would use the following acl :
12115
12116 acl script_tag payload(0,500) -m reg -i <script>
12117
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012118On systems where the regex library is much slower when using "-i", it is
12119possible to convert the sample to lowercase before matching, like this :
12120
12121 acl script_tag payload(0,500),lower -m reg <script>
12122
Willy Tarreau74ca5042013-06-11 23:12:07 +020012123All ACL-specific criteria imply a default matching method. Most often, these
12124criteria are composed by concatenating the name of the original sample fetch
12125method and the matching method. For example, "hdr_beg" applies the "beg" match
12126to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12127criteria rely on a sample fetch method, it is always possible instead to use
12128the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012129
Willy Tarreau74ca5042013-06-11 23:12:07 +020012130If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012131the matching method is simply applied to the underlying sample fetch method.
12132For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012133
Willy Tarreau74ca5042013-06-11 23:12:07 +020012134 acl short_form hdr_beg(host) www.
12135 acl alternate1 hdr_beg(host) -m beg www.
12136 acl alternate2 hdr_dom(host) -m beg www.
12137 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012138
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012139
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012140The table below summarizes the compatibility matrix between sample or converter
12141types and the pattern types to fetch against. It indicates for each compatible
12142combination the name of the matching method to be used, surrounded with angle
12143brackets ">" and "<" when the method is the default one and will work by
12144default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012145
Willy Tarreau74ca5042013-06-11 23:12:07 +020012146 +-------------------------------------------------+
12147 | Input sample type |
12148 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012149 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012150 +----------------------+---------+---------+---------+---------+---------+
12151 | none (presence only) | found | found | found | found | found |
12152 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012153 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012154 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012155 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012156 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012157 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012158 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012159 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012160 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012161 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012162 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012163 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012164 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012165 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012166 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012167 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012168 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012169 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012170 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012171 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012172 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012173 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012174 +----------------------+---------+---------+---------+---------+---------+
12175 | hex block | | | | bin | bin |
12176 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012177
12178
Willy Tarreau74ca5042013-06-11 23:12:07 +0200121797.1.1. Matching booleans
12180------------------------
12181
12182In order to match a boolean, no value is needed and all values are ignored.
12183Boolean matching is used by default for all fetch methods of type "boolean".
12184When boolean matching is used, the fetched value is returned as-is, which means
12185that a boolean "true" will always match and a boolean "false" will never match.
12186
12187Boolean matching may also be enforced using "-m bool" on fetch methods which
12188return an integer value. Then, integer value 0 is converted to the boolean
12189"false" and all other values are converted to "true".
12190
Willy Tarreau6a06a402007-07-15 20:15:28 +020012191
Willy Tarreau74ca5042013-06-11 23:12:07 +0200121927.1.2. Matching integers
12193------------------------
12194
12195Integer matching applies by default to integer fetch methods. It can also be
12196enforced on boolean fetches using "-m int". In this case, "false" is converted
12197to the integer 0, and "true" is converted to the integer 1.
12198
12199Integer matching also supports integer ranges and operators. Note that integer
12200matching only applies to positive values. A range is a value expressed with a
12201lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012202
12203For instance, "1024:65535" is a valid range to represent a range of
12204unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12205representation of privileged ports, and ":1023" would also work.
12206
Willy Tarreau62644772008-07-16 18:36:06 +020012207As a special case, some ACL functions support decimal numbers which are in fact
12208two integers separated by a dot. This is used with some version checks for
12209instance. All integer properties apply to those decimal numbers, including
12210ranges and operators.
12211
Willy Tarreau6a06a402007-07-15 20:15:28 +020012212For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012213operators with ranges does not make much sense and is strongly discouraged.
12214Similarly, it does not make much sense to perform order comparisons with a set
12215of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012216
Willy Tarreau0ba27502007-12-24 16:55:16 +010012217Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012218
12219 eq : true if the tested value equals at least one value
12220 ge : true if the tested value is greater than or equal to at least one value
12221 gt : true if the tested value is greater than at least one value
12222 le : true if the tested value is less than or equal to at least one value
12223 lt : true if the tested value is less than at least one value
12224
Willy Tarreau0ba27502007-12-24 16:55:16 +010012225For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012226
12227 acl negative-length hdr_val(content-length) lt 0
12228
Willy Tarreau62644772008-07-16 18:36:06 +020012229This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12230
12231 acl sslv3 req_ssl_ver 3:3.1
12232
Willy Tarreau6a06a402007-07-15 20:15:28 +020012233
Willy Tarreau74ca5042013-06-11 23:12:07 +0200122347.1.3. Matching strings
12235-----------------------
12236
12237String matching applies to string or binary fetch methods, and exists in 6
12238different forms :
12239
12240 - exact match (-m str) : the extracted string must exactly match the
12241 patterns ;
12242
12243 - substring match (-m sub) : the patterns are looked up inside the
12244 extracted string, and the ACL matches if any of them is found inside ;
12245
12246 - prefix match (-m beg) : the patterns are compared with the beginning of
12247 the extracted string, and the ACL matches if any of them matches.
12248
12249 - suffix match (-m end) : the patterns are compared with the end of the
12250 extracted string, and the ACL matches if any of them matches.
12251
Baptiste Assmann33db6002016-03-06 23:32:10 +010012252 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012253 string, delimited with slashes ("/"), and the ACL matches if any of them
12254 matches.
12255
12256 - domain match (-m dom) : the patterns are looked up inside the extracted
12257 string, delimited with dots ("."), and the ACL matches if any of them
12258 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012259
12260String matching applies to verbatim strings as they are passed, with the
12261exception of the backslash ("\") which makes it possible to escape some
12262characters such as the space. If the "-i" flag is passed before the first
12263string, then the matching will be performed ignoring the case. In order
12264to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012265before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012266
12267
Willy Tarreau74ca5042013-06-11 23:12:07 +0200122687.1.4. Matching regular expressions (regexes)
12269---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012270
12271Just like with string matching, regex matching applies to verbatim strings as
12272they are passed, with the exception of the backslash ("\") which makes it
12273possible to escape some characters such as the space. If the "-i" flag is
12274passed before the first regex, then the matching will be performed ignoring
12275the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012276the "--" flag before the first string. Same principle applies of course to
12277match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012278
12279
Willy Tarreau74ca5042013-06-11 23:12:07 +0200122807.1.5. Matching arbitrary data blocks
12281-------------------------------------
12282
12283It is possible to match some extracted samples against a binary block which may
12284not safely be represented as a string. For this, the patterns must be passed as
12285a series of hexadecimal digits in an even number, when the match method is set
12286to binary. Each sequence of two digits will represent a byte. The hexadecimal
12287digits may be used upper or lower case.
12288
12289Example :
12290 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12291 acl hello payload(0,6) -m bin 48656c6c6f0a
12292
12293
122947.1.6. Matching IPv4 and IPv6 addresses
12295---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012296
12297IPv4 addresses values can be specified either as plain addresses or with a
12298netmask appended, in which case the IPv4 address matches whenever it is
12299within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012300host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012301difficult to read and debug configurations. If hostnames are used, you should
12302at least ensure that they are present in /etc/hosts so that the configuration
12303does not depend on any random DNS match at the moment the configuration is
12304parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012305
Daniel Schnellereba56342016-04-13 00:26:52 +020012306The dotted IPv4 address notation is supported in both regular as well as the
12307abbreviated form with all-0-octets omitted:
12308
12309 +------------------+------------------+------------------+
12310 | Example 1 | Example 2 | Example 3 |
12311 +------------------+------------------+------------------+
12312 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12313 | 192.168.1 | 10.12 | 127.1 |
12314 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12315 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12316 +------------------+------------------+------------------+
12317
12318Notice that this is different from RFC 4632 CIDR address notation in which
12319192.168.42/24 would be equivalent to 192.168.42.0/24.
12320
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012321IPv6 may be entered in their usual form, with or without a netmask appended.
12322Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12323trouble with randomly resolved IP addresses, host names are never allowed in
12324IPv6 patterns.
12325
12326HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12327following situations :
12328 - tested address is IPv4, pattern address is IPv4, the match applies
12329 in IPv4 using the supplied mask if any.
12330 - tested address is IPv6, pattern address is IPv6, the match applies
12331 in IPv6 using the supplied mask if any.
12332 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12333 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12334 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12335 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12336 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12337 applied in IPv6 using the supplied IPv6 mask.
12338
Willy Tarreau74ca5042013-06-11 23:12:07 +020012339
123407.2. Using ACLs to form conditions
12341----------------------------------
12342
12343Some actions are only performed upon a valid condition. A condition is a
12344combination of ACLs with operators. 3 operators are supported :
12345
12346 - AND (implicit)
12347 - OR (explicit with the "or" keyword or the "||" operator)
12348 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012349
Willy Tarreau74ca5042013-06-11 23:12:07 +020012350A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012351
Willy Tarreau74ca5042013-06-11 23:12:07 +020012352 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012353
Willy Tarreau74ca5042013-06-11 23:12:07 +020012354Such conditions are generally used after an "if" or "unless" statement,
12355indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012356
Willy Tarreau74ca5042013-06-11 23:12:07 +020012357For instance, to block HTTP requests to the "*" URL with methods other than
12358"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12359requests with a content-length greater than 0, and finally every request which
12360is not either GET/HEAD/POST/OPTIONS !
12361
12362 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012363 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12364 http-request deny if METH_GET HTTP_CONTENT
12365 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012366
12367To select a different backend for requests to static contents on the "www" site
12368and to every request on the "img", "video", "download" and "ftp" hosts :
12369
12370 acl url_static path_beg /static /images /img /css
12371 acl url_static path_end .gif .png .jpg .css .js
12372 acl host_www hdr_beg(host) -i www
12373 acl host_static hdr_beg(host) -i img. video. download. ftp.
12374
12375 # now use backend "static" for all static-only hosts, and for static urls
12376 # of host "www". Use backend "www" for the rest.
12377 use_backend static if host_static or host_www url_static
12378 use_backend www if host_www
12379
12380It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
12381expressions that are built on the fly without needing to be declared. They must
12382be enclosed between braces, with a space before and after each brace (because
12383the braces must be seen as independent words). Example :
12384
12385 The following rule :
12386
12387 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012388 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020012389
12390 Can also be written that way :
12391
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012392 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020012393
12394It is generally not recommended to use this construct because it's a lot easier
12395to leave errors in the configuration when written that way. However, for very
12396simple rules matching only one source IP address for instance, it can make more
12397sense to use them than to declare ACLs with random names. Another example of
12398good use is the following :
12399
12400 With named ACLs :
12401
12402 acl site_dead nbsrv(dynamic) lt 2
12403 acl site_dead nbsrv(static) lt 2
12404 monitor fail if site_dead
12405
12406 With anonymous ACLs :
12407
12408 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
12409
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012410See section 4.2 for detailed help on the "http-request deny" and "use_backend"
12411keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012412
12413
124147.3. Fetching samples
12415---------------------
12416
12417Historically, sample fetch methods were only used to retrieve data to match
12418against patterns using ACLs. With the arrival of stick-tables, a new class of
12419sample fetch methods was created, most often sharing the same syntax as their
12420ACL counterpart. These sample fetch methods are also known as "fetches". As
12421of now, ACLs and fetches have converged. All ACL fetch methods have been made
12422available as fetch methods, and ACLs may use any sample fetch method as well.
12423
12424This section details all available sample fetch methods and their output type.
12425Some sample fetch methods have deprecated aliases that are used to maintain
12426compatibility with existing configurations. They are then explicitly marked as
12427deprecated and should not be used in new setups.
12428
12429The ACL derivatives are also indicated when available, with their respective
12430matching methods. These ones all have a well defined default pattern matching
12431method, so it is never necessary (though allowed) to pass the "-m" option to
12432indicate how the sample will be matched using ACLs.
12433
12434As indicated in the sample type versus matching compatibility matrix above,
12435when using a generic sample fetch method in an ACL, the "-m" option is
12436mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
12437the same keyword exists as an ACL keyword and as a standard fetch method, the
12438ACL engine will automatically pick the ACL-only one by default.
12439
12440Some of these keywords support one or multiple mandatory arguments, and one or
12441multiple optional arguments. These arguments are strongly typed and are checked
12442when the configuration is parsed so that there is no risk of running with an
12443incorrect argument (eg: an unresolved backend name). Fetch function arguments
12444are passed between parenthesis and are delimited by commas. When an argument
12445is optional, it will be indicated below between square brackets ('[ ]'). When
12446all arguments are optional, the parenthesis may be omitted.
12447
12448Thus, the syntax of a standard sample fetch method is one of the following :
12449 - name
12450 - name(arg1)
12451 - name(arg1,arg2)
12452
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012453
124547.3.1. Converters
12455-----------------
12456
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012457Sample fetch methods may be combined with transformations to be applied on top
12458of the fetched sample (also called "converters"). These combinations form what
12459is called "sample expressions" and the result is a "sample". Initially this
12460was only supported by "stick on" and "stick store-request" directives but this
12461has now be extended to all places where samples may be used (acls, log-format,
12462unique-id-format, add-header, ...).
12463
12464These transformations are enumerated as a series of specific keywords after the
12465sample fetch method. These keywords may equally be appended immediately after
12466the fetch keyword's argument, delimited by a comma. These keywords can also
12467support some arguments (eg: a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012468
Willy Tarreau97707872015-01-27 15:12:13 +010012469A certain category of converters are bitwise and arithmetic operators which
12470support performing basic operations on integers. Some bitwise operations are
12471supported (and, or, xor, cpl) and some arithmetic operations are supported
12472(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
12473bool) which make it possible to report a match without having to write an ACL.
12474
Willy Tarreau74ca5042013-06-11 23:12:07 +020012475The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010012476
Ben Shillitof25e8e52016-12-02 14:25:37 +00001247751d.single(<prop>[,<prop>*])
12478 Returns values for the properties requested as a string, where values are
12479 separated by the delimiter specified with "51degrees-property-separator".
12480 The device is identified using the User-Agent header passed to the
12481 converter. The function can be passed up to five property names, and if a
12482 property name can't be found, the value "NoData" is returned.
12483
12484 Example :
12485 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
12486 # containg values for the three properties requested by using the
12487 # User-Agent passed to the converter.
12488 frontend http-in
12489 bind *:8081
12490 default_backend servers
12491 http-request set-header X-51D-DeviceTypeMobileTablet \
12492 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
12493
Willy Tarreau97707872015-01-27 15:12:13 +010012494add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012495 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012496 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010012497 name. The name of the variable starts with an indication about its scope. The
12498 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012499 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012500 "sess" : the variable is shared with the whole session
12501 "txn" : the variable is shared with the transaction (request and response)
12502 "req" : the variable is shared only during request processing
12503 "res" : the variable is shared only during response processing
12504 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012505 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012506
12507and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012508 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012509 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010012510 numeric value or a variable name. The name of the variable starts with an
12511 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012512 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012513 "sess" : the variable is shared with the whole session
12514 "txn" : the variable is shared with the transaction (request and response)
12515 "req" : the variable is shared only during request processing
12516 "res" : the variable is shared only during response processing
12517 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012518 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012519
Holger Just1bfc24b2017-05-06 00:56:53 +020012520b64dec
12521 Converts (decodes) a base64 encoded input string to its binary
12522 representation. It performs the inverse operation of base64().
12523
Emeric Brun53d1a982014-04-30 18:21:37 +020012524base64
12525 Converts a binary input sample to a base64 string. It is used to log or
12526 transfer binary content in a way that can be reliably transferred (eg:
12527 an SSL ID can be copied in a header).
12528
Willy Tarreau97707872015-01-27 15:12:13 +010012529bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012530 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010012531 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
12532 used to report true/false for bit testing on input values (eg: verify the
12533 presence of a flag).
12534
Emeric Brun54c4ac82014-11-03 15:32:43 +010012535bytes(<offset>[,<length>])
12536 Extracts some bytes from an input binary sample. The result is a binary
12537 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010012538 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010012539
Willy Tarreau97707872015-01-27 15:12:13 +010012540cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012541 Takes the input value of type signed integer, applies a ones-complement
12542 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010012543
Willy Tarreau80599772015-01-20 19:35:24 +010012544crc32([<avalanche>])
12545 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
12546 hash function. Optionally, it is possible to apply a full avalanche hash
12547 function to the output if the optional <avalanche> argument equals 1. This
12548 converter uses the same functions as used by the various hash-based load
12549 balancing algorithms, so it will provide exactly the same results. It is
12550 provided for compatibility with other software which want a CRC32 to be
12551 computed on some input keys, so it follows the most common implementation as
12552 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
12553 but may provide a better or at least less predictable distribution. It must
12554 not be used for security purposes as a 32-bit hash is trivial to break. See
12555 also "djb2", "sdbm", "wt6" and the "hash-type" directive.
12556
David Carlier29b3ca32015-09-25 14:09:21 +010012557da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020012558 Asks the DeviceAtlas converter to identify the User Agent string passed on
12559 input, and to emit a string made of the concatenation of the properties
12560 enumerated in argument, delimited by the separator defined by the global
12561 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000012562 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020012563 configuration language.
12564
12565 Example:
12566 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020012567 bind *:8881
12568 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000012569 http-request set-header X-DeviceAtlas-Data %[req.fhdr(User-Agent),da-csv(primaryHardwareType,osName,osVersion,browserName,browserVersion,browserRenderingEngine)]
David Carlier4542b102015-06-01 13:54:29 +020012570
Thierry FOURNIER9687c772015-05-07 15:46:29 +020012571debug
12572 This converter is used as debug tool. It dumps on screen the content and the
12573 type of the input sample. The sample is returned as is on its output. This
12574 converter only exists when haproxy was built with debugging enabled.
12575
Willy Tarreau97707872015-01-27 15:12:13 +010012576div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012577 Divides the input value of type signed integer by <value>, and returns the
12578 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012579 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010012580 variable name. The name of the variable starts with an indication about its
12581 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012582 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012583 "sess" : the variable is shared with the whole session
12584 "txn" : the variable is shared with the transaction (request and response)
12585 "req" : the variable is shared only during request processing
12586 "res" : the variable is shared only during response processing
12587 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012588 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012589
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012590djb2([<avalanche>])
12591 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
12592 hash function. Optionally, it is possible to apply a full avalanche hash
12593 function to the output if the optional <avalanche> argument equals 1. This
12594 converter uses the same functions as used by the various hash-based load
12595 balancing algorithms, so it will provide exactly the same results. It is
12596 mostly intended for debugging, but can be used as a stick-table entry to
12597 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010012598 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6" and the
12599 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012600
Willy Tarreau97707872015-01-27 15:12:13 +010012601even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012602 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010012603 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
12604
Emeric Brunf399b0d2014-11-03 17:07:03 +010012605field(<index>,<delimiters>)
12606 Extracts the substring at the given index considering given delimiters from
12607 an input string. Indexes start at 1 and delimiters are a string formatted
12608 list of chars.
12609
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012610hex
12611 Converts a binary input sample to an hex string containing two hex digits per
12612 input byte. It is used to log or transfer hex dumps of some binary input data
12613 in a way that can be reliably transferred (eg: an SSL ID can be copied in a
12614 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010012615
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012616http_date([<offset>])
12617 Converts an integer supposed to contain a date since epoch to a string
12618 representing this date in a format suitable for use in HTTP header fields. If
12619 an offset value is specified, then it is a number of seconds that is added to
12620 the date before the conversion is operated. This is particularly useful to
12621 emit Date header fields, Expires values in responses when combined with a
12622 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012623
Willy Tarreaud9f316a2014-07-10 14:03:38 +020012624in_table(<table>)
12625 Uses the string representation of the input sample to perform a look up in
12626 the specified table. If the key is not found in the table, a boolean false
12627 is returned. Otherwise a boolean true is returned. This can be used to verify
12628 the presence of a certain key in a table tracking some elements (eg: whether
12629 or not a source IP address or an Authorization header was already seen).
12630
Willy Tarreauffcb2e42014-07-10 16:29:08 +020012631ipmask(<mask>)
12632 Apply a mask to an IPv4 address, and use the result for lookups and storage.
12633 This can be used to make all hosts within a certain mask to share the same
12634 table entries and as such use the same server. The mask can be passed in
12635 dotted form (eg: 255.255.255.0) or in CIDR form (eg: 24).
12636
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012637json([<input-code>])
12638 Escapes the input string and produces an ASCII ouput string ready to use as a
12639 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020012640 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012641 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
12642 of errors:
12643 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
12644 bytes, ...)
12645 - invalid range (the decoded value is within a UTF-8 prohibited range),
12646 - code overlong (the value is encoded with more bytes than necessary).
12647
12648 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
12649 character is greater than 0xffff because the JSON string escape specification
12650 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
12651 in 4 variants designated by a combination of two suffix letters : "p" for
12652 "permissive" and "s" for "silently ignore". The behaviors of the decoders
12653 are :
12654 - "ascii" : never fails ;
12655 - "utf8" : fails on any detected errors ;
12656 - "utf8s" : never fails, but removes characters corresponding to errors ;
12657 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
12658 error ;
12659 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
12660 characters corresponding to the other errors.
12661
12662 This converter is particularly useful for building properly escaped JSON for
12663 logging to servers which consume JSON-formated traffic logs.
12664
12665 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012666 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020012667 capture request header user-agent len 150
12668 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012669
12670 Input request from client 127.0.0.1:
12671 GET / HTTP/1.0
12672 User-Agent: Very "Ugly" UA 1/2
12673
12674 Output log:
12675 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
12676
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012677language(<value>[,<default>])
12678 Returns the value with the highest q-factor from a list as extracted from the
12679 "accept-language" header using "req.fhdr". Values with no q-factor have a
12680 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
12681 belong to the list of semi-colon delimited <values> will be considered. The
12682 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
12683 given list and a default value is provided, it is returned. Note that language
12684 names may have a variant after a dash ('-'). If this variant is present in the
12685 list, it will be matched, but if it is not, only the base language is checked.
12686 The match is case-sensitive, and the output string is always one of those
12687 provided in arguments. The ordering of arguments is meaningless, only the
12688 ordering of the values in the request counts, as the first value among
12689 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020012690
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012691 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020012692
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012693 # this configuration switches to the backend matching a
12694 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020012695
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012696 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
12697 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
12698 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
12699 use_backend spanish if es
12700 use_backend french if fr
12701 use_backend english if en
12702 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020012703
Willy Tarreauffcb2e42014-07-10 16:29:08 +020012704lower
12705 Convert a string sample to lower case. This can only be placed after a string
12706 sample fetch function or after a transformation keyword returning a string
12707 type. The result is of type string.
12708
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020012709ltime(<format>[,<offset>])
12710 Converts an integer supposed to contain a date since epoch to a string
12711 representing this date in local time using a format defined by the <format>
12712 string using strftime(3). The purpose is to allow any date format to be used
12713 in logs. An optional <offset> in seconds may be applied to the input date
12714 (positive or negative). See the strftime() man page for the format supported
12715 by your operating system. See also the utime converter.
12716
12717 Example :
12718
12719 # Emit two colons, one with the local time and another with ip:port
12720 # Eg: 20140710162350 127.0.0.1:57325
12721 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
12722
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012723map(<map_file>[,<default_value>])
12724map_<match_type>(<map_file>[,<default_value>])
12725map_<match_type>_<output_type>(<map_file>[,<default_value>])
12726 Search the input value from <map_file> using the <match_type> matching method,
12727 and return the associated value converted to the type <output_type>. If the
12728 input value cannot be found in the <map_file>, the converter returns the
12729 <default_value>. If the <default_value> is not set, the converter fails and
12730 acts as if no input value could be fetched. If the <match_type> is not set, it
12731 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
12732 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
12733 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010012734
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012735 It is important to avoid overlapping between the keys : IP addresses and
12736 strings are stored in trees, so the first of the finest match will be used.
