blob: 3d4661b6fbc07a4e572013b814991cd0476de876 [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 Tarreaub3066502017-11-26 19:50:17 +01005 version 1.9
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau96079492018-11-11 10:43:39 +01007 2018/11/11
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through 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
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 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
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. 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
Davor Ocelice9ed2812017-12-25 17:49:28 +0100595. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200605.1. 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
William Lallemand86d0df02017-11-24 21:36:45 +010011110. Cache
Cyril Bonté7b888f12017-11-26 22:24:31 +010011210.1. Limitation
11310.2. Setup
11410.2.1. Cache section
11510.2.2. Proxy section
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200116
1171. Quick reminder about HTTP
118----------------------------
119
Davor Ocelice9ed2812017-12-25 17:49:28 +0100120When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200121fully analyzed and indexed, thus it becomes possible to build matching criteria
122on almost anything found in the contents.
123
124However, it is important to understand how HTTP requests and responses are
125formed, and how HAProxy decomposes them. It will then become easier to write
126correct rules and to debug existing configurations.
127
128
1291.1. The HTTP transaction model
130-------------------------------
131
132The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100133to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100134from the client to the server, a request is sent by the client through the
135connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200136will involve a new connection :
137
138 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
139
140In this mode, called the "HTTP close" mode, there are as many connection
141establishments as there are HTTP transactions. Since the connection is closed
142by the server after the response, the client does not need to know the content
143length.
144
145Due to the transactional nature of the protocol, it was possible to improve it
146to avoid closing a connection between two subsequent transactions. In this mode
147however, it is mandatory that the server indicates the content length for each
148response so that the client does not wait indefinitely. For this, a special
149header is used: "Content-length". This mode is called the "keep-alive" mode :
150
151 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
152
153Its advantages are a reduced latency between transactions, and less processing
154power required on the server side. It is generally better than the close mode,
155but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200156a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200157
Willy Tarreau95c4e142017-11-26 12:18:55 +0100158Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200159keep-alive, but the client does not wait for the first response to send the
160second request. This is useful for fetching large number of images composing a
161page :
162
163 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
164
165This can obviously have a tremendous benefit on performance because the network
166latency is eliminated between subsequent requests. Many HTTP agents do not
167correctly support pipelining since there is no way to associate a response with
168the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100169server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170
Willy Tarreau95c4e142017-11-26 12:18:55 +0100171The next improvement is the multiplexed mode, as implemented in HTTP/2. This
172time, each transaction is assigned a single stream identifier, and all streams
173are multiplexed over an existing connection. Many requests can be sent in
174parallel by the client, and responses can arrive in any order since they also
175carry the stream identifier.
176
Willy Tarreau70dffda2014-01-30 03:07:23 +0100177By default HAProxy operates in keep-alive mode with regards to persistent
178connections: for each connection it processes each request and response, and
179leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100180start of a new request. When it receives HTTP/2 connections from a client, it
181processes all the requests in parallel and leaves the connection idling,
182waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200183
Christopher Faulet315b39c2018-09-21 16:26:19 +0200184HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100185 - keep alive : all requests and responses are processed (default)
186 - tunnel : only the first request and response are processed,
187 everything else is forwarded with no analysis.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100188 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200189 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100190
Davor Ocelice9ed2812017-12-25 17:49:28 +0100191For HTTP/2, the connection mode resembles more the "server close" mode : given
192the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100193server connection after a response, so it is closed after the response. HTTP/2
194is only supported for incoming connections, not on connections going to
195servers.
196
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197
1981.2. HTTP request
199-----------------
200
201First, let's consider this HTTP request :
202
203 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100204 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
206 2 Host: www.mydomain.com
207 3 User-agent: my small browser
208 4 Accept: image/jpeg, image/gif
209 5 Accept: image/png
210
211
2121.2.1. The Request line
213-----------------------
214
215Line 1 is the "request line". It is always composed of 3 fields :
216
217 - a METHOD : GET
218 - a URI : /serv/login.php?lang=en&profile=2
219 - a version tag : HTTP/1.1
220
221All of them are delimited by what the standard calls LWS (linear white spaces),
222which are commonly spaces, but can also be tabs or line feeds/carriage returns
223followed by spaces/tabs. The method itself cannot contain any colon (':') and
224is limited to alphabetic letters. All those various combinations make it
225desirable that HAProxy performs the splitting itself rather than leaving it to
226the user to write a complex or inaccurate regular expression.
227
228The URI itself can have several forms :
229
230 - A "relative URI" :
231
232 /serv/login.php?lang=en&profile=2
233
234 It is a complete URL without the host part. This is generally what is
235 received by servers, reverse proxies and transparent proxies.
236
237 - An "absolute URI", also called a "URL" :
238
239 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
240
241 It is composed of a "scheme" (the protocol name followed by '://'), a host
242 name or address, optionally a colon (':') followed by a port number, then
243 a relative URI beginning at the first slash ('/') after the address part.
244 This is generally what proxies receive, but a server supporting HTTP/1.1
245 must accept this form too.
246
247 - a star ('*') : this form is only accepted in association with the OPTIONS
248 method and is not relayable. It is used to inquiry a next hop's
249 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100250
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200251 - an address:port combination : 192.168.0.12:80
252 This is used with the CONNECT method, which is used to establish TCP
253 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
254 other protocols too.
255
256In a relative URI, two sub-parts are identified. The part before the question
257mark is called the "path". It is typically the relative path to static objects
258on the server. The part after the question mark is called the "query string".
259It is mostly used with GET requests sent to dynamic scripts and is very
260specific to the language, framework or application in use.
261
Willy Tarreau95c4e142017-11-26 12:18:55 +0100262HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100263assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100264However, haproxy natively processes HTTP/1.x requests and headers, so requests
265received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
266processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
267as well as in server logs.
268
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200269
2701.2.2. The request headers
271--------------------------
272
273The headers start at the second line. They are composed of a name at the
274beginning of the line, immediately followed by a colon (':'). Traditionally,
275an LWS is added after the colon but that's not required. Then come the values.
276Multiple identical headers may be folded into one single line, delimiting the
277values with commas, provided that their order is respected. This is commonly
278encountered in the "Cookie:" field. A header may span over multiple lines if
279the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
280define a total of 3 values for the "Accept:" header.
281
Davor Ocelice9ed2812017-12-25 17:49:28 +0100282Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200283their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100284"Connection:" header). In HTTP/2, header names are always sent in lower case,
285as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200286
287The end of the headers is indicated by the first empty line. People often say
288that it's a double line feed, which is not exact, even if a double line feed
289is one valid form of empty line.
290
291Fortunately, HAProxy takes care of all these complex combinations when indexing
292headers, checking values and counting them, so there is no reason to worry
293about the way they could be written, but it is important not to accuse an
294application of being buggy if it does unusual, valid things.
295
296Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000297 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200298 in the middle of headers by LWS in order to join multi-line headers. This
299 is necessary for proper analysis and helps less capable HTTP parsers to work
300 correctly and not to be fooled by such complex constructs.
301
302
3031.3. HTTP response
304------------------
305
306An HTTP response looks very much like an HTTP request. Both are called HTTP
307messages. Let's consider this HTTP response :
308
309 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100310 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200311 1 HTTP/1.1 200 OK
312 2 Content-length: 350
313 3 Content-Type: text/html
314
Willy Tarreau816b9792009-09-15 21:25:21 +0200315As a special case, HTTP supports so called "Informational responses" as status
316codes 1xx. These messages are special in that they don't convey any part of the
317response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100318continue to post its request for instance. In the case of a status 100 response
319the requested information will be carried by the next non-100 response message
320following the informational one. This implies that multiple responses may be
321sent to a single request, and that this only works when keep-alive is enabled
322(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
323correctly forward and skip them, and only process the next non-100 response. As
324such, these messages are neither logged nor transformed, unless explicitly
325state otherwise. Status 101 messages indicate that the protocol is changing
326over the same connection and that haproxy must switch to tunnel mode, just as
327if a CONNECT had occurred. Then the Upgrade header would contain additional
328information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200329
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200330
Davor Ocelice9ed2812017-12-25 17:49:28 +01003311.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200332------------------------
333
334Line 1 is the "response line". It is always composed of 3 fields :
335
336 - a version tag : HTTP/1.1
337 - a status code : 200
338 - a reason : OK
339
340The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341 - 1xx = informational message to be skipped (e.g. 100, 101)
342 - 2xx = OK, content is following (e.g. 200, 206)
343 - 3xx = OK, no content following (e.g. 302, 304)
344 - 4xx = error caused by the client (e.g. 401, 403, 404)
345 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200346
Lukas Tribus23953682017-04-28 13:24:30 +0000347Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100348"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200349found there, but it's a common practice to respect the well-established
350messages. It can be composed of one or multiple words, such as "OK", "Found",
351or "Authentication Required".
352
Davor Ocelice9ed2812017-12-25 17:49:28 +0100353HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200354
355 Code When / reason
356 200 access to stats page, and when replying to monitoring requests
357 301 when performing a redirection, depending on the configured code
358 302 when performing a redirection, depending on the configured code
359 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100360 307 when performing a redirection, depending on the configured code
361 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200362 400 for an invalid or too large request
363 401 when an authentication is required to perform the action (when
364 accessing the stats page)
365 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
366 408 when the request timeout strikes before the request is complete
367 500 when haproxy encounters an unrecoverable internal error, such as a
368 memory allocation failure, which should never happen
369 502 when the server returns an empty, invalid or incomplete response, or
370 when an "rspdeny" filter blocks the response.
371 503 when no server was available to handle the request, or in response to
372 monitoring requests which match the "monitor fail" condition
373 504 when the response timeout strikes before the server responds
374
375The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3764.2).
377
378
3791.3.2. The response headers
380---------------------------
381
382Response headers work exactly like request headers, and as such, HAProxy uses
383the same parsing function for both. Please refer to paragraph 1.2.2 for more
384details.
385
386
3872. Configuring HAProxy
388----------------------
389
3902.1. Configuration file format
391------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200392
393HAProxy's configuration process involves 3 major sources of parameters :
394
395 - the arguments from the command-line, which always take precedence
396 - the "global" section, which sets process-wide parameters
397 - the proxies sections which can take form of "defaults", "listen",
398 "frontend" and "backend".
399
Willy Tarreau0ba27502007-12-24 16:55:16 +0100400The configuration file syntax consists in lines beginning with a keyword
401referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200402delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100403
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200404
William Lallemandf9873ba2015-05-05 17:37:14 +02004052.2. Quoting and escaping
406-------------------------
407
408HAProxy's configuration introduces a quoting and escaping system similar to
409many programming languages. The configuration file supports 3 types: escaping
410with a backslash, weak quoting with double quotes, and strong quoting with
411single quotes.
412
413If spaces have to be entered in strings, then they must be escaped by preceding
414them by a backslash ('\') or by quoting them. Backslashes also have to be
415escaped by doubling or strong quoting them.
416
417Escaping is achieved by preceding a special character by a backslash ('\'):
418
419 \ to mark a space and differentiate it from a delimiter
420 \# to mark a hash and differentiate it from a comment
421 \\ to use a backslash
422 \' to use a single quote and differentiate it from strong quoting
423 \" to use a double quote and differentiate it from weak quoting
424
425Weak quoting is achieved by using double quotes (""). Weak quoting prevents
426the interpretation of:
427
428 space as a parameter separator
429 ' single quote as a strong quoting delimiter
430 # hash as a comment start
431
William Lallemandb2f07452015-05-12 14:27:13 +0200432Weak quoting permits the interpretation of variables, if you want to use a non
433-interpreted dollar within a double quoted string, you should escape it with a
434backslash ("\$"), it does not work outside weak quoting.
435
436Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200437quoting.
438
439Strong quoting is achieved by using single quotes (''). Inside single quotes,
440nothing is interpreted, it's the efficient way to quote regexes.
441
442Quoted and escaped strings are replaced in memory by their interpreted
443equivalent, it allows you to perform concatenation.
444
445 Example:
446 # those are equivalents:
447 log-format %{+Q}o\ %t\ %s\ %{-Q}r
448 log-format "%{+Q}o %t %s %{-Q}r"
449 log-format '%{+Q}o %t %s %{-Q}r'
450 log-format "%{+Q}o %t"' %s %{-Q}r'
451 log-format "%{+Q}o %t"' %s'\ %{-Q}r
452
453 # those are equivalents:
454 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
455 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
456 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
457 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
458
459
William Lallemandb2f07452015-05-12 14:27:13 +02004602.3. Environment variables
461--------------------------
462
463HAProxy's configuration supports environment variables. Those variables are
464interpreted only within double quotes. Variables are expanded during the
465configuration parsing. Variable names must be preceded by a dollar ("$") and
466optionally enclosed with braces ("{}") similarly to what is done in Bourne
467shell. Variable names can contain alphanumerical characters or the character
468underscore ("_") but should not start with a digit.
469
470 Example:
471
472 bind "fd@${FD_APP1}"
473
474 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
475
476 user "$HAPROXY_USER"
477
William Lallemanddaf4cd22018-04-17 16:46:13 +0200478A special variable $HAPROXY_LOCALPEER is defined at the startup of the process
479which contains the name of the local peer. (See "-L" in the management guide.)
480
William Lallemandb2f07452015-05-12 14:27:13 +0200481
4822.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200483----------------
484
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100485Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100486values are generally expressed in milliseconds (unless explicitly stated
487otherwise) but may be expressed in any other unit by suffixing the unit to the
488numeric value. It is important to consider this because it will not be repeated
489for every keyword. Supported units are :
490
491 - us : microseconds. 1 microsecond = 1/1000000 second
492 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
493 - s : seconds. 1s = 1000ms
494 - m : minutes. 1m = 60s = 60000ms
495 - h : hours. 1h = 60m = 3600s = 3600000ms
496 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
497
498
Lukas Tribusaa83a312017-03-21 09:25:09 +00004992.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200500-------------
501
502 # Simple configuration for an HTTP proxy listening on port 80 on all
503 # interfaces and forwarding requests to a single backend "servers" with a
504 # single server "server1" listening on 127.0.0.1:8000
505 global
506 daemon
507 maxconn 256
508
509 defaults
510 mode http
511 timeout connect 5000ms
512 timeout client 50000ms
513 timeout server 50000ms
514
515 frontend http-in
516 bind *:80
517 default_backend servers
518
519 backend servers
520 server server1 127.0.0.1:8000 maxconn 32
521
522
523 # The same configuration defined with a single listen block. Shorter but
524 # less expressive, especially in HTTP mode.
525 global
526 daemon
527 maxconn 256
528
529 defaults
530 mode http
531 timeout connect 5000ms
532 timeout client 50000ms
533 timeout server 50000ms
534
535 listen http-in
536 bind *:80
537 server server1 127.0.0.1:8000 maxconn 32
538
539
540Assuming haproxy is in $PATH, test these configurations in a shell with:
541
Willy Tarreauccb289d2010-12-11 20:19:38 +0100542 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200543
544
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005453. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200546--------------------
547
548Parameters in the "global" section are process-wide and often OS-specific. They
549are generally set once for all and do not need being changed once correct. Some
550of them have command-line equivalents.
551
552The following keywords are supported in the "global" section :
553
554 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200555 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200556 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200557 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200558 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200559 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200560 - description
561 - deviceatlas-json-file
562 - deviceatlas-log-level
563 - deviceatlas-separator
564 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900565 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200566 - gid
567 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100568 - hard-stop-after
Willy Tarreau6a06a402007-07-15 20:15:28 +0200569 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200570 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100571 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200572 - lua-load
Willy Tarreau6a06a402007-07-15 20:15:28 +0200573 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200574 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200575 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200576 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100577 - presetenv
578 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200579 - uid
580 - ulimit-n
581 - user
Willy Tarreau1d549722016-02-16 12:41:57 +0100582 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200583 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200584 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200585 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200586 - ssl-default-bind-options
587 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200588 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200589 - ssl-default-server-options
590 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100591 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100592 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100593 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100594 - 51degrees-data-file
595 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200596 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200597 - 51degrees-cache-size
scientiamobiled0027ed2016-11-04 10:55:08 +0100598 - wurfl-data-file
599 - wurfl-information-list
600 - wurfl-information-list-separator
601 - wurfl-engine-mode
602 - wurfl-cache-size
603 - wurfl-useragent-priority
Willy Tarreaud72758d2010-01-12 10:42:19 +0100604
Willy Tarreau6a06a402007-07-15 20:15:28 +0200605 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200606 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200607 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200608 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100609 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100610 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100611 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200612 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200613 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200614 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200615 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200616 - noepoll
617 - nokqueue
618 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100619 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300620 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000621 - noreuseport
Willy Tarreaufe255b72007-10-14 23:09:26 +0200622 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200623 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200624 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000625 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000626 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200627 - tune.buffers.limit
628 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200629 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200630 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100631 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200632 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200633 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200634 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100635 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200636 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200637 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100638 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100639 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100640 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100641 - tune.lua.session-timeout
642 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200643 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100644 - tune.maxaccept
645 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200646 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200647 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200648 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100649 - tune.rcvbuf.client
650 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100651 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200652 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100653 - tune.sndbuf.client
654 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100655 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100656 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200657 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100658 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200659 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200660 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100661 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200662 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100663 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200664 - tune.vars.reqres-max-size
665 - tune.vars.sess-max-size
666 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100667 - tune.zlib.memlevel
668 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100669
Willy Tarreau6a06a402007-07-15 20:15:28 +0200670 * Debugging
671 - debug
672 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200673
674
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006753.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200676------------------------------------
677
Emeric Brunc8e8d122012-10-02 18:42:10 +0200678ca-base <dir>
679 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200680 relative path is used with "ca-file" or "crl-file" directives. Absolute
681 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200682
Willy Tarreau6a06a402007-07-15 20:15:28 +0200683chroot <jail dir>
684 Changes current directory to <jail dir> and performs a chroot() there before
685 dropping privileges. This increases the security level in case an unknown
686 vulnerability would be exploited, since it would make it very hard for the
687 attacker to exploit the system. This only works when the process is started
688 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100689 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100690
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100691cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
692 On Linux 2.6 and above, it is possible to bind a process or a thread to a
693 specific CPU set. This means that the process or the thread will never run on
694 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
695 sets. The first argument is a process set, eventually followed by a thread
696 set. These sets have the format
697
698 all | odd | even | number[-[number]]
699
700 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100701 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100702 ignored. It is possible to specify a range with two such number delimited by
703 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100704 "all", only odd numbers using "odd" or even numbers using "even", just like
705 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100706 CPU sets. Each CPU set is either a unique number between 0 and 31 or 63 or a
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100707 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100708 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100709 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100710 specified. Each "cpu-map" directive will replace the previous ones when they
711 overlap. A thread will be bound on the intersection of its mapping and the
712 one of the process on which it is attached. If the intersection is null, no
713 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100714
Christopher Fauletff4121f2017-11-22 16:38:49 +0100715 Ranges can be partially defined. The higher bound can be omitted. In such
716 case, it is replaced by the corresponding maximum value, 32 or 64 depending
717 on the machine's word size.
718
Christopher Faulet26028f62017-11-22 15:01:51 +0100719 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100720 automatically bind a process or a thread to a CPU by incrementing
721 process/thread and CPU sets. To be valid, both sets must have the same
722 size. No matter the declaration order of the CPU sets, it will be bound from
723 the lowest to the highest bound. Having a process and a thread range with the
724 "auto:" prefix is not supported. Only one range is supported, the other one
725 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100726
727 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100728 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
729
730 cpu-map 1/all 0-3 # bind all threads of the first process on the
731 # first 4 CPUs
732
733 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
734 # or "cpu-map 1-32 0-31" depending on the machine's
735 # word size.
736
Christopher Faulet26028f62017-11-22 15:01:51 +0100737 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100738 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100739 cpu-map auto:1-4 0-3
740 cpu-map auto:1-4 0-1 2-3
741 cpu-map auto:1-4 3 2 1 0
742
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100743 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
744 # and so on.
745 cpu-map auto:1/1-4 0-3
746 cpu-map auto:1/1-4 0-1 2-3
747 cpu-map auto:1/1-4 3 2 1 0
748
Davor Ocelice9ed2812017-12-25 17:49:28 +0100749 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100750 cpu-map auto:all 0-63
751 cpu-map auto:even 0-31
752 cpu-map auto:odd 32-63
753
754 # invalid cpu-map because process and CPU sets have different sizes.
755 cpu-map auto:1-4 0 # invalid
756 cpu-map auto:1 0-3 # invalid
757
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100758 # invalid cpu-map because automatic binding is used with a process range
759 # and a thread range.
760 cpu-map auto:all/all 0 # invalid
761 cpu-map auto:all/1-4 0 # invalid
762 cpu-map auto:1-4/all 0 # invalid
763
Emeric Brunc8e8d122012-10-02 18:42:10 +0200764crt-base <dir>
765 Assigns a default directory to fetch SSL certificates from when a relative
766 path is used with "crtfile" directives. Absolute locations specified after
767 "crtfile" prevail and ignore "crt-base".
768
Willy Tarreau6a06a402007-07-15 20:15:28 +0200769daemon
770 Makes the process fork into background. This is the recommended mode of
771 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100772 disabled by the command line "-db" argument. This option is ignored in
773 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200774
David Carlier8167f302015-06-01 13:50:06 +0200775deviceatlas-json-file <path>
776 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100777 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200778
779deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100780 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200781 optional and set to 0 by default if not set.
782
783deviceatlas-separator <char>
784 Sets the character separator for the API properties results. This directive
785 is optional and set to | by default if not set.
786
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100787deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200788 Sets the client cookie's name used for the detection if the DeviceAtlas
789 Client-side component was used during the request. This directive is optional
790 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100791
Simon Horman98637e52014-06-20 12:30:16 +0900792external-check
793 Allows the use of an external agent to perform health checks.
794 This is disabled by default as a security precaution.
795 See "option external-check".
796
Willy Tarreau6a06a402007-07-15 20:15:28 +0200797gid <number>
798 Changes the process' group ID to <number>. It is recommended that the group
799 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
800 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100801 Note that if haproxy is started from a user having supplementary groups, it
802 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200803 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100804
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100805hard-stop-after <time>
806 Defines the maximum time allowed to perform a clean soft-stop.
807
808 Arguments :
809 <time> is the maximum time (by default in milliseconds) for which the
810 instance will remain alive when a soft-stop is received via the
811 SIGUSR1 signal.
812
813 This may be used to ensure that the instance will quit even if connections
814 remain opened during a soft-stop (for example with long timeouts for a proxy
815 in tcp mode). It applies both in TCP and HTTP mode.
816
817 Example:
818 global
819 hard-stop-after 30s
820
Willy Tarreau6a06a402007-07-15 20:15:28 +0200821group <group name>
822 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
823 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100824
Dragan Dosen7ad31542015-09-28 17:16:47 +0200825log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100826 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100827 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100828 configured with "log global".
829
830 <address> can be one of:
831
Willy Tarreau2769aa02007-12-27 18:26:09 +0100832 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100833 no port is specified, 514 is used by default (the standard syslog
834 port).
835
David du Colombier24bb5f52011-03-17 10:40:23 +0100836 - An IPv6 address followed by a colon and optionally a UDP port. If
837 no port is specified, 514 is used by default (the standard syslog
838 port).
839
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100840 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100841 considerations for chroot (be sure the path is accessible inside
842 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100843 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100844
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100845 - A file descriptor number in the form "fd@<number>", which may point
846 to a pipe, terminal, or socket. In this case unbuffered logs are used
847 and one writev() call per log is performed. This is a bit expensive
848 but acceptable for most workloads. Messages sent this way will not be
849 truncated but may be dropped, in which case the DroppedLogs counter
850 will be incremented. The writev() call is atomic even on pipes for
851 messages up to PIPE_BUF size, which POSIX recommends to be at least
852 512 and which is 4096 bytes on most modern operating systems. Any
853 larger message may be interleaved with messages from other processes.
854 Exceptionally for debugging purposes the file descriptor may also be
855 directed to a file, but doing so will significantly slow haproxy down
856 as non-blocking calls will be ignored. Also there will be no way to
857 purge nor rotate this file without restarting the process. Note that
858 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100859 for use with a TCP syslog server. See also the "short" and "raw"
860 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100861
862 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
863 "fd@2", see above.
864
William Lallemandb2f07452015-05-12 14:27:13 +0200865 You may want to reference some environment variables in the address
866 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100867
Willy Tarreau18324f52014-06-27 18:10:07 +0200868 <length> is an optional maximum line length. Log lines larger than this value
869 will be truncated before being sent. The reason is that syslog
870 servers act differently on log line length. All servers support the
871 default value of 1024, but some servers simply drop larger lines
872 while others do log them. If a server supports long lines, it may
873 make sense to set this value here in order to avoid truncating long
874 lines. Similarly, if a server drops long lines, it is preferable to
875 truncate them before sending them. Accepted values are 80 to 65535
876 inclusive. The default value of 1024 is generally fine for all
877 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100878 JSON-formatted logs may require larger values. You may also need to
879 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200880
Dragan Dosen7ad31542015-09-28 17:16:47 +0200881 <format> is the log format used when generating syslog messages. It may be
882 one of the following :
883
884 rfc3164 The RFC3164 syslog message format. This is the default.
885 (https://tools.ietf.org/html/rfc3164)
886
887 rfc5424 The RFC5424 syslog message format.
888 (https://tools.ietf.org/html/rfc5424)
889
Willy Tarreaue8746a02018-11-12 08:45:00 +0100890 short A message containing only a level between angle brackets such as
891 '<3>', followed by the text. The PID, date, time, process name
892 and system name are omitted. This is designed to be used with a
893 local log server. This format is compatible with what the systemd
894 logger consumes.
895
Willy Tarreauc1b06452018-11-12 11:57:56 +0100896 raw A message containing only the text. The level, PID, date, time,
897 process name and system name are omitted. This is designed to be
898 used in containers or during development, where the severity only
899 depends on the file descriptor used (stdout/stderr).
900
Robert Tsai81ae1952007-12-05 10:47:29 +0100901 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200902
Willy Tarreaue8746a02018-11-12 08:45:00 +0100903 kern user mail daemon auth syslog lpr news
904 uucp cron auth2 ftp ntp audit alert cron2
905 local0 local1 local2 local3 local4 local5 local6 local7
906
Willy Tarreauc1b06452018-11-12 11:57:56 +0100907 Note that the facility is ignored for the "short" and "raw"
908 formats, but still required as a positional field. It is
909 recommended to use "daemon" in this case to make it clear that
910 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200911
912 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200913 all messages are sent. If a maximum level is specified, only messages with a
914 severity at least as important as this level will be sent. An optional minimum
915 level can be specified. If it is set, logs emitted with a more severe level
916 than this one will be capped to this level. This is used to avoid sending
917 "emerg" messages on all terminals on some default syslog configurations.
918 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200919
Cyril Bontédc4d9032012-04-08 21:57:39 +0200920 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200921
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100922log-send-hostname [<string>]
923 Sets the hostname field in the syslog header. If optional "string" parameter
924 is set the header is set to the string contents, otherwise uses the hostname
925 of the system. Generally used if one is not relaying logs through an
926 intermediate syslog server or for simply customizing the hostname printed in
927 the logs.
928
Kevinm48936af2010-12-22 16:08:21 +0000929log-tag <string>
930 Sets the tag field in the syslog header to this string. It defaults to the
931 program name as launched from the command line, which usually is "haproxy".
932 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100933 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000934
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100935lua-load <file>
936 This global directive loads and executes a Lua file. This directive can be
937 used multiple times.
938
William Lallemand4cfede82017-11-24 22:02:34 +0100939master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200940 Master-worker mode. It is equivalent to the command line "-W" argument.
941 This mode will launch a "master" which will monitor the "workers". Using
942 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100943 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200944 or daemon mode. It is recommended to use this mode with multiprocess and
945 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100946 By default, if a worker exits with a bad return code, in the case of a
947 segfault for example, all workers will be killed, and the master will leave.
948 It is convenient to combine this behavior with Restart=on-failure in a
949 systemd unit file in order to relaunch the whole process. If you don't want
950 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200951
William Lallemand4cfede82017-11-24 22:02:34 +0100952 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200953
Willy Tarreau6a06a402007-07-15 20:15:28 +0200954nbproc <number>
955 Creates <number> processes when going daemon. This requires the "daemon"
956 mode. By default, only one process is created, which is the recommended mode
957 of operation. For systems limited to small sets of file descriptors per
958 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
959 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
960
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200961nbthread <number>
962 This setting is only available when support for threads was built in. It
963 creates <number> threads for each created processes. It means if HAProxy is
964 started in foreground, it only creates <number> threads for the first
965 process. FOR NOW, THREADS SUPPORT IN HAPROXY IS HIGHLY EXPERIMENTAL AND IT
966 MUST BE ENABLED WITH CAUTION AND AT YOUR OWN RISK. See also "nbproc".
967
Willy Tarreau6a06a402007-07-15 20:15:28 +0200968pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100969 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200970 the "-p" command line argument. The file must be accessible to the user
971 starting the process. See also "daemon".
972
Willy Tarreau1d549722016-02-16 12:41:57 +0100973presetenv <name> <value>
974 Sets environment variable <name> to value <value>. If the variable exists, it
975 is NOT overwritten. The changes immediately take effect so that the next line
976 in the configuration file sees the new value. See also "setenv", "resetenv",
977 and "unsetenv".
978
979resetenv [<name> ...]
980 Removes all environment variables except the ones specified in argument. It
981 allows to use a clean controlled environment before setting new values with
982 setenv or unsetenv. Please note that some internal functions may make use of
983 some environment variables, such as time manipulation functions, but also
984 OpenSSL or even external checks. This must be used with extreme care and only
985 after complete validation. The changes immediately take effect so that the
986 next line in the configuration file sees the new environment. See also
987 "setenv", "presetenv", and "unsetenv".
988
Christopher Fauletff4121f2017-11-22 16:38:49 +0100989stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200990 Limits the stats socket to a certain set of processes numbers. By default the
991 stats socket is bound to all processes, causing a warning to be emitted when
992 nbproc is greater than 1 because there is no way to select the target process
993 when connecting. However, by using this setting, it becomes possible to pin
994 the stats socket to a specific set of processes, typically the first one. The
995 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100996 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +0100997 word size (32 or 64). Ranges can be partially defined. The higher bound can
998 be omitted. In such case, it is replaced by the corresponding maximum
999 value. A better option consists in using the "process" setting of the "stats
1000 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001001
Baptiste Assmann5626f482015-08-23 10:00:10 +02001002server-state-base <directory>
1003 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001004 file names which do not start with a '/'. See also "server-state-file",
1005 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001006
1007server-state-file <file>
1008 Specifies the path to the file containing state of servers. If the path starts
1009 with a slash ('/'), it is considered absolute, otherwise it is considered
1010 relative to the directory specified using "server-state-base" (if set) or to
1011 the current directory. Before reloading HAProxy, it is possible to save the
1012 servers' current state using the stats command "show servers state". The
1013 output of this command must be written in the file pointed by <file>. When
1014 starting up, before handling traffic, HAProxy will read, load and apply state
1015 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001016 configuration. See also "server-state-base" and "show servers state",
1017 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001018
Willy Tarreau1d549722016-02-16 12:41:57 +01001019setenv <name> <value>
1020 Sets environment variable <name> to value <value>. If the variable exists, it
1021 is overwritten. The changes immediately take effect so that the next line in
1022 the configuration file sees the new value. See also "presetenv", "resetenv",
1023 and "unsetenv".
1024
Willy Tarreau610f04b2014-02-13 11:36:41 +01001025ssl-default-bind-ciphers <ciphers>
1026 This setting is only available when support for OpenSSL was built in. It sets
1027 the default string describing the list of cipher algorithms ("cipher suite")
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001028 that are negotiated during the SSL/TLS handshake except for TLSv1.3 for all
1029 "bind" lines which do not explicitly define theirs. The format of the string
1030 is defined in "man 1 ciphers" from OpenSSL man pages, and can be for instance
1031 a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). For
1032 TLSv1.3 cipher configuration, please check the "ssl-default-bind-ciphersuites"
1033 keyword. Please check the "bind" keyword for more information.
1034
1035ssl-default-bind-ciphersuites <ciphersuites>
1036 This setting is only available when support for OpenSSL was built in and
1037 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1038 describing the list of cipher algorithms ("cipher suite") that are negotiated
1039 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1040 theirs. The format of the string is defined in
1041 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites", and can
1042 be for instance a string such as
1043 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
1044 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
1045 the "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
1046 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001047
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001048ssl-default-bind-options [<option>]...
1049 This setting is only available when support for OpenSSL was built in. It sets
1050 default ssl-options to force on all "bind" lines. Please check the "bind"
1051 keyword to see available options.
1052
1053 Example:
1054 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001055 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001056
Willy Tarreau610f04b2014-02-13 11:36:41 +01001057ssl-default-server-ciphers <ciphers>
1058 This setting is only available when support for OpenSSL was built in. It
1059 sets the default string describing the list of cipher algorithms that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001060 negotiated during the SSL/TLS handshake except for TLSv1.3 with the server,
1061 for all "server" lines which do not explicitly define theirs. The format of
1062 the string is defined in "man 1 ciphers". For TLSv1.3 cipher configuration,
1063 please check the "ssl-default-server-ciphersuites" keyword. Please check the
1064 "server" keyword for more information.
1065
1066ssl-default-server-ciphersuites <ciphersuites>
1067 This setting is only available when support for OpenSSL was built in and
1068 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1069 string describing the list of cipher algorithms that are negotiated during
1070 the TLSv1.3 handshake with the server, for all "server" lines which do not
1071 explicitly define theirs. The format of the string is defined in
1072 "man 1 ciphers" under the "ciphersuites" section. For cipher configuration for
1073 TLSv1.2 and earlier, please check the "ssl-default-server-ciphers" keyword.
1074 Please check the "server" keyword for more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001075
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001076ssl-default-server-options [<option>]...
1077 This setting is only available when support for OpenSSL was built in. It sets
1078 default ssl-options to force on all "server" lines. Please check the "server"
1079 keyword to see available options.
1080
Remi Gacogne47783ef2015-05-29 15:53:22 +02001081ssl-dh-param-file <file>
1082 This setting is only available when support for OpenSSL was built in. It sets
1083 the default DH parameters that are used during the SSL/TLS handshake when
1084 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001085 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001086 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001087 are not specified either by using ssl-dh-param-file or by setting them
1088 directly in the certificate file, pre-generated DH parameters of the size
1089 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1090 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001091 Custom DH parameters may be generated by using the OpenSSL command
1092 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1093 parameters should not be considered secure anymore.
1094
Emeric Brun850efd52014-01-29 12:24:34 +01001095ssl-server-verify [none|required]
1096 The default behavior for SSL verify on servers side. If specified to 'none',
1097 servers certificates are not verified. The default is 'required' except if
1098 forced using cmdline option '-dV'.
1099
Willy Tarreauabb175f2012-09-24 12:43:26 +02001100stats socket [<address:port>|<path>] [param*]
1101 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1102 Connections to this socket will return various statistics outputs and even
1103 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001104 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001105 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001106
Willy Tarreauabb175f2012-09-24 12:43:26 +02001107 All parameters supported by "bind" lines are supported, for instance to
1108 restrict access to some users or their access rights. Please consult
1109 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001110
1111stats timeout <timeout, in milliseconds>
1112 The default timeout on the stats socket is set to 10 seconds. It is possible
1113 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001114 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001115
1116stats maxconn <connections>
1117 By default, the stats socket is limited to 10 concurrent connections. It is
1118 possible to change this value with "stats maxconn".
1119
Willy Tarreau6a06a402007-07-15 20:15:28 +02001120uid <number>
1121 Changes the process' user ID to <number>. It is recommended that the user ID
1122 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1123 be started with superuser privileges in order to be able to switch to another
1124 one. See also "gid" and "user".
1125
1126ulimit-n <number>
1127 Sets the maximum number of per-process file-descriptors to <number>. By
1128 default, it is automatically computed, so it is recommended not to use this
1129 option.
1130
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001131unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1132 [ group <group> ] [ gid <gid> ]
1133
1134 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1135 This is mainly used to simplify declaration of those UNIX sockets and reduce
1136 the risk of errors, since those settings are most commonly required but are
1137 also process-specific. The <prefix> setting can be used to force all socket
1138 path to be relative to that directory. This might be needed to access another
1139 component's chroot. Note that those paths are resolved before haproxy chroots
1140 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1141 all have the same meaning as their homonyms used by the "bind" statement. If
1142 both are specified, the "bind" statement has priority, meaning that the
1143 "unix-bind" settings may be seen as process-wide default settings.
1144
Willy Tarreau1d549722016-02-16 12:41:57 +01001145unsetenv [<name> ...]
1146 Removes environment variables specified in arguments. This can be useful to
1147 hide some sensitive information that are occasionally inherited from the
1148 user's environment during some operations. Variables which did not exist are
1149 silently ignored so that after the operation, it is certain that none of
1150 these variables remain. The changes immediately take effect so that the next
1151 line in the configuration file will not see these variables. See also
1152 "setenv", "presetenv", and "resetenv".
1153
Willy Tarreau6a06a402007-07-15 20:15:28 +02001154user <user name>
1155 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1156 See also "uid" and "group".
1157
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001158node <name>
1159 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1160
1161 This statement is useful in HA configurations where two or more processes or
1162 servers share the same IP address. By setting a different node-name on all
1163 nodes, it becomes easy to immediately spot what server is handling the
1164 traffic.
1165
1166description <text>
1167 Add a text that describes the instance.
1168
1169 Please note that it is required to escape certain characters (# for example)
1170 and this text is inserted into a html page so you should avoid using
1171 "<" and ">" characters.
1172
Thomas Holmesdb04f192015-05-18 13:21:39 +0100117351degrees-data-file <file path>
1174 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001175 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001176
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001177 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001178 compiled with USE_51DEGREES.
1179
Ben Shillitof25e8e52016-12-02 14:25:37 +0000118051degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001181 A list of 51Degrees property names to be load from the dataset. A full list
1182 of names is available on the 51Degrees website:
1183 https://51degrees.com/resources/property-dictionary
1184
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001185 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001186 compiled with USE_51DEGREES.
1187
Dragan Dosen93b38d92015-06-29 16:43:25 +0200118851degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001189 A char that will be appended to every property value in a response header
1190 containing 51Degrees results. If not set that will be set as ','.
1191
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001192 Please note that this option is only available when haproxy has been
1193 compiled with USE_51DEGREES.
1194
119551degrees-cache-size <number>
1196 Sets the size of the 51Degrees converter cache to <number> entries. This
1197 is an LRU cache which reminds previous device detections and their results.
1198 By default, this cache is disabled.
1199
1200 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001201 compiled with USE_51DEGREES.
1202
scientiamobiled0027ed2016-11-04 10:55:08 +01001203wurfl-data-file <file path>
1204 The path of the WURFL data file to provide device detection services. The
1205 file should be accessible by HAProxy with relevant permissions.
1206
1207 Please note that this option is only available when haproxy has been compiled
1208 with USE_WURFL=1.
1209
1210wurfl-information-list [<capability>]*
1211 A space-delimited list of WURFL capabilities, virtual capabilities, property
1212 names we plan to use in injected headers. A full list of capability and
1213 virtual capability names is available on the Scientiamobile website :
1214
1215 https://www.scientiamobile.com/wurflCapability
1216
1217 Valid WURFL properties are:
1218 - wurfl_id Contains the device ID of the matched device.
1219
1220 - wurfl_root_id Contains the device root ID of the matched
1221 device.
1222
1223 - wurfl_isdevroot Tells if the matched device is a root device.
1224 Possible values are "TRUE" or "FALSE".
1225
1226 - wurfl_useragent The original useragent coming with this
1227 particular web request.
1228
1229 - wurfl_api_version Contains a string representing the currently
1230 used Libwurfl API version.
1231
1232 - wurfl_engine_target Contains a string representing the currently
1233 set WURFL Engine Target. Possible values are
1234 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1235
1236 - wurfl_info A string containing information on the parsed
1237 wurfl.xml and its full path.
1238
1239 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1240 WURFL has been loaded successfully.
1241
1242 - wurfl_normalized_useragent The normalized useragent.
1243
1244 - wurfl_useragent_priority The user agent priority used by WURFL.
1245
1246 Please note that this option is only available when haproxy has been compiled
1247 with USE_WURFL=1.
1248
1249wurfl-information-list-separator <char>
1250 A char that will be used to separate values in a response header containing
1251 WURFL results. If not set that a comma (',') will be used by default.
1252
1253 Please note that this option is only available when haproxy has been compiled
1254 with USE_WURFL=1.
1255
1256wurfl-patch-file [<file path>]
1257 A list of WURFL patch file paths. Note that patches are loaded during startup
1258 thus before the chroot.
1259
1260 Please note that this option is only available when haproxy has been compiled
1261 with USE_WURFL=1.
1262
1263wurfl-engine-mode { accuracy | performance }
1264 Sets the WURFL engine target. You can choose between 'accuracy' or
1265 'performance' targets. In performance mode, desktop web browser detection is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001266 done programmatically without referencing the WURFL data. As a result, most
scientiamobiled0027ed2016-11-04 10:55:08 +01001267 desktop web browsers are returned as generic_web_browser WURFL ID for
Davor Ocelice9ed2812017-12-25 17:49:28 +01001268 performance. If either performance or accuracy are not defined, performance
scientiamobiled0027ed2016-11-04 10:55:08 +01001269 mode is enabled by default.
1270
1271 Please note that this option is only available when haproxy has been compiled
1272 with USE_WURFL=1.
1273
1274wurfl-cache-size <U>[,<D>]
1275 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1276 <D> is the internal device cache size. There are three possibilities here :
1277 - "0" : no cache is used.
1278 - <U> : the Single LRU cache is used, the size is expressed in elements.
1279 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1280 the highest performing option.
1281
1282 Please note that this option is only available when haproxy has been compiled
1283 with USE_WURFL=1.
1284
1285wurfl-useragent-priority { plain | sideloaded_browser }
1286 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1287 over the default sideloaded browser user agent ('sideloaded_browser').
1288
1289 Please note that this option is only available when haproxy has been compiled
1290 with USE_WURFL=1.
1291
Willy Tarreau6a06a402007-07-15 20:15:28 +02001292
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012933.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001294-----------------------
1295
Willy Tarreau1746eec2014-04-25 10:46:47 +02001296max-spread-checks <delay in milliseconds>
1297 By default, haproxy tries to spread the start of health checks across the
1298 smallest health check interval of all the servers in a farm. The principle is
1299 to avoid hammering services running on the same server. But when using large
1300 check intervals (10 seconds or more), the last servers in the farm take some
1301 time before starting to be tested, which can be a problem. This parameter is
1302 used to enforce an upper bound on delay between the first and the last check,
1303 even if the servers' check intervals are larger. When servers run with
1304 shorter intervals, their intervals will be respected though.
1305
Willy Tarreau6a06a402007-07-15 20:15:28 +02001306maxconn <number>
1307 Sets the maximum per-process number of concurrent connections to <number>. It
1308 is equivalent to the command-line argument "-n". Proxies will stop accepting
1309 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001310 automatically adjusted according to this value. See also "ulimit-n". Note:
1311 the "select" poller cannot reliably use more than 1024 file descriptors on
1312 some platforms. If your platform only supports select and reports "select
1313 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001314 below 500 in general). If this value is not set, it will default to the value
1315 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1316 limit is enforced, or will be computed based on the memory limit, the buffer
1317 size, memory allocated to compression, SSL cache size, and use or not of SSL
1318 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001319
Willy Tarreau81c25d02011-09-07 15:17:21 +02001320maxconnrate <number>
1321 Sets the maximum per-process number of connections per second to <number>.
1322 Proxies will stop accepting connections when this limit is reached. It can be
1323 used to limit the global capacity regardless of each frontend capacity. It is
1324 important to note that this can only be used as a service protection measure,
1325 as there will not necessarily be a fair share between frontends when the
1326 limit is reached, so it's a good idea to also limit each frontend to some
1327 value close to its expected share. Also, lowering tune.maxaccept can improve
1328 fairness.
1329
William Lallemandd85f9172012-11-09 17:05:39 +01001330maxcomprate <number>
1331 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001332 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001333 level will be decreased during the session. If the maximum is reached at the
1334 beginning of a session, the session will not compress at all. If the maximum
1335 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001336 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001337 default value.
1338
William Lallemand072a2bf2012-11-20 17:01:01 +01001339maxcompcpuusage <number>
1340 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1341 for new requests or decreasing the compression level of current requests.
1342 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1343 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1344 case of multiple processes (nbproc > 1), each process manages its individual
1345 usage. A value of 100 disable the limit. The default value is 100. Setting
1346 a lower value will prevent the compression work from slowing the whole
1347 process down and from introducing high latencies.
1348
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001349maxpipes <number>
1350 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1351 are only used by kernel-based tcp splicing. Since a pipe contains two file
1352 descriptors, the "ulimit-n" value will be increased accordingly. The default
1353 value is maxconn/4, which seems to be more than enough for most heavy usages.
1354 The splice code dynamically allocates and releases pipes, and can fall back
1355 to standard copy, so setting this value too low may only impact performance.
1356
Willy Tarreau93e7c002013-10-07 18:51:07 +02001357maxsessrate <number>
1358 Sets the maximum per-process number of sessions per second to <number>.
1359 Proxies will stop accepting connections when this limit is reached. It can be
1360 used to limit the global capacity regardless of each frontend capacity. It is
1361 important to note that this can only be used as a service protection measure,
1362 as there will not necessarily be a fair share between frontends when the
1363 limit is reached, so it's a good idea to also limit each frontend to some
1364 value close to its expected share. Also, lowering tune.maxaccept can improve
1365 fairness.
1366
Willy Tarreau403edff2012-09-06 11:58:37 +02001367maxsslconn <number>
1368 Sets the maximum per-process number of concurrent SSL connections to
1369 <number>. By default there is no SSL-specific limit, which means that the
1370 global maxconn setting will apply to all connections. Setting this limit
1371 avoids having openssl use too much memory and crash when malloc returns NULL
1372 (since it unfortunately does not reliably check for such conditions). Note
1373 that the limit applies both to incoming and outgoing connections, so one
1374 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001375 If this value is not set, but a memory limit is enforced, this value will be
1376 automatically computed based on the memory limit, maxconn, the buffer size,
1377 memory allocated to compression, SSL cache size, and use of SSL in either
1378 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1379 when there is a memory limit, haproxy will automatically adjust these values
1380 so that 100% of the connections can be made over SSL with no risk, and will
1381 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001382
Willy Tarreaue43d5322013-10-07 20:01:52 +02001383maxsslrate <number>
1384 Sets the maximum per-process number of SSL sessions per second to <number>.
1385 SSL listeners will stop accepting connections when this limit is reached. It
1386 can be used to limit the global SSL CPU usage regardless of each frontend
1387 capacity. It is important to note that this can only be used as a service
1388 protection measure, as there will not necessarily be a fair share between
1389 frontends when the limit is reached, so it's a good idea to also limit each
1390 frontend to some value close to its expected share. It is also important to
1391 note that the sessions are accounted before they enter the SSL stack and not
1392 after, which also protects the stack against bad handshakes. Also, lowering
1393 tune.maxaccept can improve fairness.
1394
William Lallemand9d5f5482012-11-07 16:12:57 +01001395maxzlibmem <number>
1396 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1397 When the maximum amount is reached, future sessions will not compress as long
1398 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001399 The default value is 0. The value is available in bytes on the UNIX socket
1400 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1401 "ZlibMemUsage" in bytes.
1402
Willy Tarreau6a06a402007-07-15 20:15:28 +02001403noepoll
1404 Disables the use of the "epoll" event polling system on Linux. It is
1405 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001406 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001407
1408nokqueue
1409 Disables the use of the "kqueue" event polling system on BSD. It is
1410 equivalent to the command-line argument "-dk". The next polling system
1411 used will generally be "poll". See also "nopoll".
1412
1413nopoll
1414 Disables the use of the "poll" event polling system. It is equivalent to the
1415 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001416 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001417 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001418
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001419nosplice
1420 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001421 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001422 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001423 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001424 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1425 be used. This option makes it easier to globally disable kernel splicing in
1426 case of doubt. See also "option splice-auto", "option splice-request" and
1427 "option splice-response".
1428
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001429nogetaddrinfo
1430 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1431 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1432
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001433noreuseport
1434 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1435 command line argument "-dR".