12737 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010012738
Tim Düsterhus4896c442016-11-29 02:15:19 +010012739 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012740 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010012741
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012742 input type | match method | output type str | output type int | output type ip
12743 -----------+--------------+-----------------+-----------------+---------------
12744 str | str | map_str | map_str_int | map_str_ip
12745 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020012746 str | beg | map_beg | map_beg_int | map_end_ip
12747 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012748 str | sub | map_sub | map_sub_int | map_sub_ip
12749 -----------+--------------+-----------------+-----------------+---------------
12750 str | dir | map_dir | map_dir_int | map_dir_ip
12751 -----------+--------------+-----------------+-----------------+---------------
12752 str | dom | map_dom | map_dom_int | map_dom_ip
12753 -----------+--------------+-----------------+-----------------+---------------
12754 str | end | map_end | map_end_int | map_end_ip
12755 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080012756 str | reg | map_reg | map_reg_int | map_reg_ip
12757 -----------+--------------+-----------------+-----------------+---------------
12758 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012759 -----------+--------------+-----------------+-----------------+---------------
12760 int | int | map_int | map_int_int | map_int_ip
12761 -----------+--------------+-----------------+-----------------+---------------
12762 ip | ip | map_ip | map_ip_int | map_ip_ip
12763 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010012764
Thierry Fournier8feaa662016-02-10 22:55:20 +010012765 The special map called "map_regm" expect matching zone in the regular
12766 expression and modify the output replacing back reference (like "\1") by
12767 the corresponding match text.
12768
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012769 The file contains one key + value per line. Lines which start with '#' are
12770 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
12771 is then the first "word" (series of non-space/tabs characters), and the value
12772 is what follows this series of space/tab till the end of the line excluding
12773 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010012774
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012775 Example :
12776
12777 # this is a comment and is ignored
12778 2.22.246.0/23 United Kingdom \n
12779 <-><-----------><--><------------><---->
12780 | | | | `- trailing spaces ignored
12781 | | | `---------- value
12782 | | `-------------------- middle spaces ignored
12783 | `---------------------------- key
12784 `------------------------------------ leading spaces ignored
12785
Willy Tarreau97707872015-01-27 15:12:13 +010012786mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012787 Divides the input value of type signed integer by <value>, and returns the
12788 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012789 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010012790 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012791 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012792 "sess" : the variable is shared with the whole session
12793 "txn" : the variable is shared with the transaction (request and response)
12794 "req" : the variable is shared only during request processing
12795 "res" : the variable is shared only during response processing
12796 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012797 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012798
12799mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012800 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020012801 the product as an signed integer. In case of overflow, the largest possible
12802 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012803 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010012804 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012805 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012806 "sess" : the variable is shared with the whole session
12807 "txn" : the variable is shared with the transaction (request and response)
12808 "req" : the variable is shared only during request processing
12809 "res" : the variable is shared only during response processing
12810 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012811 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012812
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010012813nbsrv
12814 Takes an input value of type string, interprets it as a backend name and
12815 returns the number of usable servers in that backend. Can be used in places
12816 where we want to look up a backend from a dynamic name, like a result of a
12817 map lookup.
12818
Willy Tarreau97707872015-01-27 15:12:13 +010012819neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012820 Takes the input value of type signed integer, computes the opposite value,
12821 and returns the remainder as an signed integer. 0 is identity. This operator
12822 is provided for reversed subtracts : in order to subtract the input from a
12823 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010012824
12825not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012826 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010012827 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
12828 used to report true/false for bit testing on input values (eg: verify the
12829 absence of a flag).
12830
12831odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012832 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010012833 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
12834
12835or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012836 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012837 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010012838 numeric value or a variable name. The name of the variable starts with an
12839 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012840 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012841 "sess" : the variable is shared with the whole session
12842 "txn" : the variable is shared with the transaction (request and response)
12843 "req" : the variable is shared only during request processing
12844 "res" : the variable is shared only during response processing
12845 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012846 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012847
Willy Tarreauc4dc3502015-01-23 20:39:28 +010012848regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010012849 Applies a regex-based substitution to the input string. It does the same
12850 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
12851 default it will replace in the input string the first occurrence of the
12852 largest part matching the regular expression <regex> with the substitution
12853 string <subst>. It is possible to replace all occurrences instead by adding
12854 the flag "g" in the third argument <flags>. It is also possible to make the
12855 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
12856 string, it is made up from the concatenation of all desired flags. Thus if
12857 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
12858 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010012859 configuration parser, some characters such as closing parenthesis, closing
12860 square brackets or comma are not possible to use in the arguments. The first
12861 use of this converter is to replace certain characters or sequence of
12862 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010012863
12864 Example :
12865
12866 # de-duplicate "/" in header "x-path".
12867 # input: x-path: /////a///b/c/xzxyz/
12868 # output: x-path: /a/b/c/xzxyz/
12869 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
12870
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020012871capture-req(<id>)
12872 Capture the string entry in the request slot <id> and returns the entry as
12873 is. If the slot doesn't exist, the capture fails silently.
12874
12875 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020012876 "http-response capture", "capture.req.hdr" and
12877 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020012878
12879capture-res(<id>)
12880 Capture the string entry in the response slot <id> and returns the entry as
12881 is. If the slot doesn't exist, the capture fails silently.
12882
12883 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020012884 "http-response capture", "capture.req.hdr" and
12885 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020012886
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012887sdbm([<avalanche>])
12888 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
12889 hash function. Optionally, it is possible to apply a full avalanche hash
12890 function to the output if the optional <avalanche> argument equals 1. This
12891 converter uses the same functions as used by the various hash-based load
12892 balancing algorithms, so it will provide exactly the same results. It is
12893 mostly intended for debugging, but can be used as a stick-table entry to
12894 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010012895 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6" and the
12896 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012897
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012898set-var(<var name>)
Daniel Schneller0b547052016-03-21 20:46:57 +010012899 Sets a variable with the input content and returns the content on the output as
12900 is. The variable keeps the value and the associated input type. The name of the
12901 variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012902 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012903 "sess" : the variable is shared with the whole session
12904 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012905 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010012906 "req" : the variable is shared only during request processing,
12907 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012908 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012909 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012910
Willy Tarreau97707872015-01-27 15:12:13 +010012911sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012912 Subtracts <value> from the input value of type signed integer, and returns
12913 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012914 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010012915 or a variable name. The name of the variable starts with an indication about
12916 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012917 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012918 "sess" : the variable is shared with the whole session
12919 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012920 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010012921 "req" : the variable is shared only during request processing,
12922 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012923 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012924 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012925
Willy Tarreaud9f316a2014-07-10 14:03:38 +020012926table_bytes_in_rate(<table>)
12927 Uses the string representation of the input sample to perform a look up in
12928 the specified table. If the key is not found in the table, integer value zero
12929 is returned. Otherwise the converter returns the average client-to-server
12930 bytes rate associated with the input sample in the designated table, measured
12931 in amount of bytes over the period configured in the table. See also the
12932 sc_bytes_in_rate sample fetch keyword.
12933
12934
12935table_bytes_out_rate(<table>)
12936 Uses the string representation of the input sample to perform a look up in
12937 the specified table. If the key is not found in the table, integer value zero
12938 is returned. Otherwise the converter returns the average server-to-client
12939 bytes rate associated with the input sample in the designated table, measured
12940 in amount of bytes over the period configured in the table. See also the
12941 sc_bytes_out_rate sample fetch keyword.
12942
12943table_conn_cnt(<table>)
12944 Uses the string representation of the input sample to perform a look up in
12945 the specified table. If the key is not found in the table, integer value zero
12946 is returned. Otherwise the converter returns the cumulated amount of incoming
12947 connections associated with the input sample in the designated table. See
12948 also the sc_conn_cnt sample fetch keyword.
12949
12950table_conn_cur(<table>)
12951 Uses the string representation of the input sample to perform a look up in
12952 the specified table. If the key is not found in the table, integer value zero
12953 is returned. Otherwise the converter returns the current amount of concurrent
12954 tracked connections associated with the input sample in the designated table.
12955 See also the sc_conn_cur sample fetch keyword.
12956
12957table_conn_rate(<table>)
12958 Uses the string representation of the input sample to perform a look up in
12959 the specified table. If the key is not found in the table, integer value zero
12960 is returned. Otherwise the converter returns the average incoming connection
12961 rate associated with the input sample in the designated table. See also the
12962 sc_conn_rate sample fetch keyword.
12963
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012964table_gpt0(<table>)
12965 Uses the string representation of the input sample to perform a look up in
12966 the specified table. If the key is not found in the table, boolean value zero
12967 is returned. Otherwise the converter returns the current value of the first
12968 general purpose tag associated with the input sample in the designated table.
12969 See also the sc_get_gpt0 sample fetch keyword.
12970
Willy Tarreaud9f316a2014-07-10 14:03:38 +020012971table_gpc0(<table>)
12972 Uses the string representation of the input sample to perform a look up in
12973 the specified table. If the key is not found in the table, integer value zero
12974 is returned. Otherwise the converter returns the current value of the first
12975 general purpose counter associated with the input sample in the designated
12976 table. See also the sc_get_gpc0 sample fetch keyword.
12977
12978table_gpc0_rate(<table>)
12979 Uses the string representation of the input sample to perform a look up in
12980 the specified table. If the key is not found in the table, integer value zero
12981 is returned. Otherwise the converter returns the frequency which the gpc0
12982 counter was incremented over the configured period in the table, associated
12983 with the input sample in the designated table. See also the sc_get_gpc0_rate
12984 sample fetch keyword.
12985
12986table_http_err_cnt(<table>)
12987 Uses the string representation of the input sample to perform a look up in
12988 the specified table. If the key is not found in the table, integer value zero
12989 is returned. Otherwise the converter returns the cumulated amount of HTTP
12990 errors associated with the input sample in the designated table. See also the
12991 sc_http_err_cnt sample fetch keyword.
12992
12993table_http_err_rate(<table>)
12994 Uses the string representation of the input sample to perform a look up in
12995 the specified table. If the key is not found in the table, integer value zero
12996 is returned. Otherwise the average rate of HTTP errors associated with the
12997 input sample in the designated table, measured in amount of errors over the
12998 period configured in the table. See also the sc_http_err_rate sample fetch
12999 keyword.
13000
13001table_http_req_cnt(<table>)
13002 Uses the string representation of the input sample to perform a look up in
13003 the specified table. If the key is not found in the table, integer value zero
13004 is returned. Otherwise the converter returns the cumulated amount of HTTP
13005 requests associated with the input sample in the designated table. See also
13006 the sc_http_req_cnt sample fetch keyword.
13007
13008table_http_req_rate(<table>)
13009 Uses the string representation of the input sample to perform a look up in
13010 the specified table. If the key is not found in the table, integer value zero
13011 is returned. Otherwise the average rate of HTTP requests associated with the
13012 input sample in the designated table, measured in amount of requests over the
13013 period configured in the table. See also the sc_http_req_rate sample fetch
13014 keyword.
13015
13016table_kbytes_in(<table>)
13017 Uses the string representation of the input sample to perform a look up in
13018 the specified table. If the key is not found in the table, integer value zero
13019 is returned. Otherwise the converter returns the cumulated amount of client-
13020 to-server data associated with the input sample in the designated table,
13021 measured in kilobytes. The test is currently performed on 32-bit integers,
13022 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13023 keyword.
13024
13025table_kbytes_out(<table>)
13026 Uses the string representation of the input sample to perform a look up in
13027 the specified table. If the key is not found in the table, integer value zero
13028 is returned. Otherwise the converter returns the cumulated amount of server-
13029 to-client data associated with the input sample in the designated table,
13030 measured in kilobytes. The test is currently performed on 32-bit integers,
13031 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13032 keyword.
13033
13034table_server_id(<table>)
13035 Uses the string representation of the input sample to perform a look up in
13036 the specified table. If the key is not found in the table, integer value zero
13037 is returned. Otherwise the converter returns the server ID associated with
13038 the input sample in the designated table. A server ID is associated to a
13039 sample by a "stick" rule when a connection to a server succeeds. A server ID
13040 zero means that no server is associated with this key.
13041
13042table_sess_cnt(<table>)
13043 Uses the string representation of the input sample to perform a look up in
13044 the specified table. If the key is not found in the table, integer value zero
13045 is returned. Otherwise the converter returns the cumulated amount of incoming
13046 sessions associated with the input sample in the designated table. Note that
13047 a session here refers to an incoming connection being accepted by the
13048 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13049 keyword.
13050
13051table_sess_rate(<table>)
13052 Uses the string representation of the input sample to perform a look up in
13053 the specified table. If the key is not found in the table, integer value zero
13054 is returned. Otherwise the converter returns the average incoming session
13055 rate associated with the input sample in the designated table. Note that a
13056 session here refers to an incoming connection being accepted by the
13057 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13058 keyword.
13059
13060table_trackers(<table>)
13061 Uses the string representation of the input sample to perform a look up in
13062 the specified table. If the key is not found in the table, integer value zero
13063 is returned. Otherwise the converter returns the current amount of concurrent
13064 connections tracking the same key as the input sample in the designated
13065 table. It differs from table_conn_cur in that it does not rely on any stored
13066 information but on the table's reference count (the "use" value which is
13067 returned by "show table" on the CLI). This may sometimes be more suited for
13068 layer7 tracking. It can be used to tell a server how many concurrent
13069 connections there are from a given address for example. See also the
13070 sc_trackers sample fetch keyword.
13071
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013072upper
13073 Convert a string sample to upper case. This can only be placed after a string
13074 sample fetch function or after a transformation keyword returning a string
13075 type. The result is of type string.
13076
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013077url_dec
13078 Takes an url-encoded string provided as input and returns the decoded
13079 version as output. The input and the output are of type string.
13080
Christopher Faulet85d79c92016-11-09 16:54:56 +010013081unset-var(<var name>)
13082 Unsets a variable if the input content is defined. The name of the variable
13083 starts with an indication about its scope. The scopes allowed are:
13084 "proc" : the variable is shared with the whole process
13085 "sess" : the variable is shared with the whole session
13086 "txn" : the variable is shared with the transaction (request and
13087 response),
13088 "req" : the variable is shared only during request processing,
13089 "res" : the variable is shared only during response processing.
13090 This prefix is followed by a name. The separator is a '.'. The name may only
13091 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13092
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013093utime(<format>[,<offset>])
13094 Converts an integer supposed to contain a date since epoch to a string
13095 representing this date in UTC time using a format defined by the <format>
13096 string using strftime(3). The purpose is to allow any date format to be used
13097 in logs. An optional <offset> in seconds may be applied to the input date
13098 (positive or negative). See the strftime() man page for the format supported
13099 by your operating system. See also the ltime converter.
13100
13101 Example :
13102
13103 # Emit two colons, one with the UTC time and another with ip:port
13104 # Eg: 20140710162350 127.0.0.1:57325
13105 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13106
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013107word(<index>,<delimiters>)
13108 Extracts the nth word considering given delimiters from an input string.
13109 Indexes start at 1 and delimiters are a string formatted list of chars.
13110
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013111wt6([<avalanche>])
13112 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13113 hash function. Optionally, it is possible to apply a full avalanche hash
13114 function to the output if the optional <avalanche> argument equals 1. This
13115 converter uses the same functions as used by the various hash-based load
13116 balancing algorithms, so it will provide exactly the same results. It is
13117 mostly intended for debugging, but can be used as a stick-table entry to
13118 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010013119 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", and the
13120 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013121
Willy Tarreau97707872015-01-27 15:12:13 +010013122xor(<value>)
13123 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013124 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013125 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013126 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013127 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013128 "sess" : the variable is shared with the whole session
13129 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013130 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013131 "req" : the variable is shared only during request processing,
13132 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013133 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013134 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013135
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013136xxh32([<seed>])
13137 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13138 variant of the XXHash hash function. This hash supports a seed which defaults
13139 to zero but a different value maybe passed as the <seed> argument. This hash
13140 is known to be very good and very fast so it can be used to hash URLs and/or
13141 URL parameters for use as stick-table keys to collect statistics with a low
13142 collision rate, though care must be taken as the algorithm is not considered
13143 as cryptographically secure.
13144
13145xxh64([<seed>])
13146 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13147 variant of the XXHash hash function. This hash supports a seed which defaults
13148 to zero but a different value maybe passed as the <seed> argument. This hash
13149 is known to be very good and very fast so it can be used to hash URLs and/or
13150 URL parameters for use as stick-table keys to collect statistics with a low
13151 collision rate, though care must be taken as the algorithm is not considered
13152 as cryptographically secure.
13153
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013154
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200131557.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013156--------------------------------------------
13157
13158A first set of sample fetch methods applies to internal information which does
13159not even relate to any client information. These ones are sometimes used with
13160"monitor-fail" directives to report an internal status to external watchers.
13161The sample fetch methods described in this section are usable anywhere.
13162
13163always_false : boolean
13164 Always returns the boolean "false" value. It may be used with ACLs as a
13165 temporary replacement for another one when adjusting configurations.
13166
13167always_true : boolean
13168 Always returns the boolean "true" value. It may be used with ACLs as a
13169 temporary replacement for another one when adjusting configurations.
13170
13171avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013172 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013173 divided by the number of active servers. The current backend is used if no
13174 backend is specified. This is very similar to "queue" except that the size of
13175 the farm is considered, in order to give a more accurate measurement of the
13176 time it may take for a new connection to be processed. The main usage is with
13177 ACL to return a sorry page to new users when it becomes certain they will get
13178 a degraded service, or to pass to the backend servers in a header so that
13179 they decide to work in degraded mode or to disable some functions to speed up
13180 the processing a bit. Note that in the event there would not be any active
13181 server anymore, twice the number of queued connections would be considered as
13182 the measured value. This is a fair estimate, as we expect one server to get
13183 back soon anyway, but we still prefer to send new traffic to another backend
13184 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13185 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013186
Willy Tarreau74ca5042013-06-11 23:12:07 +020013187be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013188 Applies to the number of currently established connections on the backend,
13189 possibly including the connection being evaluated. If no backend name is
13190 specified, the current one is used. But it is also possible to check another
13191 backend. It can be used to use a specific farm when the nominal one is full.
13192 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013193
Willy Tarreau74ca5042013-06-11 23:12:07 +020013194be_sess_rate([<backend>]) : integer
13195 Returns an integer value corresponding to the sessions creation rate on the
13196 backend, in number of new sessions per second. This is used with ACLs to
13197 switch to an alternate backend when an expensive or fragile one reaches too
13198 high a session rate, or to limit abuse of service (eg. prevent sucking of an
13199 online dictionary). It can also be useful to add this element to logs using a
13200 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013201
13202 Example :
13203 # Redirect to an error page if the dictionary is requested too often
13204 backend dynamic
13205 mode http
13206 acl being_scanned be_sess_rate gt 100
13207 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013208
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013209bin(<hexa>) : bin
13210 Returns a binary chain. The input is the hexadecimal representation
13211 of the string.
13212
13213bool(<bool>) : bool
13214 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
13215 'false' and '0' are the same. 'true' and '1' are the same.
13216
Willy Tarreau74ca5042013-06-11 23:12:07 +020013217connslots([<backend>]) : integer
13218 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013219 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020013220 connections on all servers and the maximum queue size. This is probably only
13221 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050013222
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013223 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020013224 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013225 usage; see "use_backend" keyword) can be redirected to a different backend.
13226
Willy Tarreau55165fe2009-05-10 12:02:55 +020013227 'connslots' = number of available server connection slots, + number of
13228 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013229
Willy Tarreaua36af912009-10-10 12:02:45 +020013230 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020013231 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020013232 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020013233 you want to be able to differentiate between different backends, and their
13234 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020013235 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020013236 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013237
Willy Tarreau55165fe2009-05-10 12:02:55 +020013238 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
13239 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020013240 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020013241 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013242
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013243date([<offset>]) : integer
13244 Returns the current date as the epoch (number of seconds since 01/01/1970).
13245 If an offset value is specified, then it is a number of seconds that is added
13246 to the current date before returning the value. This is particularly useful
13247 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020013248 It is useful combined with the http_date converter.
13249
13250 Example :
13251
13252 # set an expires header to now+1 hour in every response
13253 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013254
Willy Tarreau595ec542013-06-12 21:34:28 +020013255env(<name>) : string
13256 Returns a string containing the value of environment variable <name>. As a
13257 reminder, environment variables are per-process and are sampled when the
13258 process starts. This can be useful to pass some information to a next hop
13259 server, or with ACLs to take specific action when the process is started a
13260 certain way.
13261
13262 Examples :
13263 # Pass the Via header to next hop with the local hostname in it
13264 http-request add-header Via 1.1\ %[env(HOSTNAME)]
13265
13266 # reject cookie-less requests when the STOP environment variable is set
13267 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
13268
Willy Tarreau74ca5042013-06-11 23:12:07 +020013269fe_conn([<frontend>]) : integer
13270 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010013271 possibly including the connection being evaluated. If no frontend name is
13272 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013273 frontend. It can be used to return a sorry page before hard-blocking, or to
13274 use a specific backend to drain new requests when the farm is considered
13275 full. This is mostly used with ACLs but can also be used to pass some
13276 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
13277 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020013278
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020013279fe_req_rate([<frontend>]) : integer
13280 Returns an integer value corresponding to the number of HTTP requests per
13281 second sent to a frontend. This number can differ from "fe_sess_rate" in
13282 situations where client-side keep-alive is enabled.
13283
Willy Tarreau74ca5042013-06-11 23:12:07 +020013284fe_sess_rate([<frontend>]) : integer
13285 Returns an integer value corresponding to the sessions creation rate on the
13286 frontend, in number of new sessions per second. This is used with ACLs to
13287 limit the incoming session rate to an acceptable range in order to prevent
13288 abuse of service at the earliest moment, for example when combined with other
13289 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
13290 down below the limit. It can also be useful to add this element to logs using
13291 a log-format directive. See also the "rate-limit sessions" directive for use
13292 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013293
13294 Example :
13295 # This frontend limits incoming mails to 10/s with a max of 100
13296 # concurrent connections. We accept any connection below 10/s, and
13297 # force excess clients to wait for 100 ms. Since clients are limited to
13298 # 100 max, there cannot be more than 10 incoming mails per second.
13299 frontend mail
13300 bind :25
13301 mode tcp
13302 maxconn 100
13303 acl too_fast fe_sess_rate ge 10
13304 tcp-request inspect-delay 100ms
13305 tcp-request content accept if ! too_fast
13306 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010013307
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010013308hostname : string
13309 Returns the system hostname.
13310
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013311int(<integer>) : signed integer
13312 Returns a signed integer.
13313
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013314ipv4(<ipv4>) : ipv4
13315 Returns an ipv4.
13316
13317ipv6(<ipv6>) : ipv6
13318 Returns an ipv6.
13319
13320meth(<method>) : method
13321 Returns a method.
13322
Willy Tarreau0f30d262014-11-24 16:02:05 +010013323nbproc : integer
13324 Returns an integer value corresponding to the number of processes that were
13325 started (it equals the global "nbproc" setting). This is useful for logging
13326 and debugging purposes.
13327
Willy Tarreau74ca5042013-06-11 23:12:07 +020013328nbsrv([<backend>]) : integer
13329 Returns an integer value corresponding to the number of usable servers of
13330 either the current backend or the named backend. This is mostly used with
13331 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010013332 switch to an alternate backend when the number of servers is too low to
13333 to handle some load. It is useful to report a failure when combined with
13334 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013335
Willy Tarreau0f30d262014-11-24 16:02:05 +010013336proc : integer
13337 Returns an integer value corresponding to the position of the process calling
13338 the function, between 1 and global.nbproc. This is useful for logging and
13339 debugging purposes.
13340
Willy Tarreau74ca5042013-06-11 23:12:07 +020013341queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013342 Returns the total number of queued connections of the designated backend,
13343 including all the connections in server queues. If no backend name is
13344 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013345 one. This is useful with ACLs or to pass statistics to backend servers. This
13346 can be used to take actions when queuing goes above a known level, generally
13347 indicating a surge of traffic or a massive slowdown on the servers. One
13348 possible action could be to reject new users but still accept old ones. See
13349 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
13350
Willy Tarreau84310e22014-02-14 11:59:04 +010013351rand([<range>]) : integer
13352 Returns a random integer value within a range of <range> possible values,
13353 starting at zero. If the range is not specified, it defaults to 2^32, which
13354 gives numbers between 0 and 4294967295. It can be useful to pass some values
13355 needed to take some routing decisions for example, or just for debugging
13356 purposes. This random must not be used for security purposes.
13357
Willy Tarreau74ca5042013-06-11 23:12:07 +020013358srv_conn([<backend>/]<server>) : integer
13359 Returns an integer value corresponding to the number of currently established
13360 connections on the designated server, possibly including the connection being
13361 evaluated. If <backend> is omitted, then the server is looked up in the
13362 current backend. It can be used to use a specific farm when one server is
13363 full, or to inform the server about our view of the number of active
13364 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
13365 methods.
13366
13367srv_is_up([<backend>/]<server>) : boolean
13368 Returns true when the designated server is UP, and false when it is either
13369 DOWN or in maintenance mode. If <backend> is omitted, then the server is
13370 looked up in the current backend. It is mainly used to take action based on
13371 an external status reported via a health check (eg: a geographical site's
13372 availability). Another possible use which is more of a hack consists in
13373 using dummy servers as boolean variables that can be enabled or disabled from
13374 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
13375
13376srv_sess_rate([<backend>/]<server>) : integer
13377 Returns an integer corresponding to the sessions creation rate on the
13378 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013379 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020013380 used with ACLs but can make sense with logs too. This is used to switch to an
13381 alternate backend when an expensive or fragile one reaches too high a session
13382 rate, or to limit abuse of service (eg. prevent latent requests from
13383 overloading servers).
13384
13385 Example :
13386 # Redirect to a separate back
13387 acl srv1_full srv_sess_rate(be1/srv1) gt 50
13388 acl srv2_full srv_sess_rate(be1/srv2) gt 50
13389 use_backend be2 if srv1_full or srv2_full
13390
Willy Tarreau0f30d262014-11-24 16:02:05 +010013391stopping : boolean
13392 Returns TRUE if the process calling the function is currently stopping. This
13393 can be useful for logging, or for relaxing certain checks or helping close
13394 certain connections upon graceful shutdown.
13395
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013396str(<string>) : string
13397 Returns a string.
13398
Willy Tarreau74ca5042013-06-11 23:12:07 +020013399table_avl([<table>]) : integer
13400 Returns the total number of available entries in the current proxy's
13401 stick-table or in the designated stick-table. See also table_cnt.
13402
13403table_cnt([<table>]) : integer
13404 Returns the total number of entries currently in use in the current proxy's
13405 stick-table or in the designated stick-table. See also src_conn_cnt and
13406 table_avl for other entry counting methods.
13407
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013408var(<var-name>) : undefined
13409 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010013410 sample fetch fails. The name of the variable starts with an indication
13411 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013412 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013413 "sess" : the variable is shared with the whole session
13414 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013415 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013416 "req" : the variable is shared only during request processing,
13417 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013418 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013419 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013420
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200134217.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020013422----------------------------------
13423
13424The layer 4 usually describes just the transport layer which in haproxy is
13425closest to the connection, where no content is yet made available. The fetch
13426methods described here are usable as low as the "tcp-request connection" rule
13427sets unless they require some future information. Those generally include
13428TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013429the incoming connection. For retrieving a value from a sticky counters, the
13430counter number can be explicitly set as 0, 1, or 2 using the pre-defined
13431"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013432argument when using the "sc_" prefix. An optional table may be specified with
13433the "sc*" form, in which case the currently tracked key will be looked up into
13434this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013435
13436be_id : integer
13437 Returns an integer containing the current backend's id. It can be used in
13438 frontends with responses to check which backend processed the request.
13439
Marcin Deranekd2471c22016-12-12 14:08:05 +010013440be_name : string
13441 Returns a string containing the current backend's name. It can be used in
13442 frontends with responses to check which backend processed the request.
13443
Willy Tarreau74ca5042013-06-11 23:12:07 +020013444dst : ip
13445 This is the destination IPv4 address of the connection on the client side,
13446 which is the address the client connected to. It can be useful when running
13447 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
13448 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
13449 RFC 4291.
13450
13451dst_conn : integer
13452 Returns an integer value corresponding to the number of currently established
13453 connections on the same socket including the one being evaluated. It is
13454 normally used with ACLs but can as well be used to pass the information to
13455 servers in an HTTP header or in logs. It can be used to either return a sorry
13456 page before hard-blocking, or to use a specific backend to drain new requests
13457 when the socket is considered saturated. This offers the ability to assign
13458 different limits to different listening ports or addresses. See also the
13459 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013460
Willy Tarreau16e01562016-08-09 16:46:18 +020013461dst_is_local : boolean
13462 Returns true if the destination address of the incoming connection is local
13463 to the system, or false if the address doesn't exist on the system, meaning
13464 that it was intercepted in transparent mode. It can be useful to apply
13465 certain rules by default to forwarded traffic and other rules to the traffic
13466 targetting the real address of the machine. For example the stats page could
13467 be delivered only on this address, or SSH access could be locally redirected.
13468 Please note that the check involves a few system calls, so it's better to do
13469 it only once per connection.
13470
Willy Tarreau74ca5042013-06-11 23:12:07 +020013471dst_port : integer
13472 Returns an integer value corresponding to the destination TCP port of the
13473 connection on the client side, which is the port the client connected to.
13474 This might be used when running in transparent mode, when assigning dynamic
13475 ports to some clients for a whole application session, to stick all users to
13476 a same server, or to pass the destination port information to a server using
13477 an HTTP header.
13478
Emeric Brun4f603012017-01-05 15:11:44 +010013479fc_rcvd_proxy : boolean
13480 Returns true if the client initiated the connection with a PROXY protocol
13481 header.
13482
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020013483fc_rtt(<unit>) : integer
13484 Returns the Round Trip Time (RTT) measured by the kernel for the client
13485 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
13486 can be set to "ms" for milliseconds or "us" for microseconds. If the server
13487 connection is not established, if the connection is not TCP or if the
13488 operating system does not support TCP_INFO, for example Linux kernels before
13489 2.4, the sample fetch fails.
13490
13491fc_rttvar(<unit>) : integer
13492 Returns the Round Trip Time (RTT) variance measured by the kernel for the
13493 client connection. <unit> is facultative, by default the unit is milliseconds.
13494 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
13495 server connection is not established, if the connection is not TCP or if the
13496 operating system does not support TCP_INFO, for example Linux kernels before
13497 2.4, the sample fetch fails.
13498
Joe Williams30fcd392016-08-10 07:06:44 -070013499fc_unacked(<unit>) : integer
13500 Returns the unacked counter measured by the kernel for the client connection.
13501 If the server connection is not established, if the connection is not TCP or
13502 if the operating system does not support TCP_INFO, for example Linux kernels
13503 before 2.4, the sample fetch fails.
13504
13505fc_sacked(<unit>) : integer
13506 Returns the sacked counter measured by the kernel for the client connection.
13507 If the server connection is not established, if the connection is not TCP or
13508 if the operating system does not support TCP_INFO, for example Linux kernels
13509 before 2.4, the sample fetch fails.
13510
13511fc_retrans(<unit>) : integer
13512 Returns the retransmits counter measured by the kernel for the client
13513 connection. If the server connection is not established, if the connection is
13514 not TCP or if the operating system does not support TCP_INFO, for example
13515 Linux kernels before 2.4, the sample fetch fails.
13516
13517fc_fackets(<unit>) : integer
13518 Returns the fack counter measured by the kernel for the client
13519 connection. If the server connection is not established, if the connection is
13520 not TCP or if the operating system does not support TCP_INFO, for example
13521 Linux kernels before 2.4, the sample fetch fails.
13522
13523fc_lost(<unit>) : integer
13524 Returns the lost counter measured by the kernel for the client
13525 connection. If the server connection is not established, if the connection is
13526 not TCP or if the operating system does not support TCP_INFO, for example
13527 Linux kernels before 2.4, the sample fetch fails.
13528
13529fc_reordering(<unit>) : integer
13530 Returns the reordering counter measured by the kernel for the client
13531 connection. If the server connection is not established, if the connection is
13532 not TCP or if the operating system does not support TCP_INFO, for example
13533 Linux kernels before 2.4, the sample fetch fails.
13534
Willy Tarreau74ca5042013-06-11 23:12:07 +020013535fe_id : integer
13536 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010013537 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020013538 coming via a same frontend to the same server.
13539
Marcin Deranekd2471c22016-12-12 14:08:05 +010013540fe_name : string
13541 Returns a string containing the current frontend's name. It can be used in
13542 backends to check from which frontend it was called, or to stick all users
13543 coming via a same frontend to the same server.
13544
Cyril Bonté62ba8702014-04-22 23:52:25 +020013545sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013546sc0_bytes_in_rate([<table>]) : integer
13547sc1_bytes_in_rate([<table>]) : integer
13548sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013549 Returns the average client-to-server bytes rate from the currently tracked
13550 counters, measured in amount of bytes over the period configured in the
13551 table. See also src_bytes_in_rate.
13552
Cyril Bonté62ba8702014-04-22 23:52:25 +020013553sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013554sc0_bytes_out_rate([<table>]) : integer
13555sc1_bytes_out_rate([<table>]) : integer
13556sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013557 Returns the average server-to-client bytes rate from the currently tracked
13558 counters, measured in amount of bytes over the period configured in the
13559 table. See also src_bytes_out_rate.
13560
Cyril Bonté62ba8702014-04-22 23:52:25 +020013561sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013562sc0_clr_gpc0([<table>]) : integer
13563sc1_clr_gpc0([<table>]) : integer
13564sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020013565 Clears the first General Purpose Counter associated to the currently tracked
13566 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010013567 stored value is zero, so first invocation will always return zero. This is
13568 typically used as a second ACL in an expression in order to mark a connection
13569 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020013570
Jarno Huuskonen676f6222017-03-30 09:19:45 +030013571 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020013572 # block if 5 consecutive requests continue to come faster than 10 sess
13573 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020013574 acl abuse sc0_http_req_rate gt 10
13575 acl kill sc0_inc_gpc0 gt 5
13576 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020013577 tcp-request connection accept if !abuse save
13578 tcp-request connection reject if abuse kill
13579
Cyril Bonté62ba8702014-04-22 23:52:25 +020013580sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013581sc0_conn_cnt([<table>]) : integer
13582sc1_conn_cnt([<table>]) : integer
13583sc2_conn_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013584 Returns the cumulated number of incoming connections from currently tracked
13585 counters. See also src_conn_cnt.
13586
Cyril Bonté62ba8702014-04-22 23:52:25 +020013587sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013588sc0_conn_cur([<table>]) : integer
13589sc1_conn_cur([<table>]) : integer
13590sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013591 Returns the current amount of concurrent connections tracking the same
13592 tracked counters. This number is automatically incremented when tracking
13593 begins and decremented when tracking stops. See also src_conn_cur.
13594
Cyril Bonté62ba8702014-04-22 23:52:25 +020013595sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013596sc0_conn_rate([<table>]) : integer
13597sc1_conn_rate([<table>]) : integer
13598sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013599 Returns the average connection rate from the currently tracked counters,
13600 measured in amount of connections over the period configured in the table.
13601 See also src_conn_rate.
13602
Cyril Bonté62ba8702014-04-22 23:52:25 +020013603sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013604sc0_get_gpc0([<table>]) : integer
13605sc1_get_gpc0([<table>]) : integer
13606sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013607 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013608 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013609
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013610sc_get_gpt0(<ctr>[,<table>]) : integer
13611sc0_get_gpt0([<table>]) : integer
13612sc1_get_gpt0([<table>]) : integer
13613sc2_get_gpt0([<table>]) : integer
13614 Returns the value of the first General Purpose Tag associated to the
13615 currently tracked counters. See also src_get_gpt0.
13616
Cyril Bonté62ba8702014-04-22 23:52:25 +020013617sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013618sc0_gpc0_rate([<table>]) : integer
13619sc1_gpc0_rate([<table>]) : integer
13620sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013621 Returns the average increment rate of the first General Purpose Counter
13622 associated to the currently tracked counters. It reports the frequency
13623 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013624 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
13625 that the "gpc0_rate" counter must be stored in the stick-table for a value to
13626 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020013627
Cyril Bonté62ba8702014-04-22 23:52:25 +020013628sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013629sc0_http_err_cnt([<table>]) : integer
13630sc1_http_err_cnt([<table>]) : integer
13631sc2_http_err_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013632 Returns the cumulated number of HTTP errors from the currently tracked
13633 counters. This includes the both request errors and 4xx error responses.
13634 See also src_http_err_cnt.
13635
Cyril Bonté62ba8702014-04-22 23:52:25 +020013636sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013637sc0_http_err_rate([<table>]) : integer
13638sc1_http_err_rate([<table>]) : integer
13639sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013640 Returns the average rate of HTTP errors from the currently tracked counters,
13641 measured in amount of errors over the period configured in the table. This
13642 includes the both request errors and 4xx error responses. See also
13643 src_http_err_rate.
13644
Cyril Bonté62ba8702014-04-22 23:52:25 +020013645sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013646sc0_http_req_cnt([<table>]) : integer
13647sc1_http_req_cnt([<table>]) : integer
13648sc2_http_req_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013649 Returns the cumulated number of HTTP requests from the currently tracked
13650 counters. This includes every started request, valid or not. See also
13651 src_http_req_cnt.
13652
Cyril Bonté62ba8702014-04-22 23:52:25 +020013653sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013654sc0_http_req_rate([<table>]) : integer
13655sc1_http_req_rate([<table>]) : integer
13656sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013657 Returns the average rate of HTTP requests from the currently tracked
13658 counters, measured in amount of requests over the period configured in
13659 the table. This includes every started request, valid or not. See also
13660 src_http_req_rate.
13661
Cyril Bonté62ba8702014-04-22 23:52:25 +020013662sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013663sc0_inc_gpc0([<table>]) : integer
13664sc1_inc_gpc0([<table>]) : integer
13665sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013666 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010013667 tracked counters, and returns its new value. Before the first invocation,
13668 the stored value is zero, so first invocation will increase it to 1 and will
13669 return 1. This is typically used as a second ACL in an expression in order
13670 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020013671
Jarno Huuskonen676f6222017-03-30 09:19:45 +030013672 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020013673 acl abuse sc0_http_req_rate gt 10
13674 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020013675 tcp-request connection reject if abuse kill
13676
Cyril Bonté62ba8702014-04-22 23:52:25 +020013677sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013678sc0_kbytes_in([<table>]) : integer
13679sc1_kbytes_in([<table>]) : integer
13680sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020013681 Returns the total amount of client-to-server data from the currently tracked
13682 counters, measured in kilobytes. The test is currently performed on 32-bit
13683 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020013684
Cyril Bonté62ba8702014-04-22 23:52:25 +020013685sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013686sc0_kbytes_out([<table>]) : integer
13687sc1_kbytes_out([<table>]) : integer
13688sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020013689 Returns the total amount of server-to-client data from the currently tracked
13690 counters, measured in kilobytes. The test is currently performed on 32-bit
13691 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020013692
Cyril Bonté62ba8702014-04-22 23:52:25 +020013693sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013694sc0_sess_cnt([<table>]) : integer
13695sc1_sess_cnt([<table>]) : integer
13696sc2_sess_cnt([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013697 Returns the cumulated number of incoming connections that were transformed
13698 into sessions, which means that they were accepted by a "tcp-request
13699 connection" rule, from the currently tracked counters. A backend may count
13700 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040013701 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020013702 with the client. See also src_sess_cnt.
13703
Cyril Bonté62ba8702014-04-22 23:52:25 +020013704sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013705sc0_sess_rate([<table>]) : integer
13706sc1_sess_rate([<table>]) : integer
13707sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013708 Returns the average session rate from the currently tracked counters,
13709 measured in amount of sessions over the period configured in the table. A
13710 session is a connection that got past the early "tcp-request connection"
13711 rules. A backend may count more sessions than connections because each
13712 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040013713 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020013714
Cyril Bonté62ba8702014-04-22 23:52:25 +020013715sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020013716sc0_tracked([<table>]) : boolean
13717sc1_tracked([<table>]) : boolean
13718sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020013719 Returns true if the designated session counter is currently being tracked by
13720 the current session. This can be useful when deciding whether or not we want
13721 to set some values in a header passed to the server.
13722
Cyril Bonté62ba8702014-04-22 23:52:25 +020013723sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013724sc0_trackers([<table>]) : integer
13725sc1_trackers([<table>]) : integer
13726sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010013727 Returns the current amount of concurrent connections tracking the same
13728 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020013729 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010013730 that it does not rely on any stored information but on the table's reference
13731 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020013732 may sometimes be more suited for layer7 tracking. It can be used to tell a
13733 server how many concurrent connections there are from a given address for
13734 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010013735
Willy Tarreau74ca5042013-06-11 23:12:07 +020013736so_id : integer
13737 Returns an integer containing the current listening socket's id. It is useful
13738 in frontends involving many "bind" lines, or to stick all users coming via a
13739 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013740
Willy Tarreau74ca5042013-06-11 23:12:07 +020013741src : ip
13742 This is the source IPv4 address of the client of the session. It is of type
13743 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
13744 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
13745 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013746 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
13747 directive is used, it can be the address of a client behind another
13748 PROXY-protocol compatible component for all rule sets except
13749 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013750
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013751 Example:
13752 # add an HTTP header in requests with the originating address' country
13753 http-request set-header X-Country %[src,map_ip(geoip.lst)]
13754
Willy Tarreau74ca5042013-06-11 23:12:07 +020013755src_bytes_in_rate([<table>]) : integer
13756 Returns the average bytes rate from the incoming connection's source address
13757 in the current proxy's stick-table or in the designated stick-table, measured
13758 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013759 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013760
Willy Tarreau74ca5042013-06-11 23:12:07 +020013761src_bytes_out_rate([<table>]) : integer
13762 Returns the average bytes rate to the incoming connection's source address in
13763 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020013764 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013765 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013766
Willy Tarreau74ca5042013-06-11 23:12:07 +020013767src_clr_gpc0([<table>]) : integer
13768 Clears the first General Purpose Counter associated to the incoming
13769 connection's source address in the current proxy's stick-table or in the
13770 designated stick-table, and returns its previous value. If the address is not
13771 found, an entry is created and 0 is returned. This is typically used as a
13772 second ACL in an expression in order to mark a connection when a first ACL
13773 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020013774
Jarno Huuskonen676f6222017-03-30 09:19:45 +030013775 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020013776 # block if 5 consecutive requests continue to come faster than 10 sess
13777 # per second, and reset the counter as soon as the traffic slows down.
13778 acl abuse src_http_req_rate gt 10
13779 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010013780 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020013781 tcp-request connection accept if !abuse save
13782 tcp-request connection reject if abuse kill
13783
Willy Tarreau74ca5042013-06-11 23:12:07 +020013784src_conn_cnt([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020013785 Returns the cumulated number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020013786 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020013787 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013788 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013789
Willy Tarreau74ca5042013-06-11 23:12:07 +020013790src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020013791 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020013792 current incoming connection's source address in the current proxy's
13793 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013794 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013795
Willy Tarreau74ca5042013-06-11 23:12:07 +020013796src_conn_rate([<table>]) : integer
13797 Returns the average connection rate from the incoming connection's source
13798 address in the current proxy's stick-table or in the designated stick-table,
13799 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013800 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013801
Willy Tarreau74ca5042013-06-11 23:12:07 +020013802src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020013803 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020013804 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020013805 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013806 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013807
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013808src_get_gpt0([<table>]) : integer
13809 Returns the value of the first General Purpose Tag associated to the
13810 incoming connection's source address in the current proxy's stick-table or in
13811 the designated stick-table. If the address is not found, zero is returned.
13812 See also sc/sc0/sc1/sc2_get_gpt0.
13813
Willy Tarreau74ca5042013-06-11 23:12:07 +020013814src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013815 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020013816 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013817 stick-table or in the designated stick-table. It reports the frequency
13818 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013819 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
13820 that the "gpc0_rate" counter must be stored in the stick-table for a value to
13821 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013822
Willy Tarreau74ca5042013-06-11 23:12:07 +020013823src_http_err_cnt([<table>]) : integer
13824 Returns the cumulated number of HTTP errors from the incoming connection's
13825 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020013826 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013827 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020013828 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013829
Willy Tarreau74ca5042013-06-11 23:12:07 +020013830src_http_err_rate([<table>]) : integer
13831 Returns the average rate of HTTP errors from the incoming connection's source
13832 address in the current proxy's stick-table or in the designated stick-table,
13833 measured in amount of errors over the period configured in the table. This
13834 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013835 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013836
Willy Tarreau74ca5042013-06-11 23:12:07 +020013837src_http_req_cnt([<table>]) : integer
13838 Returns the cumulated number of HTTP requests from the incoming connection's
13839 source address in the current proxy's stick-table or in the designated stick-
13840 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013841 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013842
Willy Tarreau74ca5042013-06-11 23:12:07 +020013843src_http_req_rate([<table>]) : integer
13844 Returns the average rate of HTTP requests from the incoming connection's
13845 source address in the current proxy's stick-table or in the designated stick-
13846 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020013847 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013848 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013849
Willy Tarreau74ca5042013-06-11 23:12:07 +020013850src_inc_gpc0([<table>]) : integer
13851 Increments the first General Purpose Counter associated to the incoming
13852 connection's source address in the current proxy's stick-table or in the
13853 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020013854 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013855 This is typically used as a second ACL in an expression in order to mark a
13856 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020013857
Jarno Huuskonen676f6222017-03-30 09:19:45 +030013858 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020013859 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010013860 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020013861 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020013862
Willy Tarreau16e01562016-08-09 16:46:18 +020013863src_is_local : boolean
13864 Returns true if the source address of the incoming connection is local to the
13865 system, or false if the address doesn't exist on the system, meaning that it
13866 comes from a remote machine. Note that UNIX addresses are considered local.
13867 It can be useful to apply certain access restrictions based on where the
13868 client comes from (eg: require auth or https for remote machines). Please
13869 note that the check involves a few system calls, so it's better to do it only
13870 once per connection.
13871
Willy Tarreau74ca5042013-06-11 23:12:07 +020013872src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020013873 Returns the total amount of data received from the incoming connection's
13874 source address in the current proxy's stick-table or in the designated
13875 stick-table, measured in kilobytes. If the address is not found, zero is
13876 returned. The test is currently performed on 32-bit integers, which limits
13877 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013878
Willy Tarreau74ca5042013-06-11 23:12:07 +020013879src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020013880 Returns the total amount of data sent to the incoming connection's source
13881 address in the current proxy's stick-table or in the designated stick-table,
13882 measured in kilobytes. If the address is not found, zero is returned. The
13883 test is currently performed on 32-bit integers, which limits values to 4
13884 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020013885
Willy Tarreau74ca5042013-06-11 23:12:07 +020013886src_port : integer
13887 Returns an integer value corresponding to the TCP source port of the
13888 connection on the client side, which is the port the client connected from.