1436
Willy Tarreaufe255b72007-10-14 23:09:26 +02001437spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001438 Sometimes it is desirable to avoid sending agent and health checks to
1439 servers at exact intervals, for instance when many logical servers are
1440 located on the same physical server. With the help of this parameter, it
1441 becomes possible to add some randomness in the check interval between 0
1442 and +/- 50%. A value between 2 and 5 seems to show good results. The
1443 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001444
Davor Ocelice9ed2812017-12-25 17:49:28 +01001445ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001446 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001447 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001448 multiple times, it will simply enable multiple crypto engines. Referencing an
1449 unsupported engine will prevent haproxy from starting. Note that many engines
1450 will lead to lower HTTPS performance than pure software with recent
1451 processors. The optional command "algo" sets the default algorithms an ENGINE
1452 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001453 of "ALL" uses the engine for all cryptographic operations. If no list of
1454 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001455 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1456 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1457 openssl configuration file uses:
1458 https://www.openssl.org/docs/man1.0.2/apps/config.html
1459
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001460ssl-mode-async
1461 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001462 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001463 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1464 doesn't support moving read/write buffers and is not compliant with
1465 haproxy's buffer management. So the asynchronous mode is disabled on
1466 read/write operations (it is only enabled during initial and reneg
1467 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001468
Willy Tarreau33cb0652014-12-23 22:52:37 +01001469tune.buffers.limit <number>
1470 Sets a hard limit on the number of buffers which may be allocated per process.
1471 The default value is zero which means unlimited. The minimum non-zero value
1472 will always be greater than "tune.buffers.reserve" and should ideally always
1473 be about twice as large. Forcing this value can be particularly useful to
1474 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001475 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001476 another one to be released by another session. Since buffers are dynamically
1477 allocated and released, the waiting time is very short and not perceptible
1478 provided that limits remain reasonable. In fact sometimes reducing the limit
1479 may even increase performance by increasing the CPU cache's efficiency. Tests
1480 have shown good results on average HTTP traffic with a limit to 1/10 of the
1481 expected global maxconn setting, which also significantly reduces memory
1482 usage. The memory savings come from the fact that a number of connections
1483 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1484 advised to do so by an haproxy core developer.
1485
Willy Tarreau1058ae72014-12-23 22:40:40 +01001486tune.buffers.reserve <number>
1487 Sets the number of buffers which are pre-allocated and reserved for use only
1488 during memory shortage conditions resulting in failed memory allocations. The
1489 minimum value is 2 and is also the default. There is no reason a user would
1490 want to change this value, it's mostly aimed at haproxy core developers.
1491
Willy Tarreau27a674e2009-08-17 07:23:33 +02001492tune.bufsize <number>
1493 Sets the buffer size to this size (in bytes). Lower values allow more
1494 sessions to coexist in the same amount of RAM, and higher values allow some
1495 applications with very large cookies to work. The default value is 16384 and
1496 can be changed at build time. It is strongly recommended not to change this
1497 from the default value, as very low values will break some services such as
1498 statistics, and values larger than default size will increase memory usage,
1499 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001500 parameter should be decreased by the same factor as this one is increased. In
1501 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1502 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001503 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
1504 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +02001505
Willy Tarreau43961d52010-10-04 20:39:20 +02001506tune.chksize <number>
1507 Sets the check buffer size to this size (in bytes). Higher values may help
1508 find string or regex patterns in very large pages, though doing so may imply
1509 more memory and CPU usage. The default value is 16384 and can be changed at
1510 build time. It is not recommended to change this value, but to use better
1511 checks whenever possible.
1512
William Lallemandf3747832012-11-09 12:33:10 +01001513tune.comp.maxlevel <number>
1514 Sets the maximum compression level. The compression level affects CPU
1515 usage during compression. This value affects CPU usage during compression.
1516 Each session using compression initializes the compression algorithm with
1517 this value. The default value is 1.
1518
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001519tune.h2.header-table-size <number>
1520 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1521 cannot be larger than 65536 bytes. A larger value may help certain clients
1522 send more compact requests, depending on their capabilities. This amount of
1523 memory is consumed for each HTTP/2 connection. It is recommended not to
1524 change it.
1525
Willy Tarreaue6baec02017-07-27 11:45:11 +02001526tune.h2.initial-window-size <number>
1527 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001528 can upload before waiting for an acknowledgment from haproxy. This setting
1529 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001530 The default value is 65535, which roughly allows up to 5 Mbps of upload
1531 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1532 over a 1-ms local network. It can make sense to increase this value to allow
1533 faster uploads, or to reduce it to increase fairness when dealing with many
1534 clients. It doesn't affect resource usage.
1535
Willy Tarreau5242ef82017-07-27 11:47:28 +02001536tune.h2.max-concurrent-streams <number>
1537 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1538 number of outstanding requests on a single connection). The default value is
1539 100. A larger one may slightly improve page load time for complex sites when
1540 visited over high latency networks, but increases the amount of resources a
1541 single client may allocate. A value of zero disables the limit so a single
1542 client may create as many streams as allocatable by haproxy. It is highly
1543 recommended not to change this value.
1544
Willy Tarreau193b8c62012-11-22 00:17:38 +01001545tune.http.cookielen <number>
1546 Sets the maximum length of captured cookies. This is the maximum value that
1547 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1548 will automatically be truncated to this one. It is important not to set too
1549 high a value because all cookie captures still allocate this size whatever
1550 their configured value (they share a same pool). This value is per request
1551 per response, so the memory allocated is twice this value per connection.
1552 When not specified, the limit is set to 63 characters. It is recommended not
1553 to change this value.
1554
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001555tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001556 Sets the maximum length of request URI in logs. This prevents truncating long
1557 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001558 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001559 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001560 configuration directives too.
1561 The default value is 1024.
1562
Willy Tarreauac1932d2011-10-24 19:14:41 +02001563tune.http.maxhdr <number>
1564 Sets the maximum number of headers in a request. When a request comes with a
1565 number of headers greater than this value (including the first line), it is
1566 rejected with a "400 Bad Request" status code. Similarly, too large responses
1567 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1568 for all usages, considering that the widely deployed Apache server uses the
1569 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001570 a buggy application to work by the time it gets fixed. The accepted range is
1571 1..32767. Keep in mind that each new header consumes 32bits of memory for
1572 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001573
Willy Tarreau7e312732014-02-12 16:35:14 +01001574tune.idletimer <timeout>
1575 Sets the duration after which haproxy will consider that an empty buffer is
1576 probably associated with an idle stream. This is used to optimally adjust
1577 some packet sizes while forwarding large and small data alternatively. The
1578 decision to use splice() or to send large buffers in SSL is modulated by this
1579 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1580 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001581 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001582 clicking). There should be not reason for changing this value. Please check
1583 tune.ssl.maxrecord below.
1584
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001585tune.lua.forced-yield <number>
1586 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001587 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001588 HAProxy scheduler to process other tasks like accepting connections or
1589 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001590 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001591 lowered. If the Lua code is quite long and its result is absolutely required
1592 to process the data, the <number> can be increased.
1593
Willy Tarreau32f61e22015-03-18 17:54:59 +01001594tune.lua.maxmem
1595 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1596 default it is zero which means unlimited. It is important to set a limit to
1597 ensure that a bug in a script will not result in the system running out of
1598 memory.
1599
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001600tune.lua.session-timeout <timeout>
1601 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001602 preventing infinite loops or spending too much time in Lua. This timeout
1603 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001604 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001605
1606tune.lua.task-timeout <timeout>
1607 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1608 dedicated to the tasks. By default, this timeout isn't set because a task may
1609 remain alive during of the lifetime of HAProxy. For example, a task used to
1610 check servers.
1611
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001612tune.lua.service-timeout <timeout>
1613 This is the execution timeout for the Lua services. This is useful for
1614 preventing infinite loops or spending too much time in Lua. This timeout
1615 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001616 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001617
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001618tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001619 Sets the maximum number of consecutive connections a process may accept in a
1620 row before switching to other work. In single process mode, higher numbers
1621 give better performance at high connection rates. However in multi-process
1622 modes, keeping a bit of fairness between processes generally is better to
1623 increase performance. This value applies individually to each listener, so
1624 that the number of processes a listener is bound to is taken into account.
1625 This value defaults to 64. In multi-process mode, it is divided by twice
1626 the number of processes the listener is bound to. Setting this value to -1
1627 completely disables the limitation. It should normally not be needed to tweak
1628 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001629
1630tune.maxpollevents <number>
1631 Sets the maximum amount of events that can be processed at once in a call to
1632 the polling system. The default value is adapted to the operating system. It
1633 has been noticed that reducing it below 200 tends to slightly decrease
1634 latency at the expense of network bandwidth, and increasing it above 200
1635 tends to trade latency for slightly increased bandwidth.
1636
Willy Tarreau27a674e2009-08-17 07:23:33 +02001637tune.maxrewrite <number>
1638 Sets the reserved buffer space to this size in bytes. The reserved space is
1639 used for header rewriting or appending. The first reads on sockets will never
1640 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1641 bufsize, though that does not make much sense since there are rarely large
1642 numbers of headers to add. Setting it too high prevents processing of large
1643 requests or responses. Setting it too low prevents addition of new headers
1644 to already large requests or to POST requests. It is generally wise to set it
1645 to about 1024. It is automatically readjusted to half of bufsize if it is
1646 larger than that. This means you don't have to worry about it when changing
1647 bufsize.
1648
Willy Tarreauf3045d22015-04-29 16:24:50 +02001649tune.pattern.cache-size <number>
1650 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1651 cache which reminds previous lookups and their results. It is used by ACLs
1652 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1653 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1654 strings. It applies to pattern expressions which means that it will be able
1655 to memorize the result of a lookup among all the patterns specified on a
1656 configuration line (including all those loaded from files). It automatically
1657 invalidates entries which are updated using HTTP actions or on the CLI. The
1658 default cache size is set to 10000 entries, which limits its footprint to
1659 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1660 risk of collision in this cache, which is in the order of the size of the
1661 cache divided by 2^64. Typically, at 10000 requests per second with the
1662 default cache size of 10000 entries, there's 1% chance that a brute force
1663 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1664 This is considered much lower than the risk of a memory corruption caused by
1665 aging components. If this is not acceptable, the cache can be disabled by
1666 setting this parameter to 0.
1667
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001668tune.pipesize <number>
1669 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1670 are the default size for the system. But sometimes when using TCP splicing,
1671 it can improve performance to increase pipe sizes, especially if it is
1672 suspected that pipes are not filled and that many calls to splice() are
1673 performed. This has an impact on the kernel's memory footprint, so this must
1674 not be changed if impacts are not understood.
1675
Willy Tarreaue803de22010-01-21 17:43:04 +01001676tune.rcvbuf.client <number>
1677tune.rcvbuf.server <number>
1678 Forces the kernel socket receive buffer size on the client or the server side
1679 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1680 and backends. It should normally never be set, and the default size (0) lets
1681 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001682 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001683 order to save kernel memory by preventing it from buffering too large amounts
1684 of received data. Lower values will significantly increase CPU usage though.
1685
Willy Tarreaub22fc302015-12-14 12:04:35 +01001686tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001687 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001688 socket buffers. One of them is that a read returns more than <recv_enough>
1689 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1690 may be changed by this setting to better deal with workloads involving lots
1691 of short messages such as telnet or SSH sessions.
1692
Olivier Houchard1599b802018-05-24 18:59:04 +02001693tune.runqueue-depth <number>
1694 Sets the maxinum amount of task that can be processed at once when running
1695 tasks. The default value is 200. Increasing it may incur latency when
1696 dealing with I/Os, making it too small can incur extra overhead.
1697
Willy Tarreaue803de22010-01-21 17:43:04 +01001698tune.sndbuf.client <number>
1699tune.sndbuf.server <number>
1700 Forces the kernel socket send buffer size on the client or the server side to
1701 the specified value in bytes. This value applies to all TCP/HTTP frontends
1702 and backends. It should normally never be set, and the default size (0) lets
1703 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001704 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001705 order to save kernel memory by preventing it from buffering too large amounts
1706 of received data. Lower values will significantly increase CPU usage though.
1707 Another use case is to prevent write timeouts with extremely slow clients due
1708 to the kernel waiting for a large part of the buffer to be read before
1709 notifying haproxy again.
1710
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001711tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001712 Sets the size of the global SSL session cache, in a number of blocks. A block
1713 is large enough to contain an encoded session without peer certificate.
1714 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001715 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001716 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001717 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001718 and reassigned. Higher values reduce the occurrence of such a purge, hence
1719 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1720 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001721 and are shared between all processes if "nbproc" is greater than 1. Setting
1722 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001723
Emeric Brun8dc60392014-05-09 13:52:00 +02001724tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001725 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001726 should normally not be used since it will force many renegotiations due to
1727 clients hitting a random process. But it may be required on some operating
1728 systems where none of the SSL cache synchronization method may be used. In
1729 this case, adding a first layer of hash-based load balancing before the SSL
1730 layer might limit the impact of the lack of session sharing.
1731
Emeric Brun4f65bff2012-11-16 15:11:00 +01001732tune.ssl.lifetime <timeout>
1733 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001734 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001735 does not guarantee that sessions will last that long, because if the cache is
1736 full, the longest idle sessions will be purged despite their configured
1737 lifetime. The real usefulness of this setting is to prevent sessions from
1738 being used for too long.
1739
Willy Tarreaubfd59462013-02-21 07:46:09 +01001740tune.ssl.maxrecord <number>
1741 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1742 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1743 data only once it has received a full record. With large records, it means
1744 that clients might have to download up to 16kB of data before starting to
1745 process them. Limiting the value can improve page load times on browsers
1746 located over high latency or low bandwidth networks. It is suggested to find
1747 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1748 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1749 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1750 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001751 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001752 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001753
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001754tune.ssl.default-dh-param <number>
1755 Sets the maximum size of the Diffie-Hellman parameters used for generating
1756 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1757 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1758 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1759 this maximum value. Default value if 1024. Only 1024 or higher values are
1760 allowed. Higher values will increase the CPU load, and values greater than
1761 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001762 used if static Diffie-Hellman parameters are supplied either directly
1763 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001764
Christopher Faulet31af49d2015-06-09 17:29:50 +02001765tune.ssl.ssl-ctx-cache-size <number>
1766 Sets the size of the cache used to store generated certificates to <number>
1767 entries. This is a LRU cache. Because generating a SSL certificate
1768 dynamically is expensive, they are cached. The default cache size is set to
1769 1000 entries.
1770
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001771tune.ssl.capture-cipherlist-size <number>
1772 Sets the maximum size of the buffer used for capturing client-hello cipher
1773 list. If the value is 0 (default value) the capture is disabled, otherwise
1774 a buffer is allocated for each SSL/TLS connection.
1775
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001776tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001777tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001778tune.vars.reqres-max-size <size>
1779tune.vars.sess-max-size <size>
1780tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001781 These five tunes help to manage the maximum amount of memory used by the
1782 variables system. "global" limits the overall amount of memory available for
1783 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1784 memory for the session scope, "txn" for the transaction scope, and "reqres"
1785 limits the memory for each request or response processing.
1786 Memory accounting is hierarchical, meaning more coarse grained limits include
1787 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1788 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001789
Daniel Schneller0b547052016-03-21 20:46:57 +01001790 For example, when "tune.vars.sess-max-size" is limited to 100,
1791 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1792 100 either. If we create a variable "txn.var" that contains 100 bytes,
1793 all available space is consumed.
1794 Notice that exceeding the limits at runtime will not result in an error
1795 message, but values might be cut off or corrupted. So make sure to accurately
1796 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001797
William Lallemanda509e4c2012-11-07 16:54:34 +01001798tune.zlib.memlevel <number>
1799 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001800 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001801 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001802 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001803 between 1 and 9. The default value is 8.
1804
1805tune.zlib.windowsize <number>
1806 Sets the window size (the size of the history buffer) as a parameter of the
1807 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001808 in better compression at the expense of memory usage. Can be a value between
1809 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001810
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018113.3. Debugging
1812--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001813
1814debug
1815 Enables debug mode which dumps to stdout all exchanges, and disables forking
1816 into background. It is the equivalent of the command-line argument "-d". It
1817 should never be used in a production configuration since it may prevent full
1818 system startup.
1819
1820quiet
1821 Do not display any message during startup. It is equivalent to the command-
1822 line argument "-q".
1823
Emeric Brunf099e792010-09-27 12:05:28 +02001824
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018253.4. Userlists
1826--------------
1827It is possible to control access to frontend/backend/listen sections or to
1828http stats by allowing only authenticated and authorized users. To do this,
1829it is required to create at least one userlist and to define users.
1830
1831userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001832 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001833 used to store authentication & authorization data for independent customers.
1834
1835group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001836 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001837 attach users to this group by using a comma separated list of names
1838 proceeded by "users" keyword.
1839
Cyril Bontéf0c60612010-02-06 14:44:47 +01001840user <username> [password|insecure-password <password>]
1841 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001842 Adds user <username> to the current userlist. Both secure (encrypted) and
1843 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001844 evaluated using the crypt(3) function, so depending on the system's
1845 capabilities, different algorithms are supported. For example, modern Glibc
1846 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1847 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001848
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001849 Attention: Be aware that using encrypted passwords might cause significantly
1850 increased CPU usage, depending on the number of requests, and the algorithm
1851 used. For any of the hashed variants, the password for each request must
1852 be processed through the chosen algorithm, before it can be compared to the
1853 value specified in the config file. Most current algorithms are deliberately
1854 designed to be expensive to compute to achieve resistance against brute
1855 force attacks. They do not simply salt/hash the clear text password once,
1856 but thousands of times. This can quickly become a major factor in haproxy's
1857 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001858
1859 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001860 userlist L1
1861 group G1 users tiger,scott
1862 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001863
Cyril Bontéf0c60612010-02-06 14:44:47 +01001864 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1865 user scott insecure-password elgato
1866 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001867
Cyril Bontéf0c60612010-02-06 14:44:47 +01001868 userlist L2
1869 group G1
1870 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001871
Cyril Bontéf0c60612010-02-06 14:44:47 +01001872 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1873 user scott insecure-password elgato groups G1,G2
1874 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001875
1876 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001877
Emeric Brunf099e792010-09-27 12:05:28 +02001878
18793.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001880----------
Emeric Brun94900952015-06-11 18:25:54 +02001881It is possible to propagate entries of any data-types in stick-tables between
1882several haproxy instances over TCP connections in a multi-master fashion. Each
1883instance pushes its local updates and insertions to remote peers. The pushed
1884values overwrite remote ones without aggregation. Interrupted exchanges are
1885automatically detected and recovered from the last known point.
1886In addition, during a soft restart, the old process connects to the new one
1887using such a TCP connection to push all its entries before the new process
1888tries to connect to other peers. That ensures very fast replication during a
1889reload, it typically takes a fraction of a second even for large tables.
1890Note that Server IDs are used to identify servers remotely, so it is important
1891that configurations look similar or at least that the same IDs are forced on
1892each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001893
1894peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001895 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001896 which is referenced by one or more stick-tables.
1897
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001898disabled
1899 Disables a peers section. It disables both listening and any synchronization
1900 related to this section. This is provided to disable synchronization of stick
1901 tables without having to comment out all "peers" references.
1902
1903enable
1904 This re-enables a disabled peers section which was previously disabled.
1905
Emeric Brunf099e792010-09-27 12:05:28 +02001906peer <peername> <ip>:<port>
1907 Defines a peer inside a peers section.
1908 If <peername> is set to the local peer name (by default hostname, or forced
1909 using "-L" command line option), haproxy will listen for incoming remote peer
1910 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1911 to join the remote peer, and <peername> is used at the protocol level to
1912 identify and validate the remote peer on the server side.
1913
1914 During a soft restart, local peer <ip>:<port> is used by the old instance to
1915 connect the new one and initiate a complete replication (teaching process).
1916
1917 It is strongly recommended to have the exact same peers declaration on all
1918 peers and to only rely on the "-L" command line argument to change the local
1919 peer name. This makes it easier to maintain coherent configuration files
1920 across all peers.
1921
William Lallemandb2f07452015-05-12 14:27:13 +02001922 You may want to reference some environment variables in the address
1923 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001924
Cyril Bontédc4d9032012-04-08 21:57:39 +02001925 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001926 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001927 peer haproxy1 192.168.0.1:1024
1928 peer haproxy2 192.168.0.2:1024
1929 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001930
1931 backend mybackend
1932 mode tcp
1933 balance roundrobin
1934 stick-table type ip size 20k peers mypeers
1935 stick on src
1936
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001937 server srv1 192.168.0.30:80
1938 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001939
1940
Simon Horman51a1cf62015-02-03 13:00:44 +090019413.6. Mailers
1942------------
1943It is possible to send email alerts when the state of servers changes.
1944If configured email alerts are sent to each mailer that is configured
1945in a mailers section. Email is sent to mailers using SMTP.
1946
Pieter Baauw386a1272015-08-16 15:26:24 +02001947mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001948 Creates a new mailer list with the name <mailersect>. It is an
1949 independent section which is referenced by one or more proxies.
1950
1951mailer <mailername> <ip>:<port>
1952 Defines a mailer inside a mailers section.
1953
1954 Example:
1955 mailers mymailers
1956 mailer smtp1 192.168.0.1:587
1957 mailer smtp2 192.168.0.2:587
1958
1959 backend mybackend
1960 mode tcp
1961 balance roundrobin
1962
1963 email-alert mailers mymailers
1964 email-alert from test1@horms.org
1965 email-alert to test2@horms.org
1966
1967 server srv1 192.168.0.30:80
1968 server srv2 192.168.0.31:80
1969
Pieter Baauw235fcfc2016-02-13 15:33:40 +01001970timeout mail <time>
1971 Defines the time available for a mail/connection to be made and send to
1972 the mail-server. If not defined the default value is 10 seconds. To allow
1973 for at least two SYN-ACK packets to be send during initial TCP handshake it
1974 is advised to keep this value above 4 seconds.
1975
1976 Example:
1977 mailers mymailers
1978 timeout mail 20s
1979 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09001980
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019814. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001982----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001983
Willy Tarreau6a06a402007-07-15 20:15:28 +02001984Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02001985 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001986 - frontend <name>
1987 - backend <name>
1988 - listen <name>
1989
1990A "defaults" section sets default parameters for all other sections following
1991its declaration. Those default parameters are reset by the next "defaults"
1992section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001993section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001994
1995A "frontend" section describes a set of listening sockets accepting client
1996connections.
1997
1998A "backend" section describes a set of servers to which the proxy will connect
1999to forward incoming connections.
2000
2001A "listen" section defines a complete proxy with its frontend and backend
2002parts combined in one section. It is generally useful for TCP-only traffic.
2003
Willy Tarreau0ba27502007-12-24 16:55:16 +01002004All proxy names must be formed from upper and lower case letters, digits,
2005'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2006case-sensitive, which means that "www" and "WWW" are two different proxies.
2007
2008Historically, all proxy names could overlap, it just caused troubles in the
2009logs. Since the introduction of content switching, it is mandatory that two
2010proxies with overlapping capabilities (frontend/backend) have different names.
2011However, it is still permitted that a frontend and a backend share the same
2012name, as this configuration seems to be commonly encountered.
2013
2014Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2015and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002016bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002017protocol, and can interact with it by allowing, blocking, switching, adding,
2018modifying, or removing arbitrary contents in requests or responses, based on
2019arbitrary criteria.
2020
Willy Tarreau70dffda2014-01-30 03:07:23 +01002021In HTTP mode, the processing applied to requests and responses flowing over
2022a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002023the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002024
2025 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2026 requests and responses are processed, and connections remain open but idle
2027 between responses and new requests.
2028
2029 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2030 1.0 to 1.5-dev21 : only the first request and response are processed, and
2031 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet4212a302018-09-21 10:42:19 +02002032 be used as it creates lots of trouble with logging and HTTP processing. It
2033 is supported only on frontends.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002034
Willy Tarreau70dffda2014-01-30 03:07:23 +01002035 - SCL: server close ("option http-server-close") : the server-facing
2036 connection is closed after the end of the response is received, but the
2037 client-facing connection remains open.
2038
Christopher Faulet315b39c2018-09-21 16:26:19 +02002039 - CLO: close ("option httpclose"): the connection is closed after the end of
2040 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002041
2042The effective mode that will be applied to a connection passing through a
2043frontend and a backend can be determined by both proxy modes according to the
2044following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002045weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002046
Christopher Faulet315b39c2018-09-21 16:26:19 +02002047 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002048
Christopher Faulet315b39c2018-09-21 16:26:19 +02002049 | KAL | SCL | CLO
2050 ----+-----+-----+----
2051 KAL | KAL | SCL | CLO
2052 ----+-----+-----+----
2053 TUN | TUN | SCL | CLO
2054 Frontend ----+-----+-----+----
2055 mode SCL | SCL | SCL | CLO
2056 ----+-----+-----+----
2057 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002058
Willy Tarreau0ba27502007-12-24 16:55:16 +01002059
Willy Tarreau70dffda2014-01-30 03:07:23 +01002060
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020614.1. Proxy keywords matrix
2062--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002063
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002064The following list of keywords is supported. Most of them may only be used in a
2065limited set of section types. Some of them are marked as "deprecated" because
2066they are inherited from an old syntax which may be confusing or functionally
2067limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002068marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002069option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002070and must be disabled for a specific instance. Such options may also be prefixed
2071with "default" in order to restore default settings regardless of what has been
2072specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002073
Willy Tarreau6a06a402007-07-15 20:15:28 +02002074
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002075 keyword defaults frontend listen backend
2076------------------------------------+----------+----------+---------+---------
2077acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002078appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002079backlog X X X -
2080balance X - X X
2081bind - X X -
2082bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002083block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002084capture cookie - X X -
2085capture request header - X X -
2086capture response header - X X -
2087clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002088compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002089contimeout (deprecated) X - X X
2090cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002091declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002092default-server X - X X
2093default_backend X X X -
2094description - X X X
2095disabled X X X X
2096dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002097email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002098email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002099email-alert mailers X X X X
2100email-alert myhostname X X X X
2101email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002102enabled X X X X
2103errorfile X X X X
2104errorloc X X X X
2105errorloc302 X X X X
2106-- keyword -------------------------- defaults - frontend - listen -- backend -
2107errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002108force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002109filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002110fullconn X - X X
2111grace X X X X
2112hash-type X - X X
2113http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002114http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002115http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002116http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002117http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002118http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002119http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002120id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002121ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002122load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002123log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002124log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002125log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002126log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002127max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002128maxconn X X X -
2129mode X X X X
2130monitor fail - X X -
2131monitor-net X X X -
2132monitor-uri X X X -
2133option abortonclose (*) X - X X
2134option accept-invalid-http-request (*) X X X -
2135option accept-invalid-http-response (*) X - X X
2136option allbackups (*) X - X X
2137option checkcache (*) X - X X
2138option clitcpka (*) X X X -
2139option contstats (*) X X X -
2140option dontlog-normal (*) X X X -
2141option dontlognull (*) X X X -
Christopher Faulet315b39c2018-09-21 16:26:19 +02002142option forceclose (deprectated) (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002143-- keyword -------------------------- defaults - frontend - listen -- backend -
2144option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002145option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002146option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002147option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002148option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002149option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002150option http-server-close (*) X X X X
Christopher Faulet4212a302018-09-21 10:42:19 +02002151option http-tunnel (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002152option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002153option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002154option httpchk X - X X
2155option httpclose (*) X X X X
2156option httplog X X X X
2157option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002158option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002159option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002160option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002161option log-health-checks (*) X - X X
2162option log-separate-errors (*) X X X -
2163option logasap (*) X X X -
2164option mysql-check X - X X
2165option nolinger (*) X X X X
2166option originalto X X X X
2167option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002168option pgsql-check X - X X
2169option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002170option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002171option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002172option smtpchk X - X X
2173option socket-stats (*) X X X -
2174option splice-auto (*) X X X X
2175option splice-request (*) X X X X
2176option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002177option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002178option srvtcpka (*) X - X X
2179option ssl-hello-chk X - X X
2180-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002181option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002182option tcp-smart-accept (*) X X X -
2183option tcp-smart-connect (*) X - X X
2184option tcpka X X X X
2185option tcplog X X X X
2186option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002187external-check command X - X X
2188external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002189persist rdp-cookie X - X X
2190rate-limit sessions X X X -
2191redirect - X X X
2192redisp (deprecated) X - X X
2193redispatch (deprecated) X - X X
2194reqadd - X X X
2195reqallow - X X X
2196reqdel - X X X
2197reqdeny - X X X
2198reqiallow - X X X
2199reqidel - X X X
2200reqideny - X X X
2201reqipass - X X X
2202reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002203reqitarpit - X X X
2204reqpass - X X X
2205reqrep - X X X
2206-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002207reqtarpit - X X X
2208retries X - X X
2209rspadd - X X X
2210rspdel - X X X
2211rspdeny - X X X
2212rspidel - X X X
2213rspideny - X X X
2214rspirep - X X X
2215rsprep - X X X
2216server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002217server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002218server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002219source X - X X
2220srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002221stats admin - X X X
2222stats auth X X X X
2223stats enable X X X X
2224stats hide-version X X X X
2225stats http-request - X X X
2226stats realm X X X X
2227stats refresh X X X X
2228stats scope X X X X
2229stats show-desc X X X X
2230stats show-legends X X X X
2231stats show-node X X X X
2232stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002233-- keyword -------------------------- defaults - frontend - listen -- backend -
2234stick match - - X X
2235stick on - - X X
2236stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002237stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002238stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002239tcp-check connect - - X X
2240tcp-check expect - - X X
2241tcp-check send - - X X
2242tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002243tcp-request connection - X X -
2244tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002245tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002246tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002247tcp-response content - - X X
2248tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002249timeout check X - X X
2250timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002251timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002252timeout clitimeout (deprecated) X X X -
2253timeout connect X - X X
2254timeout contimeout (deprecated) X - X X
2255timeout http-keep-alive X X X X
2256timeout http-request X X X X
2257timeout queue X - X X
2258timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002259timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002260timeout srvtimeout (deprecated) X - X X
2261timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002262timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002263transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002264unique-id-format X X X -
2265unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002266use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002267use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002268------------------------------------+----------+----------+---------+---------
2269 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002270
Willy Tarreau0ba27502007-12-24 16:55:16 +01002271
Willy Tarreauc57f0e22009-05-10 13:12:33 +020022724.2. Alphabetically sorted keywords reference
2273---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002274
2275This section provides a description of each keyword and its usage.
2276
2277
2278acl <aclname> <criterion> [flags] [operator] <value> ...
2279 Declare or complete an access list.
2280 May be used in sections : defaults | frontend | listen | backend
2281 no | yes | yes | yes
2282 Example:
2283 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2284 acl invalid_src src_port 0:1023
2285 acl local_dst hdr(host) -i localhost
2286
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002287 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002288
2289
Cyril Bontéb21570a2009-11-29 20:04:48 +01002290appsession <cookie> len <length> timeout <holdtime>
2291 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002292 Define session stickiness on an existing application cookie.
2293 May be used in sections : defaults | frontend | listen | backend
2294 no | no | yes | yes
2295 Arguments :
2296 <cookie> this is the name of the cookie used by the application and which
2297 HAProxy will have to learn for each new session.
2298
Cyril Bontéb21570a2009-11-29 20:04:48 +01002299 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002300 checked in each cookie value.
2301
2302 <holdtime> this is the time after which the cookie will be removed from
2303 memory if unused. If no unit is specified, this time is in
2304 milliseconds.
2305
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002306 request-learn
2307 If this option is specified, then haproxy will be able to learn
2308 the cookie found in the request in case the server does not
2309 specify any in response. This is typically what happens with
2310 PHPSESSID cookies, or when haproxy's session expires before
2311 the application's session and the correct server is selected.
2312 It is recommended to specify this option to improve reliability.
2313
Cyril Bontéb21570a2009-11-29 20:04:48 +01002314 prefix When this option is specified, haproxy will match on the cookie
2315 prefix (or URL parameter prefix). The appsession value is the
2316 data following this prefix.
2317
2318 Example :
2319 appsession ASPSESSIONID len 64 timeout 3h prefix
2320
Davor Ocelice9ed2812017-12-25 17:49:28 +01002321 This will match the cookie ASPSESSIONIDXXX=XXXX,
2322 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002323
2324 mode This option allows to change the URL parser mode.
2325 2 modes are currently supported :
2326 - path-parameters :
2327 The parser looks for the appsession in the path parameters
2328 part (each parameter is separated by a semi-colon), which is
2329 convenient for JSESSIONID for example.
2330 This is the default mode if the option is not set.
2331 - query-string :
2332 In this mode, the parser will look for the appsession in the
2333 query string.
2334
Willy Tarreau294d0f02015-08-10 19:40:12 +02002335 As of version 1.6, appsessions was removed. It is more flexible and more
2336 convenient to use stick-tables instead, and stick-tables support multi-master
2337 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002338
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002339 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2340 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002341
2342
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002343backlog <conns>
2344 Give hints to the system about the approximate listen backlog desired size
2345 May be used in sections : defaults | frontend | listen | backend
2346 yes | yes | yes | no
2347 Arguments :
2348 <conns> is the number of pending connections. Depending on the operating
2349 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002350 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002351
2352 In order to protect against SYN flood attacks, one solution is to increase
2353 the system's SYN backlog size. Depending on the system, sometimes it is just
2354 tunable via a system parameter, sometimes it is not adjustable at all, and
2355 sometimes the system relies on hints given by the application at the time of
2356 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2357 to the listen() syscall. On systems which can make use of this value, it can
2358 sometimes be useful to be able to specify a different value, hence this
2359 backlog parameter.
2360
2361 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2362 used as a hint and the system accepts up to the smallest greater power of
2363 two, and never more than some limits (usually 32768).
2364
2365 See also : "maxconn" and the target operating system's tuning guide.
2366
2367
Willy Tarreau0ba27502007-12-24 16:55:16 +01002368balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002369balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002370 Define the load balancing algorithm to be used in a backend.
2371 May be used in sections : defaults | frontend | listen | backend
2372 yes | no | yes | yes
2373 Arguments :
2374 <algorithm> is the algorithm used to select a server when doing load
2375 balancing. This only applies when no persistence information
2376 is available, or when a connection is redispatched to another
2377 server. <algorithm> may be one of the following :
2378
2379 roundrobin Each server is used in turns, according to their weights.
2380 This is the smoothest and fairest algorithm when the server's
2381 processing time remains equally distributed. This algorithm
2382 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002383 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002384 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002385 large farms, when a server becomes up after having been down
2386 for a very short time, it may sometimes take a few hundreds
2387 requests for it to be re-integrated into the farm and start
2388 receiving traffic. This is normal, though very rare. It is
2389 indicated here in case you would have the chance to observe
2390 it, so that you don't worry.
2391
2392 static-rr Each server is used in turns, according to their weights.
2393 This algorithm is as similar to roundrobin except that it is
2394 static, which means that changing a server's weight on the
2395 fly will have no effect. On the other hand, it has no design
2396 limitation on the number of servers, and when a server goes
2397 up, it is always immediately reintroduced into the farm, once
2398 the full map is recomputed. It also uses slightly less CPU to
2399 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002400
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002401 leastconn The server with the lowest number of connections receives the
2402 connection. Round-robin is performed within groups of servers
2403 of the same load to ensure that all servers will be used. Use
2404 of this algorithm is recommended where very long sessions are
2405 expected, such as LDAP, SQL, TSE, etc... but is not very well
2406 suited for protocols using short sessions such as HTTP. This
2407 algorithm is dynamic, which means that server weights may be
2408 adjusted on the fly for slow starts for instance.
2409
Willy Tarreauf09c6602012-02-13 17:12:08 +01002410 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002411 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002412 identifier to the highest (see server parameter "id"), which
2413 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002414 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002415 not make sense to use this algorithm without setting maxconn.
2416 The purpose of this algorithm is to always use the smallest
2417 number of servers so that extra servers can be powered off
2418 during non-intensive hours. This algorithm ignores the server
2419 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002420 or IMAP than HTTP, though it can be useful there too. In
2421 order to use this algorithm efficiently, it is recommended
2422 that a cloud controller regularly checks server usage to turn
2423 them off when unused, and regularly checks backend queue to
2424 turn new servers on when the queue inflates. Alternatively,
2425 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002426
Willy Tarreau0ba27502007-12-24 16:55:16 +01002427 source The source IP address is hashed and divided by the total
2428 weight of the running servers to designate which server will
2429 receive the request. This ensures that the same client IP
2430 address will always reach the same server as long as no
2431 server goes down or up. If the hash result changes due to the
2432 number of running servers changing, many clients will be
2433 directed to a different server. This algorithm is generally
2434 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002435 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002436 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002437 static by default, which means that changing a server's
2438 weight on the fly will have no effect, but this can be
2439 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002440
Oskar Stolc8dc41842012-05-19 10:19:54 +01002441 uri This algorithm hashes either the left part of the URI (before
2442 the question mark) or the whole URI (if the "whole" parameter
2443 is present) and divides the hash value by the total weight of
2444 the running servers. The result designates which server will
2445 receive the request. This ensures that the same URI will
2446 always be directed to the same server as long as no server
2447 goes up or down. This is used with proxy caches and
2448 anti-virus proxies in order to maximize the cache hit rate.
2449 Note that this algorithm may only be used in an HTTP backend.
2450 This algorithm is static by default, which means that
2451 changing a server's weight on the fly will have no effect,
2452 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002453
Oskar Stolc8dc41842012-05-19 10:19:54 +01002454 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002455 "depth", both followed by a positive integer number. These
2456 options may be helpful when it is needed to balance servers
2457 based on the beginning of the URI only. The "len" parameter
2458 indicates that the algorithm should only consider that many
2459 characters at the beginning of the URI to compute the hash.
2460 Note that having "len" set to 1 rarely makes sense since most
2461 URIs start with a leading "/".
2462
2463 The "depth" parameter indicates the maximum directory depth
2464 to be used to compute the hash. One level is counted for each
2465 slash in the request. If both parameters are specified, the
2466 evaluation stops when either is reached.
2467
Willy Tarreau0ba27502007-12-24 16:55:16 +01002468 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002469 the query string of each HTTP GET request.
2470
2471 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002472 request entity will be searched for the parameter argument,
2473 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002474 ('?') in the URL. The message body will only start to be
2475 analyzed once either the advertised amount of data has been
2476 received or the request buffer is full. In the unlikely event
2477 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002478 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002479 be randomly balanced if at all. This keyword used to support
2480 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002481
2482 If the parameter is found followed by an equal sign ('=') and
2483 a value, then the value is hashed and divided by the total
2484 weight of the running servers. The result designates which
2485 server will receive the request.
2486
2487 This is used to track user identifiers in requests and ensure
2488 that a same user ID will always be sent to the same server as
2489 long as no server goes up or down. If no value is found or if
2490 the parameter is not found, then a round robin algorithm is
2491 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002492 backend. This algorithm is static by default, which means
2493 that changing a server's weight on the fly will have no
2494 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002495
Cyril Bontédc4d9032012-04-08 21:57:39 +02002496 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2497 request. Just as with the equivalent ACL 'hdr()' function,
2498 the header name in parenthesis is not case sensitive. If the
2499 header is absent or if it does not contain any value, the
2500 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002501
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002502 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002503 reducing the hash algorithm to the main domain part with some
2504 specific headers such as 'Host'. For instance, in the Host
2505 value "haproxy.1wt.eu", only "1wt" will be considered.
2506
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002507 This algorithm is static by default, which means that
2508 changing a server's weight on the fly will have no effect,
2509 but this can be changed using "hash-type".
2510
Willy Tarreau760e81d2018-05-03 07:20:40 +02002511 random A random number will be used as the key for the consistent
2512 hashing function. This means that the servers' weights are
2513 respected, dynamic weight changes immediately take effect, as
2514 well as new server additions. Random load balancing can be
2515 useful with large farms or when servers are frequently added
2516 or removed. The hash-balance-factor directive can be used to
2517 further improve fairness of the load balancing, especially
2518 in situations where servers show highly variable response
2519 times.
2520
Emeric Brun736aa232009-06-30 17:56:00 +02002521 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002522 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002523 The RDP cookie <name> (or "mstshash" if omitted) will be
2524 looked up and hashed for each incoming TCP request. Just as
2525 with the equivalent ACL 'req_rdp_cookie()' function, the name
2526 is not case-sensitive. This mechanism is useful as a degraded
2527 persistence mode, as it makes it possible to always send the
2528 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002529 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002530 used instead.
2531
2532 Note that for this to work, the frontend must ensure that an
2533 RDP cookie is already present in the request buffer. For this
2534 you must use 'tcp-request content accept' rule combined with
2535 a 'req_rdp_cookie_cnt' ACL.
2536
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002537 This algorithm is static by default, which means that
2538 changing a server's weight on the fly will have no effect,
2539 but this can be changed using "hash-type".
2540
Cyril Bontédc4d9032012-04-08 21:57:39 +02002541 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002542
Willy Tarreau0ba27502007-12-24 16:55:16 +01002543 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002544 algorithms. Right now, only "url_param" and "uri" support an
2545 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002546
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002547 The load balancing algorithm of a backend is set to roundrobin when no other
2548 algorithm, mode nor option have been set. The algorithm may only be set once
2549 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002550
Lukas Tribus80512b12018-10-27 20:07:40 +02002551 With authentication schemes that require the same connection like NTLM, URI
2552 based alghoritms must not be used, as they would cause subsequent requests
2553 to be routed to different backend servers, breaking the invalid assumptions
2554 NTLM relies on.
2555
Willy Tarreau0ba27502007-12-24 16:55:16 +01002556 Examples :
2557 balance roundrobin
2558 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002559 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002560 balance hdr(User-Agent)
2561 balance hdr(host)
2562 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002563
2564 Note: the following caveats and limitations on using the "check_post"
2565 extension with "url_param" must be considered :
2566
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002567 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002568 to determine if the parameters will be found in the body or entity which
2569 may contain binary data. Therefore another method may be required to
2570 restrict consideration of POST requests that have no URL parameters in
2571 the body. (see acl reqideny http_end)
2572
2573 - using a <max_wait> value larger than the request buffer size does not
2574 make sense and is useless. The buffer size is set at build time, and
2575 defaults to 16 kB.
2576
2577 - Content-Encoding is not supported, the parameter search will probably
2578 fail; and load balancing will fall back to Round Robin.
2579
2580 - Expect: 100-continue is not supported, load balancing will fall back to
2581 Round Robin.
2582
Lukas Tribus23953682017-04-28 13:24:30 +00002583 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002584 If the entire parameter value is not present in the first chunk, the
2585 selection of server is undefined (actually, defined by how little
2586 actually appeared in the first chunk).
2587
2588 - This feature does not support generation of a 100, 411 or 501 response.
2589
2590 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002591 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002592 white space or control characters are found, indicating the end of what
2593 might be a URL parameter list. This is probably not a concern with SGML
2594 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002595
Willy Tarreau294d0f02015-08-10 19:40:12 +02002596 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002597
2598
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002599bind [<address>]:<port_range> [, ...] [param*]
2600bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002601 Define one or several listening addresses and/or ports in a frontend.
2602 May be used in sections : defaults | frontend | listen | backend
2603 no | yes | yes | no
2604 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002605 <address> is optional and can be a host name, an IPv4 address, an IPv6
2606 address, or '*'. It designates the address the frontend will
2607 listen on. If unset, all IPv4 addresses of the system will be
2608 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002609 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002610 Optionally, an address family prefix may be used before the
2611 address to force the family regardless of the address format,
2612 which can be useful to specify a path to a unix socket with
2613 no slash ('/'). Currently supported prefixes are :
2614 - 'ipv4@' -> address is always IPv4
2615 - 'ipv6@' -> address is always IPv6
2616 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002617 - 'abns@' -> address is in abstract namespace (Linux only).
2618 Note: since abstract sockets are not "rebindable", they
2619 do not cope well with multi-process mode during
2620 soft-restart, so it is better to avoid them if
2621 nbproc is greater than 1. The effect is that if the
2622 new process fails to start, only one of the old ones
2623 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002624 - 'fd@<n>' -> use file descriptor <n> inherited from the
2625 parent. The fd must be bound and may or may not already
2626 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002627 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2628 connected unix socket or of a socketpair. The bind waits
2629 to receive a FD over the unix socket and uses it as if it
2630 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002631 You may want to reference some environment variables in the
2632 address parameter, see section 2.3 about environment
2633 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002634
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002635 <port_range> is either a unique TCP port, or a port range for which the
2636 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002637 above. The port is mandatory for TCP listeners. Note that in
2638 the case of an IPv6 address, the port is always the number
2639 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002640 - a numerical port (ex: '80')
2641 - a dash-delimited ports range explicitly stating the lower
2642 and upper bounds (ex: '2000-2100') which are included in
2643 the range.
2644
2645 Particular care must be taken against port ranges, because
2646 every <address:port> couple consumes one socket (= a file
2647 descriptor), so it's easy to consume lots of descriptors
2648 with a simple range, and to run out of sockets. Also, each
2649 <address:port> couple must be used only once among all
2650 instances running on a same system. Please note that binding
2651 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002652 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002653 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002654
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002655 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002656 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002657 receive UNIX connections on the socket located at this place.
2658 The path must begin with a slash and by default is absolute.
2659 It can be relative to the prefix defined by "unix-bind" in
2660 the global section. Note that the total length of the prefix
2661 followed by the socket path cannot exceed some system limits
2662 for UNIX sockets, which commonly are set to 107 characters.
2663
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002664 <param*> is a list of parameters common to all sockets declared on the
2665 same line. These numerous parameters depend on OS and build
2666 options and have a complete section dedicated to them. Please
2667 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002668
Willy Tarreau0ba27502007-12-24 16:55:16 +01002669 It is possible to specify a list of address:port combinations delimited by
2670 commas. The frontend will then listen on all of these addresses. There is no
2671 fixed limit to the number of addresses and ports which can be listened on in
2672 a frontend, as well as there is no limit to the number of "bind" statements
2673 in a frontend.
2674
2675 Example :
2676 listen http_proxy
2677 bind :80,:443
2678 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002679 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002680
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002681 listen http_https_proxy
2682 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002683 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002684
Willy Tarreau24709282013-03-10 21:32:12 +01002685 listen http_https_proxy_explicit
2686 bind ipv6@:80
2687 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2688 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2689
Willy Tarreaudad36a32013-03-11 01:20:04 +01002690 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002691 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002692
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002693 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2694 sun_path length is used for the address length. Some other programs
2695 such as socat use the string length only by default. Pass the option
2696 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2697 make it compatible with HAProxy's.
2698
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002699 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002700 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002701
2702
Christopher Fauletff4121f2017-11-22 16:38:49 +01002703bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002704 Limit visibility of an instance to a certain set of processes numbers.
2705 May be used in sections : defaults | frontend | listen | backend
2706 yes | yes | yes | yes
2707 Arguments :
2708 all All process will see this instance. This is the default. It
2709 may be used to override a default value.
2710
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002711 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002712 option may be combined with other numbers.
2713
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002714 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002715 option may be combined with other numbers. Do not use it
2716 with less than 2 processes otherwise some instances might be
2717 missing from all processes.
2718
Christopher Fauletff4121f2017-11-22 16:38:49 +01002719 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002720 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002721 the machine's word size. Ranges can be partially defined. The
2722 higher bound can be omitted. In such case, it is replaced by
2723 the corresponding maximum value. If a proxy is bound to
2724 process numbers greater than the configured global.nbproc, it
2725 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002726 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002727
2728 This keyword limits binding of certain instances to certain processes. This
2729 is useful in order not to have too many processes listening to the same
2730 ports. For instance, on a dual-core machine, it might make sense to set
2731 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2732 and 'even' instances.
2733
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002734 At the moment, it is not possible to reference more than 32 or 64 processes
2735 using this keyword, but this should be more than enough for most setups.
2736 Please note that 'all' really means all processes regardless of the machine's
2737 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002738
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002739 Each "bind" line may further be limited to a subset of the proxy's processes,
2740 please consult the "process" bind keyword in section 5.1.
2741
Willy Tarreaub369a042014-09-16 13:21:03 +02002742 When a frontend has no explicit "bind-process" line, it tries to bind to all
2743 the processes referenced by its "bind" lines. That means that frontends can
2744 easily adapt to their listeners' processes.
2745
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002746 If some backends are referenced by frontends bound to other processes, the
2747 backend automatically inherits the frontend's processes.
2748
2749 Example :
2750 listen app_ip1
2751 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002752 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002753
2754 listen app_ip2
2755 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002756 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002757
2758 listen management
2759 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002760 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002761
Willy Tarreau110ecc12012-11-15 17:50:01 +01002762 listen management
2763 bind 10.0.0.4:80
2764 bind-process 1-4
2765
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002766 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002767
2768
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002769block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002770 Block a layer 7 request if/unless a condition is matched
2771 May be used in sections : defaults | frontend | listen | backend
2772 no | yes | yes | yes
2773
2774 The HTTP request will be blocked very early in the layer 7 processing
2775 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002776 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002777 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002778 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002779 "block" statements per instance. To block connections at layer 4 (without
2780 sending a 403 error) see "tcp-request connection reject" and
2781 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002782
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002783 This form is deprecated, do not use it in any new configuration, use the new
2784 "http-request deny" instead.
2785
Willy Tarreau0ba27502007-12-24 16:55:16 +01002786 Example:
2787 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2788 acl invalid_src src_port 0:1023
2789 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002790 # block is deprecated. Use http-request deny instead:
2791 #block if invalid_src || local_dst
2792 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002793
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002794 See also : section 7 about ACL usage, "http-request deny",
2795 "http-response deny", "tcp-request connection reject" and
2796 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002797
2798capture cookie <name> len <length>
2799 Capture and log a cookie in the request and in the response.
2800 May be used in sections : defaults | frontend | listen | backend
2801 no | yes | yes | no
2802 Arguments :
2803 <name> is the beginning of the name of the cookie to capture. In order
2804 to match the exact name, simply suffix the name with an equal
2805 sign ('='). The full name will appear in the logs, which is
2806 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002807 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002808
2809 <length> is the maximum number of characters to report in the logs, which
2810 include the cookie name, the equal sign and the value, all in the
2811 standard "name=value" form. The string will be truncated on the
2812 right if it exceeds <length>.
2813
2814 Only the first cookie is captured. Both the "cookie" request headers and the
2815 "set-cookie" response headers are monitored. This is particularly useful to
2816 check for application bugs causing session crossing or stealing between
2817 users, because generally the user's cookies can only change on a login page.