13889 Usage of this function is very limited as modern protocols do not care much
13890 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013891
Willy Tarreau74ca5042013-06-11 23:12:07 +020013892src_sess_cnt([<table>]) : integer
13893 Returns the cumulated number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020013894 connection's source IPv4 address in the current proxy's stick-table or in the
13895 designated stick-table, that were transformed into sessions, which means that
13896 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013897 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013898
Willy Tarreau74ca5042013-06-11 23:12:07 +020013899src_sess_rate([<table>]) : integer
13900 Returns the average session rate from the incoming connection's source
13901 address in the current proxy's stick-table or in the designated stick-table,
13902 measured in amount of sessions over the period configured in the table. A
13903 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013904 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020013905
Willy Tarreau74ca5042013-06-11 23:12:07 +020013906src_updt_conn_cnt([<table>]) : integer
13907 Creates or updates the entry associated to the incoming connection's source
13908 address in the current proxy's stick-table or in the designated stick-table.
13909 This table must be configured to store the "conn_cnt" data type, otherwise
13910 the match will be ignored. The current count is incremented by one, and the
13911 expiration timer refreshed. The updated count is returned, so this match
13912 can't return zero. This was used to reject service abusers based on their
13913 source address. Note: it is recommended to use the more complete "track-sc*"
13914 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020013915
13916 Example :
13917 # This frontend limits incoming SSH connections to 3 per 10 second for
13918 # each source address, and rejects excess connections until a 10 second
13919 # silence is observed. At most 20 addresses are tracked.
13920 listen ssh
13921 bind :22
13922 mode tcp
13923 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020013924 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020013925 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020013926 server local 127.0.0.1:22
13927
Willy Tarreau74ca5042013-06-11 23:12:07 +020013928srv_id : integer
13929 Returns an integer containing the server's id when processing the response.
13930 While it's almost only used with ACLs, it may be used for logging or
13931 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020013932
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200139337.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020013934----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020013935
Willy Tarreau74ca5042013-06-11 23:12:07 +020013936The layer 5 usually describes just the session layer which in haproxy is
13937closest to the session once all the connection handshakes are finished, but
13938when no content is yet made available. The fetch methods described here are
13939usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013940future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020013941
Ben Shillitof25e8e52016-12-02 14:25:37 +00001394251d.all(<prop>[,<prop>*]) : string
13943 Returns values for the properties requested as a string, where values are
13944 separated by the delimiter specified with "51degrees-property-separator".
13945 The device is identified using all the important HTTP headers from the
13946 request. The function can be passed up to five property names, and if a
13947 property name can't be found, the value "NoData" is returned.
13948
13949 Example :
13950 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
13951 # containing the three properties requested using all relevant headers from
13952 # the request.
13953 frontend http-in
13954 bind *:8081
13955 default_backend servers
13956 http-request set-header X-51D-DeviceTypeMobileTablet \
13957 %[51d.all(DeviceType,IsMobile,IsTablet)]
13958
Emeric Brun645ae792014-04-30 14:21:06 +020013959ssl_bc : boolean
13960 Returns true when the back connection was made via an SSL/TLS transport
13961 layer and is locally deciphered. This means the outgoing connection was made
13962 other a server with the "ssl" option.
13963
13964ssl_bc_alg_keysize : integer
13965 Returns the symmetric cipher key size supported in bits when the outgoing
13966 connection was made over an SSL/TLS transport layer.
13967
13968ssl_bc_cipher : string
13969 Returns the name of the used cipher when the outgoing connection was made
13970 over an SSL/TLS transport layer.
13971
13972ssl_bc_protocol : string
13973 Returns the name of the used protocol when the outgoing connection was made
13974 over an SSL/TLS transport layer.
13975
Emeric Brunb73a9b02014-04-30 18:49:19 +020013976ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020013977 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020013978 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
13979 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020013980
13981ssl_bc_session_id : binary
13982 Returns the SSL ID of the back connection when the outgoing connection was
13983 made over an SSL/TLS transport layer. It is useful to log if we want to know
13984 if session was reused or not.
13985
13986ssl_bc_use_keysize : integer
13987 Returns the symmetric cipher key size used in bits when the outgoing
13988 connection was made over an SSL/TLS transport layer.
13989
Willy Tarreau74ca5042013-06-11 23:12:07 +020013990ssl_c_ca_err : integer
13991 When the incoming connection was made over an SSL/TLS transport layer,
13992 returns the ID of the first error detected during verification of the client
13993 certificate at depth > 0, or 0 if no error was encountered during this
13994 verification process. Please refer to your SSL library's documentation to
13995 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020013996
Willy Tarreau74ca5042013-06-11 23:12:07 +020013997ssl_c_ca_err_depth : integer
13998 When the incoming connection was made over an SSL/TLS transport layer,
13999 returns the depth in the CA chain of the first error detected during the
14000 verification of the client certificate. If no error is encountered, 0 is
14001 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014002
Emeric Brun43e79582014-10-29 19:03:26 +010014003ssl_c_der : binary
14004 Returns the DER formatted certificate presented by the client when the
14005 incoming connection was made over an SSL/TLS transport layer. When used for
14006 an ACL, the value(s) to match against can be passed in hexadecimal form.
14007
Willy Tarreau74ca5042013-06-11 23:12:07 +020014008ssl_c_err : integer
14009 When the incoming connection was made over an SSL/TLS transport layer,
14010 returns the ID of the first error detected during verification at depth 0, or
14011 0 if no error was encountered during this verification process. Please refer
14012 to your SSL library's documentation to find the exhaustive list of error
14013 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020014014
Willy Tarreau74ca5042013-06-11 23:12:07 +020014015ssl_c_i_dn([<entry>[,<occ>]]) : string
14016 When the incoming connection was made over an SSL/TLS transport layer,
14017 returns the full distinguished name of the issuer of the certificate
14018 presented by the client when no <entry> is specified, or the value of the
14019 first given entry found from the beginning of the DN. If a positive/negative
14020 occurrence number is specified as the optional second argument, it returns
14021 the value of the nth given entry value from the beginning/end of the DN.
14022 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
14023 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020014024
Willy Tarreau74ca5042013-06-11 23:12:07 +020014025ssl_c_key_alg : string
14026 Returns the name of the algorithm used to generate the key of the certificate
14027 presented by the client when the incoming connection was made over an SSL/TLS
14028 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020014029
Willy Tarreau74ca5042013-06-11 23:12:07 +020014030ssl_c_notafter : string
14031 Returns the end date presented by the client as a formatted string
14032 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14033 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020014034
Willy Tarreau74ca5042013-06-11 23:12:07 +020014035ssl_c_notbefore : string
14036 Returns the start date presented by the client as a formatted string
14037 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14038 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014039
Willy Tarreau74ca5042013-06-11 23:12:07 +020014040ssl_c_s_dn([<entry>[,<occ>]]) : string
14041 When the incoming connection was made over an SSL/TLS transport layer,
14042 returns the full distinguished name of the subject of the certificate
14043 presented by the client when no <entry> is specified, or the value of the
14044 first given entry found from the beginning of the DN. If a positive/negative
14045 occurrence number is specified as the optional second argument, it returns
14046 the value of the nth given entry value from the beginning/end of the DN.
14047 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
14048 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014049
Willy Tarreau74ca5042013-06-11 23:12:07 +020014050ssl_c_serial : binary
14051 Returns the serial of the certificate presented by the client when the
14052 incoming connection was made over an SSL/TLS transport layer. When used for
14053 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014054
Willy Tarreau74ca5042013-06-11 23:12:07 +020014055ssl_c_sha1 : binary
14056 Returns the SHA-1 fingerprint of the certificate presented by the client when
14057 the incoming connection was made over an SSL/TLS transport layer. This can be
14058 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014059 Note that the output is binary, so if you want to pass that signature to the
14060 server, you need to encode it in hex or base64, such as in the example below:
14061
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014062 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014063 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020014064
Willy Tarreau74ca5042013-06-11 23:12:07 +020014065ssl_c_sig_alg : string
14066 Returns the name of the algorithm used to sign the certificate presented by
14067 the client when the incoming connection was made over an SSL/TLS transport
14068 layer.
Emeric Brun87855892012-10-17 17:39:35 +020014069
Willy Tarreau74ca5042013-06-11 23:12:07 +020014070ssl_c_used : boolean
14071 Returns true if current SSL session uses a client certificate even if current
14072 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020014073
Willy Tarreau74ca5042013-06-11 23:12:07 +020014074ssl_c_verify : integer
14075 Returns the verify result error ID when the incoming connection was made over
14076 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
14077 refer to your SSL library's documentation for an exhaustive list of error
14078 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020014079
Willy Tarreau74ca5042013-06-11 23:12:07 +020014080ssl_c_version : integer
14081 Returns the version of the certificate presented by the client when the
14082 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020014083
Emeric Brun43e79582014-10-29 19:03:26 +010014084ssl_f_der : binary
14085 Returns the DER formatted certificate presented by the frontend when the
14086 incoming connection was made over an SSL/TLS transport layer. When used for
14087 an ACL, the value(s) to match against can be passed in hexadecimal form.
14088
Willy Tarreau74ca5042013-06-11 23:12:07 +020014089ssl_f_i_dn([<entry>[,<occ>]]) : string
14090 When the incoming connection was made over an SSL/TLS transport layer,
14091 returns the full distinguished name of the issuer of the certificate
14092 presented by the frontend when no <entry> is specified, or the value of the
14093 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020014094 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020014095 the value of the nth given entry value from the beginning/end of the DN.
14096 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
14097 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020014098
Willy Tarreau74ca5042013-06-11 23:12:07 +020014099ssl_f_key_alg : string
14100 Returns the name of the algorithm used to generate the key of the certificate
14101 presented by the frontend when the incoming connection was made over an
14102 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014103
Willy Tarreau74ca5042013-06-11 23:12:07 +020014104ssl_f_notafter : string
14105 Returns the end date presented by the frontend as a formatted string
14106 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14107 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014108
Willy Tarreau74ca5042013-06-11 23:12:07 +020014109ssl_f_notbefore : string
14110 Returns the start date presented by the frontend as a formatted string
14111 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14112 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020014113
Willy Tarreau74ca5042013-06-11 23:12:07 +020014114ssl_f_s_dn([<entry>[,<occ>]]) : string
14115 When the incoming connection was made over an SSL/TLS transport layer,
14116 returns the full distinguished name of the subject of the certificate
14117 presented by the frontend when no <entry> is specified, or the value of the
14118 first given entry found from the beginning of the DN. If a positive/negative
14119 occurrence number is specified as the optional second argument, it returns
14120 the value of the nth given entry value from the beginning/end of the DN.
14121 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
14122 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020014123
Willy Tarreau74ca5042013-06-11 23:12:07 +020014124ssl_f_serial : binary
14125 Returns the serial of the certificate presented by the frontend when the
14126 incoming connection was made over an SSL/TLS transport layer. When used for
14127 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020014128
Emeric Brun55f4fa82014-04-30 17:11:25 +020014129ssl_f_sha1 : binary
14130 Returns the SHA-1 fingerprint of the certificate presented by the frontend
14131 when the incoming connection was made over an SSL/TLS transport layer. This
14132 can be used to know which certificate was chosen using SNI.
14133
Willy Tarreau74ca5042013-06-11 23:12:07 +020014134ssl_f_sig_alg : string
14135 Returns the name of the algorithm used to sign the certificate presented by
14136 the frontend when the incoming connection was made over an SSL/TLS transport
14137 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014138
Willy Tarreau74ca5042013-06-11 23:12:07 +020014139ssl_f_version : integer
14140 Returns the version of the certificate presented by the frontend when the
14141 incoming connection was made over an SSL/TLS transport layer.
14142
14143ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014144 Returns true when the front connection was made via an SSL/TLS transport
14145 layer and is locally deciphered. This means it has matched a socket declared
14146 with a "bind" line having the "ssl" option.
14147
Willy Tarreau74ca5042013-06-11 23:12:07 +020014148 Example :
14149 # This passes "X-Proto: https" to servers when client connects over SSL
14150 listen http-https
14151 bind :80
14152 bind :443 ssl crt /etc/haproxy.pem
14153 http-request add-header X-Proto https if { ssl_fc }
14154
14155ssl_fc_alg_keysize : integer
14156 Returns the symmetric cipher key size supported in bits when the incoming
14157 connection was made over an SSL/TLS transport layer.
14158
14159ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014160 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014161 incoming connection made via a TLS transport layer and locally deciphered by
14162 haproxy. The result is a string containing the protocol name advertised by
14163 the client. The SSL library must have been built with support for TLS
14164 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
14165 not advertised unless the "alpn" keyword on the "bind" line specifies a
14166 protocol list. Also, nothing forces the client to pick a protocol from this
14167 list, any other one may be requested. The TLS ALPN extension is meant to
14168 replace the TLS NPN extension. See also "ssl_fc_npn".
14169
Willy Tarreau74ca5042013-06-11 23:12:07 +020014170ssl_fc_cipher : string
14171 Returns the name of the used cipher when the incoming connection was made
14172 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020014173
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014174ssl_fc_cipherlist_bin : binary
14175 Returns the binary form of the client hello cipher list. The maximum returned
14176 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014177 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014178
14179ssl_fc_cipherlist_hex : string
14180 Returns the binary form of the client hello cipher list encoded as
14181 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014182 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014183
14184ssl_fc_cipherlist_str : string
14185 Returns the decoded text form of the client hello cipher list. The maximum
14186 number of ciphers returned is according with the value of
14187 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020014188 avaible with OpenSSL >= 1.0.2. If the function is not enabled, this
14189 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014190
14191ssl_fc_cipherlist_xxh : integer
14192 Returns a xxh64 of the cipher list. This hash can be return only is the value
14193 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014194 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014195
Willy Tarreau74ca5042013-06-11 23:12:07 +020014196ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014197 Returns true if a client certificate is present in an incoming connection over
14198 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010014199 Note: on SSL session resumption with Session ID or TLS ticket, client
14200 certificate is not present in the current connection but may be retrieved
14201 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
14202 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014203
Willy Tarreau74ca5042013-06-11 23:12:07 +020014204ssl_fc_has_sni : boolean
14205 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020014206 in an incoming connection was made over an SSL/TLS transport layer. Returns
14207 true when the incoming connection presents a TLS SNI field. This requires
14208 that the SSL library is build with support for TLS extensions enabled (check
14209 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020014210
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020014211ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020014212 Returns true if the SSL/TLS session has been resumed through the use of
14213 SSL session cache or TLS tickets.
14214
Willy Tarreau74ca5042013-06-11 23:12:07 +020014215ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014216 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020014217 made via a TLS transport layer and locally deciphered by haproxy. The result
14218 is a string containing the protocol name advertised by the client. The SSL
14219 library must have been built with support for TLS extensions enabled (check
14220 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
14221 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
14222 forces the client to pick a protocol from this list, any other one may be
14223 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020014224
Willy Tarreau74ca5042013-06-11 23:12:07 +020014225ssl_fc_protocol : string
14226 Returns the name of the used protocol when the incoming connection was made
14227 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020014228
Emeric Brunb73a9b02014-04-30 18:49:19 +020014229ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040014230 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014231 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14232 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040014233
Willy Tarreau74ca5042013-06-11 23:12:07 +020014234ssl_fc_session_id : binary
14235 Returns the SSL ID of the front connection when the incoming connection was
14236 made over an SSL/TLS transport layer. It is useful to stick a given client to
14237 a server. It is important to note that some browsers refresh their session ID
14238 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020014239
Willy Tarreau74ca5042013-06-11 23:12:07 +020014240ssl_fc_sni : string
14241 This extracts the Server Name Indication TLS extension (SNI) field from an
14242 incoming connection made via an SSL/TLS transport layer and locally
14243 deciphered by haproxy. The result (when present) typically is a string
14244 matching the HTTPS host name (253 chars or less). The SSL library must have
14245 been built with support for TLS extensions enabled (check haproxy -vv).
14246
14247 This fetch is different from "req_ssl_sni" above in that it applies to the
14248 connection being deciphered by haproxy and not to SSL contents being blindly
14249 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020014250 requires that the SSL library is build with support for TLS extensions
14251 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020014252
Willy Tarreau74ca5042013-06-11 23:12:07 +020014253 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020014254 ssl_fc_sni_end : suffix match
14255 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020014256
Willy Tarreau74ca5042013-06-11 23:12:07 +020014257ssl_fc_use_keysize : integer
14258 Returns the symmetric cipher key size used in bits when the incoming
14259 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014260
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014261
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200142627.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020014263------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014264
Willy Tarreau74ca5042013-06-11 23:12:07 +020014265Fetching samples from buffer contents is a bit different from the previous
14266sample fetches above because the sampled data are ephemeral. These data can
14267only be used when they're available and will be lost when they're forwarded.
14268For this reason, samples fetched from buffer contents during a request cannot
14269be used in a response for example. Even while the data are being fetched, they
14270can change. Sometimes it is necessary to set some delays or combine multiple
14271sample fetch methods to ensure that the expected data are complete and usable,
14272for example through TCP request content inspection. Please see the "tcp-request
14273content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020014274
Willy Tarreau74ca5042013-06-11 23:12:07 +020014275payload(<offset>,<length>) : binary (deprecated)
14276 This is an alias for "req.payload" when used in the context of a request (eg:
14277 "stick on", "stick match"), and for "res.payload" when used in the context of
14278 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014279
Willy Tarreau74ca5042013-06-11 23:12:07 +020014280payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
14281 This is an alias for "req.payload_lv" when used in the context of a request
14282 (eg: "stick on", "stick match"), and for "res.payload_lv" when used in the
14283 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014284
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020014285req.hdrs : string
14286 Returns the current request headers as string including the last empty line
14287 separating headers from the request body. The last empty line can be used to
14288 detect a truncated header block. This sample fetch is useful for some SPOE
14289 headers analyzers and for advanced logging.
14290
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020014291req.hdrs_bin : binary
14292 Returns the current request headers contained in preparsed binary form. This
14293 is useful for offloading some processing with SPOE. Each string is described
14294 by a length followed by the number of bytes indicated in the length. The
14295 length is represented using the variable integer encoding detailed in the
14296 SPOE documentation. The end of the list is marked by a couple of empty header
14297 names and values (length of 0 for both).
14298
14299 *(<str:header-name><str:header-value>)<empty string><empty string>
14300
14301 int: refer to the SPOE documentation for the encoding
14302 str: <int:length><bytes>
14303
Willy Tarreau74ca5042013-06-11 23:12:07 +020014304req.len : integer
14305req_len : integer (deprecated)
14306 Returns an integer value corresponding to the number of bytes present in the
14307 request buffer. This is mostly used in ACL. It is important to understand
14308 that this test does not return false as long as the buffer is changing. This
14309 means that a check with equality to zero will almost always immediately match
14310 at the beginning of the session, while a test for more data will wait for
14311 that data to come in and return false only when haproxy is certain that no
14312 more data will come in. This test was designed to be used with TCP request
14313 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014314
Willy Tarreau74ca5042013-06-11 23:12:07 +020014315req.payload(<offset>,<length>) : binary
14316 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020014317 in the request buffer. As a special case, if the <length> argument is zero,
14318 the the whole buffer from <offset> to the end is extracted. This can be used
14319 with ACLs in order to check for the presence of some content in a buffer at
14320 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014321
Willy Tarreau74ca5042013-06-11 23:12:07 +020014322 ACL alternatives :
14323 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014324
Willy Tarreau74ca5042013-06-11 23:12:07 +020014325req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
14326 This extracts a binary block whose size is specified at <offset1> for <length>
14327 bytes, and which starts at <offset2> if specified or just after the length in
14328 the request buffer. The <offset2> parameter also supports relative offsets if
14329 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014330
Willy Tarreau74ca5042013-06-11 23:12:07 +020014331 ACL alternatives :
14332 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014333
Willy Tarreau74ca5042013-06-11 23:12:07 +020014334 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014335
Willy Tarreau74ca5042013-06-11 23:12:07 +020014336req.proto_http : boolean
14337req_proto_http : boolean (deprecated)
14338 Returns true when data in the request buffer look like HTTP and correctly
14339 parses as such. It is the same parser as the common HTTP request parser which
14340 is used so there should be no surprises. The test does not match until the
14341 request is complete, failed or timed out. This test may be used to report the
14342 protocol in TCP logs, but the biggest use is to block TCP request analysis
14343 until a complete HTTP request is present in the buffer, for example to track
14344 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014345
Willy Tarreau74ca5042013-06-11 23:12:07 +020014346 Example:
14347 # track request counts per "base" (concatenation of Host+URL)
14348 tcp-request inspect-delay 10s
14349 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014350 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014351
Willy Tarreau74ca5042013-06-11 23:12:07 +020014352req.rdp_cookie([<name>]) : string
14353rdp_cookie([<name>]) : string (deprecated)
14354 When the request buffer looks like the RDP protocol, extracts the RDP cookie
14355 <name>, or any cookie if unspecified. The parser only checks for the first
14356 cookie, as illustrated in the RDP protocol specification. The cookie name is
14357 case insensitive. Generally the "MSTS" cookie name will be used, as it can
14358 contain the user name of the client connecting to the server if properly
14359 configured on the client. The "MSTSHASH" cookie is often used as well for
14360 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014361
Willy Tarreau74ca5042013-06-11 23:12:07 +020014362 This differs from "balance rdp-cookie" in that any balancing algorithm may be
14363 used and thus the distribution of clients to backend servers is not linked to
14364 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
14365 such as "balance roundrobin" or "balance leastconn" will lead to a more even
14366 distribution of clients to backend servers than the hash used by "balance
14367 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014368
Willy Tarreau74ca5042013-06-11 23:12:07 +020014369 ACL derivatives :
14370 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014371
Willy Tarreau74ca5042013-06-11 23:12:07 +020014372 Example :
14373 listen tse-farm
14374 bind 0.0.0.0:3389
14375 # wait up to 5s for an RDP cookie in the request
14376 tcp-request inspect-delay 5s
14377 tcp-request content accept if RDP_COOKIE
14378 # apply RDP cookie persistence
14379 persist rdp-cookie
14380 # Persist based on the mstshash cookie
14381 # This is only useful makes sense if
14382 # balance rdp-cookie is not used
14383 stick-table type string size 204800
14384 stick on req.rdp_cookie(mstshash)
14385 server srv1 1.1.1.1:3389
14386 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014387
Willy Tarreau74ca5042013-06-11 23:12:07 +020014388 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
14389 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014390
Willy Tarreau74ca5042013-06-11 23:12:07 +020014391req.rdp_cookie_cnt([name]) : integer
14392rdp_cookie_cnt([name]) : integer (deprecated)
14393 Tries to parse the request buffer as RDP protocol, then returns an integer
14394 corresponding to the number of RDP cookies found. If an optional cookie name
14395 is passed, only cookies matching this name are considered. This is mostly
14396 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014397
Willy Tarreau74ca5042013-06-11 23:12:07 +020014398 ACL derivatives :
14399 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014400
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020014401req.ssl_ec_ext : boolean
14402 Returns a boolean identifying if client sent the Supported Elliptic Curves
14403 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020014404 message. This can be used to present ECC compatible clients with EC
14405 certificate and to use RSA for all others, on the same IP address. Note that
14406 this only applies to raw contents found in the request buffer and not to
14407 contents deciphered via an SSL data layer, so this will not work with "bind"
14408 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020014409
Willy Tarreau74ca5042013-06-11 23:12:07 +020014410req.ssl_hello_type : integer
14411req_ssl_hello_type : integer (deprecated)
14412 Returns an integer value containing the type of the SSL hello message found
14413 in the request buffer if the buffer contains data that parse as a complete
14414 SSL (v3 or superior) client hello message. Note that this only applies to raw
14415 contents found in the request buffer and not to contents deciphered via an
14416 SSL data layer, so this will not work with "bind" lines having the "ssl"
14417 option. This is mostly used in ACL to detect presence of an SSL hello message
14418 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014419
Willy Tarreau74ca5042013-06-11 23:12:07 +020014420req.ssl_sni : string
14421req_ssl_sni : string (deprecated)
14422 Returns a string containing the value of the Server Name TLS extension sent
14423 by a client in a TLS stream passing through the request buffer if the buffer
14424 contains data that parse as a complete SSL (v3 or superior) client hello
14425 message. Note that this only applies to raw contents found in the request
14426 buffer and not to contents deciphered via an SSL data layer, so this will not
14427 work with "bind" lines having the "ssl" option. SNI normally contains the
14428 name of the host the client tries to connect to (for recent browsers). SNI is
14429 useful for allowing or denying access to certain hosts when SSL/TLS is used
14430 by the client. This test was designed to be used with TCP request content
14431 inspection. If content switching is needed, it is recommended to first wait
14432 for a complete client hello (type 1), like in the example below. See also
14433 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014434
Willy Tarreau74ca5042013-06-11 23:12:07 +020014435 ACL derivatives :
14436 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014437
Willy Tarreau74ca5042013-06-11 23:12:07 +020014438 Examples :
14439 # Wait for a client hello for at most 5 seconds
14440 tcp-request inspect-delay 5s
14441 tcp-request content accept if { req_ssl_hello_type 1 }
14442 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
14443 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014444
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053014445req.ssl_st_ext : integer
14446 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
14447 Returns 1 if the client sent SessionTicket TLS Extension
14448 Returns 2 if the client also sent non-zero length TLS SessionTicket
14449 Note that this only applies to raw contents found in the request buffer and
14450 not to contents deciphered via an SSL data layer, so this will not work with
14451 "bind" lines having the "ssl" option. This can for example be used to detect
14452 whether the client sent a SessionTicket or not and stick it accordingly, if
14453 no SessionTicket then stick on SessionID or don't stick as there's no server
14454 side state is there when SessionTickets are in use.
14455
Willy Tarreau74ca5042013-06-11 23:12:07 +020014456req.ssl_ver : integer
14457req_ssl_ver : integer (deprecated)
14458 Returns an integer value containing the version of the SSL/TLS protocol of a
14459 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
14460 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
14461 composed of the major version multiplied by 65536, added to the minor
14462 version. Note that this only applies to raw contents found in the request
14463 buffer and not to contents deciphered via an SSL data layer, so this will not
14464 work with "bind" lines having the "ssl" option. The ACL version of the test
14465 matches against a decimal notation in the form MAJOR.MINOR (eg: 3.1). This
14466 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014467
Willy Tarreau74ca5042013-06-11 23:12:07 +020014468 ACL derivatives :
14469 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010014470
Willy Tarreau47e8eba2013-09-11 23:28:46 +020014471res.len : integer
14472 Returns an integer value corresponding to the number of bytes present in the
14473 response buffer. This is mostly used in ACL. It is important to understand
14474 that this test does not return false as long as the buffer is changing. This
14475 means that a check with equality to zero will almost always immediately match
14476 at the beginning of the session, while a test for more data will wait for
14477 that data to come in and return false only when haproxy is certain that no
14478 more data will come in. This test was designed to be used with TCP response
14479 content inspection.
14480
Willy Tarreau74ca5042013-06-11 23:12:07 +020014481res.payload(<offset>,<length>) : binary
14482 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020014483 in the response buffer. As a special case, if the <length> argument is zero,
14484 the the whole buffer from <offset> to the end is extracted. This can be used
14485 with ACLs in order to check for the presence of some content in a buffer at
14486 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014487
Willy Tarreau74ca5042013-06-11 23:12:07 +020014488res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
14489 This extracts a binary block whose size is specified at <offset1> for <length>
14490 bytes, and which starts at <offset2> if specified or just after the length in
14491 the response buffer. The <offset2> parameter also supports relative offsets
14492 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014493
Willy Tarreau74ca5042013-06-11 23:12:07 +020014494 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014495
Willy Tarreau971f7b62015-09-29 14:06:59 +020014496res.ssl_hello_type : integer
14497rep_ssl_hello_type : integer (deprecated)
14498 Returns an integer value containing the type of the SSL hello message found
14499 in the response buffer if the buffer contains data that parses as a complete
14500 SSL (v3 or superior) hello message. Note that this only applies to raw
14501 contents found in the response buffer and not to contents deciphered via an
14502 SSL data layer, so this will not work with "server" lines having the "ssl"
14503 option. This is mostly used in ACL to detect presence of an SSL hello message
14504 that is supposed to contain an SSL session ID usable for stickiness.
14505
Willy Tarreau74ca5042013-06-11 23:12:07 +020014506wait_end : boolean
14507 This fetch either returns true when the inspection period is over, or does
14508 not fetch. It is only used in ACLs, in conjunction with content analysis to
14509 avoid returning a wrong verdict early. It may also be used to delay some
14510 actions, such as a delayed reject for some special addresses. Since it either
14511 stops the rules evaluation or immediately returns true, it is recommended to
14512 use this acl as the last one in a rule. Please note that the default ACL
14513 "WAIT_END" is always usable without prior declaration. This test was designed
14514 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014515
Willy Tarreau74ca5042013-06-11 23:12:07 +020014516 Examples :
14517 # delay every incoming request by 2 seconds
14518 tcp-request inspect-delay 2s
14519 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010014520
Willy Tarreau74ca5042013-06-11 23:12:07 +020014521 # don't immediately tell bad guys they are rejected
14522 tcp-request inspect-delay 10s
14523 acl goodguys src 10.0.0.0/24
14524 acl badguys src 10.0.1.0/24
14525 tcp-request content accept if goodguys
14526 tcp-request content reject if badguys WAIT_END
14527 tcp-request content reject
14528
14529
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200145307.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020014531--------------------------------------
14532
14533It is possible to fetch samples from HTTP contents, requests and responses.
14534This application layer is also called layer 7. It is only possible to fetch the
14535data in this section when a full HTTP request or response has been parsed from
14536its respective request or response buffer. This is always the case with all
14537HTTP specific rules and for sections running with "mode http". When using TCP
14538content inspection, it may be necessary to support an inspection delay in order
14539to let the request or response come in first. These fetches may require a bit
14540more CPU resources than the layer 4 ones, but not much since the request and
14541response are indexed.
14542
14543base : string
14544 This returns the concatenation of the first Host header and the path part of
14545 the request, which starts at the first slash and ends before the question
14546 mark. It can be useful in virtual hosted environments to detect URL abuses as
14547 well as to improve shared caches efficiency. Using this with a limited size
14548 stick table also allows one to collect statistics about most commonly
14549 requested objects by host/path. With ACLs it can allow simple content
14550 switching rules involving the host and the path at the same time, such as
14551 "www.example.com/favicon.ico". See also "path" and "uri".
14552
14553 ACL derivatives :
14554 base : exact string match
14555 base_beg : prefix match
14556 base_dir : subdir match
14557 base_dom : domain match
14558 base_end : suffix match
14559 base_len : length match
14560 base_reg : regex match
14561 base_sub : substring match
14562
14563base32 : integer
14564 This returns a 32-bit hash of the value returned by the "base" fetch method
14565 above. This is useful to track per-URL activity on high traffic sites without
14566 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014567 memory. The output type is an unsigned integer. The hash function used is
14568 SDBM with full avalanche on the output. Technically, base32 is exactly equal
14569 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020014570
14571base32+src : binary
14572 This returns the concatenation of the base32 fetch above and the src fetch
14573 below. The resulting type is of type binary, with a size of 8 or 20 bytes
14574 depending on the source address family. This can be used to track per-IP,
14575 per-URL counters.
14576
William Lallemand65ad6e12014-01-31 15:08:02 +010014577capture.req.hdr(<idx>) : string
14578 This extracts the content of the header captured by the "capture request
14579 header", idx is the position of the capture keyword in the configuration.
14580 The first entry is an index of 0. See also: "capture request header".
14581
14582capture.req.method : string
14583 This extracts the METHOD of an HTTP request. It can be used in both request
14584 and response. Unlike "method", it can be used in both request and response
14585 because it's allocated.
14586
14587capture.req.uri : string
14588 This extracts the request's URI, which starts at the first slash and ends
14589 before the first space in the request (without the host part). Unlike "path"
14590 and "url", it can be used in both request and response because it's
14591 allocated.
14592
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020014593capture.req.ver : string
14594 This extracts the request's HTTP version and returns either "HTTP/1.0" or
14595 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
14596 logs because it relies on a persistent flag.
14597
William Lallemand65ad6e12014-01-31 15:08:02 +010014598capture.res.hdr(<idx>) : string
14599 This extracts the content of the header captured by the "capture response
14600 header", idx is the position of the capture keyword in the configuration.
14601 The first entry is an index of 0.
14602 See also: "capture response header"
14603
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020014604capture.res.ver : string
14605 This extracts the response's HTTP version and returns either "HTTP/1.0" or
14606 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
14607 persistent flag.
14608
Willy Tarreaua5910cc2015-05-02 00:46:08 +020014609req.body : binary
14610 This returns the HTTP request's available body as a block of data. It
14611 requires that the request body has been buffered made available using
14612 "option http-buffer-request". In case of chunked-encoded body, currently only
14613 the first chunk is analyzed.
14614
Thierry FOURNIER9826c772015-05-20 15:50:54 +020014615req.body_param([<name>) : string
14616 This fetch assumes that the body of the POST request is url-encoded. The user
14617 can check if the "content-type" contains the value
14618 "application/x-www-form-urlencoded". This extracts the first occurrence of the
14619 parameter <name> in the body, which ends before '&'. The parameter name is
14620 case-sensitive. If no name is given, any parameter will match, and the first
14621 one will be returned. The result is a string corresponding to the value of the
14622 parameter <name> as presented in the request body (no URL decoding is
14623 performed). Note that the ACL version of this fetch iterates over multiple
14624 parameters and will iteratively report all parameters values if no name is
14625 given.
14626
Willy Tarreaua5910cc2015-05-02 00:46:08 +020014627req.body_len : integer
14628 This returns the length of the HTTP request's available body in bytes. It may
14629 be lower than the advertised length if the body is larger than the buffer. It
14630 requires that the request body has been buffered made available using
14631 "option http-buffer-request".
14632
14633req.body_size : integer
14634 This returns the advertised length of the HTTP request's body in bytes. It
14635 will represent the advertised Content-Length header, or the size of the first
14636 chunk in case of chunked encoding. In order to parse the chunks, it requires
14637 that the request body has been buffered made available using
14638 "option http-buffer-request".
14639
Willy Tarreau74ca5042013-06-11 23:12:07 +020014640req.cook([<name>]) : string
14641cook([<name>]) : string (deprecated)
14642 This extracts the last occurrence of the cookie name <name> on a "Cookie"
14643 header line from the request, and returns its value as string. If no name is
14644 specified, the first cookie value is returned. When used with ACLs, all
14645 matching cookies are evaluated. Spaces around the name and the value are
14646 ignored as requested by the Cookie header specification (RFC6265). The cookie
14647 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
14648 well return an empty value if it is present. Use the "found" match to detect
14649 presence. Use the res.cook() variant for response cookies sent by the server.
14650
14651 ACL derivatives :
14652 cook([<name>]) : exact string match
14653 cook_beg([<name>]) : prefix match
14654 cook_dir([<name>]) : subdir match
14655 cook_dom([<name>]) : domain match
14656 cook_end([<name>]) : suffix match
14657 cook_len([<name>]) : length match
14658 cook_reg([<name>]) : regex match
14659 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010014660
Willy Tarreau74ca5042013-06-11 23:12:07 +020014661req.cook_cnt([<name>]) : integer
14662cook_cnt([<name>]) : integer (deprecated)
14663 Returns an integer value representing the number of occurrences of the cookie
14664 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014665
Willy Tarreau74ca5042013-06-11 23:12:07 +020014666req.cook_val([<name>]) : integer
14667cook_val([<name>]) : integer (deprecated)
14668 This extracts the last occurrence of the cookie name <name> on a "Cookie"
14669 header line from the request, and converts its value to an integer which is
14670 returned. If no name is specified, the first cookie value is returned. When
14671 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020014672
Willy Tarreau74ca5042013-06-11 23:12:07 +020014673cookie([<name>]) : string (deprecated)
14674 This extracts the last occurrence of the cookie name <name> on a "Cookie"
14675 header line from the request, or a "Set-Cookie" header from the response, and
14676 returns its value as a string. A typical use is to get multiple clients
14677 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020014678 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020014679 multi-peer synchronization and state keeping across restarts. If no name is
14680 specified, the first cookie value is returned. This fetch should not be used
14681 anymore and should be replaced by req.cook() or res.cook() instead as it
14682 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014683
Willy Tarreau74ca5042013-06-11 23:12:07 +020014684hdr([<name>[,<occ>]]) : string
14685 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
14686 used on responses. Please refer to these respective fetches for more details.
14687 In case of doubt about the fetch direction, please use the explicit ones.
14688 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014689 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014690
Willy Tarreau74ca5042013-06-11 23:12:07 +020014691req.fhdr(<name>[,<occ>]) : string
14692 This extracts the last occurrence of header <name> in an HTTP request. When
14693 used from an ACL, all occurrences are iterated over until a match is found.
14694 Optionally, a specific occurrence might be specified as a position number.
14695 Positive values indicate a position from the first occurrence, with 1 being
14696 the first one. Negative values indicate positions relative to the last one,
14697 with -1 being the last one. It differs from req.hdr() in that any commas
14698 present in the value are returned and are not used as delimiters. This is
14699 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014700
Willy Tarreau74ca5042013-06-11 23:12:07 +020014701req.fhdr_cnt([<name>]) : integer
14702 Returns an integer value representing the number of occurrences of request
14703 header field name <name>, or the total number of header fields if <name> is
14704 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
14705 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014706
Willy Tarreau74ca5042013-06-11 23:12:07 +020014707req.hdr([<name>[,<occ>]]) : string
14708 This extracts the last occurrence of header <name> in an HTTP request. When
14709 used from an ACL, all occurrences are iterated over until a match is found.
14710 Optionally, a specific occurrence might be specified as a position number.
14711 Positive values indicate a position from the first occurrence, with 1 being
14712 the first one. Negative values indicate positions relative to the last one,
14713 with -1 being the last one. A typical use is with the X-Forwarded-For header
14714 once converted to IP, associated with an IP stick-table. The function
14715 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000014716 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020014717 how certain headers are supposed to be parsed. Also, some of them are case
14718 insensitive (eg: Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010014719
Willy Tarreau74ca5042013-06-11 23:12:07 +020014720 ACL derivatives :
14721 hdr([<name>[,<occ>]]) : exact string match
14722 hdr_beg([<name>[,<occ>]]) : prefix match
14723 hdr_dir([<name>[,<occ>]]) : subdir match
14724 hdr_dom([<name>[,<occ>]]) : domain match
14725 hdr_end([<name>[,<occ>]]) : suffix match
14726 hdr_len([<name>[,<occ>]]) : length match
14727 hdr_reg([<name>[,<occ>]]) : regex match
14728 hdr_sub([<name>[,<occ>]]) : substring match
14729
14730req.hdr_cnt([<name>]) : integer
14731hdr_cnt([<header>]) : integer (deprecated)
14732 Returns an integer value representing the number of occurrences of request
14733 header field name <name>, or the total number of header field values if
14734 <name> is not specified. It is important to remember that one header line may
14735 count as several headers if it has several values. The function considers any
14736 comma as a delimiter for distinct values. If full-line headers are desired
14737 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
14738 detect presence, absence or abuse of a specific header, as well as to block
14739 request smuggling attacks by rejecting requests which contain more than one
14740 of certain headers. See "req.hdr" for more information on header matching.
14741
14742req.hdr_ip([<name>[,<occ>]]) : ip
14743hdr_ip([<name>[,<occ>]]) : ip (deprecated)
14744 This extracts the last occurrence of header <name> in an HTTP request,
14745 converts it to an IPv4 or IPv6 address and returns this address. When used
14746 with ACLs, all occurrences are checked, and if <name> is omitted, every value
14747 of every header is checked. Optionally, a specific occurrence might be
14748 specified as a position number. Positive values indicate a position from the
14749 first occurrence, with 1 being the first one. Negative values indicate
14750 positions relative to the last one, with -1 being the last one. A typical use
14751 is with the X-Forwarded-For and X-Client-IP headers.
14752
14753req.hdr_val([<name>[,<occ>]]) : integer
14754hdr_val([<name>[,<occ>]]) : integer (deprecated)
14755 This extracts the last occurrence of header <name> in an HTTP request, and
14756 converts it to an integer value. When used with ACLs, all occurrences are
14757 checked, and if <name> is omitted, every value of every header is checked.
14758 Optionally, a specific occurrence might be specified as a position number.
14759 Positive values indicate a position from the first occurrence, with 1 being
14760 the first one. Negative values indicate positions relative to the last one,
14761 with -1 being the last one. A typical use is with the X-Forwarded-For header.
14762
14763http_auth(<userlist>) : boolean
14764 Returns a boolean indicating whether the authentication data received from
14765 the client match a username & password stored in the specified userlist. This
14766 fetch function is not really useful outside of ACLs. Currently only http
14767 basic auth is supported.
14768
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010014769http_auth_group(<userlist>) : string
14770 Returns a string corresponding to the user name found in the authentication
14771 data received from the client if both the user name and password are valid
14772 according to the specified userlist. The main purpose is to use it in ACLs
14773 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014774 This fetch function is not really useful outside of ACLs. Currently only http
14775 basic auth is supported.
14776
14777 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010014778 http_auth_group(<userlist>) : group ...
14779 Returns true when the user extracted from the request and whose password is
14780 valid according to the specified userlist belongs to at least one of the
14781 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014782
14783http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020014784 Returns true when the request being processed is the first one of the
14785 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020014786 from some requests when a request is not the first one, or to help grouping
14787 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020014788
Willy Tarreau74ca5042013-06-11 23:12:07 +020014789method : integer + string
14790 Returns an integer value corresponding to the method in the HTTP request. For
14791 example, "GET" equals 1 (check sources to establish the matching). Value 9
14792 means "other method" and may be converted to a string extracted from the
14793 stream. This should not be used directly as a sample, this is only meant to
14794 be used from ACLs, which transparently convert methods from patterns to these
14795 integer + string values. Some predefined ACL already check for most common
14796 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014797
Willy Tarreau74ca5042013-06-11 23:12:07 +020014798 ACL derivatives :
14799 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020014800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801 Example :
14802 # only accept GET and HEAD requests
14803 acl valid_method method GET HEAD
14804 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020014805
Willy Tarreau74ca5042013-06-11 23:12:07 +020014806path : string
14807 This extracts the request's URL path, which starts at the first slash and
14808 ends before the question mark (without the host part). A typical use is with
14809 prefetch-capable caches, and with portals which need to aggregate multiple
14810 information from databases and keep them in caches. Note that with outgoing
14811 caches, it would be wiser to use "url" instead. With ACLs, it's typically
14812 used to match exact file names (eg: "/login.php"), or directory parts using
14813 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014814
Willy Tarreau74ca5042013-06-11 23:12:07 +020014815 ACL derivatives :
14816 path : exact string match
14817 path_beg : prefix match
14818 path_dir : subdir match
14819 path_dom : domain match
14820 path_end : suffix match
14821 path_len : length match
14822 path_reg : regex match
14823 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020014824
Willy Tarreau49ad95c2015-01-19 15:06:26 +010014825query : string
14826 This extracts the request's query string, which starts after the first
14827 question mark. If no question mark is present, this fetch returns nothing. If
14828 a question mark is present but nothing follows, it returns an empty string.
14829 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010014830 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010014831 which stops before the question mark.
14832
Willy Tarreaueb27ec72015-02-20 13:55:29 +010014833req.hdr_names([<delim>]) : string
14834 This builds a string made from the concatenation of all header names as they
14835 appear in the request when the rule is evaluated. The default delimiter is
14836 the comma (',') but it may be overridden as an optional argument <delim>. In
14837 this case, only the first character of <delim> is considered.
14838
Willy Tarreau74ca5042013-06-11 23:12:07 +020014839req.ver : string
14840req_ver : string (deprecated)
14841 Returns the version string from the HTTP request, for example "1.1". This can
14842 be useful for logs, but is mostly there for ACL. Some predefined ACL already
14843 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014844
Willy Tarreau74ca5042013-06-11 23:12:07 +020014845 ACL derivatives :
14846 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020014847
Willy Tarreau74ca5042013-06-11 23:12:07 +020014848res.comp : boolean
14849 Returns the boolean "true" value if the response has been compressed by
14850 HAProxy, otherwise returns boolean "false". This may be used to add
14851 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014852
Willy Tarreau74ca5042013-06-11 23:12:07 +020014853res.comp_algo : string
14854 Returns a string containing the name of the algorithm used if the response
14855 was compressed by HAProxy, for example : "deflate". This may be used to add
14856 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014857
Willy Tarreau74ca5042013-06-11 23:12:07 +020014858res.cook([<name>]) : string
14859scook([<name>]) : string (deprecated)
14860 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
14861 header line from the response, and returns its value as string. If no name is
14862 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020014863
Willy Tarreau74ca5042013-06-11 23:12:07 +020014864 ACL derivatives :
14865 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020014866
Willy Tarreau74ca5042013-06-11 23:12:07 +020014867res.cook_cnt([<name>]) : integer
14868scook_cnt([<name>]) : integer (deprecated)
14869 Returns an integer value representing the number of occurrences of the cookie
14870 <name> in the response, or all cookies if <name> is not specified. This is
14871 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014872
Willy Tarreau74ca5042013-06-11 23:12:07 +020014873res.cook_val([<name>]) : integer
14874scook_val([<name>]) : integer (deprecated)
14875 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
14876 header line from the response, and converts its value to an integer which is
14877 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014878
Willy Tarreau74ca5042013-06-11 23:12:07 +020014879res.fhdr([<name>[,<occ>]]) : string
14880 This extracts the last occurrence of header <name> in an HTTP response, or of
14881 the last header if no <name> is specified. Optionally, a specific occurrence
14882 might be specified as a position number. Positive values indicate a position
14883 from the first occurrence, with 1 being the first one. Negative values
14884 indicate positions relative to the last one, with -1 being the last one. It
14885 differs from res.hdr() in that any commas present in the value are returned
14886 and are not used as delimiters. If this is not desired, the res.hdr() fetch
14887 should be used instead. This is sometimes useful with headers such as Date or
14888 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014889
Willy Tarreau74ca5042013-06-11 23:12:07 +020014890res.fhdr_cnt([<name>]) : integer
14891 Returns an integer value representing the number of occurrences of response
14892 header field name <name>, or the total number of header fields if <name> is
14893 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
14894 the number of full line headers and does not stop on commas. If this is not
14895 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014896
Willy Tarreau74ca5042013-06-11 23:12:07 +020014897res.hdr([<name>[,<occ>]]) : string
14898shdr([<name>[,<occ>]]) : string (deprecated)
14899 This extracts the last occurrence of header <name> in an HTTP response, or of
14900 the last header if no <name> is specified. Optionally, a specific occurrence
14901 might be specified as a position number. Positive values indicate a position
14902 from the first occurrence, with 1 being the first one. Negative values
14903 indicate positions relative to the last one, with -1 being the last one. This
14904 can be useful to learn some data into a stick-table. The function considers
14905 any comma as a delimiter for distinct values. If this is not desired, the
14906 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014907
Willy Tarreau74ca5042013-06-11 23:12:07 +020014908 ACL derivatives :
14909 shdr([<name>[,<occ>]]) : exact string match
14910 shdr_beg([<name>[,<occ>]]) : prefix match
14911 shdr_dir([<name>[,<occ>]]) : subdir match
14912 shdr_dom([<name>[,<occ>]]) : domain match
14913 shdr_end([<name>[,<occ>]]) : suffix match
14914 shdr_len([<name>[,<occ>]]) : length match
14915 shdr_reg([<name>[,<occ>]]) : regex match
14916 shdr_sub([<name>[,<occ>]]) : substring match
14917
14918res.hdr_cnt([<name>]) : integer
14919shdr_cnt([<name>]) : integer (deprecated)
14920 Returns an integer value representing the number of occurrences of response
14921 header field name <name>, or the total number of header fields if <name> is
14922 not specified. The function considers any comma as a delimiter for distinct
14923 values. If this is not desired, the res.fhdr_cnt() fetch should be used
14924 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014925
Willy Tarreau74ca5042013-06-11 23:12:07 +020014926res.hdr_ip([<name>[,<occ>]]) : ip
14927shdr_ip([<name>[,<occ>]]) : ip (deprecated)
14928 This extracts the last occurrence of header <name> in an HTTP response,
14929 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
14930 specific occurrence might be specified as a position number. Positive values
14931 indicate a position from the first occurrence, with 1 being the first one.