2818
2819 When the cookie was not presented by the client, the associated log column
2820 will report "-". When a request does not cause a cookie to be assigned by the
2821 server, a "-" is reported in the response column.
2822
2823 The capture is performed in the frontend only because it is necessary that
2824 the log format does not change for a given frontend depending on the
2825 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002826 "capture cookie" statement in a frontend. The maximum capture length is set
2827 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2828 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002829
2830 Example:
2831 capture cookie ASPSESSION len 32
2832
2833 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002834 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002835
2836
2837capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002838 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002839 May be used in sections : defaults | frontend | listen | backend
2840 no | yes | yes | no
2841 Arguments :
2842 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002843 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002844 appear in the requests, with the first letter of each word in
2845 upper case. The header name will not appear in the logs, only the
2846 value is reported, but the position in the logs is respected.
2847
2848 <length> is the maximum number of characters to extract from the value and
2849 report in the logs. The string will be truncated on the right if
2850 it exceeds <length>.
2851
Willy Tarreau4460d032012-11-21 23:37:37 +01002852 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002853 value will be added to the logs between braces ('{}'). If multiple headers
2854 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002855 in the same order they were declared in the configuration. Non-existent
2856 headers will be logged just as an empty string. Common uses for request
2857 header captures include the "Host" field in virtual hosting environments, the
2858 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002859 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002860 environments to find where the request came from.
2861
2862 Note that when capturing headers such as "User-agent", some spaces may be
2863 logged, making the log analysis more difficult. Thus be careful about what
2864 you log if you know your log parser is not smart enough to rely on the
2865 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002866
Willy Tarreau0900abb2012-11-22 00:21:46 +01002867 There is no limit to the number of captured request headers nor to their
2868 length, though it is wise to keep them low to limit memory usage per session.
2869 In order to keep log format consistent for a same frontend, header captures
2870 can only be declared in a frontend. It is not possible to specify a capture
2871 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002872
2873 Example:
2874 capture request header Host len 15
2875 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002876 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002877
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002878 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002879 about logging.
2880
2881
2882capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002883 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002884 May be used in sections : defaults | frontend | listen | backend
2885 no | yes | yes | no
2886 Arguments :
2887 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002888 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002889 appear in the response, with the first letter of each word in
2890 upper case. The header name will not appear in the logs, only the
2891 value is reported, but the position in the logs is respected.
2892
2893 <length> is the maximum number of characters to extract from the value and
2894 report in the logs. The string will be truncated on the right if
2895 it exceeds <length>.
2896
Willy Tarreau4460d032012-11-21 23:37:37 +01002897 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002898 result will be added to the logs between braces ('{}') after the captured
2899 request headers. If multiple headers are captured, they will be delimited by
2900 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002901 the configuration. Non-existent headers will be logged just as an empty
2902 string. Common uses for response header captures include the "Content-length"
2903 header which indicates how many bytes are expected to be returned, the
2904 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002905
Willy Tarreau0900abb2012-11-22 00:21:46 +01002906 There is no limit to the number of captured response headers nor to their
2907 length, though it is wise to keep them low to limit memory usage per session.
2908 In order to keep log format consistent for a same frontend, header captures
2909 can only be declared in a frontend. It is not possible to specify a capture
2910 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002911
2912 Example:
2913 capture response header Content-length len 9
2914 capture response header Location len 15
2915
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002916 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002917 about logging.
2918
2919
Cyril Bontéf0c60612010-02-06 14:44:47 +01002920clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002921 Set the maximum inactivity time on the client side.
2922 May be used in sections : defaults | frontend | listen | backend
2923 yes | yes | yes | no
2924 Arguments :
2925 <timeout> is the timeout value is specified in milliseconds by default, but
2926 can be in any other unit if the number is suffixed by the unit,
2927 as explained at the top of this document.
2928
2929 The inactivity timeout applies when the client is expected to acknowledge or
2930 send data. In HTTP mode, this timeout is particularly important to consider
2931 during the first phase, when the client sends the request, and during the
2932 response while it is reading data sent by the server. The value is specified
2933 in milliseconds by default, but can be in any other unit if the number is
2934 suffixed by the unit, as specified at the top of this document. In TCP mode
2935 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2936 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002937 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002938 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002939 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002940
2941 This parameter is specific to frontends, but can be specified once for all in
2942 "defaults" sections. This is in fact one of the easiest solutions not to
2943 forget about it. An unspecified timeout results in an infinite timeout, which
2944 is not recommended. Such a usage is accepted and works but reports a warning
2945 during startup because it may results in accumulation of expired sessions in
2946 the system if the system's timeouts are not configured either.
2947
2948 This parameter is provided for compatibility but is currently deprecated.
2949 Please use "timeout client" instead.
2950
Willy Tarreau036fae02008-01-06 13:24:40 +01002951 See also : "timeout client", "timeout http-request", "timeout server", and
2952 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002953
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002954compression algo <algorithm> ...
2955compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002956compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002957 Enable HTTP compression.
2958 May be used in sections : defaults | frontend | listen | backend
2959 yes | yes | yes | yes
2960 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002961 algo is followed by the list of supported compression algorithms.
2962 type is followed by the list of MIME types that will be compressed.
2963 offload makes haproxy work as a compression offloader only (see notes).
2964
2965 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002966 identity this is mostly for debugging, and it was useful for developing
2967 the compression feature. Identity does not apply any change on
2968 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002969
Willy Tarreauc91840a2015-03-28 17:00:39 +01002970 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01002971 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01002972
2973 deflate same as "gzip", but with deflate algorithm and zlib format.
2974 Note that this algorithm has ambiguous support on many
2975 browsers and no support at all from recent ones. It is
2976 strongly recommended not to use it for anything else than
2977 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01002978 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002979
Willy Tarreauc91840a2015-03-28 17:00:39 +01002980 raw-deflate same as "deflate" without the zlib wrapper, and used as an
2981 alternative when the browser wants "deflate". All major
2982 browsers understand it and despite violating the standards,
2983 it is known to work better than "deflate", at least on MSIE
2984 and some versions of Safari. Do not use it in conjunction
2985 with "deflate", use either one or the other since both react
2986 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01002987 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002988
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002989 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002990 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002991 If backend servers support HTTP compression, these directives
2992 will be no-op: haproxy will see the compressed response and will not
2993 compress again. If backend servers do not support HTTP compression and
2994 there is Accept-Encoding header in request, haproxy will compress the
2995 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002996
2997 The "offload" setting makes haproxy remove the Accept-Encoding header to
2998 prevent backend servers from compressing responses. It is strongly
2999 recommended not to do this because this means that all the compression work
3000 will be done on the single point where haproxy is located. However in some
3001 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003002 with broken HTTP compression implementation which can't be turned off.
3003 In that case haproxy can be used to prevent that gateway from emitting
3004 invalid payloads. In this case, simply removing the header in the
3005 configuration does not work because it applies before the header is parsed,
3006 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003007 then be used for such scenarios. Note: for now, the "offload" setting is
3008 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003009
William Lallemand05097442012-11-20 12:14:28 +01003010 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003011 * the request does not advertise a supported compression algorithm in the
3012 "Accept-Encoding" header
3013 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01003014 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01003015 * response header "Transfer-Encoding" contains "chunked" (Temporary
3016 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003017 * response contain neither a "Content-Length" header nor a
3018 "Transfer-Encoding" whose last value is "chunked"
3019 * response contains a "Content-Type" header whose first value starts with
3020 "multipart"
3021 * the response contains the "no-transform" value in the "Cache-control"
3022 header
3023 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3024 and later
3025 * The response contains a "Content-Encoding" header, indicating that the
3026 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01003027
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003028 Note: The compression does not rewrite Etag headers, and does not emit the
3029 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003030
William Lallemand82fe75c2012-10-23 10:25:10 +02003031 Examples :
3032 compression algo gzip
3033 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003034
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003035
Cyril Bontéf0c60612010-02-06 14:44:47 +01003036contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003037 Set the maximum time to wait for a connection attempt to a server to succeed.
3038 May be used in sections : defaults | frontend | listen | backend
3039 yes | no | yes | yes
3040 Arguments :
3041 <timeout> is the timeout value is specified in milliseconds by default, but
3042 can be in any other unit if the number is suffixed by the unit,
3043 as explained at the top of this document.
3044
3045 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003046 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003047 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003048 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003049 connect timeout also presets the queue timeout to the same value if this one
3050 has not been specified. Historically, the contimeout was also used to set the
3051 tarpit timeout in a listen section, which is not possible in a pure frontend.
3052
3053 This parameter is specific to backends, but can be specified once for all in
3054 "defaults" sections. This is in fact one of the easiest solutions not to
3055 forget about it. An unspecified timeout results in an infinite timeout, which
3056 is not recommended. Such a usage is accepted and works but reports a warning
3057 during startup because it may results in accumulation of failed sessions in
3058 the system if the system's timeouts are not configured either.
3059
3060 This parameter is provided for backwards compatibility but is currently
3061 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3062 instead.
3063
3064 See also : "timeout connect", "timeout queue", "timeout tarpit",
3065 "timeout server", "contimeout".
3066
3067
Willy Tarreau55165fe2009-05-10 12:02:55 +02003068cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003069 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3070 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003071 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003072 Enable cookie-based persistence in a backend.
3073 May be used in sections : defaults | frontend | listen | backend
3074 yes | no | yes | yes
3075 Arguments :
3076 <name> is the name of the cookie which will be monitored, modified or
3077 inserted in order to bring persistence. This cookie is sent to
3078 the client via a "Set-Cookie" header in the response, and is
3079 brought back by the client in a "Cookie" header in all requests.
3080 Special care should be taken to choose a name which does not
3081 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003082 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003083 HTTP/HTTPS), care should be taken to use different cookie names
3084 between all backends if persistence between them is not desired.
3085
3086 rewrite This keyword indicates that the cookie will be provided by the
3087 server and that haproxy will have to modify its value to set the
3088 server's identifier in it. This mode is handy when the management
3089 of complex combinations of "Set-cookie" and "Cache-control"
3090 headers is left to the application. The application can then
3091 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003092 cookie. Since all responses should be monitored, this mode
3093 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003094 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003095 start with this mode for new deployments. This keyword is
3096 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003097
3098 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003099 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003100
Willy Tarreaua79094d2010-08-31 22:54:15 +02003101 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003102 server. When used without the "preserve" option, if the server
3103 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003104 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003105 existing configurations running in the "rewrite" mode. The cookie
3106 will only be a session cookie and will not be stored on the
3107 client's disk. By default, unless the "indirect" option is added,
3108 the server will see the cookies emitted by the client. Due to
3109 caching effects, it is generally wise to add the "nocache" or
3110 "postonly" keywords (see below). The "insert" keyword is not
3111 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003112
3113 prefix This keyword indicates that instead of relying on a dedicated
3114 cookie for the persistence, an existing one will be completed.
3115 This may be needed in some specific environments where the client
3116 does not support more than one single cookie and the application
3117 already needs it. In this case, whenever the server sets a cookie
3118 named <name>, it will be prefixed with the server's identifier
3119 and a delimiter. The prefix will be removed from all client
3120 requests so that the server still finds the cookie it emitted.
3121 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003122 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003123 not compatible with "rewrite" and "insert". Note: it is highly
3124 recommended not to use "indirect" with "prefix", otherwise server
3125 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003126
Willy Tarreaua79094d2010-08-31 22:54:15 +02003127 indirect When this option is specified, no cookie will be emitted to a
3128 client which already has a valid one for the server which has
3129 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003130 it will be removed, unless the "preserve" option is also set. In
3131 "insert" mode, this will additionally remove cookies from the
3132 requests transmitted to the server, making the persistence
3133 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003134 Note: it is highly recommended not to use "indirect" with
3135 "prefix", otherwise server cookie updates would not be sent to
3136 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003137
3138 nocache This option is recommended in conjunction with the insert mode
3139 when there is a cache between the client and HAProxy, as it
3140 ensures that a cacheable response will be tagged non-cacheable if
3141 a cookie needs to be inserted. This is important because if all
3142 persistence cookies are added on a cacheable home page for
3143 instance, then all customers will then fetch the page from an
3144 outer cache and will all share the same persistence cookie,
3145 leading to one server receiving much more traffic than others.
3146 See also the "insert" and "postonly" options.
3147
3148 postonly This option ensures that cookie insertion will only be performed
3149 on responses to POST requests. It is an alternative to the
3150 "nocache" option, because POST responses are not cacheable, so
3151 this ensures that the persistence cookie will never get cached.
3152 Since most sites do not need any sort of persistence before the
3153 first POST which generally is a login request, this is a very
3154 efficient method to optimize caching without risking to find a
3155 persistence cookie in the cache.
3156 See also the "insert" and "nocache" options.
3157
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003158 preserve This option may only be used with "insert" and/or "indirect". It
3159 allows the server to emit the persistence cookie itself. In this
3160 case, if a cookie is found in the response, haproxy will leave it
3161 untouched. This is useful in order to end persistence after a
3162 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003163 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003164 the past. By combining this mechanism with the "disable-on-404"
3165 check option, it is possible to perform a completely graceful
3166 shutdown because users will definitely leave the server after
3167 they logout.
3168
Willy Tarreau4992dd22012-05-31 21:02:17 +02003169 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3170 when a cookie is inserted. This attribute is used so that a
3171 user agent doesn't share the cookie with non-HTTP components.
3172 Please check RFC6265 for more information on this attribute.
3173
3174 secure This option tells haproxy to add a "Secure" cookie attribute when
3175 a cookie is inserted. This attribute is used so that a user agent
3176 never emits this cookie over non-secure channels, which means
3177 that a cookie learned with this flag will be presented only over
3178 SSL/TLS connections. Please check RFC6265 for more information on
3179 this attribute.
3180
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003181 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003182 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003183 name. If the domain begins with a dot, the browser is allowed to
3184 use it for any host ending with that name. It is also possible to
3185 specify several domain names by invoking this option multiple
3186 times. Some browsers might have small limits on the number of
3187 domains, so be careful when doing that. For the record, sending
3188 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003189
Willy Tarreau996a92c2010-10-13 19:30:47 +02003190 maxidle This option allows inserted cookies to be ignored after some idle
3191 time. It only works with insert-mode cookies. When a cookie is
3192 sent to the client, the date this cookie was emitted is sent too.
3193 Upon further presentations of this cookie, if the date is older
3194 than the delay indicated by the parameter (in seconds), it will
3195 be ignored. Otherwise, it will be refreshed if needed when the
3196 response is sent to the client. This is particularly useful to
3197 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003198 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003199 this option is set and a cookie has no date, it is always
3200 accepted, but gets refreshed in the response. This maintains the
3201 ability for admins to access their sites. Cookies that have a
3202 date in the future further than 24 hours are ignored. Doing so
3203 lets admins fix timezone issues without risking kicking users off
3204 the site.
3205
3206 maxlife This option allows inserted cookies to be ignored after some life
3207 time, whether they're in use or not. It only works with insert
3208 mode cookies. When a cookie is first sent to the client, the date
3209 this cookie was emitted is sent too. Upon further presentations
3210 of this cookie, if the date is older than the delay indicated by
3211 the parameter (in seconds), it will be ignored. If the cookie in
3212 the request has no date, it is accepted and a date will be set.
3213 Cookies that have a date in the future further than 24 hours are
3214 ignored. Doing so lets admins fix timezone issues without risking
3215 kicking users off the site. Contrary to maxidle, this value is
3216 not refreshed, only the first visit date counts. Both maxidle and
3217 maxlife may be used at the time. This is particularly useful to
3218 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003219 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003220 is stronger than the maxidle method in that it forces a
3221 redispatch after some absolute delay.
3222
Olivier Houchard4e694042017-03-14 20:01:29 +01003223 dynamic Activate dynamic cookies. When used, a session cookie is
3224 dynamically created for each server, based on the IP and port
3225 of the server, and a secret key, specified in the
3226 "dynamic-cookie-key" backend directive.
3227 The cookie will be regenerated each time the IP address change,
3228 and is only generated for IPv4/IPv6.
3229
Willy Tarreau0ba27502007-12-24 16:55:16 +01003230 There can be only one persistence cookie per HTTP backend, and it can be
3231 declared in a defaults section. The value of the cookie will be the value
3232 indicated after the "cookie" keyword in a "server" statement. If no cookie
3233 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003234
Willy Tarreau0ba27502007-12-24 16:55:16 +01003235 Examples :
3236 cookie JSESSIONID prefix
3237 cookie SRV insert indirect nocache
3238 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003239 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003240
Willy Tarreau294d0f02015-08-10 19:40:12 +02003241 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003242
Willy Tarreau983e01e2010-01-11 18:42:06 +01003243
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003244declare capture [ request | response ] len <length>
3245 Declares a capture slot.
3246 May be used in sections : defaults | frontend | listen | backend
3247 no | yes | yes | no
3248 Arguments:
3249 <length> is the length allowed for the capture.
3250
3251 This declaration is only available in the frontend or listen section, but the
3252 reserved slot can be used in the backends. The "request" keyword allocates a
3253 capture slot for use in the request, and "response" allocates a capture slot
3254 for use in the response.
3255
3256 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003257 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003258 "http-request capture" and "http-response capture".
3259
3260
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003261default-server [param*]
3262 Change default options for a server in a backend
3263 May be used in sections : defaults | frontend | listen | backend
3264 yes | no | yes | yes
3265 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003266 <param*> is a list of parameters for this server. The "default-server"
3267 keyword accepts an important number of options and has a complete
3268 section dedicated to it. Please refer to section 5 for more
3269 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003270
Willy Tarreau983e01e2010-01-11 18:42:06 +01003271 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003272 default-server inter 1000 weight 13
3273
3274 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003275
Willy Tarreau983e01e2010-01-11 18:42:06 +01003276
Willy Tarreau0ba27502007-12-24 16:55:16 +01003277default_backend <backend>
3278 Specify the backend to use when no "use_backend" rule has been matched.
3279 May be used in sections : defaults | frontend | listen | backend
3280 yes | yes | yes | no
3281 Arguments :
3282 <backend> is the name of the backend to use.
3283
3284 When doing content-switching between frontend and backends using the
3285 "use_backend" keyword, it is often useful to indicate which backend will be
3286 used when no rule has matched. It generally is the dynamic backend which
3287 will catch all undetermined requests.
3288
Willy Tarreau0ba27502007-12-24 16:55:16 +01003289 Example :
3290
3291 use_backend dynamic if url_dyn
3292 use_backend static if url_css url_img extension_img
3293 default_backend dynamic
3294
Willy Tarreau98d04852015-05-26 12:18:29 +02003295 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003296
Willy Tarreau0ba27502007-12-24 16:55:16 +01003297
Baptiste Assmann27f51342013-10-09 06:51:49 +02003298description <string>
3299 Describe a listen, frontend or backend.
3300 May be used in sections : defaults | frontend | listen | backend
3301 no | yes | yes | yes
3302 Arguments : string
3303
3304 Allows to add a sentence to describe the related object in the HAProxy HTML
3305 stats page. The description will be printed on the right of the object name
3306 it describes.
3307 No need to backslash spaces in the <string> arguments.
3308
3309
Willy Tarreau0ba27502007-12-24 16:55:16 +01003310disabled
3311 Disable a proxy, frontend or backend.
3312 May be used in sections : defaults | frontend | listen | backend
3313 yes | yes | yes | yes
3314 Arguments : none
3315
3316 The "disabled" keyword is used to disable an instance, mainly in order to
3317 liberate a listening port or to temporarily disable a service. The instance
3318 will still be created and its configuration will be checked, but it will be
3319 created in the "stopped" state and will appear as such in the statistics. It
3320 will not receive any traffic nor will it send any health-checks or logs. It
3321 is possible to disable many instances at once by adding the "disabled"
3322 keyword in a "defaults" section.
3323
3324 See also : "enabled"
3325
3326
Willy Tarreau5ce94572010-06-07 14:35:41 +02003327dispatch <address>:<port>
3328 Set a default server address
3329 May be used in sections : defaults | frontend | listen | backend
3330 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003331 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003332
3333 <address> is the IPv4 address of the default server. Alternatively, a
3334 resolvable hostname is supported, but this name will be resolved
3335 during start-up.
3336
3337 <ports> is a mandatory port specification. All connections will be sent
3338 to this port, and it is not permitted to use port offsets as is
3339 possible with normal servers.
3340
Willy Tarreau787aed52011-04-15 06:45:37 +02003341 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003342 server can take the connection. In the past it was used to forward non
3343 persistent connections to an auxiliary load balancer. Due to its simple
3344 syntax, it has also been used for simple TCP relays. It is recommended not to
3345 use it for more clarity, and to use the "server" directive instead.
3346
3347 See also : "server"
3348
Olivier Houchard4e694042017-03-14 20:01:29 +01003349
3350dynamic-cookie-key <string>
3351 Set the dynamic cookie secret key for a backend.
3352 May be used in sections : defaults | frontend | listen | backend
3353 yes | no | yes | yes
3354 Arguments : The secret key to be used.
3355
3356 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003357 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003358 specified on the "server" line), using a hash of the IP address of the
3359 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003360 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003361 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003362
Willy Tarreau0ba27502007-12-24 16:55:16 +01003363enabled
3364 Enable a proxy, frontend or backend.
3365 May be used in sections : defaults | frontend | listen | backend
3366 yes | yes | yes | yes
3367 Arguments : none
3368
3369 The "enabled" keyword is used to explicitly enable an instance, when the
3370 defaults has been set to "disabled". This is very rarely used.
3371
3372 See also : "disabled"
3373
3374
3375errorfile <code> <file>
3376 Return a file contents instead of errors generated by HAProxy
3377 May be used in sections : defaults | frontend | listen | backend
3378 yes | yes | yes | yes
3379 Arguments :
3380 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003381 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3382 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003383
3384 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003385 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003386 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003387 error pages, and to use absolute paths, since files are read
3388 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003389
3390 It is important to understand that this keyword is not meant to rewrite
3391 errors returned by the server, but errors detected and returned by HAProxy.
3392 This is why the list of supported errors is limited to a small set.
3393
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003394 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3395
Willy Tarreau0ba27502007-12-24 16:55:16 +01003396 The files are returned verbatim on the TCP socket. This allows any trick such
3397 as redirections to another URL or site, as well as tricks to clean cookies,
3398 force enable or disable caching, etc... The package provides default error
3399 files returning the same contents as default errors.
3400
Willy Tarreau59140a22009-02-22 12:02:30 +01003401 The files should not exceed the configured buffer size (BUFSIZE), which
3402 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003403 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003404 loops between the client and HAProxy when all servers are down, causing an
3405 error to be returned instead of an image. For better HTTP compliance, it is
3406 recommended that all header lines end with CR-LF and not LF alone.
3407
Willy Tarreau0ba27502007-12-24 16:55:16 +01003408 The files are read at the same time as the configuration and kept in memory.
3409 For this reason, the errors continue to be returned even when the process is
3410 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003411 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003412 403 status code and interrogating a blocked URL.
3413
3414 See also : "errorloc", "errorloc302", "errorloc303"
3415
Willy Tarreau59140a22009-02-22 12:02:30 +01003416 Example :
3417 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003418 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003419 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3420 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3421
Willy Tarreau2769aa02007-12-27 18:26:09 +01003422
3423errorloc <code> <url>
3424errorloc302 <code> <url>
3425 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3426 May be used in sections : defaults | frontend | listen | backend
3427 yes | yes | yes | yes
3428 Arguments :
3429 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003430 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3431 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003432
3433 <url> it is the exact contents of the "Location" header. It may contain
3434 either a relative URI to an error page hosted on the same site,
3435 or an absolute URI designating an error page on another site.
3436 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003437 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003438
3439 It is important to understand that this keyword is not meant to rewrite
3440 errors returned by the server, but errors detected and returned by HAProxy.
3441 This is why the list of supported errors is limited to a small set.
3442
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003443 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3444
Willy Tarreau2769aa02007-12-27 18:26:09 +01003445 Note that both keyword return the HTTP 302 status code, which tells the
3446 client to fetch the designated URL using the same HTTP method. This can be
3447 quite problematic in case of non-GET methods such as POST, because the URL
3448 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003449 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003450 status code, indicating to the client that the URL must be fetched with a GET
3451 request.
3452
3453 See also : "errorfile", "errorloc303"
3454
3455
3456errorloc303 <code> <url>
3457 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3458 May be used in sections : defaults | frontend | listen | backend
3459 yes | yes | yes | yes
3460 Arguments :
3461 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003462 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3463 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003464
3465 <url> it is the exact contents of the "Location" header. It may contain
3466 either a relative URI to an error page hosted on the same site,
3467 or an absolute URI designating an error page on another site.
3468 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003469 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003470
3471 It is important to understand that this keyword is not meant to rewrite
3472 errors returned by the server, but errors detected and returned by HAProxy.
3473 This is why the list of supported errors is limited to a small set.
3474
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003475 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3476
Willy Tarreau2769aa02007-12-27 18:26:09 +01003477 Note that both keyword return the HTTP 303 status code, which tells the
3478 client to fetch the designated URL using the same HTTP GET method. This
3479 solves the usual problems associated with "errorloc" and the 302 code. It is
3480 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003481 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003482
3483 See also : "errorfile", "errorloc", "errorloc302"
3484
3485
Simon Horman51a1cf62015-02-03 13:00:44 +09003486email-alert from <emailaddr>
3487 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003488 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003489 May be used in sections: defaults | frontend | listen | backend
3490 yes | yes | yes | yes
3491
3492 Arguments :
3493
3494 <emailaddr> is the from email address to use when sending email alerts
3495
3496 Also requires "email-alert mailers" and "email-alert to" to be set
3497 and if so sending email alerts is enabled for the proxy.
3498
Simon Horman64e34162015-02-06 11:11:57 +09003499 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003500 "email-alert myhostname", "email-alert to", section 3.6 about
3501 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003502
3503
3504email-alert level <level>
3505 Declare the maximum log level of messages for which email alerts will be
3506 sent. This acts as a filter on the sending of email alerts.
3507 May be used in sections: defaults | frontend | listen | backend
3508 yes | yes | yes | yes
3509
3510 Arguments :
3511
3512 <level> One of the 8 syslog levels:
3513 emerg alert crit err warning notice info debug
3514 The above syslog levels are ordered from lowest to highest.
3515
3516 By default level is alert
3517
3518 Also requires "email-alert from", "email-alert mailers" and
3519 "email-alert to" to be set and if so sending email alerts is enabled
3520 for the proxy.
3521
Simon Horman1421e212015-04-30 13:10:35 +09003522 Alerts are sent when :
3523
3524 * An un-paused server is marked as down and <level> is alert or lower
3525 * A paused server is marked as down and <level> is notice or lower
3526 * A server is marked as up or enters the drain state and <level>
3527 is notice or lower
3528 * "option log-health-checks" is enabled, <level> is info or lower,
3529 and a health check status update occurs
3530
Simon Horman64e34162015-02-06 11:11:57 +09003531 See also : "email-alert from", "email-alert mailers",
3532 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003533 section 3.6 about mailers.
3534
3535
3536email-alert mailers <mailersect>
3537 Declare the mailers to be used when sending email alerts
3538 May be used in sections: defaults | frontend | listen | backend
3539 yes | yes | yes | yes
3540
3541 Arguments :
3542
3543 <mailersect> is the name of the mailers section to send email alerts.
3544
3545 Also requires "email-alert from" and "email-alert to" to be set
3546 and if so sending email alerts is enabled for the proxy.
3547
Simon Horman64e34162015-02-06 11:11:57 +09003548 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3549 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003550
3551
3552email-alert myhostname <hostname>
3553 Declare the to hostname address to be used when communicating with
3554 mailers.
3555 May be used in sections: defaults | frontend | listen | backend
3556 yes | yes | yes | yes
3557
3558 Arguments :
3559
Baptiste Assmann738bad92015-12-21 15:27:53 +01003560 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003561
3562 By default the systems hostname is used.
3563
3564 Also requires "email-alert from", "email-alert mailers" and
3565 "email-alert to" to be set and if so sending email alerts is enabled
3566 for the proxy.
3567
Simon Horman64e34162015-02-06 11:11:57 +09003568 See also : "email-alert from", "email-alert level", "email-alert mailers",
3569 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003570
3571
3572email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003573 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003574 header of email alerts. This is the address that email alerts are sent to.
3575 May be used in sections: defaults | frontend | listen | backend
3576 yes | yes | yes | yes
3577
3578 Arguments :
3579
3580 <emailaddr> is the to email address to use when sending email alerts
3581
3582 Also requires "email-alert mailers" and "email-alert to" to be set
3583 and if so sending email alerts is enabled for the proxy.
3584
Simon Horman64e34162015-02-06 11:11:57 +09003585 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003586 "email-alert myhostname", section 3.6 about mailers.
3587
3588
Willy Tarreau4de91492010-01-22 19:10:05 +01003589force-persist { if | unless } <condition>
3590 Declare a condition to force persistence on down servers
3591 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003592 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003593
3594 By default, requests are not dispatched to down servers. It is possible to
3595 force this using "option persist", but it is unconditional and redispatches
3596 to a valid server if "option redispatch" is set. That leaves with very little
3597 possibilities to force some requests to reach a server which is artificially
3598 marked down for maintenance operations.
3599
3600 The "force-persist" statement allows one to declare various ACL-based
3601 conditions which, when met, will cause a request to ignore the down status of
3602 a server and still try to connect to it. That makes it possible to start a
3603 server, still replying an error to the health checks, and run a specially
3604 configured browser to test the service. Among the handy methods, one could
3605 use a specific source IP address, or a specific cookie. The cookie also has
3606 the advantage that it can easily be added/removed on the browser from a test
3607 page. Once the service is validated, it is then possible to open the service
3608 to the world by returning a valid response to health checks.
3609
3610 The forced persistence is enabled when an "if" condition is met, or unless an
3611 "unless" condition is met. The final redispatch is always disabled when this
3612 is used.
3613
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003614 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003615 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003616
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003617
3618filter <name> [param*]
3619 Add the filter <name> in the filter list attached to the proxy.
3620 May be used in sections : defaults | frontend | listen | backend
3621 no | yes | yes | yes
3622 Arguments :
3623 <name> is the name of the filter. Officially supported filters are
3624 referenced in section 9.
3625
Tim Düsterhus4896c442016-11-29 02:15:19 +01003626 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003627 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003628 filter. Please refer to the documentation of the corresponding
3629 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003630
3631 Multiple occurrences of the filter line can be used for the same proxy. The
3632 same filter can be referenced many times if needed.
3633
3634 Example:
3635 listen
3636 bind *:80
3637
3638 filter trace name BEFORE-HTTP-COMP
3639 filter compression
3640 filter trace name AFTER-HTTP-COMP
3641
3642 compression algo gzip
3643 compression offload
3644
3645 server srv1 192.168.0.1:80
3646
3647 See also : section 9.
3648
Willy Tarreau4de91492010-01-22 19:10:05 +01003649
Willy Tarreau2769aa02007-12-27 18:26:09 +01003650fullconn <conns>
3651 Specify at what backend load the servers will reach their maxconn
3652 May be used in sections : defaults | frontend | listen | backend
3653 yes | no | yes | yes
3654 Arguments :
3655 <conns> is the number of connections on the backend which will make the
3656 servers use the maximal number of connections.
3657
Willy Tarreau198a7442008-01-17 12:05:32 +01003658 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003659 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003660 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003661 load. The server will then always accept at least <minconn> connections,
3662 never more than <maxconn>, and the limit will be on the ramp between both
3663 values when the backend has less than <conns> concurrent connections. This
3664 makes it possible to limit the load on the servers during normal loads, but
3665 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003666 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003667
Willy Tarreaufbb78422011-06-05 15:38:35 +02003668 Since it's hard to get this value right, haproxy automatically sets it to
3669 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003670 backend (based on "use_backend" and "default_backend" rules). That way it's
3671 safe to leave it unset. However, "use_backend" involving dynamic names are
3672 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003673
Willy Tarreau2769aa02007-12-27 18:26:09 +01003674 Example :
3675 # The servers will accept between 100 and 1000 concurrent connections each
3676 # and the maximum of 1000 will be reached when the backend reaches 10000
3677 # connections.
3678 backend dynamic
3679 fullconn 10000
3680 server srv1 dyn1:80 minconn 100 maxconn 1000
3681 server srv2 dyn2:80 minconn 100 maxconn 1000
3682
3683 See also : "maxconn", "server"
3684
3685
3686grace <time>
3687 Maintain a proxy operational for some time after a soft stop
3688 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003689 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003690 Arguments :
3691 <time> is the time (by default in milliseconds) for which the instance
3692 will remain operational with the frontend sockets still listening
3693 when a soft-stop is received via the SIGUSR1 signal.
3694
3695 This may be used to ensure that the services disappear in a certain order.
3696 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003697 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003698 needed by the equipment to detect the failure.
3699
3700 Note that currently, there is very little benefit in using this parameter,
3701 and it may in fact complicate the soft-reconfiguration process more than
3702 simplify it.
3703
Willy Tarreau0ba27502007-12-24 16:55:16 +01003704
Andrew Rodland17be45e2016-10-25 17:04:12 -04003705hash-balance-factor <factor>
3706 Specify the balancing factor for bounded-load consistent hashing
3707 May be used in sections : defaults | frontend | listen | backend
3708 yes | no | no | yes
3709 Arguments :
3710 <factor> is the control for the maximum number of concurrent requests to
3711 send to a server, expressed as a percentage of the average number
3712 of concurrent requests across all of the active servers.
3713
3714 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3715 enables an algorithm that prevents any one server from getting too many
3716 requests at once, even if some hash buckets receive many more requests than
3717 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3718 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3719 then no server will be allowed to have a load more than 1.5 times the average.
3720 If server weights are used, they will be respected.
3721
3722 If the first-choice server is disqualified, the algorithm will choose another
3723 server based on the request hash, until a server with additional capacity is
3724 found. A higher <factor> allows more imbalance between the servers, while a
3725 lower <factor> means that more servers will be checked on average, affecting
3726 performance. Reasonable values are from 125 to 200.
3727
Willy Tarreau760e81d2018-05-03 07:20:40 +02003728 This setting is also used by "balance random" which internally relies on the
3729 consistent hashing mechanism.
3730
Andrew Rodland17be45e2016-10-25 17:04:12 -04003731 See also : "balance" and "hash-type".
3732
3733
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003734hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003735 Specify a method to use for mapping hashes to servers
3736 May be used in sections : defaults | frontend | listen | backend
3737 yes | no | yes | yes
3738 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003739 <method> is the method used to select a server from the hash computed by
3740 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003741
Bhaskar98634f02013-10-29 23:30:51 -04003742 map-based the hash table is a static array containing all alive servers.
3743 The hashes will be very smooth, will consider weights, but
3744 will be static in that weight changes while a server is up
3745 will be ignored. This means that there will be no slow start.
3746 Also, since a server is selected by its position in the array,
3747 most mappings are changed when the server count changes. This
3748 means that when a server goes up or down, or when a server is
3749 added to a farm, most connections will be redistributed to
3750 different servers. This can be inconvenient with caches for
3751 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003752
Bhaskar98634f02013-10-29 23:30:51 -04003753 consistent the hash table is a tree filled with many occurrences of each
3754 server. The hash key is looked up in the tree and the closest
3755 server is chosen. This hash is dynamic, it supports changing
3756 weights while the servers are up, so it is compatible with the
3757 slow start feature. It has the advantage that when a server
3758 goes up or down, only its associations are moved. When a
3759 server is added to the farm, only a few part of the mappings
3760 are redistributed, making it an ideal method for caches.
3761 However, due to its principle, the distribution will never be
3762 very smooth and it may sometimes be necessary to adjust a
3763 server's weight or its ID to get a more balanced distribution.
3764 In order to get the same distribution on multiple load
3765 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003766 same IDs. Note: consistent hash uses sdbm and avalanche if no
3767 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003768
3769 <function> is the hash function to be used :
3770
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003771 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003772 reimplementation of ndbm) database library. It was found to do
3773 well in scrambling bits, causing better distribution of the keys
3774 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003775 function with good distribution, unless the total server weight
3776 is a multiple of 64, in which case applying the avalanche
3777 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003778
3779 djb2 this function was first proposed by Dan Bernstein many years ago
3780 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003781 function provides a better distribution than sdbm. It generally
3782 works well with text-based inputs though it can perform extremely
3783 poorly with numeric-only input or when the total server weight is
3784 a multiple of 33, unless the avalanche modifier is also used.
3785
Willy Tarreaua0f42712013-11-14 14:30:35 +01003786 wt6 this function was designed for haproxy while testing other
3787 functions in the past. It is not as smooth as the other ones, but
3788 is much less sensible to the input data set or to the number of
3789 servers. It can make sense as an alternative to sdbm+avalanche or
3790 djb2+avalanche for consistent hashing or when hashing on numeric
3791 data such as a source IP address or a visitor identifier in a URL
3792 parameter.
3793
Willy Tarreau324f07f2015-01-20 19:44:50 +01003794 crc32 this is the most common CRC32 implementation as used in Ethernet,
3795 gzip, PNG, etc. It is slower than the other ones but may provide
3796 a better distribution or less predictable results especially when
3797 used on strings.
3798
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003799 <modifier> indicates an optional method applied after hashing the key :
3800
3801 avalanche This directive indicates that the result from the hash
3802 function above should not be used in its raw form but that
3803 a 4-byte full avalanche hash must be applied first. The
3804 purpose of this step is to mix the resulting bits from the
3805 previous hash in order to avoid any undesired effect when
3806 the input contains some limited values or when the number of
3807 servers is a multiple of one of the hash's components (64
3808 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3809 result less predictable, but it's also not as smooth as when
3810 using the original function. Some testing might be needed
3811 with some workloads. This hash is one of the many proposed
3812 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003813
Bhaskar98634f02013-10-29 23:30:51 -04003814 The default hash type is "map-based" and is recommended for most usages. The
3815 default function is "sdbm", the selection of a function should be based on
3816 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003817
Andrew Rodland17be45e2016-10-25 17:04:12 -04003818 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003819
3820
Willy Tarreau0ba27502007-12-24 16:55:16 +01003821http-check disable-on-404
3822 Enable a maintenance mode upon HTTP/404 response to health-checks
3823 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003824 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003825 Arguments : none
3826
3827 When this option is set, a server which returns an HTTP code 404 will be
3828 excluded from further load-balancing, but will still receive persistent
3829 connections. This provides a very convenient method for Web administrators
3830 to perform a graceful shutdown of their servers. It is also important to note
3831 that a server which is detected as failed while it was in this mode will not
3832 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3833 will immediately be reinserted into the farm. The status on the stats page
3834 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003835 option only works in conjunction with the "httpchk" option. If this option
3836 is used with "http-check expect", then it has precedence over it so that 404
3837 responses will still be considered as soft-stop.
3838
3839 See also : "option httpchk", "http-check expect"
3840
3841
3842http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003843 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003844 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003845 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003846 Arguments :
3847 <match> is a keyword indicating how to look for a specific pattern in the
3848 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003849 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003850 exclamation mark ("!") to negate the match. Spaces are allowed
3851 between the exclamation mark and the keyword. See below for more
3852 details on the supported keywords.
3853
3854 <pattern> is the pattern to look for. It may be a string or a regular
3855 expression. If the pattern contains spaces, they must be escaped
3856 with the usual backslash ('\').
3857
3858 By default, "option httpchk" considers that response statuses 2xx and 3xx
3859 are valid, and that others are invalid. When "http-check expect" is used,
3860 it defines what is considered valid or invalid. Only one "http-check"
3861 statement is supported in a backend. If a server fails to respond or times
3862 out, the check obviously fails. The available matches are :
3863
3864 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003865 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003866 response's status code is exactly this string. If the
3867 "status" keyword is prefixed with "!", then the response
3868 will be considered invalid if the status code matches.
3869
3870 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003871 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003872 response's status code matches the expression. If the
3873 "rstatus" keyword is prefixed with "!", then the response
3874 will be considered invalid if the status code matches.
3875 This is mostly used to check for multiple codes.
3876
3877 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003878 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003879 response's body contains this exact string. If the
3880 "string" keyword is prefixed with "!", then the response
3881 will be considered invalid if the body contains this
3882 string. This can be used to look for a mandatory word at
3883 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003884 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003885 trace).
3886
3887 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003888 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003889 response's body matches this expression. If the "rstring"
3890 keyword is prefixed with "!", then the response will be
3891 considered invalid if the body matches the expression.
3892 This can be used to look for a mandatory word at the end
3893 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003894 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003895
3896 It is important to note that the responses will be limited to a certain size
3897 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3898 Thus, too large responses may not contain the mandatory pattern when using
3899 "string" or "rstring". If a large response is absolutely required, it is
3900 possible to change the default max size by setting the global variable.
3901 However, it is worth keeping in mind that parsing very large responses can
3902 waste some CPU cycles, especially when regular expressions are used, and that
3903 it is always better to focus the checks on smaller resources.
3904
Cyril Bonté32602d22015-01-30 00:07:07 +01003905 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3906 will automatically append a "Connection: close" header, meaning that this
3907 header should not be present in the request provided by "option httpchk".
3908
Willy Tarreaubd741542010-03-16 18:46:54 +01003909 Last, if "http-check expect" is combined with "http-check disable-on-404",
3910 then this last one has precedence when the server responds with 404.
3911
3912 Examples :
3913 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003914 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003915
3916 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003917 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003918
3919 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003920 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003921
3922 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003923 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003924
Willy Tarreaubd741542010-03-16 18:46:54 +01003925 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003926
3927
Willy Tarreauef781042010-01-27 11:53:01 +01003928http-check send-state
3929 Enable emission of a state header with HTTP health checks
3930 May be used in sections : defaults | frontend | listen | backend
3931 yes | no | yes | yes
3932 Arguments : none
3933
3934 When this option is set, haproxy will systematically send a special header
3935 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3936 how they are seen by haproxy. This can be used for instance when a server is
3937 manipulated without access to haproxy and the operator needs to know whether
3938 haproxy still sees it up or not, or if the server is the last one in a farm.
3939
3940 The header is composed of fields delimited by semi-colons, the first of which
3941 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3942 checks on the total number before transition, just as appears in the stats
3943 interface. Next headers are in the form "<variable>=<value>", indicating in
3944 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003945 - a variable "address", containing the address of the backend server.
3946 This corresponds to the <address> field in the server declaration. For
3947 unix domain sockets, it will read "unix".
3948
3949 - a variable "port", containing the port of the backend server. This
3950 corresponds to the <port> field in the server declaration. For unix
3951 domain sockets, it will read "unix".
3952
Willy Tarreauef781042010-01-27 11:53:01 +01003953 - a variable "name", containing the name of the backend followed by a slash
3954 ("/") then the name of the server. This can be used when a server is
3955 checked in multiple backends.
3956
3957 - a variable "node" containing the name of the haproxy node, as set in the
3958 global "node" variable, otherwise the system's hostname if unspecified.
3959
3960 - a variable "weight" indicating the weight of the server, a slash ("/")
3961 and the total weight of the farm (just counting usable servers). This
3962 helps to know if other servers are available to handle the load when this
3963 one fails.
3964
3965 - a variable "scur" indicating the current number of concurrent connections
3966 on the server, followed by a slash ("/") then the total number of
3967 connections on all servers of the same backend.
3968
3969 - a variable "qcur" indicating the current number of requests in the
3970 server's queue.
3971
3972 Example of a header received by the application server :
3973 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
3974 scur=13/22; qcur=0
3975
3976 See also : "option httpchk", "http-check disable-on-404"
3977
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02003978
3979http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003980 Access control for Layer 7 requests
3981
3982 May be used in sections: defaults | frontend | listen | backend
3983 no | yes | yes | yes
3984
Willy Tarreau20b0de52012-12-24 15:45:22 +01003985 The http-request statement defines a set of rules which apply to layer 7
3986 processing. The rules are evaluated in their declaration order when they are
3987 met in a frontend, listen or backend section. Any rule may optionally be
3988 followed by an ACL-based condition, in which case it will only be evaluated
3989 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003990
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02003991 The first keyword is the rule's action. The supported actions are described
3992 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003993
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02003994 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003995
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02003996 It is important to know that http-request rules are processed very early in
3997 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
3998 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
3999 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004000
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004001 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4002 in newer versions (>= 1.5). But if you need to use regular expression to
4003 delete headers, you can still use "reqdel". Also please use
4004 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004005
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004006 Example:
4007 acl nagios src 192.168.129.3
4008 acl local_net src 192.168.0.0/16
4009 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004010
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004011 http-request allow if nagios
4012 http-request allow if local_net auth_ok
4013 http-request auth realm Gimme if local_net auth_ok
4014 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004015
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004016 Example:
4017 acl key req.hdr(X-Add-Acl-Key) -m found
4018 acl add path /addacl
4019 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004020
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004021 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004022
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004023 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4024 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004025
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004026 Example:
4027 acl value req.hdr(X-Value) -m found
4028 acl setmap path /setmap
4029 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004030
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004031 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004032
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004033 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4034 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004035
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004036 See also : "stats http-request", section 3.4 about userlists and section 7
4037 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004038
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004039http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004040
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004041 This is used to add a new entry into an ACL. The ACL must be loaded from a
4042 file (even a dummy empty file). The file name of the ACL to be updated is
4043 passed between parentheses. It takes one argument: <key fmt>, which follows
4044 log-format rules, to collect content of the new entry. It performs a lookup
4045 in the ACL before insertion, to avoid duplicated (or more) values. This
4046 lookup is done by a linear search and can be expensive with large lists!
4047 It is the equivalent of the "add acl" command from the stats socket, but can
4048 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004049
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004050http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004051
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004052 This appends an HTTP header field whose name is specified in <name> and
4053 whose value is defined by <fmt> which follows the log-format rules (see
4054 Custom Log Format in section 8.2.4). This is particularly useful to pass
4055 connection-specific information to the server (e.g. the client's SSL
4056 certificate), or to combine several headers into one. This rule is not
4057 final, so it is possible to add other similar rules. Note that header
4058 addition is performed immediately, so one rule might reuse the resulting
4059 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004060
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004061http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004062
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004063 This stops the evaluation of the rules and lets the request pass the check.
4064 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004065
Sasha Pachev218f0642014-06-16 12:05:59 -06004066
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004067http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004068
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004069 This stops the evaluation of the rules and immediately responds with an
4070 HTTP 401 or 407 error code to invite the user to present a valid user name
4071 and password. No further "http-request" rules are evaluated. An optional
4072 "realm" parameter is supported, it sets the authentication realm that is
4073 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004074
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004075 Example:
4076 acl auth_ok http_auth_group(L1) G1
4077 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004078
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004079http-request cache-use [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004080
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004081 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004082
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004083http-request capture <sample> [ len <length> | id <id> ]
4084 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004085
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004086 This captures sample expression <sample> from the request buffer, and
4087 converts it to a string of at most <len> characters. The resulting string is
4088 stored into the next request "capture" slot, so it will possibly appear next
4089 to some captured HTTP headers. It will then automatically appear in the logs,
4090 and it will be possible to extract it using sample fetch rules to feed it
4091 into headers or anything. The length should be limited given that this size
4092 will be allocated for each capture during the whole session life.
4093 Please check section 7.3 (Fetching samples) and "capture request header" for
4094 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004095
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004096 If the keyword "id" is used instead of "len", the action tries to store the
4097 captured string in a previously declared capture slot. This is useful to run
4098 captures in backends. The slot id can be declared by a previous directive
4099 "http-request capture" or with the "declare capture" keyword. If the slot
4100 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4101 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004102
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004103http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004104
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004105 This is used to delete an entry from an ACL. The ACL must be loaded from a
4106 file (even a dummy empty file). The file name of the ACL to be updated is
4107 passed between parentheses. It takes one argument: <key fmt>, which follows
4108 log-format rules, to collect content of the entry to delete.
4109 It is the equivalent of the "del acl" command from the stats socket, but can
4110 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004111
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004112http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004113
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004114 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004115
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004116http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004117
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004118 This is used to delete an entry from a MAP. The MAP must be loaded from a
4119 file (even a dummy empty file). The file name of the MAP to be updated is
4120 passed between parentheses. It takes one argument: <key fmt>, which follows
4121 log-format rules, to collect content of the entry to delete.
4122 It takes one argument: "file name" It is the equivalent of the "del map"
4123 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004124
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004125http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004126
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004127 This stops the evaluation of the rules and immediately rejects the request
4128 and emits an HTTP 403 error, or optionally the status code specified as an
4129 argument to "deny_status". The list of permitted status codes is limited to
4130 those that can be overridden by the "errorfile" directive.
4131 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004132
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004133http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4134
4135 This is used to build an HTTP 103 Early Hints response prior to any other one.
4136 This appends an HTTP header field to this response whose name is specified in
4137 <name> and whose value is defined by <fmt> which follows the log-format rules
4138 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004139 to the client some Link headers to preload resources required to render the
4140 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004141
4142 See RFC 8297 for more information.
4143
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004144http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004145
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004146 This performs an HTTP redirection based on a redirect rule. This is exactly
4147 the same as the "redirect" statement except that it inserts a redirect rule
4148 which can be processed in the middle of other "http-request" rules and that
4149 these rules use the "log-format" strings. See the "redirect" keyword for the
4150 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004151
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004152http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004153
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004154 This stops the evaluation of the rules and immediately closes the connection
4155 without sending any response. It acts similarly to the
4156 "tcp-request content reject" rules. It can be useful to force an immediate
4157 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004158
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004159http-request replace-header <name> <match-regex> <replace-fmt>
4160 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004161
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004162 This matches the regular expression in all occurrences of header field
4163 <name> according to <match-regex>, and replaces them with the
4164 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4165 work like in <fmt> arguments in "http-request add-header". The match is
4166 only case-sensitive. It is important to understand that this action only
4167 considers whole header lines, regardless of the number of values they may
4168 contain. This usage is suited to headers naturally containing commas in
4169 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004170
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004171 Example:
4172 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004173
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004174 # applied to:
4175 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004176
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004177 # outputs:
4178 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004179
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004180 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004181
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004182http-request replace-value <name> <match-regex> <replace-fmt>
4183 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004184
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004185 This works like "replace-header" except that it matches the regex against
4186 every comma-delimited value of the header field <name> instead of the
4187 entire header. This is suited for all headers which are allowed to carry
4188 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004189
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004190 Example:
4191 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004192
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004193 # applied to:
4194 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004195
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004196 # outputs:
4197 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004198
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004199http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4200http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004201
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004202 This actions increments the GPC0 or GPC1 counter according with the sticky
4203 counter designated by <sc-id>. If an error occurs, this action silently fails
4204 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004205
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004206http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004207
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004208 This action sets the GPT0 tag according to the sticky counter designated by
4209 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4210 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004211
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004212http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004213
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004214 This is used to set the destination IP address to the value of specified
4215 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4216 but provides the correct IP in a HTTP header; or you want to mask the IP for
4217 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4218 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004219
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004220 Arguments:
4221 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4222 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004223
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004224 Example:
4225 http-request set-dst hdr(x-dst)
4226 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004227
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004228 When possible, set-dst preserves the original destination port as long as the
4229 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004230
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004231http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004232
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004233 This is used to set the destination port address to the value of specified
4234 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4235 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004236
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004237 Arguments:
4238 <expr> Is a standard HAProxy expression formed by a sample-fetch
4239 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004240
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004241 Example:
4242 http-request set-dst-port hdr(x-port)
4243 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004244
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004245 When possible, set-dst-port preserves the original destination address as
4246 long as the address family supports a port, otherwise it forces the
4247 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004248
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004249http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004250
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004251 This does the same as "http-request add-header" except that the header name
4252 is first removed if it existed. This is useful when passing security
4253 information to the server, where the header must not be manipulated by
4254 external users. Note that the new value is computed before the removal so it
4255 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004256
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004257 Example:
4258 http-request set-header X-Haproxy-Current-Date %T
4259 http-request set-header X-SSL %[ssl_fc]
4260 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4261 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4262 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4263 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4264 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4265 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4266 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004267
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004268http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004269
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004270 This is used to change the log level of the current request when a certain
4271 condition is met. Valid levels are the 8 syslog levels (see the "log"
4272 keyword) plus the special level "silent" which disables logging for this
4273 request. This rule is not final so the last matching rule wins. This rule
4274 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004275
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004276http-request set-map(<file-name>) <key fmt> <value fmt>
4277 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004278
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004279 This is used to add a new entry into a MAP. The MAP must be loaded from a
4280 file (even a dummy empty file). The file name of the MAP to be updated is
4281 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4282 log-format rules, used to collect MAP key, and <value fmt>, which follows
4283 log-format rules, used to collect content for the new entry.
4284 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4285 more) values. This lookup is done by a linear search and can be expensive
4286 with large lists! It is the equivalent of the "set map" command from the
4287 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004288
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004289http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004290
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004291 This is used to set the Netfilter MARK on all packets sent to the client to
4292 the value passed in <mark> on platforms which support it. This value is an
4293 unsigned 32 bit value which can be matched by netfilter and by the routing
4294 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4295 "0x"). This can be useful to force certain packets to take a different route
4296 (for example a cheaper network path for bulk downloads). This works on Linux
4297 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004298
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004299http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004300
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004301 This rewrites the request method with the result of the evaluation of format
4302 string <fmt>. There should be very few valid reasons for having to do so as
4303 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004304
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004305http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004306
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004307 This sets the "nice" factor of the current request being processed. It only
4308 has effect against the other requests being processed at the same time.
4309 The default value is 0, unless altered by the "nice" setting on the "bind"
4310 line. The accepted range is -1024..1024. The higher the value, the nicest
4311 the request will be. Lower values will make the request more important than
4312 other ones. This can be useful to improve the speed of some requests, or
4313 lower the priority of non-important requests. Using this setting without
4314 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004315
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004316http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004317
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004318 This rewrites the request path with the result of the evaluation of format
4319 string <fmt>. The query string, if any, is left intact. If a scheme and
4320 authority is found before the path, they are left intact as well. If the
4321 request doesn't have a path ("*"), this one is replaced with the format.
4322 This can be used to prepend a directory component in front of a path for
4323 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004324
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004325 Example :
4326 # prepend the host name before the path
4327 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004328
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004329http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004331 This is used to set the queue priority class of the current request.
4332 The value must be a sample expression which converts to an integer in the
4333 range -2047..2047. Results outside this range will be truncated.
4334 The priority class determines the order in which queued requests are
4335 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004336
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004337http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004338
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004339 This is used to set the queue priority timestamp offset of the current
4340 request. The value must be a sample expression which converts to an integer
4341 in the range -524287..524287. Results outside this range will be truncated.
4342 When a request is queued, it is ordered first by the priority class, then by
4343 the current timestamp adjusted by the given offset in milliseconds. Lower
4344 values have higher priority.
4345 Note that the resulting timestamp is is only tracked with enough precision
4346 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4347 adjusted timestamp exceeds this value, it will be misidentified as highest
4348 priority. Thus it is important to set "timeout queue" to a value, where when
4349 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004350
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004351http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004352
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004353 This rewrites the request's query string which appears after the first
4354 question mark ("?") with the result of the evaluation of format string <fmt>.
4355 The part prior to the question mark is left intact. If the request doesn't
4356 contain a question mark and the new value is not empty, then one is added at
4357 the end of the URI, followed by the new value. If a question mark was
4358 present, it will never be removed even if the value is empty. This can be
4359 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004360
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004361 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004362
4363 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004364 # replace "%3D" with "=" in the query string
4365 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004366
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004367http-request set-src <expr> [ { if | unless } <condition> ]
4368 This is used to set the source IP address to the value of specified
4369 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4370 provides the correct IP in a HTTP header; or you want to mask source IP for
4371 privacy.
4372
4373 Arguments :
4374 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4375 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004376
Cyril Bonté78caf842010-03-10 22:41:43 +01004377 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004378 http-request set-src hdr(x-forwarded-for)
4379 http-request set-src src,ipmask(24)
4380
4381 When possible, set-src preserves the original source port as long as the
4382 address family allows it, otherwise the source port is set to 0.
4383
4384http-request set-src-port <expr> [ { if | unless } <condition> ]
4385
4386 This is used to set the source port address to the value of specified
4387 expression.
4388
4389 Arguments:
4390 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4391 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004392
Willy Tarreau20b0de52012-12-24 15:45:22 +01004393 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004394 http-request set-src-port hdr(x-port)
4395 http-request set-src-port int(4000)
4396
4397 When possible, set-src-port preserves the original source address as long as
4398 the address family supports a port, otherwise it forces the source address to
4399 IPv4 "0.0.0.0" before rewriting the port.
4400
4401http-request set-tos <tos> [ { if | unless } <condition> ]
4402
4403 This is used to set the TOS or DSCP field value of packets sent to the client
4404 to the value passed in <tos> on platforms which support this. This value
4405 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4406 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4407 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4408 be used to adjust some routing behavior on border routers based on some
4409 information from the request.
4410
4411 See RFC 2474, 2597, 3260 and 4594 for more information.
4412
4413http-request set-uri <fmt> [ { if | unless } <condition> ]
4414
4415 This rewrites the request URI with the result of the evaluation of format
4416 string <fmt>. The scheme, authority, path and query string are all replaced
4417 at once. This can be used to rewrite hosts in front of proxies, or to
4418 perform complex modifications to the URI such as moving parts between the
4419 path and the query string.
4420 See also "http-request set-path" and "http-request set-query".
4421
4422http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4423
4424 This is used to set the contents of a variable. The variable is declared
4425 inline.
4426
4427 Arguments:
4428 <var-name> The name of the variable starts with an indication about its
4429 scope. The scopes allowed are:
4430 "proc" : the variable is shared with the whole process
4431 "sess" : the variable is shared with the whole session
4432 "txn" : the variable is shared with the transaction
4433 (request and response)
4434 "req" : the variable is shared only during request
4435 processing
4436 "res" : the variable is shared only during response
4437 processing
4438 This prefix is followed by a name. The separator is a '.'.
4439 The name may only contain characters 'a-z', 'A-Z', '0-9'
4440 and '_'.
4441
4442 <expr> Is a standard HAProxy expression formed by a sample-fetch
4443 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004444
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004445 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004446 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004447
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004448http-request send-spoe-group <engine-name> <group-name>
4449 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004450
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004451 This action is used to trigger sending of a group of SPOE messages. To do so,
4452 the SPOE engine used to send messages must be defined, as well as the SPOE
4453 group to send. Of course, the SPOE engine must refer to an existing SPOE
4454 filter. If not engine name is provided on the SPOE filter line, the SPOE
4455 agent name must be used.
4456
4457 Arguments:
4458 <engine-name> The SPOE engine name.
4459
4460 <group-name> The SPOE group name as specified in the engine
4461 configuration.
4462
4463http-request silent-drop [ { if | unless } <condition> ]
4464
4465 This stops the evaluation of the rules and makes the client-facing connection
4466 suddenly disappear using a system-dependent way that tries to prevent the
4467 client from being notified. The effect it then that the client still sees an
4468 established connection while there's none on HAProxy. The purpose is to
4469 achieve a comparable effect to "tarpit" except that it doesn't use any local
4470 resource at all on the machine running HAProxy. It can resist much higher
4471 loads than "tarpit", and slow down stronger attackers. It is important to
4472 understand the impact of using this mechanism. All stateful equipment placed
4473 between the client and HAProxy (firewalls, proxies, load balancers) will also
4474 keep the established connection for a long time and may suffer from this
4475 action.
4476 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4477 option is used to block the emission of a TCP reset. On other systems, the
4478 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4479 router, though it's still delivered to local networks. Do not use it unless
4480 you fully understand how it works.
4481
4482http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4483
4484 This stops the evaluation of the rules and immediately blocks the request
4485 without responding for a delay specified by "timeout tarpit" or
4486 "timeout connect" if the former is not set. After that delay, if the client
4487 is still connected, an HTTP error 500 (or optionally the status code
4488 specified as an argument to "deny_status") is returned so that the client
4489 does not suspect it has been tarpitted. Logs will report the flags "PT".
4490 The goal of the tarpit rule is to slow down robots during an attack when
4491 they're limited on the number of concurrent requests. It can be very
4492 efficient against very dumb robots, and will significantly reduce the load
4493 on firewalls compared to a "deny" rule. But when facing "correctly"
4494 developed robots, it can make things worse by forcing haproxy and the front
4495 firewall to support insane number of concurrent connections.
4496 See also the "silent-drop" action.
4497
4498http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4499http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4500http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4501
4502 This enables tracking of sticky counters from current request. These rules do
4503 not stop evaluation and do not change default action. The number of counters
4504 that may be simultaneously tracked by the same connection is set in
4505 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4506 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4507 "track-sc0" rule executed enables tracking of the counters of the specified
4508 table as the first set. The first "track-sc1" rule executed enables tracking
4509 of the counters of the specified table as the second set. The first
4510 "track-sc2" rule executed enables tracking of the counters of the specified
4511 table as the third set. It is a recommended practice to use the first set of
4512 counters for the per-frontend counters and the second set for the per-backend
4513 ones. But this is just a guideline, all may be used everywhere.
4514
4515 Arguments :
4516 <key> is mandatory, and is a sample expression rule as described in
4517 section 7.3. It describes what elements of the incoming request or
4518 connection will be analyzed, extracted, combined, and used to
4519 select which table entry to update the counters.
4520
4521 <table> is an optional table to be used instead of the default one, which
4522 is the stick-table declared in the current proxy. All the counters
4523 for the matches and updates for the key will then be performed in
4524 that table until the session ends.
4525
4526 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4527 it is not found, an entry is allocated for it. Then a pointer to that entry
4528 is kept during all the session's life, and this entry's counters are updated
4529 as often as possible, every time the session's counters are updated, and also
4530 systematically when the session ends. Counters are only updated for events
4531 that happen after the tracking has been started. As an exception, connection
4532 counters and request counters are systematically updated so that they reflect
4533 useful information.
4534
4535 If the entry tracks concurrent connection counters, one connection is counted
4536 for as long as the entry is tracked, and the entry will not expire during
4537 that time. Tracking counters also provides a performance advantage over just
4538 checking the keys, because only one table lookup is performed for all ACL
4539 checks that make use of it.
4540
4541http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4542
4543 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004544
4545 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004546 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004547
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004548http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004549
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004550 This will delay the processing of the request until the SSL handshake
4551 happened. This is mostly useful to delay processing early data until we're
4552 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004553
Willy Tarreauef781042010-01-27 11:53:01 +01004554
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004555http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004556 Access control for Layer 7 responses
4557
4558 May be used in sections: defaults | frontend | listen | backend
4559 no | yes | yes | yes
4560
4561 The http-response statement defines a set of rules which apply to layer 7
4562 processing. The rules are evaluated in their declaration order when they are
4563 met in a frontend, listen or backend section. Any rule may optionally be
4564 followed by an ACL-based condition, in which case it will only be evaluated
4565 if the condition is true. Since these rules apply on responses, the backend
4566 rules are applied first, followed by the frontend's rules.
4567
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004568 The first keyword is the rule's action. The supported actions are described
4569 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004570
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004571 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004572
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004573 It is important to know that http-response rules are processed very early in
4574 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4575 headers added by "add-header"/"set-header" are visible by almost all further
4576 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004577
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004578 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4579 in newer versions (>= 1.5). But if you need to use regular expression to
4580 delete headers, you can still use "rspdel". Also please use
4581 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004582
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004583 Example:
4584 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004585
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004586 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004587
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004588 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4589 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004590
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004591 Example:
4592 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004593
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004594 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004595
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004596 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4597 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004598
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004599 See also : "http-request", section 3.4 about userlists and section 7 about
4600 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004601
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004602http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004603
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004604 This is used to add a new entry into an ACL. The ACL must be loaded from a
4605 file (even a dummy empty file). The file name of the ACL to be updated is
4606 passed between parentheses. It takes one argument: <key fmt>, which follows
4607 log-format rules, to collect content of the new entry. It performs a lookup
4608 in the ACL before insertion, to avoid duplicated (or more) values.
4609 This lookup is done by a linear search and can be expensive with large lists!
4610 It is the equivalent of the "add acl" command from the stats socket, but can
4611 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004612
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004613http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004614
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004615 This appends an HTTP header field whose name is specified in <name> and whose
4616 value is defined by <fmt> which follows the log-format rules (see Custom Log
4617 Format in section 8.2.4). This may be used to send a cookie to a client for
4618 example, or to pass some internal information.
4619 This rule is not final, so it is possible to add other similar rules.
4620 Note that header addition is performed immediately, so one rule might reuse
4621 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004622
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004623http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004624
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004625 This stops the evaluation of the rules and lets the response pass the check.
4626 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004627
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004628http-response cache-store [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004629
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004630 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004631
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004632http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004633
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004634 This captures sample expression <sample> from the response buffer, and
4635 converts it to a string. The resulting string is stored into the next request
4636 "capture" slot, so it will possibly appear next to some captured HTTP
4637 headers. It will then automatically appear in the logs, and it will be
4638 possible to extract it using sample fetch rules to feed it into headers or
4639 anything. Please check section 7.3 (Fetching samples) and
4640 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004641
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004642 The keyword "id" is the id of the capture slot which is used for storing the
4643 string. The capture slot must be defined in an associated frontend.
4644 This is useful to run captures in backends. The slot id can be declared by a
4645 previous directive "http-response capture" or with the "declare capture"
4646 keyword.
4647 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4648 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004649
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004650http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004651
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004652 This is used to delete an entry from an ACL. The ACL must be loaded from a
4653 file (even a dummy empty file). The file name of the ACL to be updated is
4654 passed between parentheses. It takes one argument: <key fmt>, which follows
4655 log-format rules, to collect content of the entry to delete.
4656 It is the equivalent of the "del acl" command from the stats socket, but can
4657 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004658
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004659http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004660
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004661 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004662
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004663http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004664
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004665 This is used to delete an entry from a MAP. The MAP must be loaded from a
4666 file (even a dummy empty file). The file name of the MAP to be updated is
4667 passed between parentheses. It takes one argument: <key fmt>, which follows
4668 log-format rules, to collect content of the entry to delete.
4669 It takes one argument: "file name" It is the equivalent of the "del map"
4670 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004671
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004672http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004673
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004674 This stops the evaluation of the rules and immediately rejects the response
4675 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004676
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004677http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004678
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004679 This performs an HTTP redirection based on a redirect rule.
4680 This supports a format string similarly to "http-request redirect" rules,
4681 with the exception that only the "location" type of redirect is possible on
4682 the response. See the "redirect" keyword for the rule's syntax. When a
4683 redirect rule is applied during a response, connections to the server are
4684 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004685
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004686http-response replace-header <name> <regex-match> <replace-fmt>
4687 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004688
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004689 This matches the regular expression in all occurrences of header field <name>
4690 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4691 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4692 in "add-header". The match is only case-sensitive. It is important to
4693 understand that this action only considers whole header lines, regardless of
4694 the number of values they may contain. This usage is suited to headers
4695 naturally containing commas in their value, such as Set-Cookie, Expires and
4696 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004697
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004698 Example:
4699 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004700
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004701 # applied to:
4702 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004703
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004704 # outputs:
4705 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004706
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004707 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004708
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004709http-response replace-value <name> <regex-match> <replace-fmt>
4710 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004711
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004712 This works like "replace-header" except that it matches the regex against
4713 every comma-delimited value of the header field <name> instead of the entire
4714 header. This is suited for all headers which are allowed to carry more than
4715 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004716
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004717 Example:
4718 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004719
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004720 # applied to:
4721 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004722
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004723 # outputs:
4724 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004725
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004726http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4727http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004728
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004729 This action increments the GPC0 or GPC1 counter according with the sticky
4730 counter designated by <sc-id>. If an error occurs, this action silently fails
4731 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004732
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004733http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004734
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004735 This action sets the GPT0 tag according to the sticky counter designated by
4736 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4737 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004738
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004739http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004740
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004741 This action is used to trigger sending of a group of SPOE messages. To do so,
4742 the SPOE engine used to send messages must be defined, as well as the SPOE
4743 group to send. Of course, the SPOE engine must refer to an existing SPOE
4744 filter. If not engine name is provided on the SPOE filter line, the SPOE
4745 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004746
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004747 Arguments:
4748 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004749
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004750 <group-name> The SPOE group name as specified in the engine
4751 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004752
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004753http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004754
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004755 This does the same as "add-header" except that the header name is first
4756 removed if it existed. This is useful when passing security information to
4757 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004759http-response set-log-level <level> [ { if | unless } <condition> ]
4760
4761 This is used to change the log level of the current request when a certain
4762 condition is met. Valid levels are the 8 syslog levels (see the "log"
4763 keyword) plus the special level "silent" which disables logging for this
4764 request. This rule is not final so the last matching rule wins. This rule can
4765 be useful to disable health checks coming from another equipment.
4766
4767http-response set-map(<file-name>) <key fmt> <value fmt>
4768
4769 This is used to add a new entry into a MAP. The MAP must be loaded from a
4770 file (even a dummy empty file). The file name of the MAP to be updated is
4771 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4772 log-format rules, used to collect MAP key, and <value fmt>, which follows
4773 log-format rules, used to collect content for the new entry. It performs a
4774 lookup in the MAP before insertion, to avoid duplicated (or more) values.
4775 This lookup is done by a linear search and can be expensive with large lists!
4776 It is the equivalent of the "set map" command from the stats socket, but can
4777 be triggered by an HTTP response.
4778
4779http-response set-mark <mark> [ { if | unless } <condition> ]
4780
4781 This is used to set the Netfilter MARK on all packets sent to the client to
4782 the value passed in <mark> on platforms which support it. This value is an
4783 unsigned 32 bit value which can be matched by netfilter and by the routing
4784 table. It can be expressed both in decimal or hexadecimal format (prefixed
4785 by "0x"). This can be useful to force certain packets to take a different
4786 route (for example a cheaper network path for bulk downloads). This works on
4787 Linux kernels 2.6.32 and above and requires admin privileges.
4788
4789http-response set-nice <nice> [ { if | unless } <condition> ]
4790
4791 This sets the "nice" factor of the current request being processed.
4792 It only has effect against the other requests being processed at the same
4793 time. The default value is 0, unless altered by the "nice" setting on the
4794 "bind" line. The accepted range is -1024..1024. The higher the value, the
4795 nicest the request will be. Lower values will make the request more important
4796 than other ones. This can be useful to improve the speed of some requests, or
4797 lower the priority of non-important requests. Using this setting without
4798 prior experimentation can cause some major slowdown.
4799
4800http-response set-status <status> [reason <str>]
4801 [ { if | unless } <condition> ]
4802
4803 This replaces the response status code with <status> which must be an integer
4804 between 100 and 999. Optionally, a custom reason text can be provided defined
4805 by <str>, or the default reason for the specified code will be used as a
4806 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004807
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004808 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004809 # return "431 Request Header Fields Too Large"
4810 http-response set-status 431
4811 # return "503 Slow Down", custom reason
4812 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004813
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004814http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004815
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004816 This is used to set the TOS or DSCP field value of packets sent to the client
4817 to the value passed in <tos> on platforms which support this.
4818 This value represents the whole 8 bits of the IP TOS field, and can be
4819 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
4820 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
4821 always 0. This can be used to adjust some routing behavior on border routers
4822 based on some information from the request.
4823
4824 See RFC 2474, 2597, 3260 and 4594 for more information.
4825
4826http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4827
4828 This is used to set the contents of a variable. The variable is declared
4829 inline.
4830
4831 Arguments:
4832 <var-name> The name of the variable starts with an indication about its
4833 scope. The scopes allowed are:
4834 "proc" : the variable is shared with the whole process
4835 "sess" : the variable is shared with the whole session
4836 "txn" : the variable is shared with the transaction
4837 (request and response)
4838 "req" : the variable is shared only during request
4839 processing
4840 "res" : the variable is shared only during response
4841 processing
4842 This prefix is followed by a name. The separator is a '.'.
4843 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4844 and '_'.
4845
4846 <expr> Is a standard HAProxy expression formed by a sample-fetch
4847 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004848
4849 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004850 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004851
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004852http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004853
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004854 This stops the evaluation of the rules and makes the client-facing connection
4855 suddenly disappear using a system-dependent way that tries to prevent the
4856 client from being notified. The effect it then that the client still sees an
4857 established connection while there's none on HAProxy. The purpose is to
4858 achieve a comparable effect to "tarpit" except that it doesn't use any local
4859 resource at all on the machine running HAProxy. It can resist much higher
4860 loads than "tarpit", and slow down stronger attackers. It is important to
4861 understand the impact of using this mechanism. All stateful equipment placed
4862 between the client and HAProxy (firewalls, proxies, load balancers) will also
4863 keep the established connection for a long time and may suffer from this
4864 action.
4865 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4866 option is used to block the emission of a TCP reset. On other systems, the
4867 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4868 router, though it's still delivered to local networks. Do not use it unless
4869 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004870
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004871http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4872http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4873http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004874
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004875 This enables tracking of sticky counters from current response. Please refer
4876 to "http-request track-sc" for a complete description. The only difference
4877 from "http-request track-sc" is the <key> sample expression can only make use
4878 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
4879 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
4880 supported, haproxy will fail and warn while parsing the config.
4881
4882http-response unset-var(<var-name>) [ { if | unless } <condition> ]
4883
4884 This is used to unset a variable. See "http-response set-var" for details
4885 about <var-name>.
4886
4887 Example:
4888 http-response unset-var(sess.last_redir)
4889
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004890
Willy Tarreau30631952015-08-06 15:05:24 +02004891http-reuse { never | safe | aggressive | always }
4892 Declare how idle HTTP connections may be shared between requests
4893
4894 May be used in sections: defaults | frontend | listen | backend
4895 yes | no | yes | yes
4896
4897 By default, a connection established between haproxy and the backend server
4898 belongs to the session that initiated it. The downside is that between the
4899 response and the next request, the connection remains idle and is not used.
4900 In many cases for performance reasons it is desirable to make it possible to
4901 reuse these idle connections to serve other requests from different sessions.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004902 This directive allows to tune this behavior.
Willy Tarreau30631952015-08-06 15:05:24 +02004903
4904 The argument indicates the desired connection reuse strategy :
4905
4906 - "never" : idle connections are never shared between sessions. This is
4907 the default choice. It may be enforced to cancel a different
4908 strategy inherited from a defaults section or for
4909 troubleshooting. For example, if an old bogus application
4910 considers that multiple requests over the same connection come
4911 from the same client and it is not possible to fix the
4912 application, it may be desirable to disable connection sharing
4913 in a single backend. An example of such an application could
4914 be an old haproxy using cookie insertion in tunnel mode and
4915 not checking any request past the first one.
4916
4917 - "safe" : this is the recommended strategy. The first request of a
4918 session is always sent over its own connection, and only
4919 subsequent requests may be dispatched over other existing
4920 connections. This ensures that in case the server closes the
4921 connection when the request is being sent, the browser can
4922 decide to silently retry it. Since it is exactly equivalent to
4923 regular keep-alive, there should be no side effects.
4924
4925 - "aggressive" : this mode may be useful in webservices environments where
4926 all servers are not necessarily known and where it would be
4927 appreciable to deliver most first requests over existing
4928 connections. In this case, first requests are only delivered
4929 over existing connections that have been reused at least once,
4930 proving that the server correctly supports connection reuse.
4931 It should only be used when it's sure that the client can
4932 retry a failed request once in a while and where the benefit
4933 of aggressive connection reuse significantly outweights the
4934 downsides of rare connection failures.
4935
4936 - "always" : this mode is only recommended when the path to the server is
4937 known for never breaking existing connections quickly after
4938 releasing them. It allows the first request of a session to be
4939 sent to an existing connection. This can provide a significant
4940 performance increase over the "safe" strategy when the backend
4941 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004942 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004943 sharing. It is recommended that the "http-keep-alive" timeout
4944 remains low in this mode so that no dead connections remain
4945 usable. In most cases, this will lead to the same performance
4946 gains as "aggressive" but with more risks. It should only be
4947 used when it improves the situation over "aggressive".
4948
4949 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004950 connection properties and compatibility. Specifically :
4951 - connections made with "usesrc" followed by a client-dependent value
4952 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004953
4954 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004955 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004956
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02004957 - connections with certain bogus authentication schemes (relying on the
4958 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004959
4960 No connection pool is involved, once a session dies, the last idle connection
4961 it was attached to is deleted at the same time. This ensures that connections
4962 may not last after all sessions are closed.
4963
4964 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4965 because almost no new connection will be established while idle connections
4966 remain available. This is particularly true with the "always" strategy.
4967
4968 See also : "option http-keep-alive", "server maxconn"
4969
4970
Mark Lamourinec2247f02012-01-04 13:02:01 -05004971http-send-name-header [<header>]
4972 Add the server name to a request. Use the header string given by <header>
4973
4974 May be used in sections: defaults | frontend | listen | backend
4975 yes | no | yes | yes
4976
4977 Arguments :
4978
4979 <header> The header string to use to send the server name
4980
4981 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01004982 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05004983 is added with the header string proved.
4984
4985 See also : "server"
4986
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004987id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02004988 Set a persistent ID to a proxy.
4989 May be used in sections : defaults | frontend | listen | backend
4990 no | yes | yes | yes
4991 Arguments : none
4992
4993 Set a persistent ID for the proxy. This ID must be unique and positive.
4994 An unused ID will automatically be assigned if unset. The first assigned
4995 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004996
4997
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004998ignore-persist { if | unless } <condition>
4999 Declare a condition to ignore persistence
5000 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005001 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005002
5003 By default, when cookie persistence is enabled, every requests containing
5004 the cookie are unconditionally persistent (assuming the target server is up
5005 and running).
5006
5007 The "ignore-persist" statement allows one to declare various ACL-based
5008 conditions which, when met, will cause a request to ignore persistence.
5009 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005010 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005011 persistence for a specific User-Agent (for example, some web crawler bots).
5012
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005013 The persistence is ignored when an "if" condition is met, or unless an
5014 "unless" condition is met.
5015
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005016 Example:
5017 acl url_static path_beg /static /images /img /css
5018 acl url_static path_end .gif .png .jpg .css .js
5019 ignore-persist if url_static
5020
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005021 See also : "force-persist", "cookie", and section 7 about ACL usage.
5022
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005023load-server-state-from-file { global | local | none }
5024 Allow seamless reload of HAProxy
5025 May be used in sections: defaults | frontend | listen | backend
5026 yes | no | yes | yes
5027
5028 This directive points HAProxy to a file where server state from previous
5029 running process has been saved. That way, when starting up, before handling
5030 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005031 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005032 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5033 loading state or load states from a file containing all backends and servers.
5034 The state file can be generated by running the command "show servers state"
5035 over the stats socket and redirect output.
5036
Davor Ocelice9ed2812017-12-25 17:49:28 +01005037 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005038 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005039 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005040
5041 Arguments:
5042 global load the content of the file pointed by the global directive
5043 named "server-state-file".
5044
5045 local load the content of the file pointed by the directive
5046 "server-state-file-name" if set. If not set, then the backend
5047 name is used as a file name.
5048
5049 none don't load any stat for this backend
5050
5051 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005052 - server's IP address is preserved across reloads by default, but the
5053 order can be changed thanks to the server's "init-addr" setting. This
5054 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005055 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005056 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005057
5058 - server's weight is applied from previous running process unless it has
5059 has changed between previous and new configuration files.
5060
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005061 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005062
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005063 global
5064 stats socket /tmp/socket
5065 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005066
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005067 defaults
5068 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005069
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005070 backend bk
5071 server s1 127.0.0.1:22 check weight 11
5072 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005073
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005074
5075 Then one can run :
5076
5077 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5078
5079 Content of the file /tmp/server_state would be like this:
5080
5081 1
5082 # <field names skipped for the doc example>
5083 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5084 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5085
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005086 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005087
5088 global
5089 stats socket /tmp/socket
5090 server-state-base /etc/haproxy/states
5091
5092 defaults
5093 load-server-state-from-file local
5094
5095 backend bk
5096 server s1 127.0.0.1:22 check weight 11
5097 server s2 127.0.0.1:22 check weight 12
5098
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005099
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005100 Then one can run :
5101
5102 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5103
5104 Content of the file /etc/haproxy/states/bk would be like this:
5105
5106 1
5107 # <field names skipped for the doc example>
5108 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5109 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5110
5111 See also: "server-state-file", "server-state-file-name", and
5112 "show servers state"
5113
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005114
Willy Tarreau2769aa02007-12-27 18:26:09 +01005115log global
Willy Tarreauadb345d2018-11-12 07:56:13 +01005116log <address> [len <length>] [format <format>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005117no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005118 Enable per-instance logging of events and traffic.
5119 May be used in sections : defaults | frontend | listen | backend
5120 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005121
5122 Prefix :
5123 no should be used when the logger list must be flushed. For example,
5124 if you don't want to inherit from the default logger list. This
5125 prefix does not allow arguments.
5126
Willy Tarreau2769aa02007-12-27 18:26:09 +01005127 Arguments :
5128 global should be used when the instance's logging parameters are the
5129 same as the global ones. This is the most common usage. "global"
5130 replaces <address>, <facility> and <level> with those of the log
5131 entries found in the "global" section. Only one "log global"
5132 statement may be used per instance, and this form takes no other
5133 parameter.
5134
5135 <address> indicates where to send the logs. It takes the same format as
5136 for the "global" section's logs, and can be one of :
5137
5138 - An IPv4 address optionally followed by a colon (':') and a UDP
5139 port. If no port is specified, 514 is used by default (the
5140 standard syslog port).
5141
David du Colombier24bb5f52011-03-17 10:40:23 +01005142 - An IPv6 address followed by a colon (':') and optionally a UDP
5143 port. If no port is specified, 514 is used by default (the
5144 standard syslog port).
5145
Willy Tarreau2769aa02007-12-27 18:26:09 +01005146 - A filesystem path to a UNIX domain socket, keeping in mind
5147 considerations for chroot (be sure the path is accessible
5148 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005149 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005150
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005151 - A file descriptor number in the form "fd@<number>", which may
5152 point to a pipe, terminal, or socket. In this case unbuffered
5153 logs are used and one writev() call per log is performed. This
5154 is a bit expensive but acceptable for most workloads. Messages
5155 sent this way will not be truncated but may be dropped, in
5156 which case the DroppedLogs counter will be incremented. The
5157 writev() call is atomic even on pipes for messages up to
5158 PIPE_BUF size, which POSIX recommends to be at least 512 and
5159 which is 4096 bytes on most modern operating systems. Any
5160 larger message may be interleaved with messages from other
5161 processes. Exceptionally for debugging purposes the file
5162 descriptor may also be directed to a file, but doing so will
5163 significantly slow haproxy down as non-blocking calls will be
5164 ignored. Also there will be no way to purge nor rotate this
5165 file without restarting the process. Note that the configured
5166 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005167 with a TCP syslog server. See also the "short" and "raw"
5168 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005169
5170 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5171 and "fd@2", see above.
5172
5173 You may want to reference some environment variables in the
5174 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005175
Willy Tarreau18324f52014-06-27 18:10:07 +02005176 <length> is an optional maximum line length. Log lines larger than this
5177 value will be truncated before being sent. The reason is that
5178 syslog servers act differently on log line length. All servers
5179 support the default value of 1024, but some servers simply drop
5180 larger lines while others do log them. If a server supports long
5181 lines, it may make sense to set this value here in order to avoid
5182 truncating long lines. Similarly, if a server drops long lines,
5183 it is preferable to truncate them before sending them. Accepted
5184 values are 80 to 65535 inclusive. The default value of 1024 is
5185 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005186 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005187
Willy Tarreauadb345d2018-11-12 07:56:13 +01005188 <format> is the log format used when generating syslog messages. It may be
5189 one of the following :
5190
5191 rfc3164 The RFC3164 syslog message format. This is the default.
5192 (https://tools.ietf.org/html/rfc3164)
5193
5194 rfc5424 The RFC5424 syslog message format.
5195 (https://tools.ietf.org/html/rfc5424)
5196
Willy Tarreaue8746a02018-11-12 08:45:00 +01005197 short A message containing only a level between angle brackets such as
5198 '<3>', followed by the text. The PID, date, time, process name
5199 and system name are omitted. This is designed to be used with a
5200 local log server. This format is compatible with what the
5201 systemd logger consumes.
5202
Willy Tarreauc1b06452018-11-12 11:57:56 +01005203 raw A message containing only the text. The level, PID, date, time,
5204 process name and system name are omitted. This is designed to
5205 be used in containers or during development, where the severity
5206 only depends on the file descriptor used (stdout/stderr).
5207
Willy Tarreau2769aa02007-12-27 18:26:09 +01005208 <facility> must be one of the 24 standard syslog facilities :
5209
Willy Tarreaue8746a02018-11-12 08:45:00 +01005210 kern user mail daemon auth syslog lpr news
5211 uucp cron auth2 ftp ntp audit alert cron2
5212 local0 local1 local2 local3 local4 local5 local6 local7
5213
Willy Tarreauc1b06452018-11-12 11:57:56 +01005214 Note that the facility is ignored for the "short" and "raw"
5215 formats, but still required as a positional field. It is
5216 recommended to use "daemon" in this case to make it clear that
5217 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005218
5219 <level> is optional and can be specified to filter outgoing messages. By
5220 default, all messages are sent. If a level is specified, only
5221 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005222 will be sent. An optional minimum level can be specified. If it
5223 is set, logs emitted with a more severe level than this one will
5224 be capped to this level. This is used to avoid sending "emerg"
5225 messages on all terminals on some default syslog configurations.
5226 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005227
5228 emerg alert crit err warning notice info debug
5229
William Lallemand0f99e342011-10-12 17:50:54 +02005230 It is important to keep in mind that it is the frontend which decides what to
5231 log from a connection, and that in case of content switching, the log entries
5232 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005233
5234 However, backend log declaration define how and where servers status changes
5235 will be logged. Level "notice" will be used to indicate a server going up,
5236 "warning" will be used for termination signals and definitive service
5237 termination, and "alert" will be used for when a server goes down.
5238
5239 Note : According to RFC3164, messages are truncated to 1024 bytes before
5240 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005241
5242 Example :
5243 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005244 log stdout format short daemon # send log to systemd
5245 log stdout format raw daemon # send everything to stdout
5246 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005247 log 127.0.0.1:514 local0 notice # only send important events
5248 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005249 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005250
Willy Tarreau2769aa02007-12-27 18:26:09 +01005251
William Lallemand48940402012-01-30 16:47:22 +01005252log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005253 Specifies the log format string to use for traffic logs
5254 May be used in sections: defaults | frontend | listen | backend
5255 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005256
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005257 This directive specifies the log format string that will be used for all logs
5258 resulting from traffic passing through the frontend using this line. If the
5259 directive is used in a defaults section, all subsequent frontends will use
5260 the same log format. Please see section 8.2.4 which covers the log format
5261 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005262
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005263 "log-format" directive overrides previous "option tcplog", "log-format" and
5264 "option httplog" directives.
5265
Dragan Dosen7ad31542015-09-28 17:16:47 +02005266log-format-sd <string>
5267 Specifies the RFC5424 structured-data log format string
5268 May be used in sections: defaults | frontend | listen | backend
5269 yes | yes | yes | no
5270
5271 This directive specifies the RFC5424 structured-data log format string that
5272 will be used for all logs resulting from traffic passing through the frontend
5273 using this line. If the directive is used in a defaults section, all
5274 subsequent frontends will use the same log format. Please see section 8.2.4
5275 which covers the log format string in depth.
5276
5277 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5278 about the RFC5424 structured-data part.
5279
5280 Note : This log format string will be used only for loggers that have set
5281 log format to "rfc5424".
5282
5283 Example :
5284 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5285
5286
Willy Tarreau094af4e2015-01-07 15:03:42 +01005287log-tag <string>
5288 Specifies the log tag to use for all outgoing logs
5289 May be used in sections: defaults | frontend | listen | backend
5290 yes | yes | yes | yes
5291
5292 Sets the tag field in the syslog header to this string. It defaults to the
5293 log-tag set in the global section, otherwise the program name as launched
5294 from the command line, which usually is "haproxy". Sometimes it can be useful
5295 to differentiate between multiple processes running on the same host, or to
5296 differentiate customer instances running in the same process. In the backend,
5297 logs about servers up/down will use this tag. As a hint, it can be convenient
5298 to set a log-tag related to a hosted customer in a defaults section then put
5299 all the frontends and backends for that customer, then start another customer
5300 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005301
Willy Tarreauc35362a2014-04-25 13:58:37 +02005302max-keep-alive-queue <value>
5303 Set the maximum server queue size for maintaining keep-alive connections
5304 May be used in sections: defaults | frontend | listen | backend
5305 yes | no | yes | yes
5306
5307 HTTP keep-alive tries to reuse the same server connection whenever possible,
5308 but sometimes it can be counter-productive, for example if a server has a lot
5309 of connections while other ones are idle. This is especially true for static
5310 servers.
5311
5312 The purpose of this setting is to set a threshold on the number of queued
5313 connections at which haproxy stops trying to reuse the same server and prefers
5314 to find another one. The default value, -1, means there is no limit. A value
5315 of zero means that keep-alive requests will never be queued. For very close
5316 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005317 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005318 use a value of 10 or less). For remote servers suffering from a high latency,
5319 higher values might be needed to cover for the latency and/or the cost of
5320 picking a different server.
5321
5322 Note that this has no impact on responses which are maintained to the same
5323 server consecutively to a 401 response. They will still go to the same server
5324 even if they have to be queued.
5325
5326 See also : "option http-server-close", "option prefer-last-server", server
5327 "maxconn" and cookie persistence.
5328
5329
Willy Tarreau2769aa02007-12-27 18:26:09 +01005330maxconn <conns>
5331 Fix the maximum number of concurrent connections on a frontend
5332 May be used in sections : defaults | frontend | listen | backend
5333 yes | yes | yes | no
5334 Arguments :
5335 <conns> is the maximum number of concurrent connections the frontend will
5336 accept to serve. Excess connections will be queued by the system
5337 in the socket's listen queue and will be served once a connection
5338 closes.
5339
5340 If the system supports it, it can be useful on big sites to raise this limit
5341 very high so that haproxy manages connection queues, instead of leaving the
5342 clients with unanswered connection attempts. This value should not exceed the
5343 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005344 of tune.bufsize (16kB by default) each, as well as some other data resulting
5345 in about 33 kB of RAM being consumed per established connection. That means
5346 that a medium system equipped with 1GB of RAM can withstand around
5347 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005348
5349 Also, when <conns> is set to large values, it is possible that the servers
5350 are not sized to accept such loads, and for this reason it is generally wise
5351 to assign them some reasonable connection limits.
5352
Vincent Bernat6341be52012-06-27 17:18:30 +02005353 By default, this value is set to 2000.
5354
Willy Tarreau2769aa02007-12-27 18:26:09 +01005355 See also : "server", global section's "maxconn", "fullconn"
5356
5357
5358mode { tcp|http|health }
5359 Set the running mode or protocol of the instance
5360 May be used in sections : defaults | frontend | listen | backend
5361 yes | yes | yes | yes
5362 Arguments :
5363 tcp The instance will work in pure TCP mode. A full-duplex connection
5364 will be established between clients and servers, and no layer 7
5365 examination will be performed. This is the default mode. It
5366 should be used for SSL, SSH, SMTP, ...
5367
5368 http The instance will work in HTTP mode. The client request will be
5369 analyzed in depth before connecting to any server. Any request
5370 which is not RFC-compliant will be rejected. Layer 7 filtering,
5371 processing and switching will be possible. This is the mode which
5372 brings HAProxy most of its value.
5373
5374 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005375 to incoming connections and close the connection. Alternatively,
5376 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5377 instead. Nothing will be logged in either case. This mode is used
5378 to reply to external components health checks. This mode is
5379 deprecated and should not be used anymore as it is possible to do
5380 the same and even better by combining TCP or HTTP modes with the
5381 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005382
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005383 When doing content switching, it is mandatory that the frontend and the
5384 backend are in the same mode (generally HTTP), otherwise the configuration
5385 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005386
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005387 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005388 defaults http_instances
5389 mode http
5390
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005391 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005392
Willy Tarreau0ba27502007-12-24 16:55:16 +01005393
Cyril Bontéf0c60612010-02-06 14:44:47 +01005394monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005395 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005396 May be used in sections : defaults | frontend | listen | backend
5397 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005398 Arguments :
5399 if <cond> the monitor request will fail if the condition is satisfied,
5400 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005401 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005402 are met, for instance a low number of servers both in a
5403 backend and its backup.
5404
5405 unless <cond> the monitor request will succeed only if the condition is
5406 satisfied, and will fail otherwise. Such a condition may be
5407 based on a test on the presence of a minimum number of active
5408 servers in a list of backends.
5409
5410 This statement adds a condition which can force the response to a monitor
5411 request to report a failure. By default, when an external component queries
5412 the URI dedicated to monitoring, a 200 response is returned. When one of the
5413 conditions above is met, haproxy will return 503 instead of 200. This is
5414 very useful to report a site failure to an external component which may base
5415 routing advertisements between multiple sites on the availability reported by
5416 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005417 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5418 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005419
5420 Example:
5421 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005422 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005423 acl site_dead nbsrv(dynamic) lt 2
5424 acl site_dead nbsrv(static) lt 2
5425 monitor-uri /site_alive
5426 monitor fail if site_dead
5427
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005428 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005429
5430
5431monitor-net <source>
5432 Declare a source network which is limited to monitor requests
5433 May be used in sections : defaults | frontend | listen | backend
5434 yes | yes | yes | no
5435 Arguments :
5436 <source> is the source IPv4 address or network which will only be able to
5437 get monitor responses to any request. It can be either an IPv4
5438 address, a host name, or an address followed by a slash ('/')
5439 followed by a mask.
5440
5441 In TCP mode, any connection coming from a source matching <source> will cause
5442 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005443 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005444 forwarding the connection to a remote server.
5445
5446 In HTTP mode, a connection coming from a source matching <source> will be
5447 accepted, the following response will be sent without waiting for a request,
5448 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5449 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005450 running without forwarding the request to a backend server. Note that this
5451 response is sent in raw format, without any transformation. This is important
5452 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005453
Willy Tarreau82569f92012-09-27 23:48:56 +02005454 Monitor requests are processed very early, just after tcp-request connection
5455 ACLs which are the only ones able to block them. These connections are short
5456 lived and never wait for any data from the client. They cannot be logged, and
5457 it is the intended purpose. They are only used to report HAProxy's health to
5458 an upper component, nothing more. Please note that "monitor fail" rules do
5459 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005460
Willy Tarreau95cd2832010-03-04 23:36:33 +01005461 Last, please note that only one "monitor-net" statement can be specified in
5462 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005463
Willy Tarreau2769aa02007-12-27 18:26:09 +01005464 Example :
5465 # addresses .252 and .253 are just probing us.
5466 frontend www
5467 monitor-net 192.168.0.252/31
5468
5469 See also : "monitor fail", "monitor-uri"
5470
5471
5472monitor-uri <uri>
5473 Intercept a URI used by external components' monitor requests
5474 May be used in sections : defaults | frontend | listen | backend
5475 yes | yes | yes | no
5476 Arguments :
5477 <uri> is the exact URI which we want to intercept to return HAProxy's
5478 health status instead of forwarding the request.
5479
5480 When an HTTP request referencing <uri> will be received on a frontend,
5481 HAProxy will not forward it nor log it, but instead will return either
5482 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5483 conditions defined with "monitor fail". This is normally enough for any
5484 front-end HTTP probe to detect that the service is UP and running without
5485 forwarding the request to a backend server. Note that the HTTP method, the
5486 version and all headers are ignored, but the request must at least be valid
5487 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5488
Willy Tarreau721d8e02017-12-01 18:25:08 +01005489 Monitor requests are processed very early, just after the request is parsed
5490 and even before any "http-request" or "block" rulesets. The only rulesets
5491 applied before are the tcp-request ones. They cannot be logged either, and it
5492 is the intended purpose. They are only used to report HAProxy's health to an
5493 upper component, nothing more. However, it is possible to add any number of
5494 conditions using "monitor fail" and ACLs so that the result can be adjusted
5495 to whatever check can be imagined (most often the number of available servers
5496 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005497
5498 Example :
5499 # Use /haproxy_test to report haproxy's status
5500 frontend www
5501 mode http
5502 monitor-uri /haproxy_test
5503
5504 See also : "monitor fail", "monitor-net"
5505
Willy Tarreau0ba27502007-12-24 16:55:16 +01005506
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005507option abortonclose
5508no option abortonclose
5509 Enable or disable early dropping of aborted requests pending in queues.
5510 May be used in sections : defaults | frontend | listen | backend
5511 yes | no | yes | yes
5512 Arguments : none
5513
5514 In presence of very high loads, the servers will take some time to respond.
5515 The per-instance connection queue will inflate, and the response time will
5516 increase respective to the size of the queue times the average per-session
5517 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005518 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005519 the queue, and slowing down other users, and the servers as well, because the
5520 request will eventually be served, then aborted at the first error
5521 encountered while delivering the response.
5522
5523 As there is no way to distinguish between a full STOP and a simple output
5524 close on the client side, HTTP agents should be conservative and consider
5525 that the client might only have closed its output channel while waiting for
5526 the response. However, this introduces risks of congestion when lots of users
5527 do the same, and is completely useless nowadays because probably no client at
5528 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005529 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005530 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005531 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005532 of being the single component to break rare but valid traffic is extremely
5533 low, which adds to the temptation to be able to abort a session early while
5534 still not served and not pollute the servers.
5535
Davor Ocelice9ed2812017-12-25 17:49:28 +01005536 In HAProxy, the user can choose the desired behavior using the option
5537 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005538 compliant and aborted requests will be served. But when the option is
5539 specified, a session with an incoming channel closed will be aborted while
5540 it is still possible, either pending in the queue for a connection slot, or
5541 during the connection establishment if the server has not yet acknowledged
5542 the connection request. This considerably reduces the queue size and the load
5543 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005544 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005545
5546 If this option has been enabled in a "defaults" section, it can be disabled
5547 in a specific instance by prepending the "no" keyword before it.
5548
5549 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5550
5551
Willy Tarreau4076a152009-04-02 15:18:36 +02005552option accept-invalid-http-request
5553no option accept-invalid-http-request
5554 Enable or disable relaxing of HTTP request parsing
5555 May be used in sections : defaults | frontend | listen | backend
5556 yes | yes | yes | no
5557 Arguments : none
5558
Willy Tarreau91852eb2015-05-01 13:26:00 +02005559 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005560 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005561 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005562 forbidden characters are essentially used to build attacks exploiting server
5563 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5564 server will emit invalid header names for whatever reason (configuration,
5565 implementation) and the issue will not be immediately fixed. In such a case,
5566 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005567 even if that does not make sense, by specifying this option. Similarly, the
5568 list of characters allowed to appear in a URI is well defined by RFC3986, and
5569 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5570 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005571 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005572 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005573 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5574 to pass through (no version specified) and multiple digits for both the major
5575 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005576
5577 This option should never be enabled by default as it hides application bugs
5578 and open security breaches. It should only be deployed after a problem has
5579 been confirmed.