14932 Negative values indicate positions relative to the last one, with -1 being
14933 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014934
Willy Tarreaueb27ec72015-02-20 13:55:29 +010014935res.hdr_names([<delim>]) : string
14936 This builds a string made from the concatenation of all header names as they
14937 appear in the response when the rule is evaluated. The default delimiter is
14938 the comma (',') but it may be overridden as an optional argument <delim>. In
14939 this case, only the first character of <delim> is considered.
14940
Willy Tarreau74ca5042013-06-11 23:12:07 +020014941res.hdr_val([<name>[,<occ>]]) : integer
14942shdr_val([<name>[,<occ>]]) : integer (deprecated)
14943 This extracts the last occurrence of header <name> in an HTTP response, and
14944 converts it to an integer value. Optionally, a specific occurrence might be
14945 specified as a position number. Positive values indicate a position from the
14946 first occurrence, with 1 being the first one. Negative values indicate
14947 positions relative to the last one, with -1 being the last one. This can be
14948 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010014949
Willy Tarreau74ca5042013-06-11 23:12:07 +020014950res.ver : string
14951resp_ver : string (deprecated)
14952 Returns the version string from the HTTP response, for example "1.1". This
14953 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020014954
Willy Tarreau74ca5042013-06-11 23:12:07 +020014955 ACL derivatives :
14956 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010014957
Willy Tarreau74ca5042013-06-11 23:12:07 +020014958set-cookie([<name>]) : string (deprecated)
14959 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
14960 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020014961 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020014962 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010014963
Willy Tarreau74ca5042013-06-11 23:12:07 +020014964 This fetch function is deprecated and has been superseded by the "res.cook"
14965 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010014966
Willy Tarreau74ca5042013-06-11 23:12:07 +020014967status : integer
14968 Returns an integer containing the HTTP status code in the HTTP response, for
14969 example, 302. It is mostly used within ACLs and integer ranges, for example,
14970 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020014971
Thierry Fournier0e00dca2016-04-07 15:47:40 +020014972unique-id : string
14973 Returns the unique-id attached to the request. The directive
14974 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
14975 fails. Note that the unique-id is usually used with HTTP requests, however this
14976 sample fetch can be used with other protocols. Obviously, if it is used with
14977 other protocols than HTTP, the unique-id-format directive must not contain
14978 HTTP parts. See: unique-id-format and unique-id-header
14979
Willy Tarreau74ca5042013-06-11 23:12:07 +020014980url : string
14981 This extracts the request's URL as presented in the request. A typical use is
14982 with prefetch-capable caches, and with portals which need to aggregate
14983 multiple information from databases and keep them in caches. With ACLs, using
14984 "path" is preferred over using "url", because clients may send a full URL as
14985 is normally done with proxies. The only real use is to match "*" which does
14986 not match in "path", and for which there is already a predefined ACL. See
14987 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020014988
Willy Tarreau74ca5042013-06-11 23:12:07 +020014989 ACL derivatives :
14990 url : exact string match
14991 url_beg : prefix match
14992 url_dir : subdir match
14993 url_dom : domain match
14994 url_end : suffix match
14995 url_len : length match
14996 url_reg : regex match
14997 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020014998
Willy Tarreau74ca5042013-06-11 23:12:07 +020014999url_ip : ip
15000 This extracts the IP address from the request's URL when the host part is
15001 presented as an IP address. Its use is very limited. For instance, a
15002 monitoring system might use this field as an alternative for the source IP in
15003 order to test what path a given source address would follow, or to force an
15004 entry in a table for a given source address. With ACLs it can be used to
15005 restrict access to certain systems through a proxy, for example when combined
15006 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015007
Willy Tarreau74ca5042013-06-11 23:12:07 +020015008url_port : integer
15009 This extracts the port part from the request's URL. Note that if the port is
15010 not specified in the request, port 80 is assumed. With ACLs it can be used to
15011 restrict access to certain systems through a proxy, for example when combined
15012 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015013
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015014urlp([<name>[,<delim>]]) : string
15015url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015016 This extracts the first occurrence of the parameter <name> in the query
15017 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015018 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
15019 any parameter will match, and the first one will be returned. The result is
15020 a string corresponding to the value of the parameter <name> as presented in
15021 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020015022 stickiness based on a client ID, to extract an application cookie passed as a
15023 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015024 this fetch iterates over multiple parameters and will iteratively report all
15025 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015026
Willy Tarreau74ca5042013-06-11 23:12:07 +020015027 ACL derivatives :
15028 urlp(<name>[,<delim>]) : exact string match
15029 urlp_beg(<name>[,<delim>]) : prefix match
15030 urlp_dir(<name>[,<delim>]) : subdir match
15031 urlp_dom(<name>[,<delim>]) : domain match
15032 urlp_end(<name>[,<delim>]) : suffix match
15033 urlp_len(<name>[,<delim>]) : length match
15034 urlp_reg(<name>[,<delim>]) : regex match
15035 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015036
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015037
Willy Tarreau74ca5042013-06-11 23:12:07 +020015038 Example :
15039 # match http://example.com/foo?PHPSESSIONID=some_id
15040 stick on urlp(PHPSESSIONID)
15041 # match http://example.com/foo;JSESSIONID=some_id
15042 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015043
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015044urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020015045 See "urlp" above. This one extracts the URL parameter <name> in the request
15046 and converts it to an integer value. This can be used for session stickiness
15047 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020015048
Dragan Dosen0070cd52016-06-16 12:19:49 +020015049url32 : integer
15050 This returns a 32-bit hash of the value obtained by concatenating the first
15051 Host header and the whole URL including parameters (not only the path part of
15052 the request, as in the "base32" fetch above). This is useful to track per-URL
15053 activity. A shorter hash is stored, saving a lot of memory. The output type
15054 is an unsigned integer.
15055
15056url32+src : binary
15057 This returns the concatenation of the "url32" fetch and the "src" fetch. The
15058 resulting type is of type binary, with a size of 8 or 20 bytes depending on
15059 the source address family. This can be used to track per-IP, per-URL counters.
15060
Willy Tarreau198a7442008-01-17 12:05:32 +010015061
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150627.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015063---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015064
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015065Some predefined ACLs are hard-coded so that they do not have to be declared in
15066every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020015067order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010015068
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015069ACL name Equivalent to Usage
15070---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015071FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020015072HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015073HTTP_1.0 req_ver 1.0 match HTTP version 1.0
15074HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010015075HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
15076HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
15077HTTP_URL_SLASH url_beg / match URL beginning with "/"
15078HTTP_URL_STAR url * match URL equal to "*"
15079LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015080METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015081METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015082METH_GET method GET HEAD match HTTP GET or HEAD method
15083METH_HEAD method HEAD match HTTP HEAD method
15084METH_OPTIONS method OPTIONS match HTTP OPTIONS method
15085METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015086METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015087METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020015088RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015089REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015090TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015091WAIT_END wait_end wait for end of content analysis
15092---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015093
Willy Tarreaub937b7e2010-01-12 15:27:54 +010015094
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200150958. Logging
15096----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010015097
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015098One of HAProxy's strong points certainly lies is its precise logs. It probably
15099provides the finest level of information available for such a product, which is
15100very important for troubleshooting complex environments. Standard information
15101provided in logs include client ports, TCP/HTTP state timers, precise session
15102state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010015103to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015104headers.
15105
15106In order to improve administrators reactivity, it offers a great transparency
15107about encountered problems, both internal and external, and it is possible to
15108send logs to different sources at the same time with different level filters :
15109
15110 - global process-level logs (system errors, start/stop, etc..)
15111 - per-instance system and internal errors (lack of resource, bugs, ...)
15112 - per-instance external troubles (servers up/down, max connections)
15113 - per-instance activity (client connections), either at the establishment or
15114 at the termination.
Jim Freeman9e8714b2015-05-26 09:16:34 -060015115 - per-request control of log-level, eg:
15116 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015117
15118The ability to distribute different levels of logs to different log servers
15119allow several production teams to interact and to fix their problems as soon
15120as possible. For example, the system team might monitor system-wide errors,
15121while the application team might be monitoring the up/down for their servers in
15122real time, and the security team might analyze the activity logs with one hour
15123delay.
15124
15125
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200151268.1. Log levels
15127---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015128
Simon Hormandf791f52011-05-29 15:01:10 +090015129TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015130source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090015131HTTP request, HTTP return code, number of bytes transmitted, conditions
15132in which the session ended, and even exchanged cookies values. For example
15133track a particular user's problems. All messages may be sent to up to two
15134syslog servers. Check the "log" keyword in section 4.2 for more information
15135about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015136
15137
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200151388.2. Log formats
15139----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015140
William Lallemand48940402012-01-30 16:47:22 +010015141HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090015142and will be detailed in the following sections. A few of them may vary
15143slightly with the configuration, due to indicators specific to certain
15144options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015145
15146 - the default format, which is very basic and very rarely used. It only
15147 provides very basic information about the incoming connection at the moment
15148 it is accepted : source IP:port, destination IP:port, and frontend-name.
15149 This mode will eventually disappear so it will not be described to great
15150 extents.
15151
15152 - the TCP format, which is more advanced. This format is enabled when "option
15153 tcplog" is set on the frontend. HAProxy will then usually wait for the
15154 connection to terminate before logging. This format provides much richer
15155 information, such as timers, connection counts, queue size, etc... This
15156 format is recommended for pure TCP proxies.
15157
15158 - the HTTP format, which is the most advanced for HTTP proxying. This format
15159 is enabled when "option httplog" is set on the frontend. It provides the
15160 same information as the TCP format with some HTTP-specific fields such as
15161 the request, the status code, and captures of headers and cookies. This
15162 format is recommended for HTTP proxies.
15163
Emeric Brun3a058f32009-06-30 18:26:00 +020015164 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
15165 fields arranged in the same order as the CLF format. In this mode, all
15166 timers, captures, flags, etc... appear one per field after the end of the
15167 common fields, in the same order they appear in the standard HTTP format.
15168
William Lallemand48940402012-01-30 16:47:22 +010015169 - the custom log format, allows you to make your own log line.
15170
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015171Next sections will go deeper into details for each of these formats. Format
15172specification will be performed on a "field" basis. Unless stated otherwise, a
15173field is a portion of text delimited by any number of spaces. Since syslog
15174servers are susceptible of inserting fields at the beginning of a line, it is
15175always assumed that the first field is the one containing the process name and
15176identifier.
15177
15178Note : Since log lines may be quite long, the log examples in sections below
15179 might be broken into multiple lines. The example log lines will be
15180 prefixed with 3 closing angle brackets ('>>>') and each time a log is
15181 broken into multiple lines, each non-final line will end with a
15182 backslash ('\') and the next line will start indented by two characters.
15183
15184
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200151858.2.1. Default log format
15186-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015187
15188This format is used when no specific option is set. The log is emitted as soon
15189as the connection is accepted. One should note that this currently is the only
15190format which logs the request's destination IP and ports.
15191
15192 Example :
15193 listen www
15194 mode http
15195 log global
15196 server srv1 127.0.0.1:8000
15197
15198 >>> Feb 6 12:12:09 localhost \
15199 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
15200 (www/HTTP)
15201
15202 Field Format Extract from the example above
15203 1 process_name '[' pid ']:' haproxy[14385]:
15204 2 'Connect from' Connect from
15205 3 source_ip ':' source_port 10.0.1.2:33312
15206 4 'to' to
15207 5 destination_ip ':' destination_port 10.0.3.31:8012
15208 6 '(' frontend_name '/' mode ')' (www/HTTP)
15209
15210Detailed fields description :
15211 - "source_ip" is the IP address of the client which initiated the connection.
15212 - "source_port" is the TCP port of the client which initiated the connection.
15213 - "destination_ip" is the IP address the client connected to.
15214 - "destination_port" is the TCP port the client connected to.
15215 - "frontend_name" is the name of the frontend (or listener) which received
15216 and processed the connection.
15217 - "mode is the mode the frontend is operating (TCP or HTTP).
15218
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015219In case of a UNIX socket, the source and destination addresses are marked as
15220"unix:" and the ports reflect the internal ID of the socket which accepted the
15221connection (the same ID as reported in the stats).
15222
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015223It is advised not to use this deprecated format for newer installations as it
15224will eventually disappear.
15225
15226
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200152278.2.2. TCP log format
15228---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015229
15230The TCP format is used when "option tcplog" is specified in the frontend, and
15231is the recommended format for pure TCP proxies. It provides a lot of precious
15232information for troubleshooting. Since this format includes timers and byte
15233counts, the log is normally emitted at the end of the session. It can be
15234emitted earlier if "option logasap" is specified, which makes sense in most
15235environments with long sessions such as remote terminals. Sessions which match
15236the "monitor" rules are never logged. It is also possible not to emit logs for
15237sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015238specifying "option dontlognull" in the frontend. Successful connections will
15239not be logged if "option dontlog-normal" is specified in the frontend. A few
15240fields may slightly vary depending on some configuration options, those are
15241marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015242
15243 Example :
15244 frontend fnt
15245 mode tcp
15246 option tcplog
15247 log global
15248 default_backend bck
15249
15250 backend bck
15251 server srv1 127.0.0.1:8000
15252
15253 >>> Feb 6 12:12:56 localhost \
15254 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
15255 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
15256
15257 Field Format Extract from the example above
15258 1 process_name '[' pid ']:' haproxy[14387]:
15259 2 client_ip ':' client_port 10.0.1.2:33313
15260 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
15261 4 frontend_name fnt
15262 5 backend_name '/' server_name bck/srv1
15263 6 Tw '/' Tc '/' Tt* 0/0/5007
15264 7 bytes_read* 212
15265 8 termination_state --
15266 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
15267 10 srv_queue '/' backend_queue 0/0
15268
15269Detailed fields description :
15270 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015271 connection to haproxy. If the connection was accepted on a UNIX socket
15272 instead, the IP address would be replaced with the word "unix". Note that
15273 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015274 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
15275 and the NetScaler Client IP insetion protocol is correctly used, then the
15276 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015277
15278 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015279 If the connection was accepted on a UNIX socket instead, the port would be
15280 replaced with the ID of the accepting socket, which is also reported in the
15281 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015282
15283 - "accept_date" is the exact date when the connection was received by haproxy
15284 (which might be very slightly different from the date observed on the
15285 network if there was some queuing in the system's backlog). This is usually
15286 the same date which may appear in any upstream firewall's log.
15287
15288 - "frontend_name" is the name of the frontend (or listener) which received
15289 and processed the connection.
15290
15291 - "backend_name" is the name of the backend (or listener) which was selected
15292 to manage the connection to the server. This will be the same as the
15293 frontend if no switching rule has been applied, which is common for TCP
15294 applications.
15295
15296 - "server_name" is the name of the last server to which the connection was
15297 sent, which might differ from the first one if there were connection errors
15298 and a redispatch occurred. Note that this server belongs to the backend
15299 which processed the request. If the connection was aborted before reaching
15300 a server, "<NOSRV>" is indicated instead of a server name.
15301
15302 - "Tw" is the total time in milliseconds spent waiting in the various queues.
15303 It can be "-1" if the connection was aborted before reaching the queue.
15304 See "Timers" below for more details.
15305
15306 - "Tc" is the total time in milliseconds spent waiting for the connection to
15307 establish to the final server, including retries. It can be "-1" if the
15308 connection was aborted before a connection could be established. See
15309 "Timers" below for more details.
15310
15311 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015312 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015313 "option logasap" was specified, then the time counting stops at the moment
15314 the log is emitted. In this case, a '+' sign is prepended before the value,
15315 indicating that the final one will be larger. See "Timers" below for more
15316 details.
15317
15318 - "bytes_read" is the total number of bytes transmitted from the server to
15319 the client when the log is emitted. If "option logasap" is specified, the
15320 this value will be prefixed with a '+' sign indicating that the final one
15321 may be larger. Please note that this value is a 64-bit counter, so log
15322 analysis tools must be able to handle it without overflowing.
15323
15324 - "termination_state" is the condition the session was in when the session
15325 ended. This indicates the session state, which side caused the end of
15326 session to happen, and for what reason (timeout, error, ...). The normal
15327 flags should be "--", indicating the session was closed by either end with
15328 no data remaining in buffers. See below "Session state at disconnection"
15329 for more details.
15330
15331 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015332 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015333 limits have been reached. For instance, if actconn is close to 512 when
15334 multiple connection errors occur, chances are high that the system limits
15335 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015336 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015337
15338 - "feconn" is the total number of concurrent connections on the frontend when
15339 the session was logged. It is useful to estimate the amount of resource
15340 required to sustain high loads, and to detect when the frontend's "maxconn"
15341 has been reached. Most often when this value increases by huge jumps, it is
15342 because there is congestion on the backend servers, but sometimes it can be
15343 caused by a denial of service attack.
15344
15345 - "beconn" is the total number of concurrent connections handled by the
15346 backend when the session was logged. It includes the total number of
15347 concurrent connections active on servers as well as the number of
15348 connections pending in queues. It is useful to estimate the amount of
15349 additional servers needed to support high loads for a given application.
15350 Most often when this value increases by huge jumps, it is because there is
15351 congestion on the backend servers, but sometimes it can be caused by a
15352 denial of service attack.
15353
15354 - "srv_conn" is the total number of concurrent connections still active on
15355 the server when the session was logged. It can never exceed the server's
15356 configured "maxconn" parameter. If this value is very often close or equal
15357 to the server's "maxconn", it means that traffic regulation is involved a
15358 lot, meaning that either the server's maxconn value is too low, or that
15359 there aren't enough servers to process the load with an optimal response
15360 time. When only one of the server's "srv_conn" is high, it usually means
15361 that this server has some trouble causing the connections to take longer to
15362 be processed than on other servers.
15363
15364 - "retries" is the number of connection retries experienced by this session
15365 when trying to connect to the server. It must normally be zero, unless a
15366 server is being stopped at the same moment the connection was attempted.
15367 Frequent retries generally indicate either a network problem between
15368 haproxy and the server, or a misconfigured system backlog on the server
15369 preventing new connections from being queued. This field may optionally be
15370 prefixed with a '+' sign, indicating that the session has experienced a
15371 redispatch after the maximal retry count has been reached on the initial
15372 server. In this case, the server name appearing in the log is the one the
15373 connection was redispatched to, and not the first one, though both may
15374 sometimes be the same in case of hashing for instance. So as a general rule
15375 of thumb, when a '+' is present in front of the retry count, this count
15376 should not be attributed to the logged server.
15377
15378 - "srv_queue" is the total number of requests which were processed before
15379 this one in the server queue. It is zero when the request has not gone
15380 through the server queue. It makes it possible to estimate the approximate
15381 server's response time by dividing the time spent in queue by the number of
15382 requests in the queue. It is worth noting that if a session experiences a
15383 redispatch and passes through two server queues, their positions will be
15384 cumulated. A request should not pass through both the server queue and the
15385 backend queue unless a redispatch occurs.
15386
15387 - "backend_queue" is the total number of requests which were processed before
15388 this one in the backend's global queue. It is zero when the request has not
15389 gone through the global queue. It makes it possible to estimate the average
15390 queue length, which easily translates into a number of missing servers when
15391 divided by a server's "maxconn" parameter. It is worth noting that if a
15392 session experiences a redispatch, it may pass twice in the backend's queue,
15393 and then both positions will be cumulated. A request should not pass
15394 through both the server queue and the backend queue unless a redispatch
15395 occurs.
15396
15397
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200153988.2.3. HTTP log format
15399----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015400
15401The HTTP format is the most complete and the best suited for HTTP proxies. It
15402is enabled by when "option httplog" is specified in the frontend. It provides
15403the same level of information as the TCP format with additional features which
15404are specific to the HTTP protocol. Just like the TCP format, the log is usually
15405emitted at the end of the session, unless "option logasap" is specified, which
15406generally only makes sense for download sites. A session which matches the
15407"monitor" rules will never logged. It is also possible not to log sessions for
15408which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015409frontend. Successful connections will not be logged if "option dontlog-normal"
15410is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015411
15412Most fields are shared with the TCP log, some being different. A few fields may
15413slightly vary depending on some configuration options. Those ones are marked
15414with a star ('*') after the field name below.
15415
15416 Example :
15417 frontend http-in
15418 mode http
15419 option httplog
15420 log global
15421 default_backend bck
15422
15423 backend static
15424 server srv1 127.0.0.1:8000
15425
15426 >>> Feb 6 12:14:14 localhost \
15427 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
15428 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 +010015429 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015430
15431 Field Format Extract from the example above
15432 1 process_name '[' pid ']:' haproxy[14389]:
15433 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015434 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015435 4 frontend_name http-in
15436 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015437 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015438 7 status_code 200
15439 8 bytes_read* 2750
15440 9 captured_request_cookie -
15441 10 captured_response_cookie -
15442 11 termination_state ----
15443 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
15444 13 srv_queue '/' backend_queue 0/0
15445 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
15446 15 '{' captured_response_headers* '}' {}
15447 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010015448
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015449Detailed fields description :
15450 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015451 connection to haproxy. If the connection was accepted on a UNIX socket
15452 instead, the IP address would be replaced with the word "unix". Note that
15453 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015454 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
15455 and the NetScaler Client IP insetion protocol is correctly used, then the
15456 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015457
15458 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015459 If the connection was accepted on a UNIX socket instead, the port would be
15460 replaced with the ID of the accepting socket, which is also reported in the
15461 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015462
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015463 - "request_date" is the exact date when the first byte of the HTTP request
15464 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015465
15466 - "frontend_name" is the name of the frontend (or listener) which received
15467 and processed the connection.
15468
15469 - "backend_name" is the name of the backend (or listener) which was selected
15470 to manage the connection to the server. This will be the same as the
15471 frontend if no switching rule has been applied.
15472
15473 - "server_name" is the name of the last server to which the connection was
15474 sent, which might differ from the first one if there were connection errors
15475 and a redispatch occurred. Note that this server belongs to the backend
15476 which processed the request. If the request was aborted before reaching a
15477 server, "<NOSRV>" is indicated instead of a server name. If the request was
15478 intercepted by the stats subsystem, "<STATS>" is indicated instead.