5580
5581 When this option is enabled, erroneous header names will still be accepted in
5582 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005583 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5584 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005585 also helps confirming that the issue has been solved.
5586
5587 If this option has been enabled in a "defaults" section, it can be disabled
5588 in a specific instance by prepending the "no" keyword before it.
5589
5590 See also : "option accept-invalid-http-response" and "show errors" on the
5591 stats socket.
5592
5593
5594option accept-invalid-http-response
5595no option accept-invalid-http-response
5596 Enable or disable relaxing of HTTP response parsing
5597 May be used in sections : defaults | frontend | listen | backend
5598 yes | no | yes | yes
5599 Arguments : none
5600
Willy Tarreau91852eb2015-05-01 13:26:00 +02005601 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005602 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005603 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005604 forbidden characters are essentially used to build attacks exploiting server
5605 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5606 server will emit invalid header names for whatever reason (configuration,
5607 implementation) and the issue will not be immediately fixed. In such a case,
5608 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005609 even if that does not make sense, by specifying this option. This option also
5610 relaxes the test on the HTTP version format, it allows multiple digits for
5611 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005612
5613 This option should never be enabled by default as it hides application bugs
5614 and open security breaches. It should only be deployed after a problem has
5615 been confirmed.
5616
5617 When this option is enabled, erroneous header names will still be accepted in
5618 responses, but the complete response will be captured in order to permit
5619 later analysis using the "show errors" request on the UNIX stats socket.
5620 Doing this also helps confirming that the issue has been solved.
5621
5622 If this option has been enabled in a "defaults" section, it can be disabled
5623 in a specific instance by prepending the "no" keyword before it.
5624
5625 See also : "option accept-invalid-http-request" and "show errors" on the
5626 stats socket.
5627
5628
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005629option allbackups
5630no option allbackups
5631 Use either all backup servers at a time or only the first one
5632 May be used in sections : defaults | frontend | listen | backend
5633 yes | no | yes | yes
5634 Arguments : none
5635
5636 By default, the first operational backup server gets all traffic when normal
5637 servers are all down. Sometimes, it may be preferred to use multiple backups
5638 at once, because one will not be enough. When "option allbackups" is enabled,
5639 the load balancing will be performed among all backup servers when all normal
5640 ones are unavailable. The same load balancing algorithm will be used and the
5641 servers' weights will be respected. Thus, there will not be any priority
5642 order between the backup servers anymore.
5643
5644 This option is mostly used with static server farms dedicated to return a
5645 "sorry" page when an application is completely offline.
5646
5647 If this option has been enabled in a "defaults" section, it can be disabled
5648 in a specific instance by prepending the "no" keyword before it.
5649
5650
5651option checkcache
5652no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005653 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005654 May be used in sections : defaults | frontend | listen | backend
5655 yes | no | yes | yes
5656 Arguments : none
5657
5658 Some high-level frameworks set application cookies everywhere and do not
5659 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005660 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005661 high risk of session crossing or stealing between users traversing the same
5662 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005663 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005664
5665 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005666 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005667 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005668 response to check if there's a risk of caching a cookie on a client-side
5669 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005670 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005671 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005672 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5673 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005674 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005675 - all those that result from a request using a method other than GET, HEAD,
5676 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005677 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005678 - those with a 'Pragma: no-cache' header
5679 - those with a 'Cache-control: private' header
5680 - those with a 'Cache-control: no-store' header
5681 - those with a 'Cache-control: max-age=0' header
5682 - those with a 'Cache-control: s-maxage=0' header
5683 - those with a 'Cache-control: no-cache' header
5684 - those with a 'Cache-control: no-cache="set-cookie"' header
5685 - those with a 'Cache-control: no-cache="set-cookie,' header
5686 (allowing other fields after set-cookie)
5687
5688 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005689 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005690 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005691 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005692 that admins are informed that there's something to be fixed.
5693
5694 Due to the high impact on the application, the application should be tested
5695 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005696 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005697 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005698
5699 If this option has been enabled in a "defaults" section, it can be disabled
5700 in a specific instance by prepending the "no" keyword before it.
5701
5702
5703option clitcpka
5704no option clitcpka
5705 Enable or disable the sending of TCP keepalive packets on the client side
5706 May be used in sections : defaults | frontend | listen | backend
5707 yes | yes | yes | no
5708 Arguments : none
5709
5710 When there is a firewall or any session-aware component between a client and
5711 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005712 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005713 components decides to expire a session which has remained idle for too long.
5714
5715 Enabling socket-level TCP keep-alives makes the system regularly send packets
5716 to the other end of the connection, leaving it active. The delay between
5717 keep-alive probes is controlled by the system only and depends both on the
5718 operating system and its tuning parameters.
5719
5720 It is important to understand that keep-alive packets are neither emitted nor
5721 received at the application level. It is only the network stacks which sees
5722 them. For this reason, even if one side of the proxy already uses keep-alives
5723 to maintain its connection alive, those keep-alive packets will not be
5724 forwarded to the other side of the proxy.
5725
5726 Please note that this has nothing to do with HTTP keep-alive.
5727
5728 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5729 client side of a connection, which should help when session expirations are
5730 noticed between HAProxy and a client.
5731
5732 If this option has been enabled in a "defaults" section, it can be disabled
5733 in a specific instance by prepending the "no" keyword before it.
5734
5735 See also : "option srvtcpka", "option tcpka"
5736
5737
Willy Tarreau0ba27502007-12-24 16:55:16 +01005738option contstats
5739 Enable continuous traffic statistics updates
5740 May be used in sections : defaults | frontend | listen | backend
5741 yes | yes | yes | no
5742 Arguments : none
5743
5744 By default, counters used for statistics calculation are incremented
5745 only when a session finishes. It works quite well when serving small
5746 objects, but with big ones (for example large images or archives) or
5747 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005748 a hedgehog. With this option enabled counters get incremented frequently
5749 along the session, typically every 5 seconds, which is often enough to
5750 produce clean graphs. Recounting touches a hotpath directly so it is not
5751 not enabled by default, as it can cause a lot of wakeups for very large
5752 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005753
5754
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005755option dontlog-normal
5756no option dontlog-normal
5757 Enable or disable logging of normal, successful connections
5758 May be used in sections : defaults | frontend | listen | backend
5759 yes | yes | yes | no
5760 Arguments : none
5761
5762 There are large sites dealing with several thousand connections per second
5763 and for which logging is a major pain. Some of them are even forced to turn
5764 logs off and cannot debug production issues. Setting this option ensures that
5765 normal connections, those which experience no error, no timeout, no retry nor
5766 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5767 mode, the response status code is checked and return codes 5xx will still be
5768 logged.
5769
5770 It is strongly discouraged to use this option as most of the time, the key to
5771 complex issues is in the normal logs which will not be logged here. If you
5772 need to separate logs, see the "log-separate-errors" option instead.
5773
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005774 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005775 logging.
5776
5777
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005778option dontlognull
5779no option dontlognull
5780 Enable or disable logging of null connections
5781 May be used in sections : defaults | frontend | listen | backend
5782 yes | yes | yes | no
5783 Arguments : none
5784
5785 In certain environments, there are components which will regularly connect to
5786 various systems to ensure that they are still alive. It can be the case from
5787 another load balancer as well as from monitoring systems. By default, even a
5788 simple port probe or scan will produce a log. If those connections pollute
5789 the logs too much, it is possible to enable option "dontlognull" to indicate
5790 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005791 which typically corresponds to those probes. Note that errors will still be
5792 returned to the client and accounted for in the stats. If this is not what is
5793 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005794
5795 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005796 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005797 would not be logged.
5798
5799 If this option has been enabled in a "defaults" section, it can be disabled
5800 in a specific instance by prepending the "no" keyword before it.
5801
Willy Tarreau0f228a02015-05-01 15:37:53 +02005802 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5803 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005804
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005805
Christopher Faulet315b39c2018-09-21 16:26:19 +02005806option forceclose (deprecated)
5807no option forceclose (deprecated)
5808 This is an alias for "option httpclose". Thus this option is deprecated.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005809
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005810 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005811
5812
Willy Tarreau87cf5142011-08-19 22:57:24 +02005813option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005814 Enable insertion of the X-Forwarded-For header to requests sent to servers
5815 May be used in sections : defaults | frontend | listen | backend
5816 yes | yes | yes | yes
5817 Arguments :
5818 <network> is an optional argument used to disable this option for sources
5819 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005820 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005821 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005822
5823 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5824 their client address. This is sometimes annoying when the client's IP address
5825 is expected in server logs. To solve this problem, the well-known HTTP header
5826 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5827 This header contains a value representing the client's IP address. Since this
5828 header is always appended at the end of the existing header list, the server
5829 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005830 the server's manual to find how to enable use of this standard header. Note
5831 that only the last occurrence of the header must be used, since it is really
5832 possible that the client has already brought one.
5833
Willy Tarreaud72758d2010-01-12 10:42:19 +01005834 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005835 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005836 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005837 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005838 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005839 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005840
5841 Sometimes, a same HAProxy instance may be shared between a direct client
5842 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5843 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5844 header for a known source address or network by adding the "except" keyword
5845 followed by the network address. In this case, any source IP matching the
5846 network will not cause an addition of this header. Most common uses are with
5847 private networks or 127.0.0.1.
5848
Willy Tarreau87cf5142011-08-19 22:57:24 +02005849 Alternatively, the keyword "if-none" states that the header will only be
5850 added if it is not present. This should only be used in perfectly trusted
5851 environment, as this might cause a security issue if headers reaching haproxy
5852 are under the control of the end-user.
5853
Willy Tarreauc27debf2008-01-06 08:57:02 +01005854 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005855 least one of them uses it, the header will be added. Note that the backend's
5856 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005857 both are defined. In the case of the "if-none" argument, if at least one of
5858 the frontend or the backend does not specify it, it wants the addition to be
5859 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005860
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005861 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005862 # Public HTTP address also used by stunnel on the same machine
5863 frontend www
5864 mode http
5865 option forwardfor except 127.0.0.1 # stunnel already adds the header
5866
Ross Westaf72a1d2008-08-03 10:51:45 +02005867 # Those servers want the IP Address in X-Client
5868 backend www
5869 mode http
5870 option forwardfor header X-Client
5871
Willy Tarreau87cf5142011-08-19 22:57:24 +02005872 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005873 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005874
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005875
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005876option http-buffer-request
5877no option http-buffer-request
5878 Enable or disable waiting for whole HTTP request body before proceeding
5879 May be used in sections : defaults | frontend | listen | backend
5880 yes | yes | yes | yes
5881 Arguments : none
5882
5883 It is sometimes desirable to wait for the body of an HTTP request before
5884 taking a decision. This is what is being done by "balance url_param" for
5885 example. The first use case is to buffer requests from slow clients before
5886 connecting to the server. Another use case consists in taking the routing
5887 decision based on the request body's contents. This option placed in a
5888 frontend or backend forces the HTTP processing to wait until either the whole
5889 body is received, or the request buffer is full, or the first chunk is
5890 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005891 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005892 the frontend and the backend, so this should definitely not be used by
5893 default.
5894
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005895 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005896
5897
Willy Tarreau0f228a02015-05-01 15:37:53 +02005898option http-ignore-probes
5899no option http-ignore-probes
5900 Enable or disable logging of null connections and request timeouts
5901 May be used in sections : defaults | frontend | listen | backend
5902 yes | yes | yes | no
5903 Arguments : none
5904
5905 Recently some browsers started to implement a "pre-connect" feature
5906 consisting in speculatively connecting to some recently visited web sites
5907 just in case the user would like to visit them. This results in many
5908 connections being established to web sites, which end up in 408 Request
5909 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5910 decides to close them first. These ones pollute the log and feed the error
5911 counters. There was already "option dontlognull" but it's insufficient in
5912 this case. Instead, this option does the following things :
5913 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005914 was received over a connection before it was closed;
5915 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005916 - prevent any error counter from being incremented
5917
5918 That way the empty connection is silently ignored. Note that it is better
5919 not to use this unless it is clear that it is needed, because it will hide
5920 real problems. The most common reason for not receiving a request and seeing
5921 a 408 is due to an MTU inconsistency between the client and an intermediary
5922 element such as a VPN, which blocks too large packets. These issues are
5923 generally seen with POST requests as well as GET with large cookies. The logs
5924 are often the only way to detect them.
5925
5926 If this option has been enabled in a "defaults" section, it can be disabled
5927 in a specific instance by prepending the "no" keyword before it.
5928
5929 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5930
5931
Willy Tarreau16bfb022010-01-16 19:48:41 +01005932option http-keep-alive
5933no option http-keep-alive
5934 Enable or disable HTTP keep-alive from client to server
5935 May be used in sections : defaults | frontend | listen | backend
5936 yes | yes | yes | yes
5937 Arguments : none
5938
Willy Tarreau70dffda2014-01-30 03:07:23 +01005939 By default HAProxy operates in keep-alive mode with regards to persistent
5940 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02005941 leaves the connection idle on both sides between the end of a response and
5942 the start of a new request. This mode may be changed by several options such
5943 as "option http-server-close", "option httpclose" or "option http-tunnel".
5944 This option allows to set back the keep-alive mode, which can be useful when
5945 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01005946
5947 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5948 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005949 network) and the fastest session reuse on the server side at the expense
5950 of maintaining idle connections to the servers. In general, it is possible
5951 with this option to achieve approximately twice the request rate that the
5952 "http-server-close" option achieves on small objects. There are mainly two
5953 situations where this option may be useful :
5954
5955 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005956 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005957
5958 - when the cost of establishing the connection to the server is significant
5959 compared to the cost of retrieving the associated object from the server.
5960
5961 This last case can happen when the server is a fast static server of cache.
5962 In this case, the server will need to be properly tuned to support high enough
5963 connection counts because connections will last until the client sends another
5964 request.
5965
5966 If the client request has to go to another backend or another server due to
5967 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01005968 immediately be closed and a new one re-opened. Option "prefer-last-server" is
5969 available to try optimize server selection so that if the server currently
5970 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01005971
Willy Tarreau16bfb022010-01-16 19:48:41 +01005972 At the moment, logs will not indicate whether requests came from the same
5973 session or not. The accept date reported in the logs corresponds to the end
5974 of the previous request, and the request time corresponds to the time spent
5975 waiting for a new request. The keep-alive request time is still bound to the
5976 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5977 not set.
5978
Cyril Bonté653dcd62014-02-20 00:13:15 +01005979 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02005980 http-server-close" or "option http-tunnel". When backend and frontend options
5981 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005982
Christopher Faulet315b39c2018-09-21 16:26:19 +02005983 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01005984 "option prefer-last-server", "option http-pretend-keepalive",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005985 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005986
5987
Willy Tarreau96e31212011-05-30 18:10:30 +02005988option http-no-delay
5989no option http-no-delay
5990 Instruct the system to favor low interactive delays over performance in HTTP
5991 May be used in sections : defaults | frontend | listen | backend
5992 yes | yes | yes | yes
5993 Arguments : none
5994
5995 In HTTP, each payload is unidirectional and has no notion of interactivity.
5996 Any agent is expected to queue data somewhat for a reasonably low delay.
5997 There are some very rare server-to-server applications that abuse the HTTP
5998 protocol and expect the payload phase to be highly interactive, with many
5999 interleaved data chunks in both directions within a single request. This is
6000 absolutely not supported by the HTTP specification and will not work across
6001 most proxies or servers. When such applications attempt to do this through
6002 haproxy, it works but they will experience high delays due to the network
6003 optimizations which favor performance by instructing the system to wait for
6004 enough data to be available in order to only send full packets. Typical
6005 delays are around 200 ms per round trip. Note that this only happens with
6006 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6007 affected.
6008
6009 When "option http-no-delay" is present in either the frontend or the backend
6010 used by a connection, all such optimizations will be disabled in order to
6011 make the exchanges as fast as possible. Of course this offers no guarantee on
6012 the functionality, as it may break at any other place. But if it works via
6013 HAProxy, it will work as fast as possible. This option should never be used
6014 by default, and should never be used at all unless such a buggy application
6015 is discovered. The impact of using this option is an increase of bandwidth
6016 usage and CPU usage, which may significantly lower performance in high
6017 latency environments.
6018
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006019 See also : "option http-buffer-request"
6020
Willy Tarreau96e31212011-05-30 18:10:30 +02006021
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006022option http-pretend-keepalive
6023no option http-pretend-keepalive
6024 Define whether haproxy will announce keepalive to the server or not
6025 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006026 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006027 Arguments : none
6028
Christopher Faulet315b39c2018-09-21 16:26:19 +02006029 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006030 adds a "Connection: close" header to the request forwarded to the server.
6031 Unfortunately, when some servers see this header, they automatically refrain
6032 from using the chunked encoding for responses of unknown length, while this
6033 is totally unrelated. The immediate effect is that this prevents haproxy from
6034 maintaining the client connection alive. A second effect is that a client or
6035 a cache could receive an incomplete response without being aware of it, and
6036 consider the response complete.
6037
6038 By setting "option http-pretend-keepalive", haproxy will make the server
6039 believe it will keep the connection alive. The server will then not fall back
6040 to the abnormal undesired above. When haproxy gets the whole response, it
6041 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006042 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006043 connection is correctly closed on the server side.
6044
6045 It is recommended not to enable this option by default, because most servers
6046 will more efficiently close the connection themselves after the last packet,
6047 and release its buffers slightly earlier. Also, the added packet on the
6048 network could slightly reduce the overall peak performance. However it is
6049 worth noting that when this option is enabled, haproxy will have slightly
6050 less work to do. So if haproxy is the bottleneck on the whole architecture,
6051 enabling this option might save a few CPU cycles.
6052
Christopher Faulet98db9762018-09-21 10:25:19 +02006053 This option may be set in backend and listen sections. Using it in a frontend
6054 section will be ignored and a warning will be reported during startup. It is
6055 a backend related option, so there is no real reason to set it on a
6056 frontend. This option may be combined with "option httpclose", which will
6057 cause keepalive to be announced to the server and close to be announced to
6058 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006059
6060 If this option has been enabled in a "defaults" section, it can be disabled
6061 in a specific instance by prepending the "no" keyword before it.
6062
Christopher Faulet315b39c2018-09-21 16:26:19 +02006063 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006064 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006065
Willy Tarreauc27debf2008-01-06 08:57:02 +01006066
Willy Tarreaub608feb2010-01-02 22:47:18 +01006067option http-server-close
6068no option http-server-close
6069 Enable or disable HTTP connection closing on the server side
6070 May be used in sections : defaults | frontend | listen | backend
6071 yes | yes | yes | yes
6072 Arguments : none
6073
Willy Tarreau70dffda2014-01-30 03:07:23 +01006074 By default HAProxy operates in keep-alive mode with regards to persistent
6075 connections: for each connection it processes each request and response, and
6076 leaves the connection idle on both sides between the end of a response and
6077 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006078 as "option http-server-close", "option httpclose" or "option http-tunnel".
6079 Setting "option http-server-close" enables HTTP connection-close mode on the
6080 server side while keeping the ability to support HTTP keep-alive and
6081 pipelining on the client side. This provides the lowest latency on the client
6082 side (slow network) and the fastest session reuse on the server side to save
6083 server resources, similarly to "option httpclose". It also permits
6084 non-keepalive capable servers to be served in keep-alive mode to the clients
6085 if they conform to the requirements of RFC7230. Please note that some servers
6086 do not always conform to those requirements when they see "Connection: close"
6087 in the request. The effect will be that keep-alive will never be used. A
6088 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006089
6090 At the moment, logs will not indicate whether requests came from the same
6091 session or not. The accept date reported in the logs corresponds to the end
6092 of the previous request, and the request time corresponds to the time spent
6093 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006094 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6095 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006096
6097 This option may be set both in a frontend and in a backend. It is enabled if
6098 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006099 It disables and replaces any previous "option httpclose", "option http-tunnel"
6100 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6101 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006102
6103 If this option has been enabled in a "defaults" section, it can be disabled
6104 in a specific instance by prepending the "no" keyword before it.
6105
Christopher Faulet315b39c2018-09-21 16:26:19 +02006106 See also : "option httpclose", "option http-pretend-keepalive",
6107 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006108
6109
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006110option http-tunnel
6111no option http-tunnel
6112 Disable or enable HTTP connection processing after first transaction
6113 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006114 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006115 Arguments : none
6116
Willy Tarreau70dffda2014-01-30 03:07:23 +01006117 By default HAProxy operates in keep-alive mode with regards to persistent
6118 connections: for each connection it processes each request and response, and
6119 leaves the connection idle on both sides between the end of a response and
6120 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006121 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006122
6123 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006124 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006125 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6126 is normally not needed anymore unless in very specific cases such as when
6127 using an in-house protocol that looks like HTTP but is not compatible, or
6128 just to log one request per client in order to reduce log size. Note that
6129 everything which works at the HTTP level, including header parsing/addition,
6130 cookie processing or content switching will only work for the first request
6131 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006132
Christopher Faulet4212a302018-09-21 10:42:19 +02006133 This option may be set on frontend and listen sections. Using it on a backend
6134 section will be ignored and a warning will be reported during the startup. It
6135 is a frontend related option, so there is no real reason to set it on a
6136 backend.
6137
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006138 If this option has been enabled in a "defaults" section, it can be disabled
6139 in a specific instance by prepending the "no" keyword before it.
6140
Christopher Faulet315b39c2018-09-21 16:26:19 +02006141 See also : "option httpclose", "option http-server-close",
6142 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006143
6144
Willy Tarreau88d349d2010-01-25 12:15:43 +01006145option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006146no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006147 Make use of non-standard Proxy-Connection header instead of Connection
6148 May be used in sections : defaults | frontend | listen | backend
6149 yes | yes | yes | no
6150 Arguments : none
6151
Lukas Tribus23953682017-04-28 13:24:30 +00006152 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006153 Connection header to indicate their wish of persistent or non-persistent
6154 connections, both browsers and proxies ignore this header for proxied
6155 connections and make use of the undocumented, non-standard Proxy-Connection
6156 header instead. The issue begins when trying to put a load balancer between
6157 browsers and such proxies, because there will be a difference between what
6158 haproxy understands and what the client and the proxy agree on.
6159
6160 By setting this option in a frontend, haproxy can automatically switch to use
6161 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006162 defined here as one where the URI begins with neither a '/' nor a '*'. This
6163 is incompatible with the HTTP tunnel mode. Note that this option can only be
6164 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006165
Willy Tarreau844a7e72010-01-31 21:46:18 +01006166 Also, when this option is set, a request which requires authentication will
6167 automatically switch to use proxy authentication headers if it is itself a
6168 proxied request. That makes it possible to check or enforce authentication in
6169 front of an existing proxy.
6170
Willy Tarreau88d349d2010-01-25 12:15:43 +01006171 This option should normally never be used, except in front of a proxy.
6172
Christopher Faulet315b39c2018-09-21 16:26:19 +02006173 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006174
6175
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006176option http-use-htx
6177no option http-use-htx
6178 Switch to the new HTX internal representation for HTTP protocol elements
6179 May be used in sections : defaults | frontend | listen | backend
6180 yes | yes | yes | yes
6181 Arguments : none
6182
6183 By default, the HTTP protocol is processed as-is. Inserting, deleting, or
6184 modifying a header field requires to rewrite the affected part in the buffer
6185 and to move the buffer's tail accordingly. Since this principle has deep
6186 roots in haproxy, the HTTP/2 protocol is converted to HTTP/1.1 before being
6187 processed this way. It also results in the inability to establish HTTP/2
6188 connections to servers because of the loss of HTTP/2 semantics in the HTTP/1
6189 representation.
6190
6191 HTX is the name of a totally new native internal representation for the HTTP
6192 protocol, that is agnostic to the version and aims at preserving semantics
6193 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6194 the protocol elements so that no more memory moves are necessary and that
6195 most elements are directly accessed. This mechanism is still limited to the
6196 most basic operations (no compression, filters, Lua, applets, cache, etc).
6197 But it supports using either HTTP/1 or HTTP/2 on any side regardless of the
6198 other side's version.
6199
6200 This option indicates that HTX needs to be used. It will cause errors to be
6201 emitted if incompatible features are used, but will allow H2 to be selected
6202 as a server protocol. It is recommended to use this option on new reasonably
6203 simple configurations, but since the feature still has incomplete functional
6204 coverage, it is not enabled by default.
6205
6206 See also : "mode http"
6207
6208
Willy Tarreaud63335a2010-02-26 12:56:52 +01006209option httpchk
6210option httpchk <uri>
6211option httpchk <method> <uri>
6212option httpchk <method> <uri> <version>
6213 Enable HTTP protocol to check on the servers health
6214 May be used in sections : defaults | frontend | listen | backend
6215 yes | no | yes | yes
6216 Arguments :
6217 <method> is the optional HTTP method used with the requests. When not set,
6218 the "OPTIONS" method is used, as it generally requires low server
6219 processing and is easy to filter out from the logs. Any method
6220 may be used, though it is not recommended to invent non-standard
6221 ones.
6222
6223 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6224 which is accessible by default on almost any server, but may be
6225 changed to any other URI. Query strings are permitted.
6226
6227 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6228 but some servers might behave incorrectly in HTTP 1.0, so turning
6229 it to HTTP/1.1 may sometimes help. Note that the Host field is
6230 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6231 after "\r\n" following the version string.
6232
6233 By default, server health checks only consist in trying to establish a TCP
6234 connection. When "option httpchk" is specified, a complete HTTP request is
6235 sent once the TCP connection is established, and responses 2xx and 3xx are
6236 considered valid, while all other ones indicate a server failure, including
6237 the lack of any response.
6238
6239 The port and interval are specified in the server configuration.
6240
6241 This option does not necessarily require an HTTP backend, it also works with
6242 plain TCP backends. This is particularly useful to check simple scripts bound
6243 to some dedicated ports using the inetd daemon.
6244
6245 Examples :
6246 # Relay HTTPS traffic to Apache instance and check service availability
6247 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6248 backend https_relay
6249 mode tcp
6250 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6251 server apache1 192.168.1.1:443 check port 80
6252
Simon Hormanafc47ee2013-11-25 10:46:35 +09006253 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6254 "option pgsql-check", "http-check" and the "check", "port" and
6255 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006256
6257
Willy Tarreauc27debf2008-01-06 08:57:02 +01006258option httpclose
6259no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006260 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006261 May be used in sections : defaults | frontend | listen | backend
6262 yes | yes | yes | yes
6263 Arguments : none
6264
Willy Tarreau70dffda2014-01-30 03:07:23 +01006265 By default HAProxy operates in keep-alive mode with regards to persistent
6266 connections: for each connection it processes each request and response, and
6267 leaves the connection idle on both sides between the end of a response and
6268 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006269 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006270
Christopher Faulet315b39c2018-09-21 16:26:19 +02006271 If "option httpclose" is set, HAProxy will close connections with the server
6272 and the client as soon as the request and the response are received. It will
6273 alos check if a "Connection: close" header is already set in each direction,
6274 and will add one if missing. Any "Connection" header different from "close"
6275 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006276
Christopher Faulet315b39c2018-09-21 16:26:19 +02006277 This option may also be combined with "option http-pretend-keepalive", which
6278 will disable sending of the "Connection: close" header, but will still cause
6279 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006280
6281 This option may be set both in a frontend and in a backend. It is enabled if
6282 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006283 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006284 "option http-keep-alive" or "option http-tunnel". Please check section 4
6285 ("Proxies") to see how this option combines with others when frontend and
6286 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006287
6288 If this option has been enabled in a "defaults" section, it can be disabled
6289 in a specific instance by prepending the "no" keyword before it.
6290
Christopher Faulet315b39c2018-09-21 16:26:19 +02006291 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006292
6293
Emeric Brun3a058f32009-06-30 18:26:00 +02006294option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006295 Enable logging of HTTP request, session state and timers
6296 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006297 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006298 Arguments :
6299 clf if the "clf" argument is added, then the output format will be
6300 the CLF format instead of HAProxy's default HTTP format. You can
6301 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006302 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006303 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006304
6305 By default, the log output format is very poor, as it only contains the
6306 source and destination addresses, and the instance name. By specifying
6307 "option httplog", each log line turns into a much richer format including,
6308 but not limited to, the HTTP request, the connection timers, the session
6309 status, the connections numbers, the captured headers and cookies, the
6310 frontend, backend and server name, and of course the source address and
6311 ports.
6312
PiBa-NLbd556bf2014-12-11 21:31:54 +01006313 Specifying only "option httplog" will automatically clear the 'clf' mode
6314 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006315
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006316 "option httplog" overrides any previous "log-format" directive.
6317
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006318 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006319
Willy Tarreau55165fe2009-05-10 12:02:55 +02006320
6321option http_proxy
6322no option http_proxy
6323 Enable or disable plain HTTP proxy mode
6324 May be used in sections : defaults | frontend | listen | backend
6325 yes | yes | yes | yes
6326 Arguments : none
6327
6328 It sometimes happens that people need a pure HTTP proxy which understands
6329 basic proxy requests without caching nor any fancy feature. In this case,
6330 it may be worth setting up an HAProxy instance with the "option http_proxy"
6331 set. In this mode, no server is declared, and the connection is forwarded to
6332 the IP address and port found in the URL after the "http://" scheme.
6333
6334 No host address resolution is performed, so this only works when pure IP
6335 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006336 it will probably be used only by experts who know they need exactly it. This
6337 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006338
6339 If this option has been enabled in a "defaults" section, it can be disabled
6340 in a specific instance by prepending the "no" keyword before it.
6341
6342 Example :
6343 # this backend understands HTTP proxy requests and forwards them directly.
6344 backend direct_forward
6345 option httpclose
6346 option http_proxy
6347
6348 See also : "option httpclose"
6349
Willy Tarreau211ad242009-10-03 21:45:07 +02006350
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006351option independent-streams
6352no option independent-streams
6353 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006354 May be used in sections : defaults | frontend | listen | backend
6355 yes | yes | yes | yes
6356 Arguments : none
6357
6358 By default, when data is sent over a socket, both the write timeout and the
6359 read timeout for that socket are refreshed, because we consider that there is
6360 activity on that socket, and we have no other means of guessing if we should
6361 receive data or not.
6362
Davor Ocelice9ed2812017-12-25 17:49:28 +01006363 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006364 exists a situation where it is desirable to disable it, and only refresh the
6365 read timeout if there are incoming data. This happens on sessions with large
6366 timeouts and low amounts of exchanged data such as telnet session. If the
6367 server suddenly disappears, the output data accumulates in the system's
6368 socket buffers, both timeouts are correctly refreshed, and there is no way
6369 to know the server does not receive them, so we don't timeout. However, when
6370 the underlying protocol always echoes sent data, it would be enough by itself
6371 to detect the issue using the read timeout. Note that this problem does not
6372 happen with more verbose protocols because data won't accumulate long in the
6373 socket buffers.
6374
6375 When this option is set on the frontend, it will disable read timeout updates
6376 on data sent to the client. There probably is little use of this case. When
6377 the option is set on the backend, it will disable read timeout updates on
6378 data sent to the server. Doing so will typically break large HTTP posts from
6379 slow lines, so use it with caution.
6380
Lukas Tribus745f15e2018-11-08 12:41:42 +01006381 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006382 with a spelling mistake. This spelling is still supported but
6383 deprecated.
6384
Willy Tarreauce887fd2012-05-12 12:50:00 +02006385 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006386
6387
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006388option ldap-check
6389 Use LDAPv3 health checks for server testing
6390 May be used in sections : defaults | frontend | listen | backend
6391 yes | no | yes | yes
6392 Arguments : none
6393
6394 It is possible to test that the server correctly talks LDAPv3 instead of just
6395 testing that it accepts the TCP connection. When this option is set, an
6396 LDAPv3 anonymous simple bind message is sent to the server, and the response
6397 is analyzed to find an LDAPv3 bind response message.
6398
6399 The server is considered valid only when the LDAP response contains success
6400 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6401
6402 Logging of bind requests is server dependent see your documentation how to
6403 configure it.
6404
6405 Example :
6406 option ldap-check
6407
6408 See also : "option httpchk"
6409
6410
Simon Horman98637e52014-06-20 12:30:16 +09006411option external-check
6412 Use external processes for server health checks
6413 May be used in sections : defaults | frontend | listen | backend
6414 yes | no | yes | yes
6415
6416 It is possible to test the health of a server using an external command.
6417 This is achieved by running the executable set using "external-check
6418 command".
6419
6420 Requires the "external-check" global to be set.
6421
6422 See also : "external-check", "external-check command", "external-check path"
6423
6424
Willy Tarreau211ad242009-10-03 21:45:07 +02006425option log-health-checks
6426no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006427 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006428 May be used in sections : defaults | frontend | listen | backend
6429 yes | no | yes | yes
6430 Arguments : none
6431
Willy Tarreaubef1b322014-05-13 21:01:39 +02006432 By default, failed health check are logged if server is UP and successful
6433 health checks are logged if server is DOWN, so the amount of additional
6434 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006435
Willy Tarreaubef1b322014-05-13 21:01:39 +02006436 When this option is enabled, any change of the health check status or to
6437 the server's health will be logged, so that it becomes possible to know
6438 that a server was failing occasional checks before crashing, or exactly when
6439 it failed to respond a valid HTTP status, then when the port started to
6440 reject connections, then when the server stopped responding at all.
6441
Davor Ocelice9ed2812017-12-25 17:49:28 +01006442 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006443 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006444
Willy Tarreaubef1b322014-05-13 21:01:39 +02006445 See also: "option httpchk", "option ldap-check", "option mysql-check",
6446 "option pgsql-check", "option redis-check", "option smtpchk",
6447 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006448
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006449
6450option log-separate-errors
6451no option log-separate-errors
6452 Change log level for non-completely successful connections
6453 May be used in sections : defaults | frontend | listen | backend
6454 yes | yes | yes | no
6455 Arguments : none
6456
6457 Sometimes looking for errors in logs is not easy. This option makes haproxy
6458 raise the level of logs containing potentially interesting information such
6459 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6460 level changes from "info" to "err". This makes it possible to log them
6461 separately to a different file with most syslog daemons. Be careful not to
6462 remove them from the original file, otherwise you would lose ordering which
6463 provides very important information.
6464
6465 Using this option, large sites dealing with several thousand connections per
6466 second may log normal traffic to a rotating buffer and only archive smaller
6467 error logs.
6468
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006469 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006470 logging.
6471
Willy Tarreauc27debf2008-01-06 08:57:02 +01006472
6473option logasap
6474no option logasap
6475 Enable or disable early logging of HTTP requests
6476 May be used in sections : defaults | frontend | listen | backend
6477 yes | yes | yes | no
6478 Arguments : none
6479
6480 By default, HTTP requests are logged upon termination so that the total
6481 transfer time and the number of bytes appear in the logs. When large objects
6482 are being transferred, it may take a while before the request appears in the
6483 logs. Using "option logasap", the request gets logged as soon as the server
6484 sends the complete headers. The only missing information in the logs will be
6485 the total number of bytes which will indicate everything except the amount
6486 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006487 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006488 "Content-Length" response header so that the logs at least indicate how many
6489 bytes are expected to be transferred.
6490
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006491 Examples :
6492 listen http_proxy 0.0.0.0:80
6493 mode http
6494 option httplog
6495 option logasap
6496 log 192.168.2.200 local3
6497
6498 >>> Feb 6 12:14:14 localhost \
6499 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6500 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6501 "GET /image.iso HTTP/1.0"
6502
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006503 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006504 logging.
6505
6506
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006507option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006508 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006509 May be used in sections : defaults | frontend | listen | backend
6510 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006511 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006512 <username> This is the username which will be used when connecting to MySQL
6513 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006514 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006515
6516 If you specify a username, the check consists of sending two MySQL packet,
6517 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006518 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006519 Error packet. It is a basic but useful test which does not produce error nor
6520 aborted connect on the server. However, it requires adding an authorization
6521 in the MySQL table, like this :
6522
6523 USE mysql;
6524 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6525 FLUSH PRIVILEGES;
6526
6527 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006528 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006529 Error packet, we don't send anything in this mode. It was reported that it
6530 can generate lockout if check is too frequent and/or if there is not enough
6531 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6532 value as if a connection is established successfully within fewer than MySQL
6533 "max_connect_errors" attempts after a previous connection was interrupted,
6534 the error count for the host is cleared to zero. If HAProxy's server get
6535 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6536
6537 Remember that this does not check database presence nor database consistency.
6538 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006539
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006540 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006541
6542 Most often, an incoming MySQL server needs to see the client's IP address for
6543 various purposes, including IP privilege matching and connection logging.
6544 When possible, it is often wise to masquerade the client's IP address when
6545 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006546 which requires the transparent proxy feature to be compiled in, and the MySQL
6547 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006548
6549 See also: "option httpchk"
6550
6551
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006552option nolinger
6553no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006554 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006555 May be used in sections: defaults | frontend | listen | backend
6556 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006557 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006558
Davor Ocelice9ed2812017-12-25 17:49:28 +01006559 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006560 physically disconnected), the session timeouts triggers and the session is
6561 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6562 using some resources and possibly limiting the ability to establish newer
6563 connections.
6564
6565 When this happens, it is possible to activate "option nolinger" which forces
6566 the system to immediately remove any socket's pending data on close. Thus,
6567 the session is instantly purged from the system's tables. This usually has
6568 side effects such as increased number of TCP resets due to old retransmits
6569 getting immediately rejected. Some firewalls may sometimes complain about
6570 this too.
6571
6572 For this reason, it is not recommended to use this option when not absolutely
6573 needed. You know that you need it when you have thousands of FIN_WAIT1
6574 sessions on your system (TIME_WAIT ones do not count).
6575
6576 This option may be used both on frontends and backends, depending on the side
6577 where it is required. Use it on the frontend for clients, and on the backend
6578 for servers.
6579
6580 If this option has been enabled in a "defaults" section, it can be disabled
6581 in a specific instance by prepending the "no" keyword before it.
6582
6583
Willy Tarreau55165fe2009-05-10 12:02:55 +02006584option originalto [ except <network> ] [ header <name> ]
6585 Enable insertion of the X-Original-To header to requests sent to servers
6586 May be used in sections : defaults | frontend | listen | backend
6587 yes | yes | yes | yes
6588 Arguments :
6589 <network> is an optional argument used to disable this option for sources
6590 matching <network>
6591 <name> an optional argument to specify a different "X-Original-To"
6592 header name.
6593
6594 Since HAProxy can work in transparent mode, every request from a client can
6595 be redirected to the proxy and HAProxy itself can proxy every request to a
6596 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6597 be lost. This is annoying when you want access rules based on destination ip
6598 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6599 added by HAProxy to all requests sent to the server. This header contains a
6600 value representing the original destination IP address. Since this must be
6601 configured to always use the last occurrence of this header only. Note that
6602 only the last occurrence of the header must be used, since it is really
6603 possible that the client has already brought one.
6604
6605 The keyword "header" may be used to supply a different header name to replace
6606 the default "X-Original-To". This can be useful where you might already
6607 have a "X-Original-To" header from a different application, and you need
6608 preserve it. Also if your backend server doesn't use the "X-Original-To"
6609 header and requires different one.
6610
6611 Sometimes, a same HAProxy instance may be shared between a direct client
6612 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6613 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6614 header for a known source address or network by adding the "except" keyword
6615 followed by the network address. In this case, any source IP matching the
6616 network will not cause an addition of this header. Most common uses are with
6617 private networks or 127.0.0.1.
6618
6619 This option may be specified either in the frontend or in the backend. If at
6620 least one of them uses it, the header will be added. Note that the backend's
6621 setting of the header subargument takes precedence over the frontend's if
6622 both are defined.
6623
Willy Tarreau55165fe2009-05-10 12:02:55 +02006624 Examples :
6625 # Original Destination address
6626 frontend www
6627 mode http
6628 option originalto except 127.0.0.1
6629
6630 # Those servers want the IP Address in X-Client-Dst
6631 backend www
6632 mode http
6633 option originalto header X-Client-Dst
6634
Christopher Faulet315b39c2018-09-21 16:26:19 +02006635 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006636
6637
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006638option persist
6639no option persist
6640 Enable or disable forced persistence on down servers
6641 May be used in sections: defaults | frontend | listen | backend
6642 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006643 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006644
6645 When an HTTP request reaches a backend with a cookie which references a dead
6646 server, by default it is redispatched to another server. It is possible to
6647 force the request to be sent to the dead server first using "option persist"
6648 if absolutely needed. A common use case is when servers are under extreme
6649 load and spend their time flapping. In this case, the users would still be
6650 directed to the server they opened the session on, in the hope they would be
6651 correctly served. It is recommended to use "option redispatch" in conjunction
6652 with this option so that in the event it would not be possible to connect to
6653 the server at all (server definitely dead), the client would finally be
6654 redirected to another valid server.
6655
6656 If this option has been enabled in a "defaults" section, it can be disabled
6657 in a specific instance by prepending the "no" keyword before it.
6658
Willy Tarreau4de91492010-01-22 19:10:05 +01006659 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006660
6661
Willy Tarreau0c122822013-12-15 18:49:01 +01006662option pgsql-check [ user <username> ]
6663 Use PostgreSQL health checks for server testing
6664 May be used in sections : defaults | frontend | listen | backend
6665 yes | no | yes | yes
6666 Arguments :
6667 <username> This is the username which will be used when connecting to
6668 PostgreSQL server.
6669
6670 The check sends a PostgreSQL StartupMessage and waits for either
6671 Authentication request or ErrorResponse message. It is a basic but useful
6672 test which does not produce error nor aborted connect on the server.
6673 This check is identical with the "mysql-check".
6674
6675 See also: "option httpchk"
6676
6677
Willy Tarreau9420b122013-12-15 18:58:25 +01006678option prefer-last-server
6679no option prefer-last-server
6680 Allow multiple load balanced requests to remain on the same server
6681 May be used in sections: defaults | frontend | listen | backend
6682 yes | no | yes | yes
6683 Arguments : none
6684
6685 When the load balancing algorithm in use is not deterministic, and a previous
6686 request was sent to a server to which haproxy still holds a connection, it is
6687 sometimes desirable that subsequent requests on a same session go to the same
6688 server as much as possible. Note that this is different from persistence, as
6689 we only indicate a preference which haproxy tries to apply without any form
6690 of warranty. The real use is for keep-alive connections sent to servers. When
6691 this option is used, haproxy will try to reuse the same connection that is
6692 attached to the server instead of rebalancing to another server, causing a
6693 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006694 not make much sense to use this in combination with hashing algorithms. Note,
6695 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006696 to a proxy which sends a 407 (authentication required), when the load
6697 balancing algorithm is not deterministic. This is mandatory for use with the
6698 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006699 troubleshooting some faulty applications. Option prefer-last-server might be
6700 desirable in these environments as well, to avoid redistributing the traffic
6701 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006702
6703 If this option has been enabled in a "defaults" section, it can be disabled
6704 in a specific instance by prepending the "no" keyword before it.
6705
6706 See also: "option http-keep-alive"
6707
6708
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006709option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006710option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006711no option redispatch
6712 Enable or disable session redistribution in case of connection failure
6713 May be used in sections: defaults | frontend | listen | backend
6714 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006715 Arguments :
6716 <interval> The optional integer value that controls how often redispatches
6717 occur when retrying connections. Positive value P indicates a
6718 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006719 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006720 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006721 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006722 positive value of 1 would indicate a redispatch on every retry,
6723 and a positive value of 3 would indicate a redispatch on every
6724 third retry. You can disable redispatches with a value of 0.
6725
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006726
6727 In HTTP mode, if a server designated by a cookie is down, clients may
6728 definitely stick to it because they cannot flush the cookie, so they will not
6729 be able to access the service anymore.
6730
6731 Specifying "option redispatch" will allow the proxy to break their
6732 persistence and redistribute them to a working server.
6733
Joseph Lynch726ab712015-05-11 23:25:34 -07006734 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006735 connection failures. Of course, it requires having "retries" set to a nonzero
6736 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006737
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006738 This form is the preferred form, which replaces both the "redispatch" and
6739 "redisp" keywords.
6740
6741 If this option has been enabled in a "defaults" section, it can be disabled
6742 in a specific instance by prepending the "no" keyword before it.
6743
Willy Tarreau4de91492010-01-22 19:10:05 +01006744 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006745
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006746
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006747option redis-check
6748 Use redis health checks for server testing
6749 May be used in sections : defaults | frontend | listen | backend
6750 yes | no | yes | yes
6751 Arguments : none
6752
6753 It is possible to test that the server correctly talks REDIS protocol instead
6754 of just testing that it accepts the TCP connection. When this option is set,
6755 a PING redis command is sent to the server, and the response is analyzed to
6756 find the "+PONG" response message.
6757
6758 Example :
6759 option redis-check
6760
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006761 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006762
6763
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006764option smtpchk
6765option smtpchk <hello> <domain>
6766 Use SMTP health checks for server testing
6767 May be used in sections : defaults | frontend | listen | backend
6768 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006769 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006770 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006771 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006772 values will be turned into the default command ("HELO").
6773
6774 <domain> is the domain name to present to the server. It may only be
6775 specified (and is mandatory) if the hello command has been
6776 specified. By default, "localhost" is used.
6777
6778 When "option smtpchk" is set, the health checks will consist in TCP
6779 connections followed by an SMTP command. By default, this command is
6780 "HELO localhost". The server's return code is analyzed and only return codes
6781 starting with a "2" will be considered as valid. All other responses,
6782 including a lack of response will constitute an error and will indicate a
6783 dead server.
6784
6785 This test is meant to be used with SMTP servers or relays. Depending on the
6786 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006787 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006788 25 is often easier than adjusting the configuration.
6789
6790 Most often, an incoming SMTP server needs to see the client's IP address for
6791 various purposes, including spam filtering, anti-spoofing and logging. When
6792 possible, it is often wise to masquerade the client's IP address when
6793 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006794 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006795
6796 Example :
6797 option smtpchk HELO mydomain.org
6798
6799 See also : "option httpchk", "source"
6800
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006801
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006802option socket-stats
6803no option socket-stats
6804
6805 Enable or disable collecting & providing separate statistics for each socket.
6806 May be used in sections : defaults | frontend | listen | backend
6807 yes | yes | yes | no
6808
6809 Arguments : none
6810
6811
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006812option splice-auto
6813no option splice-auto
6814 Enable or disable automatic kernel acceleration on sockets in both directions
6815 May be used in sections : defaults | frontend | listen | backend
6816 yes | yes | yes | yes
6817 Arguments : none
6818
6819 When this option is enabled either on a frontend or on a backend, haproxy
6820 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006821 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006822 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006823 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006824 are not much aggressive in order to limit excessive use of splicing. This
6825 option requires splicing to be enabled at compile time, and may be globally
6826 disabled with the global option "nosplice". Since splice uses pipes, using it
6827 requires that there are enough spare pipes.
6828
6829 Important note: kernel-based TCP splicing is a Linux-specific feature which
6830 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6831 transfer data between sockets without copying these data to user-space, thus
6832 providing noticeable performance gains and CPU cycles savings. Since many
6833 early implementations are buggy, corrupt data and/or are inefficient, this
6834 feature is not enabled by default, and it should be used with extreme care.
6835 While it is not possible to detect the correctness of an implementation,
6836 2.6.29 is the first version offering a properly working implementation. In
6837 case of doubt, splicing may be globally disabled using the global "nosplice"
6838 keyword.
6839
6840 Example :
6841 option splice-auto
6842
6843 If this option has been enabled in a "defaults" section, it can be disabled
6844 in a specific instance by prepending the "no" keyword before it.
6845
6846 See also : "option splice-request", "option splice-response", and global
6847 options "nosplice" and "maxpipes"
6848
6849
6850option splice-request
6851no option splice-request
6852 Enable or disable automatic kernel acceleration on sockets for requests
6853 May be used in sections : defaults | frontend | listen | backend
6854 yes | yes | yes | yes
6855 Arguments : none
6856
6857 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006858 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006859 the client to the server. It might still use the recv/send scheme if there
6860 are no spare pipes left. This option requires splicing to be enabled at
6861 compile time, and may be globally disabled with the global option "nosplice".
6862 Since splice uses pipes, using it requires that there are enough spare pipes.
6863
6864 Important note: see "option splice-auto" for usage limitations.
6865
6866 Example :
6867 option splice-request
6868
6869 If this option has been enabled in a "defaults" section, it can be disabled
6870 in a specific instance by prepending the "no" keyword before it.
6871
6872 See also : "option splice-auto", "option splice-response", and global options
6873 "nosplice" and "maxpipes"
6874
6875
6876option splice-response
6877no option splice-response
6878 Enable or disable automatic kernel acceleration on sockets for responses
6879 May be used in sections : defaults | frontend | listen | backend
6880 yes | yes | yes | yes
6881 Arguments : none
6882
6883 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006884 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006885 the server to the client. It might still use the recv/send scheme if there
6886 are no spare pipes left. This option requires splicing to be enabled at
6887 compile time, and may be globally disabled with the global option "nosplice".
6888 Since splice uses pipes, using it requires that there are enough spare pipes.
6889
6890 Important note: see "option splice-auto" for usage limitations.
6891
6892 Example :
6893 option splice-response
6894
6895 If this option has been enabled in a "defaults" section, it can be disabled
6896 in a specific instance by prepending the "no" keyword before it.
6897
6898 See also : "option splice-auto", "option splice-request", and global options
6899 "nosplice" and "maxpipes"
6900
6901
Christopher Fauletba7bc162016-11-07 21:07:38 +01006902option spop-check
6903 Use SPOP health checks for server testing
6904 May be used in sections : defaults | frontend | listen | backend
6905 no | no | no | yes
6906 Arguments : none
6907
6908 It is possible to test that the server correctly talks SPOP protocol instead
6909 of just testing that it accepts the TCP connection. When this option is set,
6910 a HELLO handshake is performed between HAProxy and the server, and the
6911 response is analyzed to check no error is reported.
6912
6913 Example :
6914 option spop-check
6915
6916 See also : "option httpchk"
6917
6918
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006919option srvtcpka
6920no option srvtcpka
6921 Enable or disable the sending of TCP keepalive packets on the server side
6922 May be used in sections : defaults | frontend | listen | backend
6923 yes | no | yes | yes
6924 Arguments : none
6925
6926 When there is a firewall or any session-aware component between a client and
6927 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006928 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006929 components decides to expire a session which has remained idle for too long.
6930
6931 Enabling socket-level TCP keep-alives makes the system regularly send packets
6932 to the other end of the connection, leaving it active. The delay between
6933 keep-alive probes is controlled by the system only and depends both on the
6934 operating system and its tuning parameters.
6935
6936 It is important to understand that keep-alive packets are neither emitted nor
6937 received at the application level. It is only the network stacks which sees
6938 them. For this reason, even if one side of the proxy already uses keep-alives
6939 to maintain its connection alive, those keep-alive packets will not be
6940 forwarded to the other side of the proxy.
6941
6942 Please note that this has nothing to do with HTTP keep-alive.
6943
6944 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6945 server side of a connection, which should help when session expirations are
6946 noticed between HAProxy and a server.
6947
6948 If this option has been enabled in a "defaults" section, it can be disabled
6949 in a specific instance by prepending the "no" keyword before it.
6950
6951 See also : "option clitcpka", "option tcpka"
6952
6953
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006954option ssl-hello-chk
6955 Use SSLv3 client hello health checks for server testing
6956 May be used in sections : defaults | frontend | listen | backend
6957 yes | no | yes | yes
6958 Arguments : none
6959
6960 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6961 possible to test that the server correctly talks SSL instead of just testing
6962 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6963 SSLv3 client hello messages are sent once the connection is established to
6964 the server, and the response is analyzed to find an SSL server hello message.
6965 The server is considered valid only when the response contains this server
6966 hello message.
6967
6968 All servers tested till there correctly reply to SSLv3 client hello messages,
6969 and most servers tested do not even log the requests containing only hello
6970 messages, which is appreciable.
6971
Willy Tarreau763a95b2012-10-04 23:15:39 +02006972 Note that this check works even when SSL support was not built into haproxy
6973 because it forges the SSL message. When SSL support is available, it is best
6974 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006975
Willy Tarreau763a95b2012-10-04 23:15:39 +02006976 See also: "option httpchk", "check-ssl"
6977
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006978
Willy Tarreaued179852013-12-16 01:07:00 +01006979option tcp-check
6980 Perform health checks using tcp-check send/expect sequences
6981 May be used in sections: defaults | frontend | listen | backend
6982 yes | no | yes | yes
6983
6984 This health check method is intended to be combined with "tcp-check" command
6985 lists in order to support send/expect types of health check sequences.
6986
6987 TCP checks currently support 4 modes of operations :
6988 - no "tcp-check" directive : the health check only consists in a connection
6989 attempt, which remains the default mode.
6990
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006991 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01006992 used to send a string along with a connection opening. With some
6993 protocols, it helps sending a "QUIT" message for example that prevents
6994 the server from logging a connection error for each health check. The
6995 check result will still be based on the ability to open the connection
6996 only.
6997
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006998 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01006999 The connection is opened and haproxy waits for the server to present some
7000 contents which must validate some rules. The check result will be based
7001 on the matching between the contents and the rules. This is suited for
7002 POP, IMAP, SMTP, FTP, SSH, TELNET.
7003
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007004 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007005 used to test a hello-type protocol. HAProxy sends a message, the server
7006 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007007 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007008 suited for protocols which require a binding or a request/response model.
7009 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7010 already all have their dedicated checks with a deeper understanding of
7011 the respective protocols.
7012 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007013 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007014
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007015 A fifth mode can be used to insert comments in different steps of the
7016 script.
7017
7018 For each tcp-check rule you create, you can add a "comment" directive,
7019 followed by a string. This string will be reported in the log and stderr
7020 in debug mode. It is useful to make user-friendly error reporting.
7021 The "comment" is of course optional.
7022
7023
Willy Tarreaued179852013-12-16 01:07:00 +01007024 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007025 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007026 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007027 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007028
Davor Ocelice9ed2812017-12-25 17:49:28 +01007029 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007030 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007031 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007032
7033 # look for the redis master server after ensuring it speaks well
7034 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007035 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007036 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007037 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007038 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007039 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007040 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007041 tcp-check send info\ replication\r\n
7042 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007043 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007044 tcp-check send QUIT\r\n
7045 tcp-check expect string +OK
7046
Davor Ocelice9ed2812017-12-25 17:49:28 +01007047 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007048 (send many headers before analyzing)
7049 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007050 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007051 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7052 tcp-check send Host:\ www.mydomain.com\r\n
7053 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7054 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007055 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007056
7057
7058 See also : "tcp-check expect", "tcp-check send"
7059
7060
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007061option tcp-smart-accept
7062no option tcp-smart-accept
7063 Enable or disable the saving of one ACK packet during the accept sequence
7064 May be used in sections : defaults | frontend | listen | backend
7065 yes | yes | yes | no
7066 Arguments : none
7067
7068 When an HTTP connection request comes in, the system acknowledges it on
7069 behalf of HAProxy, then the client immediately sends its request, and the
7070 system acknowledges it too while it is notifying HAProxy about the new
7071 connection. HAProxy then reads the request and responds. This means that we
7072 have one TCP ACK sent by the system for nothing, because the request could
7073 very well be acknowledged by HAProxy when it sends its response.
7074
7075 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7076 sending this useless ACK on platforms which support it (currently at least
7077 Linux). It must not cause any problem, because the system will send it anyway
7078 after 40 ms if the response takes more time than expected to come.
7079
7080 During complex network debugging sessions, it may be desirable to disable
7081 this optimization because delayed ACKs can make troubleshooting more complex
7082 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007083 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007084
7085 It is also possible to force it for non-HTTP proxies by simply specifying
7086 "option tcp-smart-accept". For instance, it can make sense with some services
7087 such as SMTP where the server speaks first.
7088
7089 It is recommended to avoid forcing this option in a defaults section. In case
7090 of doubt, consider setting it back to automatic values by prepending the
7091 "default" keyword before it, or disabling it using the "no" keyword.
7092
Willy Tarreaud88edf22009-06-14 15:48:17 +02007093 See also : "option tcp-smart-connect"
7094
7095
7096option tcp-smart-connect
7097no option tcp-smart-connect
7098 Enable or disable the saving of one ACK packet during the connect sequence
7099 May be used in sections : defaults | frontend | listen | backend
7100 yes | no | yes | yes
7101 Arguments : none
7102
7103 On certain systems (at least Linux), HAProxy can ask the kernel not to
7104 immediately send an empty ACK upon a connection request, but to directly
7105 send the buffer request instead. This saves one packet on the network and
7106 thus boosts performance. It can also be useful for some servers, because they
7107 immediately get the request along with the incoming connection.
7108
7109 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7110 It is not enabled by default because it makes network troubleshooting more
7111 complex.
7112
7113 It only makes sense to enable it with protocols where the client speaks first
7114 such as HTTP. In other situations, if there is no data to send in place of
7115 the ACK, a normal ACK is sent.
7116
7117 If this option has been enabled in a "defaults" section, it can be disabled
7118 in a specific instance by prepending the "no" keyword before it.
7119
7120 See also : "option tcp-smart-accept"
7121
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007122
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007123option tcpka
7124 Enable or disable the sending of TCP keepalive packets on both sides
7125 May be used in sections : defaults | frontend | listen | backend
7126 yes | yes | yes | yes
7127 Arguments : none
7128
7129 When there is a firewall or any session-aware component between a client and
7130 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007131 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007132 components decides to expire a session which has remained idle for too long.
7133
7134 Enabling socket-level TCP keep-alives makes the system regularly send packets
7135 to the other end of the connection, leaving it active. The delay between
7136 keep-alive probes is controlled by the system only and depends both on the
7137 operating system and its tuning parameters.
7138
7139 It is important to understand that keep-alive packets are neither emitted nor
7140 received at the application level. It is only the network stacks which sees
7141 them. For this reason, even if one side of the proxy already uses keep-alives
7142 to maintain its connection alive, those keep-alive packets will not be
7143 forwarded to the other side of the proxy.
7144
7145 Please note that this has nothing to do with HTTP keep-alive.
7146
7147 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7148 the client and server sides of a connection. Note that this is meaningful
7149 only in "defaults" or "listen" sections. If this option is used in a
7150 frontend, only the client side will get keep-alives, and if this option is
7151 used in a backend, only the server side will get keep-alives. For this
7152 reason, it is strongly recommended to explicitly use "option clitcpka" and
7153 "option srvtcpka" when the configuration is split between frontends and
7154 backends.
7155
7156 See also : "option clitcpka", "option srvtcpka"
7157
Willy Tarreau844e3c52008-01-11 16:28:18 +01007158
7159option tcplog
7160 Enable advanced logging of TCP connections with session state and timers
7161 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007162 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007163 Arguments : none
7164
7165 By default, the log output format is very poor, as it only contains the
7166 source and destination addresses, and the instance name. By specifying
7167 "option tcplog", each log line turns into a much richer format including, but
7168 not limited to, the connection timers, the session status, the connections
7169 numbers, the frontend, backend and server name, and of course the source
7170 address and ports. This option is useful for pure TCP proxies in order to
7171 find which of the client or server disconnects or times out. For normal HTTP
7172 proxies, it's better to use "option httplog" which is even more complete.
7173
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007174 "option tcplog" overrides any previous "log-format" directive.
7175
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007176 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007177
7178
Willy Tarreau844e3c52008-01-11 16:28:18 +01007179option transparent
7180no option transparent
7181 Enable client-side transparent proxying
7182 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007183 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007184 Arguments : none
7185
7186 This option was introduced in order to provide layer 7 persistence to layer 3
7187 load balancers. The idea is to use the OS's ability to redirect an incoming
7188 connection for a remote address to a local process (here HAProxy), and let
7189 this process know what address was initially requested. When this option is
7190 used, sessions without cookies will be forwarded to the original destination
7191 IP address of the incoming request (which should match that of another
7192 equipment), while requests with cookies will still be forwarded to the
7193 appropriate server.
7194
7195 Note that contrary to a common belief, this option does NOT make HAProxy
7196 present the client's IP to the server when establishing the connection.
7197
Willy Tarreaua1146052011-03-01 09:51:54 +01007198 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007199 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007200
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007201
Simon Horman98637e52014-06-20 12:30:16 +09007202external-check command <command>
7203 Executable to run when performing an external-check
7204 May be used in sections : defaults | frontend | listen | backend
7205 yes | no | yes | yes
7206
7207 Arguments :
7208 <command> is the external command to run
7209
Simon Horman98637e52014-06-20 12:30:16 +09007210 The arguments passed to the to the command are:
7211
Cyril Bonté777be862014-12-02 21:21:35 +01007212 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007213
Cyril Bonté777be862014-12-02 21:21:35 +01007214 The <proxy_address> and <proxy_port> are derived from the first listener
7215 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7216 listener the proxy_address will be the path of the socket and the
7217 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7218 possible to determine a listener, and both <proxy_address> and <proxy_port>
7219 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007220
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007221 Some values are also provided through environment variables.
7222
7223 Environment variables :
7224 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7225 applicable, for example in a "backend" section).
7226
7227 HAPROXY_PROXY_ID The backend id.
7228
7229 HAPROXY_PROXY_NAME The backend name.
7230
7231 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7232 applicable, for example in a "backend" section or
7233 for a UNIX socket).
7234
7235 HAPROXY_SERVER_ADDR The server address.
7236
7237 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7238
7239 HAPROXY_SERVER_ID The server id.
7240
7241 HAPROXY_SERVER_MAXCONN The server max connections.
7242
7243 HAPROXY_SERVER_NAME The server name.
7244
7245 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7246 socket).
7247
7248 PATH The PATH environment variable used when executing
7249 the command may be set using "external-check path".
7250
Simon Horman98637e52014-06-20 12:30:16 +09007251 If the command executed and exits with a zero status then the check is
7252 considered to have passed, otherwise the check is considered to have
7253 failed.
7254
7255 Example :
7256 external-check command /bin/true
7257
7258 See also : "external-check", "option external-check", "external-check path"
7259
7260
7261external-check path <path>
7262 The value of the PATH environment variable used when running an external-check
7263 May be used in sections : defaults | frontend | listen | backend
7264 yes | no | yes | yes
7265
7266 Arguments :
7267 <path> is the path used when executing external command to run
7268
7269 The default path is "".
7270
7271 Example :
7272 external-check path "/usr/bin:/bin"
7273
7274 See also : "external-check", "option external-check",
7275 "external-check command"
7276
7277
Emeric Brun647caf12009-06-30 17:57:00 +02007278persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007279persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007280 Enable RDP cookie-based persistence
7281 May be used in sections : defaults | frontend | listen | backend
7282 yes | no | yes | yes
7283 Arguments :
7284 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007285 default cookie name "msts" will be used. There currently is no
7286 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007287
7288 This statement enables persistence based on an RDP cookie. The RDP cookie
7289 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007290 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007291 and if an RDP cookie is found, it is decoded. If it matches a known server
7292 which is still UP (or if "option persist" is set), then the connection is
7293 forwarded to this server.
7294
7295 Note that this only makes sense in a TCP backend, but for this to work, the
7296 frontend must have waited long enough to ensure that an RDP cookie is present
7297 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007298 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007299 a single "listen" section.
7300
Willy Tarreau61e28f22010-05-16 22:31:05 +02007301 Also, it is important to understand that the terminal server will emit this
7302 RDP cookie only if it is configured for "token redirection mode", which means
7303 that the "IP address redirection" option is disabled.
7304
Emeric Brun647caf12009-06-30 17:57:00 +02007305 Example :
7306 listen tse-farm
7307 bind :3389
7308 # wait up to 5s for an RDP cookie in the request
7309 tcp-request inspect-delay 5s
7310 tcp-request content accept if RDP_COOKIE
7311 # apply RDP cookie persistence
7312 persist rdp-cookie
7313 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007314 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007315 balance rdp-cookie
7316 server srv1 1.1.1.1:3389
7317 server srv2 1.1.1.2:3389
7318
Simon Hormanab814e02011-06-24 14:50:20 +09007319 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7320 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007321
7322
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007323rate-limit sessions <rate>
7324 Set a limit on the number of new sessions accepted per second on a frontend
7325 May be used in sections : defaults | frontend | listen | backend
7326 yes | yes | yes | no
7327 Arguments :
7328 <rate> The <rate> parameter is an integer designating the maximum number
7329 of new sessions per second to accept on the frontend.
7330
7331 When the frontend reaches the specified number of new sessions per second, it
7332 stops accepting new connections until the rate drops below the limit again.
7333 During this time, the pending sessions will be kept in the socket's backlog
7334 (in system buffers) and haproxy will not even be aware that sessions are
7335 pending. When applying very low limit on a highly loaded service, it may make
7336 sense to increase the socket's backlog using the "backlog" keyword.
7337
7338 This feature is particularly efficient at blocking connection-based attacks
7339 or service abuse on fragile servers. Since the session rate is measured every
7340 millisecond, it is extremely accurate. Also, the limit applies immediately,
7341 no delay is needed at all to detect the threshold.
7342
7343 Example : limit the connection rate on SMTP to 10 per second max
7344 listen smtp
7345 mode tcp
7346 bind :25
7347 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007348 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007349
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007350 Note : when the maximum rate is reached, the frontend's status is not changed
7351 but its sockets appear as "WAITING" in the statistics if the
7352 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007353
7354 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7355
7356
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007357redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7358redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7359redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007360 Return an HTTP redirection if/unless a condition is matched
7361 May be used in sections : defaults | frontend | listen | backend
7362 no | yes | yes | yes
7363
7364 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007365 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007366
Willy Tarreau0140f252008-11-19 21:07:09 +01007367 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007368 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007369 the HTTP "Location" header. When used in an "http-request" rule,
7370 <loc> value follows the log-format rules and can include some
7371 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007372
7373 <pfx> With "redirect prefix", the "Location" header is built from the
7374 concatenation of <pfx> and the complete URI path, including the
7375 query string, unless the "drop-query" option is specified (see
7376 below). As a special case, if <pfx> equals exactly "/", then
7377 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007378 redirect to the same URL (for instance, to insert a cookie). When
7379 used in an "http-request" rule, <pfx> value follows the log-format
7380 rules and can include some dynamic values (see Custom Log Format
7381 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007382
7383 <sch> With "redirect scheme", then the "Location" header is built by
7384 concatenating <sch> with "://" then the first occurrence of the
7385 "Host" header, and then the URI path, including the query string
7386 unless the "drop-query" option is specified (see below). If no
7387 path is found or if the path is "*", then "/" is used instead. If
7388 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007389 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007390 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007391 HTTPS. When used in an "http-request" rule, <sch> value follows
7392 the log-format rules and can include some dynamic values (see
7393 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007394
7395 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007396 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7397 with 302 used by default if no code is specified. 301 means
7398 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007399 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007400 cache the redirection. 303 is equivalent to 302 except that the
7401 browser will fetch the location with a GET method. 307 is just
7402 like 302 but makes it clear that the same method must be reused.
7403 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007404
7405 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007406 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007407
7408 - "drop-query"
7409 When this keyword is used in a prefix-based redirection, then the
7410 location will be set without any possible query-string, which is useful
7411 for directing users to a non-secure page for instance. It has no effect
7412 with a location-type redirect.
7413
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007414 - "append-slash"
7415 This keyword may be used in conjunction with "drop-query" to redirect
7416 users who use a URL not ending with a '/' to the same one with the '/'.
7417 It can be useful to ensure that search engines will only see one URL.
7418 For this, a return code 301 is preferred.
7419
Willy Tarreau0140f252008-11-19 21:07:09 +01007420 - "set-cookie NAME[=value]"
7421 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7422 to the response. This is sometimes used to indicate that a user has
7423 been seen, for instance to protect against some types of DoS. No other
7424 cookie option is added, so the cookie will be a session cookie. Note
7425 that for a browser, a sole cookie name without an equal sign is
7426 different from a cookie with an equal sign.
7427
7428 - "clear-cookie NAME[=]"
7429 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7430 with the "Max-Age" attribute set to zero. This will tell the browser to
7431 delete this cookie. It is useful for instance on logout pages. It is
7432 important to note that clearing the cookie "NAME" will not remove a
7433 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7434 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007435
7436 Example: move the login URL only to HTTPS.
7437 acl clear dst_port 80
7438 acl secure dst_port 8080
7439 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007440 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007441 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007442 acl cookie_set hdr_sub(cookie) SEEN=1
7443
7444 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007445 redirect prefix https://mysite.com if login_page !secure
7446 redirect prefix http://mysite.com drop-query if login_page !uid_given
7447 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007448 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007449
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007450 Example: send redirects for request for articles without a '/'.
7451 acl missing_slash path_reg ^/article/[^/]*$
7452 redirect code 301 prefix / drop-query append-slash if missing_slash
7453
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007454 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007455 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007456
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007457 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007458 http-request redirect code 301 location \
7459 http://www.%[hdr(host)]%[capture.req.uri] \
7460 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007461
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007462 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007463
7464
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007465redisp (deprecated)
7466redispatch (deprecated)
7467 Enable or disable session redistribution in case of connection failure
7468 May be used in sections: defaults | frontend | listen | backend
7469 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007470 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007471
7472 In HTTP mode, if a server designated by a cookie is down, clients may
7473 definitely stick to it because they cannot flush the cookie, so they will not
7474 be able to access the service anymore.
7475
7476 Specifying "redispatch" will allow the proxy to break their persistence and
7477 redistribute them to a working server.
7478
7479 It also allows to retry last connection to another server in case of multiple
7480 connection failures. Of course, it requires having "retries" set to a nonzero
7481 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007482
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007483 This form is deprecated, do not use it in any new configuration, use the new
7484 "option redispatch" instead.
7485
7486 See also : "option redispatch"
7487
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007488
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007489reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007490 Add a header at the end of the HTTP request
7491 May be used in sections : defaults | frontend | listen | backend
7492 no | yes | yes | yes
7493 Arguments :
7494 <string> is the complete line to be added. Any space or known delimiter
7495 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007496 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007497
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007498 <cond> is an optional matching condition built from ACLs. It makes it
7499 possible to ignore this rule when other conditions are not met.
7500
Willy Tarreau303c0352008-01-17 19:01:39 +01007501 A new line consisting in <string> followed by a line feed will be added after
7502 the last header of an HTTP request.
7503
7504 Header transformations only apply to traffic which passes through HAProxy,
7505 and not to traffic generated by HAProxy, such as health-checks or error
7506 responses.
7507
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007508 Example : add "X-Proto: SSL" to requests coming via port 81
7509 acl is-ssl dst_port 81
7510 reqadd X-Proto:\ SSL if is-ssl
7511
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007512 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7513 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007514
7515
Willy Tarreau5321c422010-01-28 20:35:13 +01007516reqallow <search> [{if | unless} <cond>]
7517reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007518 Definitely allow an HTTP request if a line matches a regular expression
7519 May be used in sections : defaults | frontend | listen | backend
7520 no | yes | yes | yes
7521 Arguments :
7522 <search> is the regular expression applied to HTTP headers and to the
7523 request line. This is an extended regular expression. Parenthesis
7524 grouping is supported and no preliminary backslash is required.
7525 Any space or known delimiter must be escaped using a backslash
7526 ('\'). The pattern applies to a full line at a time. The
7527 "reqallow" keyword strictly matches case while "reqiallow"
7528 ignores case.
7529
Willy Tarreau5321c422010-01-28 20:35:13 +01007530 <cond> is an optional matching condition built from ACLs. It makes it
7531 possible to ignore this rule when other conditions are not met.
7532
Willy Tarreau303c0352008-01-17 19:01:39 +01007533 A request containing any line which matches extended regular expression
7534 <search> will mark the request as allowed, even if any later test would
7535 result in a deny. The test applies both to the request line and to request
7536 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007537 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007538
7539 It is easier, faster and more powerful to use ACLs to write access policies.
7540 Reqdeny, reqallow and reqpass should be avoided in new designs.
7541
7542 Example :
7543 # allow www.* but refuse *.local
7544 reqiallow ^Host:\ www\.
7545 reqideny ^Host:\ .*\.local
7546
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007547 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7548 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007549
7550
Willy Tarreau5321c422010-01-28 20:35:13 +01007551reqdel <search> [{if | unless} <cond>]
7552reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007553 Delete all headers matching a regular expression in an HTTP request
7554 May be used in sections : defaults | frontend | listen | backend
7555 no | yes | yes | yes
7556 Arguments :
7557 <search> is the regular expression applied to HTTP headers and to the
7558 request line. This is an extended regular expression. Parenthesis
7559 grouping is supported and no preliminary backslash is required.
7560 Any space or known delimiter must be escaped using a backslash
7561 ('\'). The pattern applies to a full line at a time. The "reqdel"
7562 keyword strictly matches case while "reqidel" ignores case.
7563
Willy Tarreau5321c422010-01-28 20:35:13 +01007564 <cond> is an optional matching condition built from ACLs. It makes it
7565 possible to ignore this rule when other conditions are not met.
7566
Willy Tarreau303c0352008-01-17 19:01:39 +01007567 Any header line matching extended regular expression <search> in the request
7568 will be completely deleted. Most common use of this is to remove unwanted
7569 and/or dangerous headers or cookies from a request before passing it to the
7570 next servers.
7571
7572 Header transformations only apply to traffic which passes through HAProxy,
7573 and not to traffic generated by HAProxy, such as health-checks or error
7574 responses. Keep in mind that header names are not case-sensitive.
7575
7576 Example :
7577 # remove X-Forwarded-For header and SERVER cookie
7578 reqidel ^X-Forwarded-For:.*
7579 reqidel ^Cookie:.*SERVER=
7580
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007581 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7582 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007583
7584
Willy Tarreau5321c422010-01-28 20:35:13 +01007585reqdeny <search> [{if | unless} <cond>]
7586reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007587 Deny an HTTP request if a line matches a regular expression
7588 May be used in sections : defaults | frontend | listen | backend
7589 no | yes | yes | yes
7590 Arguments :
7591 <search> is the regular expression applied to HTTP headers and to the
7592 request line. This is an extended regular expression. Parenthesis
7593 grouping is supported and no preliminary backslash is required.
7594 Any space or known delimiter must be escaped using a backslash
7595 ('\'). The pattern applies to a full line at a time. The
7596 "reqdeny" keyword strictly matches case while "reqideny" ignores
7597 case.
7598
Willy Tarreau5321c422010-01-28 20:35:13 +01007599 <cond> is an optional matching condition built from ACLs. It makes it
7600 possible to ignore this rule when other conditions are not met.
7601
Willy Tarreau303c0352008-01-17 19:01:39 +01007602 A request containing any line which matches extended regular expression
7603 <search> will mark the request as denied, even if any later test would
7604 result in an allow. The test applies both to the request line and to request
7605 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007606 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007607
Willy Tarreauced27012008-01-17 20:35:34 +01007608 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007609 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007610 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007611
Willy Tarreau303c0352008-01-17 19:01:39 +01007612 It is easier, faster and more powerful to use ACLs to write access policies.
7613 Reqdeny, reqallow and reqpass should be avoided in new designs.
7614
7615 Example :
7616 # refuse *.local, then allow www.*
7617 reqideny ^Host:\ .*\.local
7618 reqiallow ^Host:\ www\.
7619
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007620 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7621 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007622
7623
Willy Tarreau5321c422010-01-28 20:35:13 +01007624reqpass <search> [{if | unless} <cond>]
7625reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007626 Ignore any HTTP request line matching a regular expression in next rules
7627 May be used in sections : defaults | frontend | listen | backend
7628 no | yes | yes | yes
7629 Arguments :
7630 <search> is the regular expression applied to HTTP headers and to the
7631 request line. This is an extended regular expression. Parenthesis
7632 grouping is supported and no preliminary backslash is required.
7633 Any space or known delimiter must be escaped using a backslash
7634 ('\'). The pattern applies to a full line at a time. The
7635 "reqpass" keyword strictly matches case while "reqipass" ignores
7636 case.
7637
Willy Tarreau5321c422010-01-28 20:35:13 +01007638 <cond> is an optional matching condition built from ACLs. It makes it
7639 possible to ignore this rule when other conditions are not met.
7640
Willy Tarreau303c0352008-01-17 19:01:39 +01007641 A request containing any line which matches extended regular expression
7642 <search> will skip next rules, without assigning any deny or allow verdict.
7643 The test applies both to the request line and to request headers. Keep in
7644 mind that URLs in request line are case-sensitive while header names are not.
7645
7646 It is easier, faster and more powerful to use ACLs to write access policies.
7647 Reqdeny, reqallow and reqpass should be avoided in new designs.
7648
7649 Example :
7650 # refuse *.local, then allow www.*, but ignore "www.private.local"
7651 reqipass ^Host:\ www.private\.local
7652 reqideny ^Host:\ .*\.local
7653 reqiallow ^Host:\ www\.
7654
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007655 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7656 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007657
7658
Willy Tarreau5321c422010-01-28 20:35:13 +01007659reqrep <search> <string> [{if | unless} <cond>]
7660reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007661 Replace a regular expression with a string in an HTTP request line
7662 May be used in sections : defaults | frontend | listen | backend
7663 no | yes | yes | yes
7664 Arguments :
7665 <search> is the regular expression applied to HTTP headers and to the
7666 request line. This is an extended regular expression. Parenthesis
7667 grouping is supported and no preliminary backslash is required.
7668 Any space or known delimiter must be escaped using a backslash
7669 ('\'). The pattern applies to a full line at a time. The "reqrep"
7670 keyword strictly matches case while "reqirep" ignores case.
7671
7672 <string> is the complete line to be added. Any space or known delimiter
7673 must be escaped using a backslash ('\'). References to matched
7674 pattern groups are possible using the common \N form, with N
7675 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007676 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007677
Willy Tarreau5321c422010-01-28 20:35:13 +01007678 <cond> is an optional matching condition built from ACLs. It makes it
7679 possible to ignore this rule when other conditions are not met.
7680
Willy Tarreau303c0352008-01-17 19:01:39 +01007681 Any line matching extended regular expression <search> in the request (both
7682 the request line and header lines) will be completely replaced with <string>.
7683 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7684
7685 Header transformations only apply to traffic which passes through HAProxy,
7686 and not to traffic generated by HAProxy, such as health-checks or error
7687 responses. Note that for increased readability, it is suggested to add enough
7688 spaces between the request and the response. Keep in mind that URLs in
7689 request line are case-sensitive while header names are not.
7690
7691 Example :
7692 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007693 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007694 # replace "www.mydomain.com" with "www" in the host name.
7695 reqirep ^Host:\ www.mydomain.com Host:\ www
7696
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007697 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7698 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007699
7700
Willy Tarreau5321c422010-01-28 20:35:13 +01007701reqtarpit <search> [{if | unless} <cond>]
7702reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007703 Tarpit an HTTP request containing a line matching a regular expression
7704 May be used in sections : defaults | frontend | listen | backend
7705 no | yes | yes | yes
7706 Arguments :
7707 <search> is the regular expression applied to HTTP headers and to the
7708 request line. This is an extended regular expression. Parenthesis
7709 grouping is supported and no preliminary backslash is required.
7710 Any space or known delimiter must be escaped using a backslash
7711 ('\'). The pattern applies to a full line at a time. The
7712 "reqtarpit" keyword strictly matches case while "reqitarpit"
7713 ignores case.
7714
Willy Tarreau5321c422010-01-28 20:35:13 +01007715 <cond> is an optional matching condition built from ACLs. It makes it
7716 possible to ignore this rule when other conditions are not met.
7717
Willy Tarreau303c0352008-01-17 19:01:39 +01007718 A request containing any line which matches extended regular expression
7719 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007720 be kept open for a pre-defined time, then will return an HTTP error 500 so
7721 that the attacker does not suspect it has been tarpitted. The status 500 will
7722 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007723 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7724 not set.
7725
7726 The goal of the tarpit is to slow down robots attacking servers with
7727 identifiable requests. Many robots limit their outgoing number of connections
7728 and stay connected waiting for a reply which can take several minutes to
7729 come. Depending on the environment and attack, it may be particularly
7730 efficient at reducing the load on the network and firewalls.
7731
Willy Tarreau5321c422010-01-28 20:35:13 +01007732 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007733 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007734 # block all others.
7735 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7736 reqitarpit ^User-Agent:
7737
Willy Tarreau5321c422010-01-28 20:35:13 +01007738 # block bad guys
7739 acl badguys src 10.1.0.3 172.16.13.20/28
7740 reqitarpit . if badguys
7741
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007742 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7743 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007744
7745
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007746retries <value>
7747 Set the number of retries to perform on a server after a connection failure
7748 May be used in sections: defaults | frontend | listen | backend
7749 yes | no | yes | yes
7750 Arguments :
7751 <value> is the number of times a connection attempt should be retried on
7752 a server when a connection either is refused or times out. The
7753 default value is 3.
7754
7755 It is important to understand that this value applies to the number of
7756 connection attempts, not full requests. When a connection has effectively
7757 been established to a server, there will be no more retry.
7758
7759 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007760 a turn-around timer of min("timeout connect", one second) is applied before
7761 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007762
7763 When "option redispatch" is set, the last retry may be performed on another
7764 server even if a cookie references a different server.
7765
7766 See also : "option redispatch"
7767
7768
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007769rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007770 Add a header at the end of the HTTP response
7771 May be used in sections : defaults | frontend | listen | backend
7772 no | yes | yes | yes
7773 Arguments :
7774 <string> is the complete line to be added. Any space or known delimiter
7775 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007776 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007777
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007778 <cond> is an optional matching condition built from ACLs. It makes it
7779 possible to ignore this rule when other conditions are not met.
7780
Willy Tarreau303c0352008-01-17 19:01:39 +01007781 A new line consisting in <string> followed by a line feed will be added after
7782 the last header of an HTTP response.
7783
7784 Header transformations only apply to traffic which passes through HAProxy,
7785 and not to traffic generated by HAProxy, such as health-checks or error
7786 responses.
7787
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007788 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7789 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007790
7791
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007792rspdel <search> [{if | unless} <cond>]
7793rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007794 Delete all headers matching a regular expression in an HTTP response
7795 May be used in sections : defaults | frontend | listen | backend
7796 no | yes | yes | yes
7797 Arguments :
7798 <search> is the regular expression applied to HTTP headers and to the
7799 response line. This is an extended regular expression, so
7800 parenthesis grouping is supported and no preliminary backslash
7801 is required. Any space or known delimiter must be escaped using
7802 a backslash ('\'). The pattern applies to a full line at a time.
7803 The "rspdel" keyword strictly matches case while "rspidel"
7804 ignores case.
7805
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007806 <cond> is an optional matching condition built from ACLs. It makes it
7807 possible to ignore this rule when other conditions are not met.
7808
Willy Tarreau303c0352008-01-17 19:01:39 +01007809 Any header line matching extended regular expression <search> in the response
7810 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007811 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007812 client.
7813
7814 Header transformations only apply to traffic which passes through HAProxy,
7815 and not to traffic generated by HAProxy, such as health-checks or error
7816 responses. Keep in mind that header names are not case-sensitive.
7817
7818 Example :
7819 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007820 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007821
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007822 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7823 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007824
7825
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007826rspdeny <search> [{if | unless} <cond>]
7827rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007828 Block an HTTP response if a line matches a regular expression
7829 May be used in sections : defaults | frontend | listen | backend
7830 no | yes | yes | yes
7831 Arguments :
7832 <search> is the regular expression applied to HTTP headers and to the
7833 response line. This is an extended regular expression, so
7834 parenthesis grouping is supported and no preliminary backslash
7835 is required. Any space or known delimiter must be escaped using
7836 a backslash ('\'). The pattern applies to a full line at a time.
7837 The "rspdeny" keyword strictly matches case while "rspideny"
7838 ignores case.
7839
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007840 <cond> is an optional matching condition built from ACLs. It makes it
7841 possible to ignore this rule when other conditions are not met.
7842
Willy Tarreau303c0352008-01-17 19:01:39 +01007843 A response containing any line which matches extended regular expression
7844 <search> will mark the request as denied. The test applies both to the
7845 response line and to response headers. Keep in mind that header names are not
7846 case-sensitive.
7847
7848 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007849 block the response before it reaches the client. If a response is denied, it
7850 will be replaced with an HTTP 502 error so that the client never retrieves
7851 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007852
7853 It is easier, faster and more powerful to use ACLs to write access policies.
7854 Rspdeny should be avoided in new designs.
7855
7856 Example :
7857 # Ensure that no content type matching ms-word will leak
7858 rspideny ^Content-type:\.*/ms-word
7859
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007860 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7861 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007862
7863
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007864rsprep <search> <string> [{if | unless} <cond>]
7865rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007866 Replace a regular expression with a string in an HTTP response line
7867 May be used in sections : defaults | frontend | listen | backend
7868 no | yes | yes | yes
7869 Arguments :
7870 <search> is the regular expression applied to HTTP headers and to the
7871 response line. This is an extended regular expression, so
7872 parenthesis grouping is supported and no preliminary backslash
7873 is required. Any space or known delimiter must be escaped using
7874 a backslash ('\'). The pattern applies to a full line at a time.
7875 The "rsprep" keyword strictly matches case while "rspirep"
7876 ignores case.
7877
7878 <string> is the complete line to be added. Any space or known delimiter
7879 must be escaped using a backslash ('\'). References to matched
7880 pattern groups are possible using the common \N form, with N
7881 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007882 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007883
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007884 <cond> is an optional matching condition built from ACLs. It makes it
7885 possible to ignore this rule when other conditions are not met.
7886
Willy Tarreau303c0352008-01-17 19:01:39 +01007887 Any line matching extended regular expression <search> in the response (both
7888 the response line and header lines) will be completely replaced with
7889 <string>. Most common use of this is to rewrite Location headers.
7890
7891 Header transformations only apply to traffic which passes through HAProxy,
7892 and not to traffic generated by HAProxy, such as health-checks or error
7893 responses. Note that for increased readability, it is suggested to add enough
7894 spaces between the request and the response. Keep in mind that header names
7895 are not case-sensitive.
7896
7897 Example :
7898 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7899 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7900
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007901 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7902 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007903
7904
David du Colombier486df472011-03-17 10:40:26 +01007905server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007906 Declare a server in a backend
7907 May be used in sections : defaults | frontend | listen | backend
7908 no | no | yes | yes
7909 Arguments :
7910 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007911 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007912 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007913
David du Colombier486df472011-03-17 10:40:26 +01007914 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7915 resolvable hostname is supported, but this name will be resolved
7916 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7917 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007918 address as the one from the client connection. This is useful in
7919 transparent proxy architectures where the client's connection is
7920 intercepted and haproxy must forward to the original destination
7921 address. This is more or less what the "transparent" keyword does
7922 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007923 to report statistics. Optionally, an address family prefix may be
7924 used before the address to force the family regardless of the
7925 address format, which can be useful to specify a path to a unix
7926 socket with no slash ('/'). Currently supported prefixes are :
7927 - 'ipv4@' -> address is always IPv4
7928 - 'ipv6@' -> address is always IPv6
7929 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007930 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007931 - 'sockpair@' -> address is the FD of a connected unix
7932 socket or of a socketpair. During a connection, the
7933 backend creates a pair of connected sockets, and passes
7934 one of them over the FD. The bind part will use the
7935 received socket as the client FD. Should be used
7936 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007937 You may want to reference some environment variables in the
7938 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007939 variables. The "init-addr" setting can be used to modify the way
7940 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007941
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007942 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007943 be sent to this port. If unset, the same port the client
7944 connected to will be used. The port may also be prefixed by a "+"
7945 or a "-". In this case, the server's port will be determined by
7946 adding this value to the client's port.
7947
7948 <param*> is a list of parameters for this server. The "server" keywords
7949 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007950 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007951
7952 Examples :
7953 server first 10.1.1.1:1080 cookie first check inter 1000
7954 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007955 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007956 server backup "${SRV_BACKUP}:1080" backup
7957 server www1_dc1 "${LAN_DC1}.101:80"
7958 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007959
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007960 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7961 sun_path length is used for the address length. Some other programs
7962 such as socat use the string length only by default. Pass the option
7963 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7964 make it compatible with HAProxy's.
7965
Mark Lamourinec2247f02012-01-04 13:02:01 -05007966 See also: "default-server", "http-send-name-header" and section 5 about
7967 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007968
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007969server-state-file-name [<file>]
7970 Set the server state file to read, load and apply to servers available in
7971 this backend. It only applies when the directive "load-server-state-from-file"
7972 is set to "local". When <file> is not provided or if this directive is not
7973 set, then backend name is used. If <file> starts with a slash '/', then it is
7974 considered as an absolute path. Otherwise, <file> is concatenated to the
7975 global directive "server-state-file-base".
7976
7977 Example: the minimal configuration below would make HAProxy look for the
7978 state server file '/etc/haproxy/states/bk':
7979
7980 global
7981 server-state-file-base /etc/haproxy/states
7982
7983 backend bk
7984 load-server-state-from-file
7985
7986 See also: "server-state-file-base", "load-server-state-from-file", and
7987 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007988
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007989server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7990 Set a template to initialize servers with shared parameters.
7991 The names of these servers are built from <prefix> and <num | range> parameters.
7992 May be used in sections : defaults | frontend | listen | backend
7993 no | no | yes | yes
7994
7995 Arguments:
7996 <prefix> A prefix for the server names to be built.
7997
7998 <num | range>
7999 If <num> is provided, this template initializes <num> servers
8000 with 1 up to <num> as server name suffixes. A range of numbers
8001 <num_low>-<num_high> may also be used to use <num_low> up to
8002 <num_high> as server name suffixes.
8003
8004 <fqdn> A FQDN for all the servers this template initializes.
8005
8006 <port> Same meaning as "server" <port> argument (see "server" keyword).
8007
8008 <params*>
8009 Remaining server parameters among all those supported by "server"
8010 keyword.
8011
8012 Examples:
8013 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8014 # google.com as FQDN, and health-check enabled.
8015 server-template srv 1-3 google.com:80 check
8016
8017 # or
8018 server-template srv 3 google.com:80 check
8019
8020 # would be equivalent to:
8021 server srv1 google.com:80 check
8022 server srv2 google.com:80 check
8023 server srv3 google.com:80 check
8024
8025
8026
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008027source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008028source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008029source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008030 Set the source address for outgoing connections
8031 May be used in sections : defaults | frontend | listen | backend
8032 yes | no | yes | yes
8033 Arguments :
8034 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8035 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008036
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008037 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008038 the most appropriate address to reach its destination. Optionally
8039 an address family prefix may be used before the address to force
8040 the family regardless of the address format, which can be useful
8041 to specify a path to a unix socket with no slash ('/'). Currently
8042 supported prefixes are :
8043 - 'ipv4@' -> address is always IPv4
8044 - 'ipv6@' -> address is always IPv6
8045 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008046 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008047 You may want to reference some environment variables in the
8048 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008049
8050 <port> is an optional port. It is normally not needed but may be useful
8051 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008052 the system will select a free port. Note that port ranges are not
8053 supported in the backend. If you want to force port ranges, you
8054 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008055
8056 <addr2> is the IP address to present to the server when connections are
8057 forwarded in full transparent proxy mode. This is currently only
8058 supported on some patched Linux kernels. When this address is
8059 specified, clients connecting to the server will be presented
8060 with this address, while health checks will still use the address
8061 <addr>.
8062
8063 <port2> is the optional port to present to the server when connections
8064 are forwarded in full transparent proxy mode (see <addr2> above).
8065 The default value of zero means the system will select a free
8066 port.
8067
Willy Tarreaubce70882009-09-07 11:51:47 +02008068 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8069 This is the name of a comma-separated header list which can
8070 contain multiple IP addresses. By default, the last occurrence is
8071 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008072 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008073 by previous proxy, typically Stunnel. In order to use another
8074 occurrence from the last one, please see the <occ> parameter
8075 below. When the header (or occurrence) is not found, no binding
8076 is performed so that the proxy's default IP address is used. Also
8077 keep in mind that the header name is case insensitive, as for any
8078 HTTP header.
8079
8080 <occ> is the occurrence number of a value to be used in a multi-value
8081 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008082 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008083 address. Positive values indicate a position from the first
8084 occurrence, 1 being the first one. Negative values indicate
8085 positions relative to the last one, -1 being the last one. This
8086 is helpful for situations where an X-Forwarded-For header is set
8087 at the entry point of an infrastructure and must be used several
8088 proxy layers away. When this value is not specified, -1 is
8089 assumed. Passing a zero here disables the feature.
8090
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008091 <name> is an optional interface name to which to bind to for outgoing
8092 traffic. On systems supporting this features (currently, only
8093 Linux), this allows one to bind all traffic to the server to
8094 this interface even if it is not the one the system would select
8095 based on routing tables. This should be used with extreme care.
8096 Note that using this option requires root privileges.
8097
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008098 The "source" keyword is useful in complex environments where a specific
8099 address only is allowed to connect to the servers. It may be needed when a
8100 private address must be used through a public gateway for instance, and it is
8101 known that the system cannot determine the adequate source address by itself.
8102
8103 An extension which is available on certain patched Linux kernels may be used
8104 through the "usesrc" optional keyword. It makes it possible to connect to the
8105 servers with an IP address which does not belong to the system itself. This
8106 is called "full transparent proxy mode". For this to work, the destination
8107 servers have to route their traffic back to this address through the machine
8108 running HAProxy, and IP forwarding must generally be enabled on this machine.
8109
8110 In this "full transparent proxy" mode, it is possible to force a specific IP
8111 address to be presented to the servers. This is not much used in fact. A more
8112 common use is to tell HAProxy to present the client's IP address. For this,
8113 there are two methods :
8114
8115 - present the client's IP and port addresses. This is the most transparent
8116 mode, but it can cause problems when IP connection tracking is enabled on
8117 the machine, because a same connection may be seen twice with different
8118 states. However, this solution presents the huge advantage of not
8119 limiting the system to the 64k outgoing address+port couples, because all
8120 of the client ranges may be used.
8121
8122 - present only the client's IP address and select a spare port. This
8123 solution is still quite elegant but slightly less transparent (downstream
8124 firewalls logs will not match upstream's). It also presents the downside
8125 of limiting the number of concurrent connections to the usual 64k ports.
8126 However, since the upstream and downstream ports are different, local IP
8127 connection tracking on the machine will not be upset by the reuse of the
8128 same session.
8129
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008130 This option sets the default source for all servers in the backend. It may
8131 also be specified in a "defaults" section. Finer source address specification
8132 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008133 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008134
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008135 In order to work, "usesrc" requires root privileges.
8136
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008137 Examples :
8138 backend private
8139 # Connect to the servers using our 192.168.1.200 source address
8140 source 192.168.1.200
8141
8142 backend transparent_ssl1
8143 # Connect to the SSL farm from the client's source address
8144 source 192.168.1.200 usesrc clientip
8145
8146 backend transparent_ssl2
8147 # Connect to the SSL farm from the client's source address and port
8148 # not recommended if IP conntrack is present on the local machine.
8149 source 192.168.1.200 usesrc client
8150
8151 backend transparent_ssl3
8152 # Connect to the SSL farm from the client's source address. It
8153 # is more conntrack-friendly.
8154 source 192.168.1.200 usesrc clientip
8155
8156 backend transparent_smtp
8157 # Connect to the SMTP farm from the client's source address/port
8158 # with Tproxy version 4.
8159 source 0.0.0.0 usesrc clientip
8160
Willy Tarreaubce70882009-09-07 11:51:47 +02008161 backend transparent_http
8162 # Connect to the servers using the client's IP as seen by previous
8163 # proxy.
8164 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8165
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008166 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008167 the Linux kernel on www.balabit.com, the "bind" keyword.
8168
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008169
Willy Tarreau844e3c52008-01-11 16:28:18 +01008170srvtimeout <timeout> (deprecated)
8171 Set the maximum inactivity time on the server side.
8172 May be used in sections : defaults | frontend | listen | backend
8173 yes | no | yes | yes
8174 Arguments :
8175 <timeout> is the timeout value specified in milliseconds by default, but
8176 can be in any other unit if the number is suffixed by the unit,
8177 as explained at the top of this document.
8178
8179 The inactivity timeout applies when the server is expected to acknowledge or
8180 send data. In HTTP mode, this timeout is particularly important to consider
8181 during the first phase of the server's response, when it has to send the
8182 headers, as it directly represents the server's processing time for the
8183 request. To find out what value to put there, it's often good to start with
8184 what would be considered as unacceptable response times, then check the logs
8185 to observe the response time distribution, and adjust the value accordingly.
8186
8187 The value is specified in milliseconds by default, but can be in any other
8188 unit if the number is suffixed by the unit, as specified at the top of this
8189 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8190 recommended that the client timeout remains equal to the server timeout in
8191 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008192 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008193 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008194 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008195
8196 This parameter is specific to backends, but can be specified once for all in
8197 "defaults" sections. This is in fact one of the easiest solutions not to
8198 forget about it. An unspecified timeout results in an infinite timeout, which
8199 is not recommended. Such a usage is accepted and works but reports a warning
8200 during startup because it may results in accumulation of expired sessions in
8201 the system if the system's timeouts are not configured either.
8202
8203 This parameter is provided for compatibility but is currently deprecated.
8204 Please use "timeout server" instead.
8205
Willy Tarreauce887fd2012-05-12 12:50:00 +02008206 See also : "timeout server", "timeout tunnel", "timeout client" and
8207 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008208
8209
Cyril Bonté66c327d2010-10-12 00:14:37 +02008210stats admin { if | unless } <cond>
8211 Enable statistics admin level if/unless a condition is matched
8212 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008213 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008214
8215 This statement enables the statistics admin level if/unless a condition is
8216 matched.
8217
8218 The admin level allows to enable/disable servers from the web interface. By
8219 default, statistics page is read-only for security reasons.