15479
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015480 - "TR" is the total time in milliseconds spent waiting for a full HTTP
15481 request from the client (not counting body) after the first byte was
15482 received. It can be "-1" if the connection was aborted before a complete
15483 request could be received or the a bad request was received. It should
15484 always be very small because a request generally fits in one single packet.
15485 Large times here generally indicate network issues between the client and
15486 haproxy or requests being typed by hand. See "Timers" below for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015487
15488 - "Tw" is the total time in milliseconds spent waiting in the various queues.
15489 It can be "-1" if the connection was aborted before reaching the queue.
15490 See "Timers" below for more details.
15491
15492 - "Tc" is the total time in milliseconds spent waiting for the connection to
15493 establish to the final server, including retries. It can be "-1" if the
15494 request was aborted before a connection could be established. See "Timers"
15495 below for more details.
15496
15497 - "Tr" is the total time in milliseconds spent waiting for the server to send
15498 a full HTTP response, not counting data. It can be "-1" if the request was
15499 aborted before a complete response could be received. It generally matches
15500 the server's processing time for the request, though it may be altered by
15501 the amount of data sent by the client to the server. Large times here on
15502 "GET" requests generally indicate an overloaded server. See "Timers" below
15503 for more details.
15504
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015505 - "Ta" is the time the request remained active in haproxy, which is the total
15506 time in milliseconds elapsed between the first byte of the request was
15507 received and the last byte of response was sent. It covers all possible
15508 processing except the handshake (see Th) and idle time (see Ti). There is
15509 one exception, if "option logasap" was specified, then the time counting
15510 stops at the moment the log is emitted. In this case, a '+' sign is
15511 prepended before the value, indicating that the final one will be larger.
15512 See "Timers" below for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015513
15514 - "status_code" is the HTTP status code returned to the client. This status
15515 is generally set by the server, but it might also be set by haproxy when
15516 the server cannot be reached or when its response is blocked by haproxy.
15517
15518 - "bytes_read" is the total number of bytes transmitted to the client when
15519 the log is emitted. This does include HTTP headers. If "option logasap" is
15520 specified, the this value will be prefixed with a '+' sign indicating that
15521 the final one may be larger. Please note that this value is a 64-bit
15522 counter, so log analysis tools must be able to handle it without
15523 overflowing.
15524
15525 - "captured_request_cookie" is an optional "name=value" entry indicating that
15526 the client had this cookie in the request. The cookie name and its maximum
15527 length are defined by the "capture cookie" statement in the frontend
15528 configuration. The field is a single dash ('-') when the option is not
15529 set. Only one cookie may be captured, it is generally used to track session
15530 ID exchanges between a client and a server to detect session crossing
15531 between clients due to application bugs. For more details, please consult
15532 the section "Capturing HTTP headers and cookies" below.
15533
15534 - "captured_response_cookie" is an optional "name=value" entry indicating
15535 that the server has returned a cookie with its response. The cookie name
15536 and its maximum length are defined by the "capture cookie" statement in the
15537 frontend configuration. The field is a single dash ('-') when the option is
15538 not set. Only one cookie may be captured, it is generally used to track
15539 session ID exchanges between a client and a server to detect session
15540 crossing between clients due to application bugs. For more details, please
15541 consult the section "Capturing HTTP headers and cookies" below.
15542
15543 - "termination_state" is the condition the session was in when the session
15544 ended. This indicates the session state, which side caused the end of
15545 session to happen, for what reason (timeout, error, ...), just like in TCP
15546 logs, and information about persistence operations on cookies in the last
15547 two characters. The normal flags should begin with "--", indicating the
15548 session was closed by either end with no data remaining in buffers. See
15549 below "Session state at disconnection" for more details.
15550
15551 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015552 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015553 limits have been reached. For instance, if actconn is close to 512 or 1024
15554 when multiple connection errors occur, chances are high that the system
15555 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015556 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015557 system.
15558
15559 - "feconn" is the total number of concurrent connections on the frontend when
15560 the session was logged. It is useful to estimate the amount of resource
15561 required to sustain high loads, and to detect when the frontend's "maxconn"
15562 has been reached. Most often when this value increases by huge jumps, it is
15563 because there is congestion on the backend servers, but sometimes it can be
15564 caused by a denial of service attack.
15565
15566 - "beconn" is the total number of concurrent connections handled by the
15567 backend when the session was logged. It includes the total number of
15568 concurrent connections active on servers as well as the number of
15569 connections pending in queues. It is useful to estimate the amount of
15570 additional servers needed to support high loads for a given application.
15571 Most often when this value increases by huge jumps, it is because there is
15572 congestion on the backend servers, but sometimes it can be caused by a
15573 denial of service attack.
15574
15575 - "srv_conn" is the total number of concurrent connections still active on
15576 the server when the session was logged. It can never exceed the server's
15577 configured "maxconn" parameter. If this value is very often close or equal
15578 to the server's "maxconn", it means that traffic regulation is involved a
15579 lot, meaning that either the server's maxconn value is too low, or that
15580 there aren't enough servers to process the load with an optimal response
15581 time. When only one of the server's "srv_conn" is high, it usually means
15582 that this server has some trouble causing the requests to take longer to be
15583 processed than on other servers.
15584
15585 - "retries" is the number of connection retries experienced by this session
15586 when trying to connect to the server. It must normally be zero, unless a
15587 server is being stopped at the same moment the connection was attempted.
15588 Frequent retries generally indicate either a network problem between
15589 haproxy and the server, or a misconfigured system backlog on the server
15590 preventing new connections from being queued. This field may optionally be
15591 prefixed with a '+' sign, indicating that the session has experienced a
15592 redispatch after the maximal retry count has been reached on the initial
15593 server. In this case, the server name appearing in the log is the one the
15594 connection was redispatched to, and not the first one, though both may
15595 sometimes be the same in case of hashing for instance. So as a general rule
15596 of thumb, when a '+' is present in front of the retry count, this count
15597 should not be attributed to the logged server.
15598
15599 - "srv_queue" is the total number of requests which were processed before
15600 this one in the server queue. It is zero when the request has not gone
15601 through the server queue. It makes it possible to estimate the approximate
15602 server's response time by dividing the time spent in queue by the number of
15603 requests in the queue. It is worth noting that if a session experiences a
15604 redispatch and passes through two server queues, their positions will be
15605 cumulated. A request should not pass through both the server queue and the
15606 backend queue unless a redispatch occurs.
15607
15608 - "backend_queue" is the total number of requests which were processed before
15609 this one in the backend's global queue. It is zero when the request has not
15610 gone through the global queue. It makes it possible to estimate the average
15611 queue length, which easily translates into a number of missing servers when
15612 divided by a server's "maxconn" parameter. It is worth noting that if a
15613 session experiences a redispatch, it may pass twice in the backend's queue,
15614 and then both positions will be cumulated. A request should not pass
15615 through both the server queue and the backend queue unless a redispatch
15616 occurs.
15617
15618 - "captured_request_headers" is a list of headers captured in the request due
15619 to the presence of the "capture request header" statement in the frontend.
15620 Multiple headers can be captured, they will be delimited by a vertical bar
15621 ('|'). When no capture is enabled, the braces do not appear, causing a
15622 shift of remaining fields. It is important to note that this field may
15623 contain spaces, and that using it requires a smarter log parser than when
15624 it's not used. Please consult the section "Capturing HTTP headers and
15625 cookies" below for more details.
15626
15627 - "captured_response_headers" is a list of headers captured in the response
15628 due to the presence of the "capture response header" statement in the
15629 frontend. Multiple headers can be captured, they will be delimited by a
15630 vertical bar ('|'). When no capture is enabled, the braces do not appear,
15631 causing a shift of remaining fields. It is important to note that this
15632 field may contain spaces, and that using it requires a smarter log parser
15633 than when it's not used. Please consult the section "Capturing HTTP headers
15634 and cookies" below for more details.
15635
15636 - "http_request" is the complete HTTP request line, including the method,
15637 request and HTTP version string. Non-printable characters are encoded (see
15638 below the section "Non-printable characters"). This is always the last
15639 field, and it is always delimited by quotes and is the only one which can
15640 contain quotes. If new fields are added to the log format, they will be
15641 added before this field. This field might be truncated if the request is
15642 huge and does not fit in the standard syslog buffer (1024 characters). This
15643 is the reason why this field must always remain the last one.
15644
15645
Cyril Bontédc4d9032012-04-08 21:57:39 +0200156468.2.4. Custom log format
15647------------------------
William Lallemand48940402012-01-30 16:47:22 +010015648
Willy Tarreau2beef582012-12-20 17:22:52 +010015649The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010015650mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010015651
15652HAproxy understands some log format variables. % precedes log format variables.
15653Variables can take arguments using braces ('{}'), and multiple arguments are
15654separated by commas within the braces. Flags may be added or removed by
15655prefixing them with a '+' or '-' sign.
15656
15657Special variable "%o" may be used to propagate its flags to all other
15658variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010015659("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010015660
Willy Tarreauc8368452012-12-21 00:09:23 +010015661If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020015662as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010015663less common information such as the client's SSL certificate's DN, or to log
15664the key that would be used to store an entry into a stick table.
15665
William Lallemand48940402012-01-30 16:47:22 +010015666Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015667In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010015668in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010015669
Dragan Dosen835b9212016-02-12 13:23:03 +010015670Note: when using the RFC5424 syslog message format, the characters '"',
15671'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
15672https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
15673such cases, the use of the flag "E" should be considered.
15674
William Lallemand48940402012-01-30 16:47:22 +010015675Flags are :
15676 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015677 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010015678 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
15679 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010015680
15681 Example:
15682
15683 log-format %T\ %t\ Some\ Text
15684 log-format %{+Q}o\ %t\ %s\ %{-Q}r
15685
Dragan Dosen835b9212016-02-12 13:23:03 +010015686 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
15687
William Lallemand48940402012-01-30 16:47:22 +010015688At the moment, the default HTTP format is defined this way :
15689
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015690 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
15691 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010015692
William Lallemandbddd4fd2012-02-27 11:23:10 +010015693the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010015694
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015695 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
15696 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
15697 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010015698
William Lallemandbddd4fd2012-02-27 11:23:10 +010015699and the default TCP format is defined this way :
15700
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015701 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
15702 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010015703
William Lallemand48940402012-01-30 16:47:22 +010015704Please refer to the table below for currently defined variables :
15705
William Lallemandbddd4fd2012-02-27 11:23:10 +010015706 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020015707 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015708 +---+------+-----------------------------------------------+-------------+
15709 | | %o | special variable, apply flags on all next var | |
15710 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010015711 | | %B | bytes_read (from server to client) | numeric |
15712 | H | %CC | captured_request_cookie | string |
15713 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020015714 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000015715 | H | %HM | HTTP method (ex: POST) | string |
15716 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000015717 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000015718 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
15719 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010015720 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020015721 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020015722 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015723 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015724 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020015725 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080015726 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015727 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
15728 | H | %Ti | idle time before the HTTP request | numeric |
15729 | H | %Tq | Th + Ti + TR | numeric |
15730 | H | %TR | time to receive the full request from 1st byte| numeric |
15731 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020015732 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015733 | | %Tt | Tt | numeric |
15734 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010015735 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015736 | | %ac | actconn | numeric |
15737 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010015738 | | %bc | beconn (backend concurrent connections) | numeric |
15739 | | %bi | backend_source_ip (connecting address) | IP |
15740 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015741 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010015742 | | %ci | client_ip (accepted address) | IP |
15743 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015744 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010015745 | | %fc | feconn (frontend concurrent connections) | numeric |
15746 | | %fi | frontend_ip (accepting address) | IP |
15747 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020015748 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020015749 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020015750 | | %hr | captured_request_headers default style | string |
15751 | | %hrl | captured_request_headers CLF style | string list |
15752 | | %hs | captured_response_headers default style | string |
15753 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020015754 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020015755 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020015756 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015757 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010015758 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015759 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010015760 | | %sc | srv_conn (server concurrent connections) | numeric |
15761 | | %si | server_IP (target address) | IP |
15762 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015763 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020015764 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
15765 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010015766 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015767 | H | %tr | date_time of HTTP request | date |
15768 | H | %trg | gmt_date_time of start of HTTP request | date |
15769 | H | %trl | locla_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015770 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020015771 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010015772 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010015773
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020015774 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010015775
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010015776
157778.2.5. Error log format
15778-----------------------
15779
15780When an incoming connection fails due to an SSL handshake or an invalid PROXY
15781protocol header, haproxy will log the event using a shorter, fixed line format.
15782By default, logs are emitted at the LOG_INFO level, unless the option
15783"log-separate-errors" is set in the backend, in which case the LOG_ERR level
15784will be used. Connections on which no data are exchanged (eg: probes) are not
15785logged if the "dontlognull" option is set.
15786
15787The format looks like this :
15788
15789 >>> Dec 3 18:27:14 localhost \
15790 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
15791 Connection error during SSL handshake
15792
15793 Field Format Extract from the example above
15794 1 process_name '[' pid ']:' haproxy[6103]:
15795 2 client_ip ':' client_port 127.0.0.1:56059
15796 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
15797 4 frontend_name "/" bind_name ":" frt/f1:
15798 5 message Connection error during SSL handshake
15799
15800These fields just provide minimal information to help debugging connection
15801failures.
15802
15803
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158048.3. Advanced logging options
15805-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015806
15807Some advanced logging options are often looked for but are not easy to find out
15808just by looking at the various options. Here is an entry point for the few
15809options which can enable better logging. Please refer to the keywords reference
15810for more information about their usage.
15811
15812
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158138.3.1. Disabling logging of external tests
15814------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015815
15816It is quite common to have some monitoring tools perform health checks on
15817haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
15818commercial load-balancer, and sometimes it will simply be a more complete
15819monitoring system such as Nagios. When the tests are very frequent, users often
15820ask how to disable logging for those checks. There are three possibilities :
15821
15822 - if connections come from everywhere and are just TCP probes, it is often
15823 desired to simply disable logging of connections without data exchange, by
15824 setting "option dontlognull" in the frontend. It also disables logging of
15825 port scans, which may or may not be desired.
15826
15827 - if the connection come from a known source network, use "monitor-net" to
15828 declare this network as monitoring only. Any host in this network will then
15829 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015830 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015831 such as other load-balancers.
15832
15833 - if the tests are performed on a known URI, use "monitor-uri" to declare
15834 this URI as dedicated to monitoring. Any host sending this request will
15835 only get the result of a health-check, and the request will not be logged.
15836
15837
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158388.3.2. Logging before waiting for the session to terminate
15839----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015840
15841The problem with logging at end of connection is that you have no clue about
15842what is happening during very long sessions, such as remote terminal sessions
15843or large file downloads. This problem can be worked around by specifying
15844"option logasap" in the frontend. Haproxy will then log as soon as possible,
15845just before data transfer begins. This means that in case of TCP, it will still
15846log the connection status to the server, and in case of HTTP, it will log just
15847after processing the server headers. In this case, the number of bytes reported
15848is the number of header bytes sent to the client. In order to avoid confusion
15849with normal logs, the total time field and the number of bytes are prefixed
15850with a '+' sign which means that real numbers are certainly larger.
15851
15852
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158538.3.3. Raising log level upon errors
15854------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015855
15856Sometimes it is more convenient to separate normal traffic from errors logs,
15857for instance in order to ease error monitoring from log files. When the option
15858"log-separate-errors" is used, connections which experience errors, timeouts,
15859retries, redispatches or HTTP status codes 5xx will see their syslog level
15860raised from "info" to "err". This will help a syslog daemon store the log in
15861a separate file. It is very important to keep the errors in the normal traffic
15862file too, so that log ordering is not altered. You should also be careful if
15863you already have configured your syslog daemon to store all logs higher than
15864"notice" in an "admin" file, because the "err" level is higher than "notice".
15865
15866
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158678.3.4. Disabling logging of successful connections
15868--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015869
15870Although this may sound strange at first, some large sites have to deal with
15871multiple thousands of logs per second and are experiencing difficulties keeping
15872them intact for a long time or detecting errors within them. If the option
15873"dontlog-normal" is set on the frontend, all normal connections will not be
15874logged. In this regard, a normal connection is defined as one without any
15875error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
15876and a response with a status 5xx is not considered normal and will be logged
15877too. Of course, doing is is really discouraged as it will remove most of the
15878useful information from the logs. Do this only if you have no other
15879alternative.
15880
15881
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158828.4. Timing events
15883------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015884
15885Timers provide a great help in troubleshooting network problems. All values are
15886reported in milliseconds (ms). These timers should be used in conjunction with
15887the session termination flags. In TCP mode with "option tcplog" set on the
15888frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015889mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
15890addition, three other measures are provided, "Th", "Ti", and "Tq".
15891
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010015892Timings events in HTTP mode:
15893
15894 first request 2nd request
15895 |<-------------------------------->|<-------------- ...
15896 t tr t tr ...
15897 ---|----|----|----|----|----|----|----|----|--
15898 : Th Ti TR Tw Tc Tr Td : Ti ...
15899 :<---- Tq ---->: :
15900 :<-------------- Tt -------------->:
15901 :<--------- Ta --------->:
15902
15903Timings events in TCP mode:
15904
15905 TCP session
15906 |<----------------->|
15907 t t
15908 ---|----|----|----|----|---
15909 | Th Tw Tc Td |
15910 |<------ Tt ------->|
15911
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015912 - Th: total time to accept tcp connection and execute handshakes for low level
15913 protocols. Currently, these protocoles are proxy-protocol and SSL. This may
15914 only happen once during the whole connection's lifetime. A large time here
15915 may indicate that the client only pre-established the connection without
15916 speaking, that it is experiencing network issues preventing it from
15917 completing a handshake in a reasonable time (eg: MTU issues), or that an
15918 SSL handshake was very expensive to compute.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015919
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015920 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
15921 counts between the end of the handshakes and the first byte of the HTTP
15922 request. When dealing with a second request in keep-alive mode, it starts
15923 to count after the end of the transmission the previous response. Some
15924 browsers pre-establish connections to a server in order to reduce the
15925 latency of a future request, and keep them pending until they need it. This
15926 delay will be reported as the idle time. A value of -1 indicates that
15927 nothing was received on the connection.
15928
15929 - TR: total time to get the client request (HTTP mode only). It's the time
15930 elapsed between the first bytes received and the moment the proxy received
15931 the empty line marking the end of the HTTP headers. The value "-1"
15932 indicates that the end of headers has never been seen. This happens when
15933 the client closes prematurely or times out. This time is usually very short
15934 since most requests fit in a single packet. A large time may indicate a
15935 request typed by hand during a test.
15936
15937 - Tq: total time to get the client request from the accept date or since the
15938 emission of the last byte of the previous response (HTTP mode only). It's
15939 exactly equalt to Th + Ti + TR unless any of them is -1, in which case it
15940 returns -1 as well. This timer used to be very useful before the arrival of
15941 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
15942 it in favor of TR nowadays, as the idle time adds a lot of noise to the
15943 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015944
15945 - Tw: total time spent in the queues waiting for a connection slot. It
15946 accounts for backend queue as well as the server queues, and depends on the
15947 queue size, and the time needed for the server to complete previous
15948 requests. The value "-1" means that the request was killed before reaching
15949 the queue, which is generally what happens with invalid or denied requests.
15950
15951 - Tc: total time to establish the TCP connection to the server. It's the time
15952 elapsed between the moment the proxy sent the connection request, and the
15953 moment it was acknowledged by the server, or between the TCP SYN packet and
15954 the matching SYN/ACK packet in return. The value "-1" means that the
15955 connection never established.
15956
15957 - Tr: server response time (HTTP mode only). It's the time elapsed between
15958 the moment the TCP connection was established to the server and the moment
15959 the server sent its complete response headers. It purely shows its request
15960 processing time, without the network overhead due to the data transmission.
15961 It is worth noting that when the client has data to send to the server, for
15962 instance during a POST request, the time already runs, and this can distort
15963 apparent response time. For this reason, it's generally wise not to trust
15964 too much this field for POST requests initiated from clients behind an
15965 untrusted network. A value of "-1" here means that the last the response
15966 header (empty line) was never seen, most likely because the server timeout
15967 stroke before the server managed to process the request.
15968
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015969 - Ta: total active time for the HTTP request, between the moment the proxy
15970 received the first byte of the request header and the emission of the last
15971 byte of the response body. The exception is when the "logasap" option is
15972 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
15973 a '+' sign. From this field, we can deduce "Td", the data transmission time,
15974 by subtracting other timers when valid :
15975
15976 Td = Ta - (TR + Tw + Tc + Tr)
15977
15978 Timers with "-1" values have to be excluded from this equation. Note that
15979 "Ta" can never be negative.
15980
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015981 - Tt: total session duration time, between the moment the proxy accepted it
15982 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015983 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
15984 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015985 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015986
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015987 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015988
15989 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015990 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
15991 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015992
15993These timers provide precious indications on trouble causes. Since the TCP
15994protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
15995that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015996due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
15997"Tt" is close to a timeout value specified in the configuration, it often means
15998that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015999
16000Most common cases :
16001
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016002 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
16003 the client and the proxy. This is very rare on local networks but might
16004 happen when clients are on far remote networks and send large requests. It
16005 may happen that values larger than usual appear here without any network
16006 cause. Sometimes, during an attack or just after a resource starvation has
16007 ended, haproxy may accept thousands of connections in a few milliseconds.
16008 The time spent accepting these connections will inevitably slightly delay
16009 processing of other connections, and it can happen that request times in the
16010 order of a few tens of milliseconds are measured after a few thousands of
16011 new connections have been accepted at once. Using one of the keep-alive
16012 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020016013 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016014
16015 - If "Tc" is close to 3000, a packet has probably been lost between the
16016 server and the proxy during the server connection phase. This value should
16017 always be very low, such as 1 ms on local networks and less than a few tens
16018 of ms on remote networks.
16019
Willy Tarreau55165fe2009-05-10 12:02:55 +020016020 - If "Tr" is nearly always lower than 3000 except some rare values which seem
16021 to be the average majored by 3000, there are probably some packets lost
16022 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016023
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016024 - If "Ta" is large even for small byte counts, it generally is because
16025 neither the client nor the server decides to close the connection while
16026 haproxy is running in tunnel mode and both have agreed on a keep-alive
16027 connection mode. In order to solve this issue, it will be needed to specify
16028 one of the HTTP options to manipulate keep-alive or close options on either
16029 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
16030 important when connection regulation is used with the "maxconn" option on
16031 the servers, since no new connection will be sent to the server until
16032 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016033
16034Other noticeable HTTP log cases ('xx' means any value to be ignored) :
16035
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016036 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016037 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016038 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016039
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016040 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016041 or it aborted too early. Check the session termination flags
16042 then "timeout http-request" and "timeout client" settings.
16043
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016044 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016045 servers were out of order, because the request was invalid
16046 or forbidden by ACL rules. Check the session termination
16047 flags.
16048
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016049 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
16050 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016051 Check the session termination flags, then check the
16052 "timeout connect" setting. Note that the tarpit action might
16053 return similar-looking patterns, with "Tw" equal to the time
16054 the client connection was maintained open.
16055
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016056 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016057 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016058 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016059 termination flags, then check the "timeout server" setting.
16060
16061
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160628.5. Session state at disconnection
16063-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016064
16065TCP and HTTP logs provide a session termination indicator in the
16066"termination_state" field, just before the number of active connections. It is
160672-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
16068each of which has a special meaning :
16069
16070 - On the first character, a code reporting the first event which caused the
16071 session to terminate :
16072
16073 C : the TCP session was unexpectedly aborted by the client.
16074
16075 S : the TCP session was unexpectedly aborted by the server, or the
16076 server explicitly refused it.