8220
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008221 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8222 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008223 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008224
Cyril Bonté23b39d92011-02-10 22:54:44 +01008225 Currently, the POST request is limited to the buffer size minus the reserved
8226 buffer space, which means that if the list of servers is too long, the
8227 request won't be processed. It is recommended to alter few servers at a
8228 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008229
8230 Example :
8231 # statistics admin level only for localhost
8232 backend stats_localhost
8233 stats enable
8234 stats admin if LOCALHOST
8235
8236 Example :
8237 # statistics admin level always enabled because of the authentication
8238 backend stats_auth
8239 stats enable
8240 stats auth admin:AdMiN123
8241 stats admin if TRUE
8242
8243 Example :
8244 # statistics admin level depends on the authenticated user
8245 userlist stats-auth
8246 group admin users admin
8247 user admin insecure-password AdMiN123
8248 group readonly users haproxy
8249 user haproxy insecure-password haproxy
8250
8251 backend stats_auth
8252 stats enable
8253 acl AUTH http_auth(stats-auth)
8254 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8255 stats http-request auth unless AUTH
8256 stats admin if AUTH_ADMIN
8257
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008258 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8259 "bind-process", section 3.4 about userlists and section 7 about
8260 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008261
8262
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008263stats auth <user>:<passwd>
8264 Enable statistics with authentication and grant access to an account
8265 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008266 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008267 Arguments :
8268 <user> is a user name to grant access to
8269
8270 <passwd> is the cleartext password associated to this user
8271
8272 This statement enables statistics with default settings, and restricts access
8273 to declared users only. It may be repeated as many times as necessary to
8274 allow as many users as desired. When a user tries to access the statistics
8275 without a valid account, a "401 Forbidden" response will be returned so that
8276 the browser asks the user to provide a valid user and password. The real
8277 which will be returned to the browser is configurable using "stats realm".
8278
8279 Since the authentication method is HTTP Basic Authentication, the passwords
8280 circulate in cleartext on the network. Thus, it was decided that the
8281 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008282 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008283
8284 It is also possible to reduce the scope of the proxies which appear in the
8285 report using "stats scope".
8286
8287 Though this statement alone is enough to enable statistics reporting, it is
8288 recommended to set all other settings in order to avoid relying on default
8289 unobvious parameters.
8290
8291 Example :
8292 # public access (limited to this backend only)
8293 backend public_www
8294 server srv1 192.168.0.1:80
8295 stats enable
8296 stats hide-version
8297 stats scope .
8298 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008299 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008300 stats auth admin1:AdMiN123
8301 stats auth admin2:AdMiN321
8302
8303 # internal monitoring access (unlimited)
8304 backend private_monitoring
8305 stats enable
8306 stats uri /admin?stats
8307 stats refresh 5s
8308
8309 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8310
8311
8312stats enable
8313 Enable statistics reporting with default settings
8314 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008315 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008316 Arguments : none
8317
8318 This statement enables statistics reporting with default settings defined
8319 at build time. Unless stated otherwise, these settings are used :
8320 - stats uri : /haproxy?stats
8321 - stats realm : "HAProxy Statistics"
8322 - stats auth : no authentication
8323 - stats scope : no restriction
8324
8325 Though this statement alone is enough to enable statistics reporting, it is
8326 recommended to set all other settings in order to avoid relying on default
8327 unobvious parameters.
8328
8329 Example :
8330 # public access (limited to this backend only)
8331 backend public_www
8332 server srv1 192.168.0.1:80
8333 stats enable
8334 stats hide-version
8335 stats scope .
8336 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008337 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008338 stats auth admin1:AdMiN123
8339 stats auth admin2:AdMiN321
8340
8341 # internal monitoring access (unlimited)
8342 backend private_monitoring
8343 stats enable
8344 stats uri /admin?stats
8345 stats refresh 5s
8346
8347 See also : "stats auth", "stats realm", "stats uri"
8348
8349
Willy Tarreaud63335a2010-02-26 12:56:52 +01008350stats hide-version
8351 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008352 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008353 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008354 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008355
Willy Tarreaud63335a2010-02-26 12:56:52 +01008356 By default, the stats page reports some useful status information along with
8357 the statistics. Among them is HAProxy's version. However, it is generally
8358 considered dangerous to report precise version to anyone, as it can help them
8359 target known weaknesses with specific attacks. The "stats hide-version"
8360 statement removes the version from the statistics report. This is recommended
8361 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008362
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008363 Though this statement alone is enough to enable statistics reporting, it is
8364 recommended to set all other settings in order to avoid relying on default
8365 unobvious parameters.
8366
Willy Tarreaud63335a2010-02-26 12:56:52 +01008367 Example :
8368 # public access (limited to this backend only)
8369 backend public_www
8370 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008371 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008372 stats hide-version
8373 stats scope .
8374 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008375 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008376 stats auth admin1:AdMiN123
8377 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008378
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008379 # internal monitoring access (unlimited)
8380 backend private_monitoring
8381 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008382 stats uri /admin?stats
8383 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008384
Willy Tarreaud63335a2010-02-26 12:56:52 +01008385 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008386
Willy Tarreau983e01e2010-01-11 18:42:06 +01008387
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008388stats http-request { allow | deny | auth [realm <realm>] }
8389 [ { if | unless } <condition> ]
8390 Access control for statistics
8391
8392 May be used in sections: defaults | frontend | listen | backend
8393 no | no | yes | yes
8394
8395 As "http-request", these set of options allow to fine control access to
8396 statistics. Each option may be followed by if/unless and acl.
8397 First option with matched condition (or option without condition) is final.
8398 For "deny" a 403 error will be returned, for "allow" normal processing is
8399 performed, for "auth" a 401/407 error code is returned so the client
8400 should be asked to enter a username and password.
8401
8402 There is no fixed limit to the number of http-request statements per
8403 instance.
8404
8405 See also : "http-request", section 3.4 about userlists and section 7
8406 about ACL usage.
8407
8408
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008409stats realm <realm>
8410 Enable statistics and set authentication realm
8411 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008412 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008413 Arguments :
8414 <realm> is the name of the HTTP Basic Authentication realm reported to
8415 the browser. The browser uses it to display it in the pop-up
8416 inviting the user to enter a valid username and password.
8417
8418 The realm is read as a single word, so any spaces in it should be escaped
8419 using a backslash ('\').
8420
8421 This statement is useful only in conjunction with "stats auth" since it is
8422 only related to authentication.
8423
8424 Though this statement alone is enough to enable statistics reporting, it is
8425 recommended to set all other settings in order to avoid relying on default
8426 unobvious parameters.
8427
8428 Example :
8429 # public access (limited to this backend only)
8430 backend public_www
8431 server srv1 192.168.0.1:80
8432 stats enable
8433 stats hide-version
8434 stats scope .
8435 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008436 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008437 stats auth admin1:AdMiN123
8438 stats auth admin2:AdMiN321
8439
8440 # internal monitoring access (unlimited)
8441 backend private_monitoring
8442 stats enable
8443 stats uri /admin?stats
8444 stats refresh 5s
8445
8446 See also : "stats auth", "stats enable", "stats uri"
8447
8448
8449stats refresh <delay>
8450 Enable statistics with automatic refresh
8451 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008452 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008453 Arguments :
8454 <delay> is the suggested refresh delay, specified in seconds, which will
8455 be returned to the browser consulting the report page. While the
8456 browser is free to apply any delay, it will generally respect it
8457 and refresh the page this every seconds. The refresh interval may
8458 be specified in any other non-default time unit, by suffixing the
8459 unit after the value, as explained at the top of this document.
8460
8461 This statement is useful on monitoring displays with a permanent page
8462 reporting the load balancer's activity. When set, the HTML report page will
8463 include a link "refresh"/"stop refresh" so that the user can select whether
8464 he wants automatic refresh of the page or not.
8465
8466 Though this statement alone is enough to enable statistics reporting, it is
8467 recommended to set all other settings in order to avoid relying on default
8468 unobvious parameters.
8469
8470 Example :
8471 # public access (limited to this backend only)
8472 backend public_www
8473 server srv1 192.168.0.1:80
8474 stats enable
8475 stats hide-version
8476 stats scope .
8477 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008478 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008479 stats auth admin1:AdMiN123
8480 stats auth admin2:AdMiN321
8481
8482 # internal monitoring access (unlimited)
8483 backend private_monitoring
8484 stats enable
8485 stats uri /admin?stats
8486 stats refresh 5s
8487
8488 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8489
8490
8491stats scope { <name> | "." }
8492 Enable statistics and limit access scope
8493 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008494 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008495 Arguments :
8496 <name> is the name of a listen, frontend or backend section to be
8497 reported. The special name "." (a single dot) designates the
8498 section in which the statement appears.
8499
8500 When this statement is specified, only the sections enumerated with this
8501 statement will appear in the report. All other ones will be hidden. This
8502 statement may appear as many times as needed if multiple sections need to be
8503 reported. Please note that the name checking is performed as simple string
8504 comparisons, and that it is never checked that a give section name really
8505 exists.
8506
8507 Though this statement alone is enough to enable statistics reporting, it is
8508 recommended to set all other settings in order to avoid relying on default
8509 unobvious parameters.
8510
8511 Example :
8512 # public access (limited to this backend only)
8513 backend public_www
8514 server srv1 192.168.0.1:80
8515 stats enable
8516 stats hide-version
8517 stats scope .
8518 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008519 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008520 stats auth admin1:AdMiN123
8521 stats auth admin2:AdMiN321
8522
8523 # internal monitoring access (unlimited)
8524 backend private_monitoring
8525 stats enable
8526 stats uri /admin?stats
8527 stats refresh 5s
8528
8529 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8530
Willy Tarreaud63335a2010-02-26 12:56:52 +01008531
Willy Tarreauc9705a12010-07-27 20:05:50 +02008532stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008533 Enable reporting of a description on the statistics page.
8534 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008535 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008536
Willy Tarreauc9705a12010-07-27 20:05:50 +02008537 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008538 description from global section is automatically used instead.
8539
8540 This statement is useful for users that offer shared services to their
8541 customers, where node or description should be different for each customer.
8542
8543 Though this statement alone is enough to enable statistics reporting, it is
8544 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008545 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008546
8547 Example :
8548 # internal monitoring access (unlimited)
8549 backend private_monitoring
8550 stats enable
8551 stats show-desc Master node for Europe, Asia, Africa
8552 stats uri /admin?stats
8553 stats refresh 5s
8554
8555 See also: "show-node", "stats enable", "stats uri" and "description" in
8556 global section.
8557
8558
8559stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008560 Enable reporting additional information on the statistics page
8561 May be used in sections : defaults | frontend | listen | backend
8562 yes | yes | yes | yes
8563 Arguments : none
8564
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008565 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008566 - cap: capabilities (proxy)
8567 - mode: one of tcp, http or health (proxy)
8568 - id: SNMP ID (proxy, socket, server)
8569 - IP (socket, server)
8570 - cookie (backend, server)
8571
8572 Though this statement alone is enough to enable statistics reporting, it is
8573 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008574 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008575
8576 See also: "stats enable", "stats uri".
8577
8578
8579stats show-node [ <name> ]
8580 Enable reporting of a host name on the statistics page.
8581 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008582 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008583 Arguments:
8584 <name> is an optional name to be reported. If unspecified, the
8585 node name from global section is automatically used instead.
8586
8587 This statement is useful for users that offer shared services to their
8588 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008589 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008590
8591 Though this statement alone is enough to enable statistics reporting, it is
8592 recommended to set all other settings in order to avoid relying on default
8593 unobvious parameters.
8594
8595 Example:
8596 # internal monitoring access (unlimited)
8597 backend private_monitoring
8598 stats enable
8599 stats show-node Europe-1
8600 stats uri /admin?stats
8601 stats refresh 5s
8602
8603 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8604 section.
8605
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008606
8607stats uri <prefix>
8608 Enable statistics and define the URI prefix to access them
8609 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008610 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008611 Arguments :
8612 <prefix> is the prefix of any URI which will be redirected to stats. This
8613 prefix may contain a question mark ('?') to indicate part of a
8614 query string.
8615
8616 The statistics URI is intercepted on the relayed traffic, so it appears as a
8617 page within the normal application. It is strongly advised to ensure that the
8618 selected URI will never appear in the application, otherwise it will never be
8619 possible to reach it in the application.
8620
8621 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008622 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008623 It is generally a good idea to include a question mark in the URI so that
8624 intermediate proxies refrain from caching the results. Also, since any string
8625 beginning with the prefix will be accepted as a stats request, the question
8626 mark helps ensuring that no valid URI will begin with the same words.
8627
8628 It is sometimes very convenient to use "/" as the URI prefix, and put that
8629 statement in a "listen" instance of its own. That makes it easy to dedicate
8630 an address or a port to statistics only.
8631
8632 Though this statement alone is enough to enable statistics reporting, it is
8633 recommended to set all other settings in order to avoid relying on default
8634 unobvious parameters.
8635
8636 Example :
8637 # public access (limited to this backend only)
8638 backend public_www
8639 server srv1 192.168.0.1:80
8640 stats enable
8641 stats hide-version
8642 stats scope .
8643 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008644 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008645 stats auth admin1:AdMiN123
8646 stats auth admin2:AdMiN321
8647
8648 # internal monitoring access (unlimited)
8649 backend private_monitoring
8650 stats enable
8651 stats uri /admin?stats
8652 stats refresh 5s
8653
8654 See also : "stats auth", "stats enable", "stats realm"
8655
8656
Willy Tarreaud63335a2010-02-26 12:56:52 +01008657stick match <pattern> [table <table>] [{if | unless} <cond>]
8658 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008659 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008660 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008661
8662 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008663 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008664 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008665 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008666 stickiness table. This rule is mandatory.
8667
8668 <table> is an optional stickiness table name. If unspecified, the same
8669 backend's table is used. A stickiness table is declared using
8670 the "stick-table" statement.
8671
8672 <cond> is an optional matching condition. It makes it possible to match
8673 on a certain criterion only when other conditions are met (or
8674 not met). For instance, it could be used to match on a source IP
8675 address except when a request passes through a known proxy, in
8676 which case we'd match on a header containing that IP address.
8677
8678 Some protocols or applications require complex stickiness rules and cannot
8679 always simply rely on cookies nor hashing. The "stick match" statement
8680 describes a rule to extract the stickiness criterion from an incoming request
8681 or connection. See section 7 for a complete list of possible patterns and
8682 transformation rules.
8683
8684 The table has to be declared using the "stick-table" statement. It must be of
8685 a type compatible with the pattern. By default it is the one which is present
8686 in the same backend. It is possible to share a table with other backends by
8687 referencing it using the "table" keyword. If another table is referenced,
8688 the server's ID inside the backends are used. By default, all server IDs
8689 start at 1 in each backend, so the server ordering is enough. But in case of
8690 doubt, it is highly recommended to force server IDs using their "id" setting.
8691
8692 It is possible to restrict the conditions where a "stick match" statement
8693 will apply, using "if" or "unless" followed by a condition. See section 7 for
8694 ACL based conditions.
8695
8696 There is no limit on the number of "stick match" statements. The first that
8697 applies and matches will cause the request to be directed to the same server
8698 as was used for the request which created the entry. That way, multiple
8699 matches can be used as fallbacks.
8700
8701 The stick rules are checked after the persistence cookies, so they will not
8702 affect stickiness if a cookie has already been used to select a server. That
8703 way, it becomes very easy to insert cookies and match on IP addresses in
8704 order to maintain stickiness between HTTP and HTTPS.
8705
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008706 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8707 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008708 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008709
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008710 Example :
8711 # forward SMTP users to the same server they just used for POP in the
8712 # last 30 minutes
8713 backend pop
8714 mode tcp
8715 balance roundrobin
8716 stick store-request src
8717 stick-table type ip size 200k expire 30m
8718 server s1 192.168.1.1:110
8719 server s2 192.168.1.1:110
8720
8721 backend smtp
8722 mode tcp
8723 balance roundrobin
8724 stick match src table pop
8725 server s1 192.168.1.1:25
8726 server s2 192.168.1.1:25
8727
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008728 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008729 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008730
8731
8732stick on <pattern> [table <table>] [{if | unless} <condition>]
8733 Define a request pattern to associate a user to a server
8734 May be used in sections : defaults | frontend | listen | backend
8735 no | no | yes | yes
8736
8737 Note : This form is exactly equivalent to "stick match" followed by
8738 "stick store-request", all with the same arguments. Please refer
8739 to both keywords for details. It is only provided as a convenience
8740 for writing more maintainable configurations.
8741
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008742 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8743 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008744 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008745
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008746 Examples :
8747 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008748 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008749
8750 # ...is strictly equivalent to this one :
8751 stick match src table pop if !localhost
8752 stick store-request src table pop if !localhost
8753
8754
8755 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8756 # well as HTTP without cookie. Share the same table between both accesses.
8757 backend http
8758 mode http
8759 balance roundrobin
8760 stick on src table https
8761 cookie SRV insert indirect nocache
8762 server s1 192.168.1.1:80 cookie s1
8763 server s2 192.168.1.1:80 cookie s2
8764
8765 backend https
8766 mode tcp
8767 balance roundrobin
8768 stick-table type ip size 200k expire 30m
8769 stick on src
8770 server s1 192.168.1.1:443
8771 server s2 192.168.1.1:443
8772
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008773 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008774
8775
8776stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8777 Define a request pattern used to create an entry in a stickiness table
8778 May be used in sections : defaults | frontend | listen | backend
8779 no | no | yes | yes
8780
8781 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008782 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008783 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008784 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008785 server is selected.
8786
8787 <table> is an optional stickiness table name. If unspecified, the same
8788 backend's table is used. A stickiness table is declared using
8789 the "stick-table" statement.
8790
8791 <cond> is an optional storage condition. It makes it possible to store
8792 certain criteria only when some conditions are met (or not met).
8793 For instance, it could be used to store the source IP address
8794 except when the request passes through a known proxy, in which
8795 case we'd store a converted form of a header containing that IP
8796 address.
8797
8798 Some protocols or applications require complex stickiness rules and cannot
8799 always simply rely on cookies nor hashing. The "stick store-request" statement
8800 describes a rule to decide what to extract from the request and when to do
8801 it, in order to store it into a stickiness table for further requests to
8802 match it using the "stick match" statement. Obviously the extracted part must
8803 make sense and have a chance to be matched in a further request. Storing a
8804 client's IP address for instance often makes sense. Storing an ID found in a
8805 URL parameter also makes sense. Storing a source port will almost never make
8806 any sense because it will be randomly matched. See section 7 for a complete
8807 list of possible patterns and transformation rules.
8808
8809 The table has to be declared using the "stick-table" statement. It must be of
8810 a type compatible with the pattern. By default it is the one which is present
8811 in the same backend. It is possible to share a table with other backends by
8812 referencing it using the "table" keyword. If another table is referenced,
8813 the server's ID inside the backends are used. By default, all server IDs
8814 start at 1 in each backend, so the server ordering is enough. But in case of
8815 doubt, it is highly recommended to force server IDs using their "id" setting.
8816
8817 It is possible to restrict the conditions where a "stick store-request"
8818 statement will apply, using "if" or "unless" followed by a condition. This
8819 condition will be evaluated while parsing the request, so any criteria can be
8820 used. See section 7 for ACL based conditions.
8821
8822 There is no limit on the number of "stick store-request" statements, but
8823 there is a limit of 8 simultaneous stores per request or response. This
8824 makes it possible to store up to 8 criteria, all extracted from either the
8825 request or the response, regardless of the number of rules. Only the 8 first
8826 ones which match will be kept. Using this, it is possible to feed multiple
8827 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008828 another protocol or access method. Using multiple store-request rules with
8829 the same table is possible and may be used to find the best criterion to rely
8830 on, by arranging the rules by decreasing preference order. Only the first
8831 extracted criterion for a given table will be stored. All subsequent store-
8832 request rules referencing the same table will be skipped and their ACLs will
8833 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008834
8835 The "store-request" rules are evaluated once the server connection has been
8836 established, so that the table will contain the real server that processed
8837 the request.
8838
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008839 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8840 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008841 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008842
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008843 Example :
8844 # forward SMTP users to the same server they just used for POP in the
8845 # last 30 minutes
8846 backend pop
8847 mode tcp
8848 balance roundrobin
8849 stick store-request src
8850 stick-table type ip size 200k expire 30m
8851 server s1 192.168.1.1:110
8852 server s2 192.168.1.1:110
8853
8854 backend smtp
8855 mode tcp
8856 balance roundrobin
8857 stick match src table pop
8858 server s1 192.168.1.1:25
8859 server s2 192.168.1.1:25
8860
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008861 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008862 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008863
8864
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008865stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008866 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8867 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008868 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008869 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008870 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008871
8872 Arguments :
8873 ip a table declared with "type ip" will only store IPv4 addresses.
8874 This form is very compact (about 50 bytes per entry) and allows
8875 very fast entry lookup and stores with almost no overhead. This
8876 is mainly used to store client source IP addresses.
8877
David du Colombier9a6d3c92011-03-17 10:40:24 +01008878 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8879 This form is very compact (about 60 bytes per entry) and allows
8880 very fast entry lookup and stores with almost no overhead. This
8881 is mainly used to store client source IP addresses.
8882
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008883 integer a table declared with "type integer" will store 32bit integers
8884 which can represent a client identifier found in a request for
8885 instance.
8886
8887 string a table declared with "type string" will store substrings of up
8888 to <len> characters. If the string provided by the pattern
8889 extractor is larger than <len>, it will be truncated before
8890 being stored. During matching, at most <len> characters will be
8891 compared between the string in the table and the extracted
8892 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008893 to 32 characters.
8894
8895 binary a table declared with "type binary" will store binary blocks
8896 of <len> bytes. If the block provided by the pattern
8897 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008898 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008899 is shorter than <len>, it will be padded by 0. When not
8900 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008901
8902 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008903 "string" type table (See type "string" above). Or the number
8904 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008905 changing this parameter as memory usage will proportionally
8906 increase.
8907
8908 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008909 value directly impacts memory usage. Count approximately
8910 50 bytes per entry, plus the size of a string if any. The size
8911 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008912
8913 [nopurge] indicates that we refuse to purge older entries when the table
8914 is full. When not specified and the table is full when haproxy
8915 wants to store an entry in it, it will flush a few of the oldest
8916 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008917 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008918 be desirable to refuse new entries instead of purging the older
8919 ones. That may be the case when the amount of data to store is
8920 far above the hardware limits and we prefer not to offer access
8921 to new clients than to reject the ones already connected. When
8922 using this parameter, be sure to properly set the "expire"
8923 parameter (see below).
8924
Emeric Brunf099e792010-09-27 12:05:28 +02008925 <peersect> is the name of the peers section to use for replication. Entries
8926 which associate keys to server IDs are kept synchronized with
8927 the remote peers declared in this section. All entries are also
8928 automatically learned from the local peer (old process) during a
8929 soft restart.
8930
Willy Tarreau1abc6732015-05-01 19:21:02 +02008931 NOTE : each peers section may be referenced only by tables
8932 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008933
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008934 <expire> defines the maximum duration of an entry in the table since it
8935 was last created, refreshed or matched. The expiration delay is
8936 defined using the standard time format, similarly as the various
8937 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008938 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008939 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008940 be removed once full. Be sure not to use the "nopurge" parameter
8941 if not expiration delay is specified.
8942
Willy Tarreau08d5f982010-06-06 13:34:54 +02008943 <data_type> is used to store additional information in the stick-table. This
8944 may be used by ACLs in order to control various criteria related
8945 to the activity of the client matching the stick-table. For each
8946 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008947 that the additional data can fit. Several data types may be
8948 stored with an entry. Multiple data types may be specified after
8949 the "store" keyword, as a comma-separated list. Alternatively,
8950 it is possible to repeat the "store" keyword followed by one or
8951 several data types. Except for the "server_id" type which is
8952 automatically detected and enabled, all data types must be
8953 explicitly declared to be stored. If an ACL references a data
8954 type which is not stored, the ACL will simply not match. Some
8955 data types require an argument which must be passed just after
8956 the type between parenthesis. See below for the supported data
8957 types and their arguments.
8958
8959 The data types that can be stored with an entry are the following :
8960 - server_id : this is an integer which holds the numeric ID of the server a
8961 request was assigned to. It is used by the "stick match", "stick store",
8962 and "stick on" rules. It is automatically enabled when referenced.
8963
8964 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8965 integer which may be used for anything. Most of the time it will be used
8966 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008967 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008968
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008969 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8970 over a period. It is a positive 32-bit integer integer which may be used
8971 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008972 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008973 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008974 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008975
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008976 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8977 integer which may be used for anything. Most of the time it will be used
8978 to put a special tag on some entries, for instance to note that a
8979 specific behavior was detected and must be known for future matches.
8980
8981 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8982 over a period. It is a positive 32-bit integer integer which may be used
8983 for anything. Just like <gpc1>, it counts events, but instead of keeping
8984 a cumulative number, it maintains the rate at which the counter is
8985 incremented. Most of the time it will be used to measure the frequency of
8986 occurrence of certain events (e.g. requests to a specific URL).
8987
Willy Tarreauc9705a12010-07-27 20:05:50 +02008988 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8989 the absolute number of connections received from clients which matched
8990 this entry. It does not mean the connections were accepted, just that
8991 they were received.
8992
8993 - conn_cur : Current Connections. It is a positive 32-bit integer which
8994 stores the concurrent connection counts for the entry. It is incremented
8995 once an incoming connection matches the entry, and decremented once the
8996 connection leaves. That way it is possible to know at any time the exact
8997 number of concurrent connections for an entry.
8998
8999 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9000 integer parameter <period> which indicates in milliseconds the length
9001 of the period over which the average is measured. It reports the average
9002 incoming connection rate over that period, in connections per period. The
9003 result is an integer which can be matched using ACLs.
9004
9005 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9006 the absolute number of sessions received from clients which matched this
9007 entry. A session is a connection that was accepted by the layer 4 rules.
9008
9009 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9010 integer parameter <period> which indicates in milliseconds the length
9011 of the period over which the average is measured. It reports the average
9012 incoming session rate over that period, in sessions per period. The
9013 result is an integer which can be matched using ACLs.
9014
9015 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9016 counts the absolute number of HTTP requests received from clients which
9017 matched this entry. It does not matter whether they are valid requests or
9018 not. Note that this is different from sessions when keep-alive is used on
9019 the client side.
9020
9021 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9022 integer parameter <period> which indicates in milliseconds the length
9023 of the period over which the average is measured. It reports the average
9024 HTTP request rate over that period, in requests per period. The result is
9025 an integer which can be matched using ACLs. It does not matter whether
9026 they are valid requests or not. Note that this is different from sessions
9027 when keep-alive is used on the client side.
9028
9029 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9030 counts the absolute number of HTTP requests errors induced by clients
9031 which matched this entry. Errors are counted on invalid and truncated
9032 requests, as well as on denied or tarpitted requests, and on failed
9033 authentications. If the server responds with 4xx, then the request is
9034 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009035 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009036
9037 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9038 integer parameter <period> which indicates in milliseconds the length
9039 of the period over which the average is measured. It reports the average
9040 HTTP request error rate over that period, in requests per period (see
9041 http_err_cnt above for what is accounted as an error). The result is an
9042 integer which can be matched using ACLs.
9043
9044 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009045 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009046 which matched this entry. Headers are included in the count. This may be
9047 used to limit abuse of upload features on photo or video servers.
9048
9049 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9050 integer parameter <period> which indicates in milliseconds the length
9051 of the period over which the average is measured. It reports the average
9052 incoming bytes rate over that period, in bytes per period. It may be used
9053 to detect users which upload too much and too fast. Warning: with large
9054 uploads, it is possible that the amount of uploaded data will be counted
9055 once upon termination, thus causing spikes in the average transfer speed
9056 instead of having a smooth one. This may partially be smoothed with
9057 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9058 recommended for better fairness.
9059
9060 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009061 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009062 matched this entry. Headers are included in the count. This may be used
9063 to limit abuse of bots sucking the whole site.
9064
9065 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9066 an integer parameter <period> which indicates in milliseconds the length
9067 of the period over which the average is measured. It reports the average
9068 outgoing bytes rate over that period, in bytes per period. It may be used
9069 to detect users which download too much and too fast. Warning: with large
9070 transfers, it is possible that the amount of transferred data will be
9071 counted once upon termination, thus causing spikes in the average
9072 transfer speed instead of having a smooth one. This may partially be
9073 smoothed with "option contstats" though this is not perfect yet. Use of
9074 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009075
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009076 There is only one stick-table per proxy. At the moment of writing this doc,
9077 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009078 to be required, simply create a dummy backend with a stick-table in it and
9079 reference it.
9080
9081 It is important to understand that stickiness based on learning information
9082 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009083 lost upon restart unless peers are properly configured to transfer such
9084 information upon restart (recommended). In general it can be good as a
9085 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009086
Willy Tarreauc9705a12010-07-27 20:05:50 +02009087 Last, memory requirements may be important when storing many data types.
9088 Indeed, storing all indicators above at once in each entry requires 116 bytes
9089 per entry, or 116 MB for a 1-million entries table. This is definitely not
9090 something that can be ignored.
9091
9092 Example:
9093 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9094 # and store a general purpose counter and the average connection rate
9095 # computed over a sliding window of 30 seconds.
9096 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9097
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009098 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009099 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009100
9101
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009102stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009103 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009104 May be used in sections : defaults | frontend | listen | backend
9105 no | no | yes | yes
9106
9107 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009108 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009109 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009110 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009111 server is selected.
9112
9113 <table> is an optional stickiness table name. If unspecified, the same
9114 backend's table is used. A stickiness table is declared using
9115 the "stick-table" statement.
9116
9117 <cond> is an optional storage condition. It makes it possible to store
9118 certain criteria only when some conditions are met (or not met).
9119 For instance, it could be used to store the SSL session ID only
9120 when the response is a SSL server hello.
9121
9122 Some protocols or applications require complex stickiness rules and cannot
9123 always simply rely on cookies nor hashing. The "stick store-response"
9124 statement describes a rule to decide what to extract from the response and
9125 when to do it, in order to store it into a stickiness table for further
9126 requests to match it using the "stick match" statement. Obviously the
9127 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009128 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009129 See section 7 for a complete list of possible patterns and transformation
9130 rules.
9131
9132 The table has to be declared using the "stick-table" statement. It must be of
9133 a type compatible with the pattern. By default it is the one which is present
9134 in the same backend. It is possible to share a table with other backends by
9135 referencing it using the "table" keyword. If another table is referenced,
9136 the server's ID inside the backends are used. By default, all server IDs
9137 start at 1 in each backend, so the server ordering is enough. But in case of
9138 doubt, it is highly recommended to force server IDs using their "id" setting.
9139
9140 It is possible to restrict the conditions where a "stick store-response"
9141 statement will apply, using "if" or "unless" followed by a condition. This
9142 condition will be evaluated while parsing the response, so any criteria can
9143 be used. See section 7 for ACL based conditions.
9144
9145 There is no limit on the number of "stick store-response" statements, but
9146 there is a limit of 8 simultaneous stores per request or response. This
9147 makes it possible to store up to 8 criteria, all extracted from either the
9148 request or the response, regardless of the number of rules. Only the 8 first
9149 ones which match will be kept. Using this, it is possible to feed multiple
9150 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009151 another protocol or access method. Using multiple store-response rules with
9152 the same table is possible and may be used to find the best criterion to rely
9153 on, by arranging the rules by decreasing preference order. Only the first
9154 extracted criterion for a given table will be stored. All subsequent store-
9155 response rules referencing the same table will be skipped and their ACLs will
9156 not be evaluated. However, even if a store-request rule references a table, a
9157 store-response rule may also use the same table. This means that each table
9158 may learn exactly one element from the request and one element from the
9159 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009160
9161 The table will contain the real server that processed the request.
9162
9163 Example :
9164 # Learn SSL session ID from both request and response and create affinity.
9165 backend https
9166 mode tcp
9167 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009168 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009169 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009170
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009171 acl clienthello req_ssl_hello_type 1
9172 acl serverhello rep_ssl_hello_type 2
9173
9174 # use tcp content accepts to detects ssl client and server hello.
9175 tcp-request inspect-delay 5s
9176 tcp-request content accept if clienthello
9177
9178 # no timeout on response inspect delay by default.
9179 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009180
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009181 # SSL session ID (SSLID) may be present on a client or server hello.
9182 # Its length is coded on 1 byte at offset 43 and its value starts
9183 # at offset 44.
9184
9185 # Match and learn on request if client hello.
9186 stick on payload_lv(43,1) if clienthello
9187
9188 # Learn on response if server hello.
9189 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009190
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009191 server s1 192.168.1.1:443
9192 server s2 192.168.1.1:443
9193
9194 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9195 extraction.
9196
9197
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009198tcp-check connect [params*]
9199 Opens a new connection
9200 May be used in sections: defaults | frontend | listen | backend
9201 no | no | yes | yes
9202
9203 When an application lies on more than a single TCP port or when HAProxy
9204 load-balance many services in a single backend, it makes sense to probe all
9205 the services individually before considering a server as operational.
9206
9207 When there are no TCP port configured on the server line neither server port
9208 directive, then the 'tcp-check connect port <port>' must be the first step
9209 of the sequence.
9210
9211 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9212 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9213 do.
9214
9215 Parameters :
9216 They are optional and can be used to describe how HAProxy should open and
9217 use the TCP connection.
9218
9219 port if not set, check port or server port is used.
9220 It tells HAProxy where to open the connection to.
9221 <port> must be a valid TCP port source integer, from 1 to 65535.
9222
9223 send-proxy send a PROXY protocol string
9224
9225 ssl opens a ciphered connection
9226
9227 Examples:
9228 # check HTTP and HTTPs services on a server.
9229 # first open port 80 thanks to server line port directive, then
9230 # tcp-check opens port 443, ciphered and run a request on it:
9231 option tcp-check
9232 tcp-check connect
9233 tcp-check send GET\ /\ HTTP/1.0\r\n
9234 tcp-check send Host:\ haproxy.1wt.eu\r\n
9235 tcp-check send \r\n
9236 tcp-check expect rstring (2..|3..)
9237 tcp-check connect port 443 ssl
9238 tcp-check send GET\ /\ HTTP/1.0\r\n
9239 tcp-check send Host:\ haproxy.1wt.eu\r\n
9240 tcp-check send \r\n
9241 tcp-check expect rstring (2..|3..)
9242 server www 10.0.0.1 check port 80
9243
9244 # check both POP and IMAP from a single server:
9245 option tcp-check
9246 tcp-check connect port 110
9247 tcp-check expect string +OK\ POP3\ ready
9248 tcp-check connect port 143
9249 tcp-check expect string *\ OK\ IMAP4\ ready
9250 server mail 10.0.0.1 check
9251
9252 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9253
9254
9255tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009256 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009257 May be used in sections: defaults | frontend | listen | backend
9258 no | no | yes | yes
9259
9260 Arguments :
9261 <match> is a keyword indicating how to look for a specific pattern in the
9262 response. The keyword may be one of "string", "rstring" or
9263 binary.
9264 The keyword may be preceded by an exclamation mark ("!") to negate
9265 the match. Spaces are allowed between the exclamation mark and the
9266 keyword. See below for more details on the supported keywords.
9267
9268 <pattern> is the pattern to look for. It may be a string or a regular
9269 expression. If the pattern contains spaces, they must be escaped
9270 with the usual backslash ('\').
9271 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009272 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009273 two digits will represent a byte. The hexadecimal digits may be
9274 used upper or lower case.
9275
9276
9277 The available matches are intentionally similar to their http-check cousins :
9278
9279 string <string> : test the exact string matches in the response buffer.
9280 A health check response will be considered valid if the
9281 response's buffer contains this exact string. If the
9282 "string" keyword is prefixed with "!", then the response
9283 will be considered invalid if the body contains this
9284 string. This can be used to look for a mandatory pattern
9285 in a protocol response, or to detect a failure when a
9286 specific error appears in a protocol banner.
9287
9288 rstring <regex> : test a regular expression on the response buffer.
9289 A health check response will be considered valid if the
9290 response's buffer matches this expression. If the
9291 "rstring" keyword is prefixed with "!", then the response
9292 will be considered invalid if the body matches the
9293 expression.
9294
9295 binary <hexstring> : test the exact string in its hexadecimal form matches
9296 in the response buffer. A health check response will
9297 be considered valid if the response's buffer contains
9298 this exact hexadecimal string.
9299 Purpose is to match data on binary protocols.
9300
9301 It is important to note that the responses will be limited to a certain size
9302 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9303 Thus, too large responses may not contain the mandatory pattern when using
9304 "string", "rstring" or binary. If a large response is absolutely required, it
9305 is possible to change the default max size by setting the global variable.
9306 However, it is worth keeping in mind that parsing very large responses can
9307 waste some CPU cycles, especially when regular expressions are used, and that
9308 it is always better to focus the checks on smaller resources. Also, in its
9309 current state, the check will not find any string nor regex past a null
9310 character in the response. Similarly it is not possible to request matching
9311 the null character.
9312
9313 Examples :
9314 # perform a POP check
9315 option tcp-check
9316 tcp-check expect string +OK\ POP3\ ready
9317
9318 # perform an IMAP check
9319 option tcp-check
9320 tcp-check expect string *\ OK\ IMAP4\ ready
9321
9322 # look for the redis master server
9323 option tcp-check
9324 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009325 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009326 tcp-check send info\ replication\r\n
9327 tcp-check expect string role:master
9328 tcp-check send QUIT\r\n
9329 tcp-check expect string +OK
9330
9331
9332 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9333 "tcp-check send-binary", "http-check expect", tune.chksize
9334
9335
9336tcp-check send <data>
9337 Specify a string to be sent as a question during a generic health check
9338 May be used in sections: defaults | frontend | listen | backend
9339 no | no | yes | yes
9340
9341 <data> : the data to be sent as a question during a generic health check
9342 session. For now, <data> must be a string.
9343
9344 Examples :
9345 # look for the redis master server
9346 option tcp-check
9347 tcp-check send info\ replication\r\n
9348 tcp-check expect string role:master
9349
9350 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9351 "tcp-check send-binary", tune.chksize
9352
9353
Davor Ocelice9ed2812017-12-25 17:49:28 +01009354tcp-check send-binary <hexstring>
9355 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009356 tcp health check
9357 May be used in sections: defaults | frontend | listen | backend
9358 no | no | yes | yes
9359
9360 <data> : the data to be sent as a question during a generic health check
9361 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009362 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009363 response buffer. A health check response will be considered
9364 valid if the response's buffer contains this exact
9365 hexadecimal string.
9366 Purpose is to send binary data to ask on binary protocols.
9367
9368 Examples :
9369 # redis check in binary
9370 option tcp-check
9371 tcp-check send-binary 50494e470d0a # PING\r\n
9372 tcp-check expect binary 2b504F4e47 # +PONG
9373
9374
9375 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9376 "tcp-check send", tune.chksize
9377
9378
Willy Tarreaue9656522010-08-17 15:40:09 +02009379tcp-request connection <action> [{if | unless} <condition>]
9380 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009381 May be used in sections : defaults | frontend | listen | backend
9382 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009383 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009384 <action> defines the action to perform if the condition applies. See
9385 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009386
Willy Tarreaue9656522010-08-17 15:40:09 +02009387 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009388
9389 Immediately after acceptance of a new incoming connection, it is possible to
9390 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009391 or dropped or have its counters tracked. Those conditions cannot make use of
9392 any data contents because the connection has not been read from yet, and the
9393 buffers are not yet allocated. This is used to selectively and very quickly
9394 accept or drop connections from various sources with a very low overhead. If
9395 some contents need to be inspected in order to take the decision, the
9396 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009397
Willy Tarreaue9656522010-08-17 15:40:09 +02009398 The "tcp-request connection" rules are evaluated in their exact declaration
9399 order. If no rule matches or if there is no rule, the default action is to
9400 accept the incoming connection. There is no specific limit to the number of
9401 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009402
Willy Tarreaua9083d02015-05-08 15:27:59 +02009403 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009404 - accept :
9405 accepts the connection if the condition is true (when used with "if")
9406 or false (when used with "unless"). The first such rule executed ends
9407 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009408
Willy Tarreaue9656522010-08-17 15:40:09 +02009409 - reject :
9410 rejects the connection if the condition is true (when used with "if")
9411 or false (when used with "unless"). The first such rule executed ends
9412 the rules evaluation. Rejected connections do not even become a
9413 session, which is why they are accounted separately for in the stats,
9414 as "denied connections". They are not considered for the session
9415 rate-limit and are not logged either. The reason is that these rules
9416 should only be used to filter extremely high connection rates such as
9417 the ones encountered during a massive DDoS attack. Under these extreme
9418 conditions, the simple action of logging each event would make the
9419 system collapse and would considerably lower the filtering capacity. If
9420 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009421 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009422
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009423 - expect-proxy layer4 :
9424 configures the client-facing connection to receive a PROXY protocol
9425 header before any byte is read from the socket. This is equivalent to
9426 having the "accept-proxy" keyword on the "bind" line, except that using
9427 the TCP rule allows the PROXY protocol to be accepted only for certain
9428 IP address ranges using an ACL. This is convenient when multiple layers
9429 of load balancers are passed through by traffic coming from public
9430 hosts.
9431
Bertrand Jacquin90759682016-06-06 15:35:39 +01009432 - expect-netscaler-cip layer4 :
9433 configures the client-facing connection to receive a NetScaler Client
9434 IP insertion protocol header before any byte is read from the socket.
9435 This is equivalent to having the "accept-netscaler-cip" keyword on the
9436 "bind" line, except that using the TCP rule allows the PROXY protocol
9437 to be accepted only for certain IP address ranges using an ACL. This
9438 is convenient when multiple layers of load balancers are passed
9439 through by traffic coming from public hosts.
9440
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009441 - capture <sample> len <length> :
9442 This only applies to "tcp-request content" rules. It captures sample
9443 expression <sample> from the request buffer, and converts it to a
9444 string of at most <len> characters. The resulting string is stored into
9445 the next request "capture" slot, so it will possibly appear next to
9446 some captured HTTP headers. It will then automatically appear in the
9447 logs, and it will be possible to extract it using sample fetch rules to
9448 feed it into headers or anything. The length should be limited given
9449 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009450 session life. Please check section 7.3 (Fetching samples) and "capture
9451 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009452
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009453 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009454 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009455 rules do not stop evaluation and do not change default action. The
9456 number of counters that may be simultaneously tracked by the same
9457 connection is set in MAX_SESS_STKCTR at build time (reported in
9458 haproxy -vv) whichs defaults to 3, so the track-sc number is between 0
9459 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9460 tracking of the counters of the specified table as the first set. The
9461 first "track-sc1" rule executed enables tracking of the counters of the
9462 specified table as the second set. The first "track-sc2" rule executed
9463 enables tracking of the counters of the specified table as the third
9464 set. It is a recommended practice to use the first set of counters for
9465 the per-frontend counters and the second set for the per-backend ones.
9466 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009467
Willy Tarreaue9656522010-08-17 15:40:09 +02009468 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009469 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009470 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009471 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009472 and used to select which table entry to update the counters.
9473 Note that "tcp-request connection" cannot use content-based
9474 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009475
Willy Tarreaue9656522010-08-17 15:40:09 +02009476 <table> is an optional table to be used instead of the default one,
9477 which is the stick-table declared in the current proxy. All
9478 the counters for the matches and updates for the key will
9479 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009480
Willy Tarreaue9656522010-08-17 15:40:09 +02009481 Once a "track-sc*" rule is executed, the key is looked up in the table
9482 and if it is not found, an entry is allocated for it. Then a pointer to
9483 that entry is kept during all the session's life, and this entry's
9484 counters are updated as often as possible, every time the session's
9485 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009486 Counters are only updated for events that happen after the tracking has
9487 been started. For example, connection counters will not be updated when
9488 tracking layer 7 information, since the connection event happens before
9489 layer7 information is extracted.
9490
Willy Tarreaue9656522010-08-17 15:40:09 +02009491 If the entry tracks concurrent connection counters, one connection is
9492 counted for as long as the entry is tracked, and the entry will not
9493 expire during that time. Tracking counters also provides a performance
9494 advantage over just checking the keys, because only one table lookup is
9495 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009496
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009497 - sc-inc-gpc0(<sc-id>):
9498 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9499 counter designated by <sc-id>. If an error occurs, this action silently
9500 fails and the actions evaluation continues.
9501
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009502 - sc-inc-gpc1(<sc-id>):
9503 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9504 counter designated by <sc-id>. If an error occurs, this action silently
9505 fails and the actions evaluation continues.
9506
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009507 - sc-set-gpt0(<sc-id>) <int>:
9508 This action sets the GPT0 tag according to the sticky counter designated
9509 by <sc-id> and the value of <int>. The expected result is a boolean. If
9510 an error occurs, this action silently fails and the actions evaluation
9511 continues.
9512
William Lallemand2e785f22016-05-25 01:48:42 +02009513 - set-src <expr> :
9514 Is used to set the source IP address to the value of specified
9515 expression. Useful if you want to mask source IP for privacy.
9516 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009517 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009518
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009519 Arguments:
9520 <expr> Is a standard HAProxy expression formed by a sample-fetch
9521 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009522
9523 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009524 tcp-request connection set-src src,ipmask(24)
9525
Willy Tarreau0c630532016-10-21 17:52:58 +02009526 When possible, set-src preserves the original source port as long as the
9527 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009528
William Lallemand44be6402016-05-25 01:51:35 +02009529 - set-src-port <expr> :
9530 Is used to set the source port address to the value of specified
9531 expression.
9532
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009533 Arguments:
9534 <expr> Is a standard HAProxy expression formed by a sample-fetch
9535 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009536
9537 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009538 tcp-request connection set-src-port int(4000)
9539
Willy Tarreau0c630532016-10-21 17:52:58 +02009540 When possible, set-src-port preserves the original source address as long
9541 as the address family supports a port, otherwise it forces the source
9542 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009543
William Lallemand13e9b0c2016-05-25 02:34:07 +02009544 - set-dst <expr> :
9545 Is used to set the destination IP address to the value of specified
9546 expression. Useful if you want to mask IP for privacy in log.
9547 If you want to provide an IP from a HTTP header use "http-request
9548 set-dst". If you want to connect to the new address/port, use
9549 '0.0.0.0:0' as a server address in the backend.
9550
9551 <expr> Is a standard HAProxy expression formed by a sample-fetch
9552 followed by some converters.
9553
9554 Example:
9555
9556 tcp-request connection set-dst dst,ipmask(24)
9557 tcp-request connection set-dst ipv4(10.0.0.1)
9558
Willy Tarreau0c630532016-10-21 17:52:58 +02009559 When possible, set-dst preserves the original destination port as long as
9560 the address family allows it, otherwise the destination port is set to 0.
9561
William Lallemand13e9b0c2016-05-25 02:34:07 +02009562 - set-dst-port <expr> :
9563 Is used to set the destination port address to the value of specified
9564 expression. If you want to connect to the new address/port, use
9565 '0.0.0.0:0' as a server address in the backend.
9566
9567
9568 <expr> Is a standard HAProxy expression formed by a sample-fetch
9569 followed by some converters.
9570
9571 Example:
9572
9573 tcp-request connection set-dst-port int(4000)
9574
Willy Tarreau0c630532016-10-21 17:52:58 +02009575 When possible, set-dst-port preserves the original destination address as
9576 long as the address family supports a port, otherwise it forces the
9577 destination address to IPv4 "0.0.0.0" before rewriting the port.
9578
Willy Tarreau2d392c22015-08-24 01:43:45 +02009579 - "silent-drop" :
9580 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009581 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009582 to prevent the client from being notified. The effect it then that the
9583 client still sees an established connection while there's none on
9584 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9585 except that it doesn't use any local resource at all on the machine
9586 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009587 slow down stronger attackers. It is important to understand the impact
9588 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009589 client and HAProxy (firewalls, proxies, load balancers) will also keep
9590 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009591 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009592 TCP_REPAIR socket option is used to block the emission of a TCP
9593 reset. On other systems, the socket's TTL is reduced to 1 so that the
9594 TCP reset doesn't pass the first router, though it's still delivered to
9595 local networks. Do not use it unless you fully understand how it works.
9596
Willy Tarreaue9656522010-08-17 15:40:09 +02009597 Note that the "if/unless" condition is optional. If no condition is set on
9598 the action, it is simply performed unconditionally. That can be useful for
9599 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009600
Willy Tarreaue9656522010-08-17 15:40:09 +02009601 Example: accept all connections from white-listed hosts, reject too fast
9602 connection without counting them, and track accepted connections.
9603 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009604
Willy Tarreaue9656522010-08-17 15:40:09 +02009605 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009606 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009607 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009608
Willy Tarreaue9656522010-08-17 15:40:09 +02009609 Example: accept all connections from white-listed hosts, count all other
9610 connections and reject too fast ones. This results in abusive ones
9611 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009612
Willy Tarreaue9656522010-08-17 15:40:09 +02009613 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009614 tcp-request connection track-sc0 src
9615 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009616
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009617 Example: enable the PROXY protocol for traffic coming from all known proxies.
9618
9619 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9620
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009621 See section 7 about ACL usage.