16077
16078 P : the session was prematurely aborted by the proxy, because of a
16079 connection limit enforcement, because a DENY filter was matched,
16080 because of a security check which detected and blocked a dangerous
16081 error in server response which might have caused information leak
Willy Tarreau570f2212013-06-10 16:42:09 +020016082 (eg: cacheable cookie).
16083
16084 L : the session was locally processed by haproxy and was not passed to
16085 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016086
16087 R : a resource on the proxy has been exhausted (memory, sockets, source
16088 ports, ...). Usually, this appears during the connection phase, and
16089 system logs should contain a copy of the precise error. If this
16090 happens, it must be considered as a very serious anomaly which
16091 should be fixed as soon as possible by any means.
16092
16093 I : an internal error was identified by the proxy during a self-check.
16094 This should NEVER happen, and you are encouraged to report any log
16095 containing this, because this would almost certainly be a bug. It
16096 would be wise to preventively restart the process after such an
16097 event too, in case it would be caused by memory corruption.
16098
Simon Horman752dc4a2011-06-21 14:34:59 +090016099 D : the session was killed by haproxy because the server was detected
16100 as down and was configured to kill all connections when going down.
16101
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070016102 U : the session was killed by haproxy on this backup server because an
16103 active server was detected as up and was configured to kill all
16104 backup connections when going up.
16105
Willy Tarreaua2a64e92011-09-07 23:01:56 +020016106 K : the session was actively killed by an admin operating on haproxy.
16107
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016108 c : the client-side timeout expired while waiting for the client to
16109 send or receive data.
16110
16111 s : the server-side timeout expired while waiting for the server to
16112 send or receive data.
16113
16114 - : normal session completion, both the client and the server closed
16115 with nothing left in the buffers.
16116
16117 - on the second character, the TCP or HTTP session state when it was closed :
16118
Willy Tarreauf7b30a92010-12-06 22:59:17 +010016119 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016120 (HTTP mode only). Nothing was sent to any server.
16121
16122 Q : the proxy was waiting in the QUEUE for a connection slot. This can
16123 only happen when servers have a 'maxconn' parameter set. It can
16124 also happen in the global queue after a redispatch consecutive to
16125 a failed attempt to connect to a dying server. If no redispatch is
16126 reported, then no connection attempt was made to any server.
16127
16128 C : the proxy was waiting for the CONNECTION to establish on the
16129 server. The server might at most have noticed a connection attempt.
16130
16131 H : the proxy was waiting for complete, valid response HEADERS from the
16132 server (HTTP only).
16133
16134 D : the session was in the DATA phase.
16135
16136 L : the proxy was still transmitting LAST data to the client while the
16137 server had already finished. This one is very rare as it can only
16138 happen when the client dies while receiving the last packets.
16139
16140 T : the request was tarpitted. It has been held open with the client
16141 during the whole "timeout tarpit" duration or until the client
16142 closed, both of which will be reported in the "Tw" timer.
16143
16144 - : normal session completion after end of data transfer.
16145
16146 - the third character tells whether the persistence cookie was provided by
16147 the client (only in HTTP mode) :
16148
16149 N : the client provided NO cookie. This is usually the case for new
16150 visitors, so counting the number of occurrences of this flag in the
16151 logs generally indicate a valid trend for the site frequentation.
16152
16153 I : the client provided an INVALID cookie matching no known server.
16154 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020016155 cookies between HTTP/HTTPS sites, persistence conditionally
16156 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016157
16158 D : the client provided a cookie designating a server which was DOWN,
16159 so either "option persist" was used and the client was sent to
16160 this server, or it was not set and the client was redispatched to
16161 another server.
16162
Willy Tarreau996a92c2010-10-13 19:30:47 +020016163 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016164 server.
16165
Willy Tarreau996a92c2010-10-13 19:30:47 +020016166 E : the client provided a valid cookie, but with a last date which was
16167 older than what is allowed by the "maxidle" cookie parameter, so
16168 the cookie is consider EXPIRED and is ignored. The request will be
16169 redispatched just as if there was no cookie.
16170
16171 O : the client provided a valid cookie, but with a first date which was
16172 older than what is allowed by the "maxlife" cookie parameter, so
16173 the cookie is consider too OLD and is ignored. The request will be
16174 redispatched just as if there was no cookie.
16175
Willy Tarreauc89ccb62012-04-05 21:18:22 +020016176 U : a cookie was present but was not used to select the server because
16177 some other server selection mechanism was used instead (typically a
16178 "use-server" rule).
16179
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016180 - : does not apply (no cookie set in configuration).
16181
16182 - the last character reports what operations were performed on the persistence
16183 cookie returned by the server (only in HTTP mode) :
16184
16185 N : NO cookie was provided by the server, and none was inserted either.
16186
16187 I : no cookie was provided by the server, and the proxy INSERTED one.
16188 Note that in "cookie insert" mode, if the server provides a cookie,
16189 it will still be overwritten and reported as "I" here.
16190
Willy Tarreau996a92c2010-10-13 19:30:47 +020016191 U : the proxy UPDATED the last date in the cookie that was presented by
16192 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016193 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020016194 date indicated in the cookie. If any other change happens, such as
16195 a redispatch, then the cookie will be marked as inserted instead.
16196
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016197 P : a cookie was PROVIDED by the server and transmitted as-is.
16198
16199 R : the cookie provided by the server was REWRITTEN by the proxy, which
16200 happens in "cookie rewrite" or "cookie prefix" modes.
16201
16202 D : the cookie provided by the server was DELETED by the proxy.
16203
16204 - : does not apply (no cookie set in configuration).
16205
Willy Tarreau996a92c2010-10-13 19:30:47 +020016206The combination of the two first flags gives a lot of information about what
16207was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016208helpful to detect server saturation, network troubles, local system resource
16209starvation, attacks, etc...
16210
16211The most common termination flags combinations are indicated below. They are
16212alphabetically sorted, with the lowercase set just after the upper case for
16213easier finding and understanding.
16214
16215 Flags Reason
16216
16217 -- Normal termination.
16218
16219 CC The client aborted before the connection could be established to the
16220 server. This can happen when haproxy tries to connect to a recently
16221 dead (or unchecked) server, and the client aborts while haproxy is
16222 waiting for the server to respond or for "timeout connect" to expire.
16223
16224 CD The client unexpectedly aborted during data transfer. This can be
16225 caused by a browser crash, by an intermediate equipment between the
16226 client and haproxy which decided to actively break the connection,
16227 by network routing issues between the client and haproxy, or by a
16228 keep-alive session between the server and the client terminated first
16229 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010016230
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016231 cD The client did not send nor acknowledge any data for as long as the
16232 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020016233 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016234
16235 CH The client aborted while waiting for the server to start responding.
16236 It might be the server taking too long to respond or the client
16237 clicking the 'Stop' button too fast.
16238
16239 cH The "timeout client" stroke while waiting for client data during a
16240 POST request. This is sometimes caused by too large TCP MSS values
16241 for PPPoE networks which cannot transport full-sized packets. It can
16242 also happen when client timeout is smaller than server timeout and
16243 the server takes too long to respond.
16244
16245 CQ The client aborted while its session was queued, waiting for a server
16246 with enough empty slots to accept it. It might be that either all the
16247 servers were saturated or that the assigned server was taking too
16248 long a time to respond.
16249
16250 CR The client aborted before sending a full HTTP request. Most likely
16251 the request was typed by hand using a telnet client, and aborted
16252 too early. The HTTP status code is likely a 400 here. Sometimes this
16253 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020016254 and the client. "option http-ignore-probes" can be used to ignore
16255 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016256
16257 cR The "timeout http-request" stroke before the client sent a full HTTP
16258 request. This is sometimes caused by too large TCP MSS values on the
16259 client side for PPPoE networks which cannot transport full-sized
16260 packets, or by clients sending requests by hand and not typing fast
16261 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020016262 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020016263 some browsers started to implement a "pre-connect" feature consisting
16264 in speculatively connecting to some recently visited web sites just
16265 in case the user would like to visit them. This results in many
16266 connections being established to web sites, which end up in 408
16267 Request Timeout if the timeout strikes first, or 400 Bad Request when
16268 the browser decides to close them first. These ones pollute the log
16269 and feed the error counters. Some versions of some browsers have even
16270 been reported to display the error code. It is possible to work
16271 around the undesirable effects of this behaviour by adding "option
16272 http-ignore-probes" in the frontend, resulting in connections with
16273 zero data transfer to be totally ignored. This will definitely hide
16274 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016275
16276 CT The client aborted while its session was tarpitted. It is important to
16277 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020016278 wrong tarpit rules have been written. If a lot of them happen, it
16279 might make sense to lower the "timeout tarpit" value to something
16280 closer to the average reported "Tw" timer, in order not to consume
16281 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016282
Willy Tarreau570f2212013-06-10 16:42:09 +020016283 LR The request was intercepted and locally handled by haproxy. Generally
16284 it means that this was a redirect or a stats request.
16285
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016286 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016287 the TCP connection (the proxy received a TCP RST or an ICMP message
16288 in return). Under some circumstances, it can also be the network
16289 stack telling the proxy that the server is unreachable (eg: no route,
16290 or no ARP response on local network). When this happens in HTTP mode,
16291 the status code is likely a 502 or 503 here.
16292
16293 sC The "timeout connect" stroke before a connection to the server could
16294 complete. When this happens in HTTP mode, the status code is likely a
16295 503 or 504 here.
16296
16297 SD The connection to the server died with an error during the data
16298 transfer. This usually means that haproxy has received an RST from
16299 the server or an ICMP message from an intermediate equipment while
16300 exchanging data with the server. This can be caused by a server crash
16301 or by a network issue on an intermediate equipment.
16302
16303 sD The server did not send nor acknowledge any data for as long as the
16304 "timeout server" setting during the data phase. This is often caused
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016305 by too short timeouts on L4 equipments before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016306 load-balancers, ...), as well as keep-alive sessions maintained
16307 between the client and the server expiring first on haproxy.
16308
16309 SH The server aborted before sending its full HTTP response headers, or
16310 it crashed while processing the request. Since a server aborting at
16311 this moment is very rare, it would be wise to inspect its logs to
16312 control whether it crashed and why. The logged request may indicate a
16313 small set of faulty requests, demonstrating bugs in the application.
16314 Sometimes this might also be caused by an IDS killing the connection
16315 between haproxy and the server.
16316
16317 sH The "timeout server" stroke before the server could return its
16318 response headers. This is the most common anomaly, indicating too
16319 long transactions, probably caused by server or database saturation.
16320 The immediate workaround consists in increasing the "timeout server"
16321 setting, but it is important to keep in mind that the user experience
16322 will suffer from these long response times. The only long term
16323 solution is to fix the application.
16324
16325 sQ The session spent too much time in queue and has been expired. See
16326 the "timeout queue" and "timeout connect" settings to find out how to
16327 fix this if it happens too often. If it often happens massively in
16328 short periods, it may indicate general problems on the affected
16329 servers due to I/O or database congestion, or saturation caused by
16330 external attacks.
16331
16332 PC The proxy refused to establish a connection to the server because the
16333 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020016334 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016335 so that it does not happen anymore. This status is very rare and
16336 might happen when the global "ulimit-n" parameter is forced by hand.
16337
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016338 PD The proxy blocked an incorrectly formatted chunked encoded message in
16339 a request or a response, after the server has emitted its headers. In
16340 most cases, this will indicate an invalid message from the server to
Willy Tarreauf3a3e132013-08-31 08:16:26 +020016341 the client. Haproxy supports chunk sizes of up to 2GB - 1 (2147483647
16342 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016343
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016344 PH The proxy blocked the server's response, because it was invalid,
16345 incomplete, dangerous (cache control), or matched a security filter.
16346 In any case, an HTTP 502 error is sent to the client. One possible
16347 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016348 containing unauthorized characters. It is also possible but quite
16349 rare, that the proxy blocked a chunked-encoding request from the
16350 client due to an invalid syntax, before the server responded. In this
16351 case, an HTTP 400 error is sent to the client and reported in the
16352 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016353
16354 PR The proxy blocked the client's HTTP request, either because of an
16355 invalid HTTP syntax, in which case it returned an HTTP 400 error to
16356 the client, or because a deny filter matched, in which case it
16357 returned an HTTP 403 error.
16358
16359 PT The proxy blocked the client's request and has tarpitted its
16360 connection before returning it a 500 server error. Nothing was sent
16361 to the server. The connection was maintained open for as long as
16362 reported by the "Tw" timer field.
16363
16364 RC A local resource has been exhausted (memory, sockets, source ports)
16365 preventing the connection to the server from establishing. The error
16366 logs will tell precisely what was missing. This is very rare and can
16367 only be solved by proper system tuning.
16368
Willy Tarreau996a92c2010-10-13 19:30:47 +020016369The combination of the two last flags gives a lot of information about how
16370persistence was handled by the client, the server and by haproxy. This is very
16371important to troubleshoot disconnections, when users complain they have to
16372re-authenticate. The commonly encountered flags are :
16373
16374 -- Persistence cookie is not enabled.
16375
16376 NN No cookie was provided by the client, none was inserted in the
16377 response. For instance, this can be in insert mode with "postonly"
16378 set on a GET request.
16379
16380 II A cookie designating an invalid server was provided by the client,
16381 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016382 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020016383 value can be presented by a client when no other server knows it.
16384
16385 NI No cookie was provided by the client, one was inserted in the
16386 response. This typically happens for first requests from every user
16387 in "insert" mode, which makes it an easy way to count real users.
16388
16389 VN A cookie was provided by the client, none was inserted in the
16390 response. This happens for most responses for which the client has
16391 already got a cookie.
16392
16393 VU A cookie was provided by the client, with a last visit date which is
16394 not completely up-to-date, so an updated cookie was provided in
16395 response. This can also happen if there was no date at all, or if
16396 there was a date but the "maxidle" parameter was not set, so that the
16397 cookie can be switched to unlimited time.
16398
16399 EI A cookie was provided by the client, with a last visit date which is
16400 too old for the "maxidle" parameter, so the cookie was ignored and a
16401 new cookie was inserted in the response.
16402
16403 OI A cookie was provided by the client, with a first visit date which is
16404 too old for the "maxlife" parameter, so the cookie was ignored and a
16405 new cookie was inserted in the response.
16406
16407 DI The server designated by the cookie was down, a new server was
16408 selected and a new cookie was emitted in the response.
16409
16410 VI The server designated by the cookie was not marked dead but could not
16411 be reached. A redispatch happened and selected another one, which was
16412 then advertised in the response.
16413
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016414
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164158.6. Non-printable characters
16416-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016417
16418In order not to cause trouble to log analysis tools or terminals during log
16419consulting, non-printable characters are not sent as-is into log files, but are
16420converted to the two-digits hexadecimal representation of their ASCII code,
16421prefixed by the character '#'. The only characters that can be logged without
16422being escaped are comprised between 32 and 126 (inclusive). Obviously, the
16423escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
16424is the same for the character '"' which becomes "#22", as well as '{', '|' and
16425'}' when logging headers.
16426
16427Note that the space character (' ') is not encoded in headers, which can cause
16428issues for tools relying on space count to locate fields. A typical header
16429containing spaces is "User-Agent".
16430
16431Last, it has been observed that some syslog daemons such as syslog-ng escape
16432the quote ('"') with a backslash ('\'). The reverse operation can safely be
16433performed since no quote may appear anywhere else in the logs.
16434
16435
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164368.7. Capturing HTTP cookies
16437---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016438
16439Cookie capture simplifies the tracking a complete user session. This can be
16440achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016441section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016442cookie will simultaneously be checked in the request ("Cookie:" header) and in
16443the response ("Set-Cookie:" header). The respective values will be reported in
16444the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016445locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016446not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
16447user switches to a new session for example, because the server will reassign it
16448a new cookie. It is also possible to detect if a server unexpectedly sets a
16449wrong cookie to a client, leading to session crossing.
16450
16451 Examples :
16452 # capture the first cookie whose name starts with "ASPSESSION"
16453 capture cookie ASPSESSION len 32
16454
16455 # capture the first cookie whose name is exactly "vgnvisitor"
16456 capture cookie vgnvisitor= len 32
16457
16458
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164598.8. Capturing HTTP headers
16460---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016461
16462Header captures are useful to track unique request identifiers set by an upper
16463proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
16464the response, one can search for information about the response length, how the
16465server asked the cache to behave, or an object location during a redirection.
16466
16467Header captures are performed using the "capture request header" and "capture
16468response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016469section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016470
16471It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016472time. Non-existent headers are logged as empty strings, and if one header
16473appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016474are grouped within braces '{' and '}' in the same order as they were declared,
16475and delimited with a vertical bar '|' without any space. Response headers
16476follow the same representation, but are displayed after a space following the
16477request headers block. These blocks are displayed just before the HTTP request
16478in the logs.
16479
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016480As a special case, it is possible to specify an HTTP header capture in a TCP
16481frontend. The purpose is to enable logging of headers which will be parsed in
16482an HTTP backend if the request is then switched to this HTTP backend.
16483
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016484 Example :
16485 # This instance chains to the outgoing proxy
16486 listen proxy-out
16487 mode http
16488 option httplog
16489 option logasap
16490 log global
16491 server cache1 192.168.1.1:3128
16492
16493 # log the name of the virtual server
16494 capture request header Host len 20
16495
16496 # log the amount of data uploaded during a POST
16497 capture request header Content-Length len 10
16498
16499 # log the beginning of the referrer
16500 capture request header Referer len 20
16501
16502 # server name (useful for outgoing proxies only)
16503 capture response header Server len 20
16504
16505 # logging the content-length is useful with "option logasap"
16506 capture response header Content-Length len 10
16507
16508 # log the expected cache behaviour on the response
16509 capture response header Cache-Control len 8
16510
16511 # the Via header will report the next proxy's name
16512 capture response header Via len 20
16513
16514 # log the URL location during a redirection
16515 capture response header Location len 20
16516
16517 >>> Aug 9 20:26:09 localhost \
16518 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
16519 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
16520 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
16521 "GET http://fr.adserver.yahoo.com/"
16522
16523 >>> Aug 9 20:30:46 localhost \
16524 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
16525 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
16526 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016527 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016528
16529 >>> Aug 9 20:30:46 localhost \
16530 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
16531 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
16532 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
16533 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016534 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016535
16536
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165378.9. Examples of logs
16538---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016539
16540These are real-world examples of logs accompanied with an explanation. Some of
16541them have been made up by hand. The syslog part has been removed for better
16542reading. Their sole purpose is to explain how to decipher them.
16543
16544 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
16545 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
16546 "HEAD / HTTP/1.0"
16547
16548 => long request (6.5s) entered by hand through 'telnet'. The server replied
16549 in 147 ms, and the session ended normally ('----')
16550
16551 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
16552 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
16553 0/9 "HEAD / HTTP/1.0"
16554
16555 => Idem, but the request was queued in the global queue behind 9 other
16556 requests, and waited there for 1230 ms.
16557
16558 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
16559 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
16560 "GET /image.iso HTTP/1.0"
16561
16562 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016563 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016564 14 ms, 243 bytes of headers were sent to the client, and total time from
16565 accept to first data byte is 30 ms.
16566
16567 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
16568 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
16569 "GET /cgi-bin/bug.cgi? HTTP/1.0"
16570
16571 => the proxy blocked a server response either because of an "rspdeny" or
16572 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020016573 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016574 risked being cached. In this case, the response is replaced with a "502
16575 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
16576 to return the 502 and not the server.
16577
16578 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016579 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 +010016580
16581 => the client never completed its request and aborted itself ("C---") after
16582 8.5s, while the proxy was waiting for the request headers ("-R--").
16583 Nothing was sent to any server.
16584
16585 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
16586 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
16587
16588 => The client never completed its request, which was aborted by the
16589 time-out ("c---") after 50s, while the proxy was waiting for the request
16590 headers ("-R--"). Nothing was sent to any server, but the proxy could
16591 send a 408 return code to the client.
16592
16593 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
16594 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
16595
16596 => This log was produced with "option tcplog". The client timed out after
16597 5 seconds ("c----").
16598
16599 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
16600 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016601 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016602
16603 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016604 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016605 (config says 'retries 3'), and no redispatch (otherwise we would have
16606 seen "/+3"). Status code 503 was returned to the client. There were 115
16607 connections on this server, 202 connections on this proxy, and 205 on
16608 the global process. It is possible that the server refused the
16609 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010016610
Willy Tarreau52b2d222011-09-07 23:48:48 +020016611
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200166129. Supported filters
16613--------------------
16614
16615Here are listed officially supported filters with the list of parameters they
16616accept. Depending on compile options, some of these filters might be
16617unavailable. The list of available filters is reported in haproxy -vv.
16618
16619See also : "filter"
16620
166219.1. Trace
16622----------
16623
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010016624filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020016625
16626 Arguments:
16627 <name> is an arbitrary name that will be reported in
16628 messages. If no name is provided, "TRACE" is used.
16629
16630 <random-parsing> enables the random parsing of data exchanged between
16631 the client and the server. By default, this filter
16632 parses all available data. With this parameter, it
16633 only parses a random amount of the available data.
16634
16635 <random-forwarding> enables the random forwading of parsed data. By
16636 default, this filter forwards all previously parsed
16637 data. With this parameter, it only forwards a random
16638 amount of the parsed data.
16639
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010016640 <hexump> dumps all forwarded data to the server and the client.
16641
Christopher Fauletc3fe5332016-04-07 15:30:10 +020016642This filter can be used as a base to develop new filters. It defines all
16643callbacks and print a message on the standard error stream (stderr) with useful
16644information for all of them. It may be useful to debug the activity of other
16645filters or, quite simply, HAProxy's activity.
16646
16647Using <random-parsing> and/or <random-forwarding> parameters is a good way to
16648tests the behavior of a filter that parses data exchanged between a client and
16649a server by adding some latencies in the processing.
16650
16651
166529.2. HTTP compression
16653---------------------
16654
16655filter compression
16656
16657The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
16658keyword must still be used to enable and configure the HTTP compression. And
16659when no other filter is used, it is enough. But it is mandatory to explicitly
16660use a filter line to enable the HTTP compression when two or more filters are
16661used for the same listener/frontend/backend. This is important to know the
16662filters evaluation order.
16663
16664See also : "compression"
16665
16666
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200166679.3. Stream Processing Offload Engine (SPOE)
16668--------------------------------------------
16669
16670filter spoe [engine <name>] config <file>
16671
16672 Arguments :
16673
16674 <name> is the engine name that will be used to find the right scope in
16675 the configuration file. If not provided, all the file will be
16676 parsed.
16677
16678 <file> is the path of the engine configuration file. This file can
16679 contain configuration of several engines. In this case, each
16680 part must be placed in its own scope.
16681
16682The Stream Processing Offload Engine (SPOE) is a filter communicating with
16683external components. It allows the offload of some specifics processing on the
16684streams in tierce applications. These external components and information
16685exchanged with them are configured in dedicated files, for the main part. It
16686also requires dedicated backends, defined in HAProxy configuration.
16687
16688SPOE communicates with external components using an in-house binary protocol,
16689the Stream Processing Offload Protocol (SPOP).
16690
Tim Düsterhus4896c442016-11-29 02:15:19 +010016691For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020016692"doc/SPOE.txt".
16693
16694Important note:
16695 The SPOE filter is highly experimental for now and was not heavily
16696 tested. It is really not production ready. So use it carefully.
16697
Willy Tarreau0ba27502007-12-24 16:55:16 +010016698/*
16699 * Local variables:
16700 * fill-column: 79
16701 * End:
16702 */