9622
Willy Tarreau4f614292016-10-21 17:49:36 +02009623 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009624
9625
Willy Tarreaue9656522010-08-17 15:40:09 +02009626tcp-request content <action> [{if | unless} <condition>]
9627 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009628 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009629 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009630 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009631 <action> defines the action to perform if the condition applies. See
9632 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009633
Willy Tarreaue9656522010-08-17 15:40:09 +02009634 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009635
Davor Ocelice9ed2812017-12-25 17:49:28 +01009636 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009637 called "TCP content inspection". During this stage, ACL-based rules are
9638 evaluated every time the request contents are updated, until either an
9639 "accept" or a "reject" rule matches, or the TCP request inspection delay
9640 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009641
Willy Tarreaue9656522010-08-17 15:40:09 +02009642 The first difference between these rules and "tcp-request connection" rules
9643 is that "tcp-request content" rules can make use of contents to take a
9644 decision. Most often, these decisions will consider a protocol recognition or
9645 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009646 both frontends and backends. In case of HTTP keep-alive with the client, all
9647 tcp-request content rules are evaluated again, so haproxy keeps a record of
9648 what sticky counters were assigned by a "tcp-request connection" versus a
9649 "tcp-request content" rule, and flushes all the content-related ones after
9650 processing an HTTP request, so that they may be evaluated again by the rules
9651 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009652 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009653 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009654
Willy Tarreaue9656522010-08-17 15:40:09 +02009655 Content-based rules are evaluated in their exact declaration order. If no
9656 rule matches or if there is no rule, the default action is to accept the
9657 contents. There is no specific limit to the number of rules which may be
9658 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009659
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009660 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009661 - accept : the request is accepted
9662 - reject : the request is rejected and the connection is closed
9663 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009664 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009665 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009666 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009667 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009668 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009669 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009670 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009671 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009672 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009673
Willy Tarreaue9656522010-08-17 15:40:09 +02009674 They have the same meaning as their counter-parts in "tcp-request connection"
9675 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009676
Willy Tarreauf3338342014-01-28 21:40:28 +01009677 While there is nothing mandatory about it, it is recommended to use the
9678 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9679 content" rules in the frontend, and track-sc2 for "tcp-request content"
9680 rules in the backend, because that makes the configuration more readable
9681 and easier to troubleshoot, but this is just a guideline and all counters
9682 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009683
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009684 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009685 the action, it is simply performed unconditionally. That can be useful for
9686 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009687
Willy Tarreaue9656522010-08-17 15:40:09 +02009688 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009689 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9690 contents of a buffer before extracting the required data. If the buffered
9691 contents do not parse as a valid HTTP message, then the ACL does not match.
9692 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009693 processing, so there is no risk of parsing something differently. In an HTTP
9694 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9695 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009696
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009697 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009698 are present when the rule is processed. The rule processing engine is able to
9699 wait until the inspect delay expires when the data to be tracked is not yet
9700 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009701
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009702 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009703 declared inline. For "tcp-request session" rules, only session-level
9704 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009705
Daniel Schneller0b547052016-03-21 20:46:57 +01009706 <var-name> The name of the variable starts with an indication about
9707 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009708 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009709 "sess" : the variable is shared with the whole session
9710 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009711 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009712 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009713 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009714 "res" : the variable is shared only during response
9715 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009716 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009717 The name may only contain characters 'a-z', 'A-Z', '0-9',
9718 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009719
9720 <expr> Is a standard HAProxy expression formed by a sample-fetch
9721 followed by some converters.
9722
Christopher Faulet85d79c92016-11-09 16:54:56 +01009723 The "unset-var" is used to unset a variable. See above for details about
9724 <var-name>.
9725
Patrick Hemmer268a7072018-05-11 12:52:31 -04009726 The "set-priority-class" is used to set the queue priority class of the
9727 current request. The value must be a sample expression which converts to an
9728 integer in the range -2047..2047. Results outside this range will be
9729 truncated. The priority class determines the order in which queued requests
9730 are processed. Lower values have higher priority.
9731
9732 The "set-priority-offset" is used to set the queue priority timestamp offset
9733 of the current request. The value must be a sample expression which converts
9734 to an integer in the range -524287..524287. Results outside this range will be
9735 truncated. When a request is queued, it is ordered first by the priority
9736 class, then by the current timestamp adjusted by the given offset in
9737 milliseconds. Lower values have higher priority.
9738 Note that the resulting timestamp is is only tracked with enough precision for
9739 524,287ms (8m44s287ms). If the request is queued long enough to where the
9740 adjusted timestamp exceeds this value, it will be misidentified as highest
9741 priority. Thus it is important to set "timeout queue" to a value, where when
9742 combined with the offset, does not exceed this limit.
9743
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009744 The "send-spoe-group" is used to trigger sending of a group of SPOE
9745 messages. To do so, the SPOE engine used to send messages must be defined, as
9746 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9747 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9748 the SPOE agent name must be used.
9749
9750 <engine-name> The SPOE engine name.
9751
9752 <group-name> The SPOE group name as specified in the engine configuration.
9753
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009754 Example:
9755
9756 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009757 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009758
Willy Tarreau62644772008-07-16 18:36:06 +02009759 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009760 # Accept HTTP requests containing a Host header saying "example.com"
9761 # and reject everything else.
9762 acl is_host_com hdr(Host) -i example.com
9763 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009764 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009765 tcp-request content reject
9766
9767 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009768 # reject SMTP connection if client speaks first
9769 tcp-request inspect-delay 30s
9770 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009771 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009772
9773 # Forward HTTPS connection only if client speaks
9774 tcp-request inspect-delay 30s
9775 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009776 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009777 tcp-request content reject
9778
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009779 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009780 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009781 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009782 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009783 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9784 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009785
9786 Example:
9787 # track request counts per "base" (concatenation of Host+URL)
9788 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009789 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009790
Willy Tarreaue9656522010-08-17 15:40:09 +02009791 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009792 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009793
9794 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009795 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009796 # protecting all our sites
9797 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009798 tcp-request connection track-sc0 src
9799 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009800 ...
9801 use_backend http_dynamic if { path_end .php }
9802
9803 backend http_dynamic
9804 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009805 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009806 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009807 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009808 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009809 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009810 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009811
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009812 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009813
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009814 See also : "tcp-request connection", "tcp-request session",
9815 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009816
9817
9818tcp-request inspect-delay <timeout>
9819 Set the maximum allowed time to wait for data during content inspection
9820 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009821 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009822 Arguments :
9823 <timeout> is the timeout value specified in milliseconds by default, but
9824 can be in any other unit if the number is suffixed by the unit,
9825 as explained at the top of this document.
9826
9827 People using haproxy primarily as a TCP relay are often worried about the
9828 risk of passing any type of protocol to a server without any analysis. In
9829 order to be able to analyze the request contents, we must first withhold
9830 the data then analyze them. This statement simply enables withholding of
9831 data for at most the specified amount of time.
9832
Willy Tarreaufb356202010-08-03 14:02:05 +02009833 TCP content inspection applies very early when a connection reaches a
9834 frontend, then very early when the connection is forwarded to a backend. This
9835 means that a connection may experience a first delay in the frontend and a
9836 second delay in the backend if both have tcp-request rules.
9837
Willy Tarreau62644772008-07-16 18:36:06 +02009838 Note that when performing content inspection, haproxy will evaluate the whole
9839 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009840 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009841 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009842 contents are definitive. If no delay is set, haproxy will not wait at all
9843 and will immediately apply a verdict based on the available information.
9844 Obviously this is unlikely to be very useful and might even be racy, so such
9845 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009846
9847 As soon as a rule matches, the request is released and continues as usual. If
9848 the timeout is reached and no rule matches, the default policy will be to let
9849 it pass through unaffected.
9850
9851 For most protocols, it is enough to set it to a few seconds, as most clients
9852 send the full request immediately upon connection. Add 3 or more seconds to
9853 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009854 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009855 before the server (e.g. SMTP), or to wait for a client to talk before passing
9856 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009857 least the inspection delay, otherwise it will expire first. If the client
9858 closes the connection or if the buffer is full, the delay immediately expires
9859 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009860
Willy Tarreau55165fe2009-05-10 12:02:55 +02009861 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009862 "timeout client".
9863
9864
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009865tcp-response content <action> [{if | unless} <condition>]
9866 Perform an action on a session response depending on a layer 4-7 condition
9867 May be used in sections : defaults | frontend | listen | backend
9868 no | no | yes | yes
9869 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009870 <action> defines the action to perform if the condition applies. See
9871 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009872
9873 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9874
Davor Ocelice9ed2812017-12-25 17:49:28 +01009875 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009876 called "TCP content inspection". During this stage, ACL-based rules are
9877 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009878 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9879 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009880
9881 Most often, these decisions will consider a protocol recognition or validity.
9882
9883 Content-based rules are evaluated in their exact declaration order. If no
9884 rule matches or if there is no rule, the default action is to accept the
9885 contents. There is no specific limit to the number of rules which may be
9886 inserted.
9887
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009888 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009889 - accept :
9890 accepts the response if the condition is true (when used with "if")
9891 or false (when used with "unless"). The first such rule executed ends
9892 the rules evaluation.
9893
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009894 - close :
9895 immediately closes the connection with the server if the condition is
9896 true (when used with "if"), or false (when used with "unless"). The
9897 first such rule executed ends the rules evaluation. The main purpose of
9898 this action is to force a connection to be finished between a client
9899 and a server after an exchange when the application protocol expects
9900 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009901 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009902 protocols.
9903
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009904 - reject :
9905 rejects the response if the condition is true (when used with "if")
9906 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009907 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009908
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009909 - set-var(<var-name>) <expr>
9910 Sets a variable.
9911
Christopher Faulet85d79c92016-11-09 16:54:56 +01009912 - unset-var(<var-name>)
9913 Unsets a variable.
9914
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009915 - sc-inc-gpc0(<sc-id>):
9916 This action increments the GPC0 counter according to the sticky
9917 counter designated by <sc-id>. If an error occurs, this action fails
9918 silently and the actions evaluation continues.
9919
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009920 - sc-inc-gpc1(<sc-id>):
9921 This action increments the GPC1 counter according to the sticky
9922 counter designated by <sc-id>. If an error occurs, this action fails
9923 silently and the actions evaluation continues.
9924
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009925 - sc-set-gpt0(<sc-id>) <int> :
9926 This action sets the GPT0 tag according to the sticky counter designated
9927 by <sc-id> and the value of <int>. The expected result is a boolean. If
9928 an error occurs, this action silently fails and the actions evaluation
9929 continues.
9930
Willy Tarreau2d392c22015-08-24 01:43:45 +02009931 - "silent-drop" :
9932 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009933 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009934 to prevent the client from being notified. The effect it then that the
9935 client still sees an established connection while there's none on
9936 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9937 except that it doesn't use any local resource at all on the machine
9938 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009939 slow down stronger attackers. It is important to understand the impact
9940 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009941 client and HAProxy (firewalls, proxies, load balancers) will also keep
9942 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009943 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009944 TCP_REPAIR socket option is used to block the emission of a TCP
9945 reset. On other systems, the socket's TTL is reduced to 1 so that the
9946 TCP reset doesn't pass the first router, though it's still delivered to
9947 local networks. Do not use it unless you fully understand how it works.
9948
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009949 - send-spoe-group <engine-name> <group-name>
9950 Send a group of SPOE messages.
9951
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009952 Note that the "if/unless" condition is optional. If no condition is set on
9953 the action, it is simply performed unconditionally. That can be useful for
9954 for changing the default action to a reject.
9955
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009956 It is perfectly possible to match layer 7 contents with "tcp-response
9957 content" rules, but then it is important to ensure that a full response has
9958 been buffered, otherwise no contents will match. In order to achieve this,
9959 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009960 period.
9961
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009962 The "set-var" is used to set the content of a variable. The variable is
9963 declared inline.
9964
Daniel Schneller0b547052016-03-21 20:46:57 +01009965 <var-name> The name of the variable starts with an indication about
9966 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009967 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009968 "sess" : the variable is shared with the whole session
9969 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009970 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009971 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009972 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009973 "res" : the variable is shared only during response
9974 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009975 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009976 The name may only contain characters 'a-z', 'A-Z', '0-9',
9977 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009978
9979 <expr> Is a standard HAProxy expression formed by a sample-fetch
9980 followed by some converters.
9981
9982 Example:
9983
9984 tcp-request content set-var(sess.my_var) src
9985
Christopher Faulet85d79c92016-11-09 16:54:56 +01009986 The "unset-var" is used to unset a variable. See above for details about
9987 <var-name>.
9988
9989 Example:
9990
9991 tcp-request content unset-var(sess.my_var)
9992
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009993 The "send-spoe-group" is used to trigger sending of a group of SPOE
9994 messages. To do so, the SPOE engine used to send messages must be defined, as
9995 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9996 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9997 the SPOE agent name must be used.
9998
9999 <engine-name> The SPOE engine name.
10000
10001 <group-name> The SPOE group name as specified in the engine configuration.
10002
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010003 See section 7 about ACL usage.
10004
10005 See also : "tcp-request content", "tcp-response inspect-delay"
10006
10007
Willy Tarreau4f614292016-10-21 17:49:36 +020010008tcp-request session <action> [{if | unless} <condition>]
10009 Perform an action on a validated session depending on a layer 5 condition
10010 May be used in sections : defaults | frontend | listen | backend
10011 no | yes | yes | no
10012 Arguments :
10013 <action> defines the action to perform if the condition applies. See
10014 below.
10015
10016 <condition> is a standard layer5-only ACL-based condition (see section 7).
10017
Davor Ocelice9ed2812017-12-25 17:49:28 +010010018 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010019 it is possible to evaluate some conditions to decide whether this session
10020 must be accepted or dropped or have its counters tracked. Those conditions
10021 cannot make use of any data contents because no buffers are allocated yet and
10022 the processing cannot wait at this stage. The main use case it to copy some
10023 early information into variables (since variables are accessible in the
10024 session), or to keep track of some information collected after the handshake,
10025 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010026 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010027 extracted information can thus be copied to a variable or tracked using
10028 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10029 with other rulesets. Most operations performed here could also be performed
10030 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10031 for each new request, and that might not always be acceptable. For example a
10032 rule might increment a counter on each evaluation. It would also be possible
10033 that a country is resolved by geolocation from the source IP address,
10034 assigned to a session-wide variable, then the source address rewritten from
10035 an HTTP header for all requests. If some contents need to be inspected in
10036 order to take the decision, the "tcp-request content" statements must be used
10037 instead.
10038
10039 The "tcp-request session" rules are evaluated in their exact declaration
10040 order. If no rule matches or if there is no rule, the default action is to
10041 accept the incoming session. There is no specific limit to the number of
10042 rules which may be inserted.
10043
10044 Several types of actions are supported :
10045 - accept : the request is accepted
10046 - reject : the request is rejected and the connection is closed
10047 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10048 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010049 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010050 - sc-set-gpt0(<sc-id>) <int>
10051 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010052 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010053 - silent-drop
10054
10055 These actions have the same meaning as their respective counter-parts in
10056 "tcp-request connection" and "tcp-request content", so please refer to these
10057 sections for a complete description.
10058
10059 Note that the "if/unless" condition is optional. If no condition is set on
10060 the action, it is simply performed unconditionally. That can be useful for
10061 "track-sc*" actions as well as for changing the default action to a reject.
10062
10063 Example: track the original source address by default, or the one advertised
10064 in the PROXY protocol header for connection coming from the local
10065 proxies. The first connection-level rule enables receipt of the
10066 PROXY protocol for these ones, the second rule tracks whatever
10067 address we decide to keep after optional decoding.
10068
10069 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10070 tcp-request session track-sc0 src
10071
10072 Example: accept all sessions from white-listed hosts, reject too fast
10073 sessions without counting them, and track accepted sessions.
10074 This results in session rate being capped from abusive sources.
10075
10076 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10077 tcp-request session reject if { src_sess_rate gt 10 }
10078 tcp-request session track-sc0 src
10079
10080 Example: accept all sessions from white-listed hosts, count all other
10081 sessions and reject too fast ones. This results in abusive ones
10082 being blocked as long as they don't slow down.
10083
10084 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10085 tcp-request session track-sc0 src
10086 tcp-request session reject if { sc0_sess_rate gt 10 }
10087
10088 See section 7 about ACL usage.
10089
10090 See also : "tcp-request connection", "tcp-request content", "stick-table"
10091
10092
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010093tcp-response inspect-delay <timeout>
10094 Set the maximum allowed time to wait for a response during content inspection
10095 May be used in sections : defaults | frontend | listen | backend
10096 no | no | yes | yes
10097 Arguments :
10098 <timeout> is the timeout value specified in milliseconds by default, but
10099 can be in any other unit if the number is suffixed by the unit,
10100 as explained at the top of this document.
10101
10102 See also : "tcp-response content", "tcp-request inspect-delay".
10103
10104
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010105timeout check <timeout>
10106 Set additional check timeout, but only after a connection has been already
10107 established.
10108
10109 May be used in sections: defaults | frontend | listen | backend
10110 yes | no | yes | yes
10111 Arguments:
10112 <timeout> is the timeout value specified in milliseconds by default, but
10113 can be in any other unit if the number is suffixed by the unit,
10114 as explained at the top of this document.
10115
10116 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10117 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010118 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010119 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010120 (Please also note that there is no valid reason to have such long connect
10121 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10122 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010123
10124 If "timeout check" is not set haproxy uses "inter" for complete check
10125 timeout (connect + read) exactly like all <1.3.15 version.
10126
10127 In most cases check request is much simpler and faster to handle than normal
10128 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010129 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010130
10131 This parameter is specific to backends, but can be specified once for all in
10132 "defaults" sections. This is in fact one of the easiest solutions not to
10133 forget about it.
10134
Willy Tarreau41a340d2008-01-22 12:25:31 +010010135 See also: "timeout connect", "timeout queue", "timeout server",
10136 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010137
10138
Willy Tarreau0ba27502007-12-24 16:55:16 +010010139timeout client <timeout>
10140timeout clitimeout <timeout> (deprecated)
10141 Set the maximum inactivity time on the client side.
10142 May be used in sections : defaults | frontend | listen | backend
10143 yes | yes | yes | no
10144 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010145 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010146 can be in any other unit if the number is suffixed by the unit,
10147 as explained at the top of this document.
10148
10149 The inactivity timeout applies when the client is expected to acknowledge or
10150 send data. In HTTP mode, this timeout is particularly important to consider
10151 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010152 response while it is reading data sent by the server. That said, for the
10153 first phase, it is preferable to set the "timeout http-request" to better
10154 protect HAProxy from Slowloris like attacks. The value is specified in
10155 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010156 suffixed by the unit, as specified at the top of this document. In TCP mode
10157 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10158 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010159 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010160 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010161 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10162 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010163 which overrides "timeout client" and "timeout server" for tunnels, as well as
10164 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010165
10166 This parameter is specific to frontends, but can be specified once for all in
10167 "defaults" sections. This is in fact one of the easiest solutions not to
10168 forget about it. An unspecified timeout results in an infinite timeout, which
10169 is not recommended. Such a usage is accepted and works but reports a warning
10170 during startup because it may results in accumulation of expired sessions in
10171 the system if the system's timeouts are not configured either.
10172
Willy Tarreau95c4e142017-11-26 12:18:55 +010010173 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010174
Willy Tarreau0ba27502007-12-24 16:55:16 +010010175 This parameter replaces the old, deprecated "clitimeout". It is recommended
10176 to use it to write new configurations. The form "timeout clitimeout" is
10177 provided only by backwards compatibility but its use is strongly discouraged.
10178
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010179 See also : "clitimeout", "timeout server", "timeout tunnel",
10180 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010181
10182
Willy Tarreau05cdd962014-05-10 14:30:07 +020010183timeout client-fin <timeout>
10184 Set the inactivity timeout on the client side for half-closed connections.
10185 May be used in sections : defaults | frontend | listen | backend
10186 yes | yes | yes | no
10187 Arguments :
10188 <timeout> is the timeout value specified in milliseconds by default, but
10189 can be in any other unit if the number is suffixed by the unit,
10190 as explained at the top of this document.
10191
10192 The inactivity timeout applies when the client is expected to acknowledge or
10193 send data while one direction is already shut down. This timeout is different
10194 from "timeout client" in that it only applies to connections which are closed
10195 in one direction. This is particularly useful to avoid keeping connections in
10196 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10197 problem is particularly common long connections such as RDP or WebSocket.
10198 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010199 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10200 frame was sent, often indicating an expectation that the connection quickly
10201 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010202
10203 This parameter is specific to frontends, but can be specified once for all in
10204 "defaults" sections. By default it is not set, so half-closed connections
10205 will use the other timeouts (timeout.client or timeout.tunnel).
10206
10207 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10208
10209
Willy Tarreau0ba27502007-12-24 16:55:16 +010010210timeout connect <timeout>
10211timeout contimeout <timeout> (deprecated)
10212 Set the maximum time to wait for a connection attempt to a server to succeed.
10213 May be used in sections : defaults | frontend | listen | backend
10214 yes | no | yes | yes
10215 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010216 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010217 can be in any other unit if the number is suffixed by the unit,
10218 as explained at the top of this document.
10219
10220 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010221 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010222 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010223 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010224 connect timeout also presets both queue and tarpit timeouts to the same value
10225 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010226
10227 This parameter is specific to backends, but can be specified once for all in
10228 "defaults" sections. This is in fact one of the easiest solutions not to
10229 forget about it. An unspecified timeout results in an infinite timeout, which
10230 is not recommended. Such a usage is accepted and works but reports a warning
10231 during startup because it may results in accumulation of failed sessions in
10232 the system if the system's timeouts are not configured either.
10233
10234 This parameter replaces the old, deprecated "contimeout". It is recommended
10235 to use it to write new configurations. The form "timeout contimeout" is
10236 provided only by backwards compatibility but its use is strongly discouraged.
10237
Willy Tarreau41a340d2008-01-22 12:25:31 +010010238 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10239 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010240
10241
Willy Tarreaub16a5742010-01-10 14:46:16 +010010242timeout http-keep-alive <timeout>
10243 Set the maximum allowed time to wait for a new HTTP request to appear
10244 May be used in sections : defaults | frontend | listen | backend
10245 yes | yes | yes | yes
10246 Arguments :
10247 <timeout> is the timeout value specified in milliseconds by default, but
10248 can be in any other unit if the number is suffixed by the unit,
10249 as explained at the top of this document.
10250
10251 By default, the time to wait for a new request in case of keep-alive is set
10252 by "timeout http-request". However this is not always convenient because some
10253 people want very short keep-alive timeouts in order to release connections
10254 faster, and others prefer to have larger ones but still have short timeouts
10255 once the request has started to present itself.
10256
10257 The "http-keep-alive" timeout covers these needs. It will define how long to
10258 wait for a new HTTP request to start coming after a response was sent. Once
10259 the first byte of request has been seen, the "http-request" timeout is used
10260 to wait for the complete request to come. Note that empty lines prior to a
10261 new request do not refresh the timeout and are not counted as a new request.
10262
10263 There is also another difference between the two timeouts : when a connection
10264 expires during timeout http-keep-alive, no error is returned, the connection
10265 just closes. If the connection expires in "http-request" while waiting for a
10266 connection to complete, a HTTP 408 error is returned.
10267
10268 In general it is optimal to set this value to a few tens to hundreds of
10269 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010270 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010271 1 millisecond) it will probably only accept pipelined requests but not the
10272 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010273 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010274
10275 If this parameter is not set, the "http-request" timeout applies, and if both
10276 are not set, "timeout client" still applies at the lower level. It should be
10277 set in the frontend to take effect, unless the frontend is in TCP mode, in
10278 which case the HTTP backend's timeout will be used.
10279
Willy Tarreau95c4e142017-11-26 12:18:55 +010010280 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10281 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010282 (where we only have one connection per client and a connection setup).
10283
Willy Tarreaub16a5742010-01-10 14:46:16 +010010284 See also : "timeout http-request", "timeout client".
10285
10286
Willy Tarreau036fae02008-01-06 13:24:40 +010010287timeout http-request <timeout>
10288 Set the maximum allowed time to wait for a complete HTTP request
10289 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010290 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010291 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010292 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010293 can be in any other unit if the number is suffixed by the unit,
10294 as explained at the top of this document.
10295
10296 In order to offer DoS protection, it may be required to lower the maximum
10297 accepted time to receive a complete HTTP request without affecting the client
10298 timeout. This helps protecting against established connections on which
10299 nothing is sent. The client timeout cannot offer a good protection against
10300 this abuse because it is an inactivity timeout, which means that if the
10301 attacker sends one character every now and then, the timeout will not
10302 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010303 types, the request will be aborted if it does not complete in time. When the
10304 timeout expires, an HTTP 408 response is sent to the client to inform it
10305 about the problem, and the connection is closed. The logs will report
10306 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010307 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010308 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10309 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010310
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010311 By default, this timeout only applies to the header part of the request,
10312 and not to any data. As soon as the empty line is received, this timeout is
10313 not used anymore. When combined with "option http-buffer-request", this
10314 timeout also applies to the body of the request..
10315 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010316 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010317
10318 Generally it is enough to set it to a few seconds, as most clients send the
10319 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010320 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010321 generally work on local networks as long as there are no packet losses. This
10322 will prevent people from sending bare HTTP requests using telnet.
10323
10324 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010325 chunk of the incoming request. It should be set in the frontend to take
10326 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10327 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010328
Willy Tarreau0f228a02015-05-01 15:37:53 +020010329 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010330 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010331
Willy Tarreau844e3c52008-01-11 16:28:18 +010010332
10333timeout queue <timeout>
10334 Set the maximum time to wait in the queue for a connection slot to be free
10335 May be used in sections : defaults | frontend | listen | backend
10336 yes | no | yes | yes
10337 Arguments :
10338 <timeout> is the timeout value specified in milliseconds by default, but
10339 can be in any other unit if the number is suffixed by the unit,
10340 as explained at the top of this document.
10341
10342 When a server's maxconn is reached, connections are left pending in a queue
10343 which may be server-specific or global to the backend. In order not to wait
10344 indefinitely, a timeout is applied to requests pending in the queue. If the
10345 timeout is reached, it is considered that the request will almost never be
10346 served, so it is dropped and a 503 error is returned to the client.
10347
10348 The "timeout queue" statement allows to fix the maximum time for a request to
10349 be left pending in a queue. If unspecified, the same value as the backend's
10350 connection timeout ("timeout connect") is used, for backwards compatibility
10351 with older versions with no "timeout queue" parameter.
10352
10353 See also : "timeout connect", "contimeout".
10354
10355
10356timeout server <timeout>
10357timeout srvtimeout <timeout> (deprecated)
10358 Set the maximum inactivity time on the server side.
10359 May be used in sections : defaults | frontend | listen | backend
10360 yes | no | yes | yes
10361 Arguments :
10362 <timeout> is the timeout value specified in milliseconds by default, but
10363 can be in any other unit if the number is suffixed by the unit,
10364 as explained at the top of this document.
10365
10366 The inactivity timeout applies when the server is expected to acknowledge or
10367 send data. In HTTP mode, this timeout is particularly important to consider
10368 during the first phase of the server's response, when it has to send the
10369 headers, as it directly represents the server's processing time for the
10370 request. To find out what value to put there, it's often good to start with
10371 what would be considered as unacceptable response times, then check the logs
10372 to observe the response time distribution, and adjust the value accordingly.
10373
10374 The value is specified in milliseconds by default, but can be in any other
10375 unit if the number is suffixed by the unit, as specified at the top of this
10376 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10377 recommended that the client timeout remains equal to the server timeout in
10378 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010379 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010380 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010381 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10382 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010383 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10384 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010385
10386 This parameter is specific to backends, but can be specified once for all in
10387 "defaults" sections. This is in fact one of the easiest solutions not to
10388 forget about it. An unspecified timeout results in an infinite timeout, which
10389 is not recommended. Such a usage is accepted and works but reports a warning
10390 during startup because it may results in accumulation of expired sessions in
10391 the system if the system's timeouts are not configured either.
10392
10393 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10394 to use it to write new configurations. The form "timeout srvtimeout" is
10395 provided only by backwards compatibility but its use is strongly discouraged.
10396
Willy Tarreauce887fd2012-05-12 12:50:00 +020010397 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010398
Willy Tarreau05cdd962014-05-10 14:30:07 +020010399
10400timeout server-fin <timeout>
10401 Set the inactivity timeout on the server side for half-closed connections.
10402 May be used in sections : defaults | frontend | listen | backend
10403 yes | no | yes | yes
10404 Arguments :
10405 <timeout> is the timeout value specified in milliseconds by default, but
10406 can be in any other unit if the number is suffixed by the unit,
10407 as explained at the top of this document.
10408
10409 The inactivity timeout applies when the server is expected to acknowledge or
10410 send data while one direction is already shut down. This timeout is different
10411 from "timeout server" in that it only applies to connections which are closed
10412 in one direction. This is particularly useful to avoid keeping connections in
10413 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10414 This problem is particularly common long connections such as RDP or WebSocket.
10415 Note that this timeout can override "timeout tunnel" when a connection shuts
10416 down in one direction. This setting was provided for completeness, but in most
10417 situations, it should not be needed.
10418
10419 This parameter is specific to backends, but can be specified once for all in
10420 "defaults" sections. By default it is not set, so half-closed connections
10421 will use the other timeouts (timeout.server or timeout.tunnel).
10422
10423 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10424
Willy Tarreau844e3c52008-01-11 16:28:18 +010010425
10426timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010427 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010428 May be used in sections : defaults | frontend | listen | backend
10429 yes | yes | yes | yes
10430 Arguments :
10431 <timeout> is the tarpit duration specified in milliseconds by default, but
10432 can be in any other unit if the number is suffixed by the unit,
10433 as explained at the top of this document.
10434
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010435 When a connection is tarpitted using "http-request tarpit" or
10436 "reqtarpit", it is maintained open with no activity for a certain
10437 amount of time, then closed. "timeout tarpit" defines how long it will
10438 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010439
10440 The value is specified in milliseconds by default, but can be in any other
10441 unit if the number is suffixed by the unit, as specified at the top of this
10442 document. If unspecified, the same value as the backend's connection timeout
10443 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010444 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010445
10446 See also : "timeout connect", "contimeout".
10447
10448
Willy Tarreauce887fd2012-05-12 12:50:00 +020010449timeout tunnel <timeout>
10450 Set the maximum inactivity time on the client and server side for tunnels.
10451 May be used in sections : defaults | frontend | listen | backend
10452 yes | no | yes | yes
10453 Arguments :
10454 <timeout> is the timeout value specified in milliseconds by default, but
10455 can be in any other unit if the number is suffixed by the unit,
10456 as explained at the top of this document.
10457
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010458 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010459 between a client and a server, and the connection remains inactive in both
10460 directions. This timeout supersedes both the client and server timeouts once
10461 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010462 analyzer remains attached to either connection (e.g. tcp content rules are
10463 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010464 when switching to the WebSocket protocol, or forwarding a CONNECT request
10465 to a proxy), or after the first response when no keepalive/close option is
10466 specified.
10467
Willy Tarreau05cdd962014-05-10 14:30:07 +020010468 Since this timeout is usually used in conjunction with long-lived connections,
10469 it usually is a good idea to also set "timeout client-fin" to handle the
10470 situation where a client suddenly disappears from the net and does not
10471 acknowledge a close, or sends a shutdown and does not acknowledge pending
10472 data anymore. This can happen in lossy networks where firewalls are present,
10473 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10474 state.
10475
Willy Tarreauce887fd2012-05-12 12:50:00 +020010476 The value is specified in milliseconds by default, but can be in any other
10477 unit if the number is suffixed by the unit, as specified at the top of this
10478 document. Whatever the expected normal idle time, it is a good practice to
10479 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010480 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010481
10482 This parameter is specific to backends, but can be specified once for all in
10483 "defaults" sections. This is in fact one of the easiest solutions not to
10484 forget about it.
10485
10486 Example :
10487 defaults http
10488 option http-server-close
10489 timeout connect 5s
10490 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010491 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010492 timeout server 30s
10493 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10494
Willy Tarreau05cdd962014-05-10 14:30:07 +020010495 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010496
10497
Willy Tarreau844e3c52008-01-11 16:28:18 +010010498transparent (deprecated)
10499 Enable client-side transparent proxying
10500 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010501 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010502 Arguments : none
10503
10504 This keyword was introduced in order to provide layer 7 persistence to layer
10505 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10506 connection for a remote address to a local process (here HAProxy), and let
10507 this process know what address was initially requested. When this option is
10508 used, sessions without cookies will be forwarded to the original destination
10509 IP address of the incoming request (which should match that of another
10510 equipment), while requests with cookies will still be forwarded to the
10511 appropriate server.
10512
10513 The "transparent" keyword is deprecated, use "option transparent" instead.
10514
10515 Note that contrary to a common belief, this option does NOT make HAProxy
10516 present the client's IP to the server when establishing the connection.
10517
Willy Tarreau844e3c52008-01-11 16:28:18 +010010518 See also: "option transparent"
10519
William Lallemanda73203e2012-03-12 12:48:57 +010010520unique-id-format <string>
10521 Generate a unique ID for each request.
10522 May be used in sections : defaults | frontend | listen | backend
10523 yes | yes | yes | no
10524 Arguments :
10525 <string> is a log-format string.
10526
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010527 This keyword creates a ID for each request using the custom log format. A
10528 unique ID is useful to trace a request passing through many components of
10529 a complex infrastructure. The newly created ID may also be logged using the
10530 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010531
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010532 The format should be composed from elements that are guaranteed to be
10533 unique when combined together. For instance, if multiple haproxy instances
10534 are involved, it might be important to include the node name. It is often
10535 needed to log the incoming connection's source and destination addresses
10536 and ports. Note that since multiple requests may be performed over the same
10537 connection, including a request counter may help differentiate them.
10538 Similarly, a timestamp may protect against a rollover of the counter.
10539 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010540
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010541 It is recommended to use hexadecimal notation for many fields since it
10542 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010543
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010544 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010545
Julien Vehentf21be322014-03-07 08:27:34 -050010546 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010547
10548 will generate:
10549
10550 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10551
10552 See also: "unique-id-header"
10553
10554unique-id-header <name>
10555 Add a unique ID header in the HTTP request.
10556 May be used in sections : defaults | frontend | listen | backend
10557 yes | yes | yes | no
10558 Arguments :
10559 <name> is the name of the header.
10560
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010561 Add a unique-id header in the HTTP request sent to the server, using the
10562 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010563
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010564 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010565
Julien Vehentf21be322014-03-07 08:27:34 -050010566 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010567 unique-id-header X-Unique-ID
10568
10569 will generate:
10570
10571 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10572
10573 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010574
Willy Tarreauf51658d2014-04-23 01:21:56 +020010575use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010576 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010577 May be used in sections : defaults | frontend | listen | backend
10578 no | yes | yes | no
10579 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010580 <backend> is the name of a valid backend or "listen" section, or a
10581 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010582
Willy Tarreauf51658d2014-04-23 01:21:56 +020010583 <condition> is a condition composed of ACLs, as described in section 7. If
10584 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010585
10586 When doing content-switching, connections arrive on a frontend and are then
10587 dispatched to various backends depending on a number of conditions. The
10588 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010589 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010590 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010591 source address validation) or combined with a "tcp-request" rule to wait for
10592 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010593
10594 There may be as many "use_backend" rules as desired. All of these rules are
10595 evaluated in their declaration order, and the first one which matches will
10596 assign the backend.
10597
10598 In the first form, the backend will be used if the condition is met. In the
10599 second form, the backend will be used if the condition is not met. If no
10600 condition is valid, the backend defined with "default_backend" will be used.
10601 If no default backend is defined, either the servers in the same section are
10602 used (in case of a "listen" section) or, in case of a frontend, no server is
10603 used and a 503 service unavailable response is returned.
10604
Willy Tarreau51aecc72009-07-12 09:47:04 +020010605 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010606 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010607 and backend processing will immediately follow, or the backend will wait for
10608 a complete HTTP request to get in. This feature is useful when a frontend
10609 must decode several protocols on a unique port, one of them being HTTP.
10610
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010611 When <backend> is a simple name, it is resolved at configuration time, and an
10612 error is reported if the specified backend does not exist. If <backend> is
10613 a log-format string instead, no check may be done at configuration time, so
10614 the backend name is resolved dynamically at run time. If the resulting
10615 backend name does not correspond to any valid backend, no other rule is
10616 evaluated, and the default_backend directive is applied instead. Note that
10617 when using dynamic backend names, it is highly recommended to use a prefix
10618 that no other backend uses in order to ensure that an unauthorized backend
10619 cannot be forced from the request.
10620
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010621 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010622 used to detect the association between frontends and backends to compute the
10623 backend's "fullconn" setting. This cannot be done for dynamic names.
10624
10625 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10626 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010627
Willy Tarreau036fae02008-01-06 13:24:40 +010010628
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010629use-server <server> if <condition>
10630use-server <server> unless <condition>
10631 Only use a specific server if/unless an ACL-based condition is matched.
10632 May be used in sections : defaults | frontend | listen | backend
10633 no | no | yes | yes
10634 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010635 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010636
10637 <condition> is a condition composed of ACLs, as described in section 7.
10638
10639 By default, connections which arrive to a backend are load-balanced across
10640 the available servers according to the configured algorithm, unless a
10641 persistence mechanism such as a cookie is used and found in the request.
10642
10643 Sometimes it is desirable to forward a particular request to a specific
10644 server without having to declare a dedicated backend for this server. This
10645 can be achieved using the "use-server" rules. These rules are evaluated after
10646 the "redirect" rules and before evaluating cookies, and they have precedence
10647 on them. There may be as many "use-server" rules as desired. All of these
10648 rules are evaluated in their declaration order, and the first one which
10649 matches will assign the server.
10650
10651 If a rule designates a server which is down, and "option persist" is not used
10652 and no force-persist rule was validated, it is ignored and evaluation goes on
10653 with the next rules until one matches.
10654
10655 In the first form, the server will be used if the condition is met. In the
10656 second form, the server will be used if the condition is not met. If no
10657 condition is valid, the processing continues and the server will be assigned
10658 according to other persistence mechanisms.
10659
10660 Note that even if a rule is matched, cookie processing is still performed but
10661 does not assign the server. This allows prefixed cookies to have their prefix
10662 stripped.
10663
10664 The "use-server" statement works both in HTTP and TCP mode. This makes it
10665 suitable for use with content-based inspection. For instance, a server could
10666 be selected in a farm according to the TLS SNI field. And if these servers
10667 have their weight set to zero, they will not be used for other traffic.
10668
10669 Example :
10670 # intercept incoming TLS requests based on the SNI field
10671 use-server www if { req_ssl_sni -i www.example.com }
10672 server www 192.168.0.1:443 weight 0
10673 use-server mail if { req_ssl_sni -i mail.example.com }
10674 server mail 192.168.0.1:587 weight 0
10675 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010676 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010677 # all the rest is forwarded to this server
10678 server default 192.168.0.2:443 check
10679
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010680 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010681
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010682
Davor Ocelice9ed2812017-12-25 17:49:28 +0100106835. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010684--------------------------
10685
10686The "bind", "server" and "default-server" keywords support a number of settings
10687depending on some build options and on the system HAProxy was built on. These
10688settings generally each consist in one word sometimes followed by a value,
10689written on the same line as the "bind" or "server" line. All these options are
10690described in this section.
10691
10692
106935.1. Bind options
10694-----------------
10695
10696The "bind" keyword supports a certain number of settings which are all passed
10697as arguments on the same line. The order in which those arguments appear makes
10698no importance, provided that they appear after the bind address. All of these
10699parameters are optional. Some of them consist in a single words (booleans),
10700while other ones expect a value after them. In this case, the value must be
10701provided immediately after the setting name.
10702
10703The currently supported settings are the following ones.
10704
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010705accept-netscaler-cip <magic number>
10706 Enforces the use of the NetScaler Client IP insertion protocol over any
10707 connection accepted by any of the TCP sockets declared on the same line. The
10708 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10709 the incoming connection to be used everywhere an address is used, with the
10710 only exception of "tcp-request connection" rules which will only see the
10711 real connection address. Logs will reflect the addresses indicated in the
10712 protocol, unless it is violated, in which case the real address will still
10713 be used. This keyword combined with support from external components can be
10714 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010715 mechanism which is not always reliable and not even always usable. See also
10716 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10717 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010718
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010719accept-proxy
10720 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010721 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10722 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010723 3/4 addresses of the incoming connection to be used everywhere an address is
10724 used, with the only exception of "tcp-request connection" rules which will
10725 only see the real connection address. Logs will reflect the addresses
10726 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010727 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010728 components can be used as an efficient and reliable alternative to the
10729 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010730 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10731 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010732
Olivier Houchardc2aae742017-09-22 18:26:28 +020010733allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010734 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchardc2aae742017-09-22 18:26:28 +020010735 due to security considerations.
10736
Willy Tarreauab861d32013-04-02 02:30:41 +020010737alpn <protocols>
10738 This enables the TLS ALPN extension and advertises the specified protocol
10739 list as supported on top of ALPN. The protocol list consists in a comma-
10740 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10741 quotes). This requires that the SSL library is build with support for TLS
10742 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010743 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10744 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10745 now obsolete NPN extension. At the time of writing this, most browsers still
10746 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10747 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10748 are expected to be supported, both versions can be advertised, in order of
10749 preference, like below :
10750
10751 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010752
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010753backlog <backlog>
10754 Sets the socket's backlog to this value. If unspecified, the frontend's
10755 backlog is used instead, which generally defaults to the maxconn value.
10756
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010757curves <curves>
10758 This setting is only available when support for OpenSSL was built in. It sets
10759 the string describing the list of elliptic curves algorithms ("curve suite")
10760 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10761 string is a colon-delimited list of curve name.
10762 Example: "X25519:P-256" (without quote)
10763 When "curves" is set, "ecdhe" parameter is ignored.
10764
Emeric Brun7fb34422012-09-28 15:26:15 +020010765ecdhe <named curve>
10766 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010767 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10768 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010769
Emeric Brunfd33a262012-10-11 16:28:27 +020010770ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010771 This setting is only available when support for OpenSSL was built in. It
10772 designates a PEM file from which to load CA certificates used to verify
10773 client's certificate.
10774
Emeric Brunb6dc9342012-09-28 17:55:37 +020010775ca-ignore-err [all|<errorID>,...]
10776 This setting is only available when support for OpenSSL was built in.
10777 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10778 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10779 error is ignored.
10780
Christopher Faulet31af49d2015-06-09 17:29:50 +020010781ca-sign-file <cafile>
10782 This setting is only available when support for OpenSSL was built in. It
10783 designates a PEM file containing both the CA certificate and the CA private
10784 key used to create and sign server's certificates. This is a mandatory
10785 setting when the dynamic generation of certificates is enabled. See
10786 'generate-certificates' for details.
10787
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010788ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010789 This setting is only available when support for OpenSSL was built in. It is
10790 the CA private key passphrase. This setting is optional and used only when
10791 the dynamic generation of certificates is enabled. See
10792 'generate-certificates' for details.
10793
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010794ciphers <ciphers>
10795 This setting is only available when support for OpenSSL was built in. It sets
10796 the string describing the list of cipher algorithms ("cipher suite") that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010797 negotiated during the SSL/TLS handshake except for TLSv1.3. The format of the
10798 string is defined in "man 1 ciphers" from OpenSSL man pages, and can be for
10799 instance a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without
10800 quotes). Depending on the compatibility and security requirements, the list
10801 of suitable ciphers depends on a variety of variables. For background
10802 information and recommendations see e.g.
10803 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10804 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10805 cipher configuration, please check the "ciphersuites" keyword.
10806
10807ciphersuites <ciphersuites>
10808 This setting is only available when support for OpenSSL was built in and
10809 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10810 the list of cipher algorithms ("cipher suite") that are negotiated during the
10811 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
10812 OpenSSL man pages under the "ciphersuites" section, and can be for instance a
10813 string such as
10814 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
10815 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
10816 the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010817
Emeric Brunfd33a262012-10-11 16:28:27 +020010818crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010819 This setting is only available when support for OpenSSL was built in. It
10820 designates a PEM file from which to load certificate revocation list used
10821 to verify client's certificate.
10822
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010823crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010824 This setting is only available when support for OpenSSL was built in. It
10825 designates a PEM file containing both the required certificates and any
10826 associated private keys. This file can be built by concatenating multiple
10827 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10828 requires an intermediate certificate, this can also be concatenated into this
10829 file.
10830
10831 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10832 are loaded.
10833
10834 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010835 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010836 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10837 specified multiple times in order to load certificates from multiple files or
10838 directories. The certificates will be presented to clients who provide a
10839 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010840 subjects. Wildcards are supported, where a wildcard character '*' is used
10841 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010842 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010843
10844 If no SNI is provided by the client or if the SSL library does not support
10845 TLS extensions, or if the client provides an SNI hostname which does not
10846 match any certificate, then the first loaded certificate will be presented.
10847 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010848 recommended to load the default one first as a file or to ensure that it will
10849 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010850
Emeric Brune032bfa2012-09-28 13:01:45 +020010851 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010852
Davor Ocelice9ed2812017-12-25 17:49:28 +010010853 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010854 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010855 choose a web server that the CA believes requires an intermediate CA (for
10856 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010857 others, e.g. nginx, result in a wrong bundle that will not work for some
10858 clients).
10859
Emeric Brun4147b2e2014-06-16 18:36:30 +020010860 For each PEM file, haproxy checks for the presence of file at the same path
10861 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10862 Status Request extension (also known as "OCSP stapling") is automatically
10863 enabled. The content of this file is optional. If not empty, it must contain
10864 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10865 must comply with the following rules: it has to indicate a good status,
10866 it has to be a single response for the certificate of the PEM file, and it
10867 has to be valid at the moment of addition. If these rules are not respected
10868 the OCSP Response is ignored and a warning is emitted. In order to identify
10869 which certificate an OCSP Response applies to, the issuer's certificate is
10870 necessary. If the issuer's certificate is not found in the PEM file, it will
10871 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10872 if it exists otherwise it will fail with an error.
10873
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010874 For each PEM file, haproxy also checks for the presence of file at the same
10875 path suffixed by ".sctl". If such file is found, support for Certificate
10876 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10877 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10878 to check basic syntax, but no signatures are verified.
10879
yanbzhu6c25e9e2016-01-05 12:52:02 -050010880 There are cases where it is desirable to support multiple key types, e.g. RSA
10881 and ECDSA in the cipher suites offered to the clients. This allows clients
10882 that support EC certificates to be able to use EC ciphers, while
10883 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010884
10885 In order to provide this functionality, multiple PEM files, each with a
10886 different key type, are required. To associate these PEM files into a
10887 "cert bundle" that is recognized by haproxy, they must be named in the
10888 following way: All PEM files that are to be bundled must have the same base
10889 name, with a suffix indicating the key type. Currently, three suffixes are
10890 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10891 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10892 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10893 suffix matters. To load this bundle into haproxy, specify the base name only:
10894
10895 Example : bind :8443 ssl crt example.pem
10896
yanbzhu6c25e9e2016-01-05 12:52:02 -050010897 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010898 a cert bundle.
10899
Davor Ocelice9ed2812017-12-25 17:49:28 +010010900 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010901 support multiple key types. PEM files are combined based on Common Name
10902 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10903 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10904 entries in the certificates in that bundle, haproxy will not be able to
10905 provide multi-cert support.
10906
10907 Assuming bundle in the example above contained the following:
10908
10909 Filename | CN | SAN
10910 -------------------+-----------------+-------------------
10911 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010912 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010913 example.pem.ecdsa | www.example.com | ecdsa.example.com
10914 -------------------+-----------------+-------------------
10915
10916 Users connecting with an SNI of "www.example.com" will be able
10917 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10918 "rsa.example.com" will only be able to use RSA cipher suites, and users
10919 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010920 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10921 no need to bundle certificates. ECDSA certificate will be preferred if client
10922 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010923
10924 If a directory name is given as the <cert> argument, haproxy will
10925 automatically search and load bundled files in that directory.
10926
10927 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10928 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10929 time, sctl is not supported in multi-certificate bundling.
10930
Emeric Brunb6dc9342012-09-28 17:55:37 +020010931crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010932 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010933 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010934 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010935 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010936
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010937crt-list <file>
10938 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010939 designates a list of PEM file with an optional ssl configuration and a SNI
10940 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010941
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010942 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10943
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010944 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10945 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010946 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010947 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010948
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010949 Wildcards are supported in the SNI filter. Negative filter are also supported,
10950 only useful in combination with a wildcard filter to exclude a particular SNI.
10951 The certificates will be presented to clients who provide a valid TLS Server
10952 Name Indication field matching one of the SNI filters. If no SNI filter is
10953 specified, the CN and alt subjects are used. This directive may be specified
10954 multiple times. See the "crt" option for more information. The default
10955 certificate is still needed to meet OpenSSL expectations. If it is not used,
10956 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010957
yanbzhu6c25e9e2016-01-05 12:52:02 -050010958 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010959 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010960 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10961 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10962 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010963
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010964 crt-list file example:
10965 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010966 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010967 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010968 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010969
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010970defer-accept
10971 Is an optional keyword which is supported only on certain Linux kernels. It
10972 states that a connection will only be accepted once some data arrive on it,
10973 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010974 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010975 performance by ensuring that most of the request is already available when
10976 the connection is accepted. On the other hand, it will not be able to detect
10977 connections which don't talk. It is important to note that this option is
10978 broken in all kernels up to 2.6.31, as the connection is never accepted until
10979 the client talks. This can cause issues with front firewalls which would see
10980 an established connection while the proxy will only see it in SYN_RECV. This
10981 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10982
William Lallemandf6975e92017-05-26 17:42:10 +020010983expose-fd listeners
10984 This option is only usable with the stats socket. It gives your stats socket
10985 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010986 During a reload with the master-worker mode, the process is automatically
10987 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010988 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010989
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010990force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010991 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010992 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010993 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010994 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010995
10996force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010997 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010998 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010999 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011000
11001force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011002 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011003 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011004 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011005
11006force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011007 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011008 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011009 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011010
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011011force-tlsv13
11012 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11013 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011014 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011015
Christopher Faulet31af49d2015-06-09 17:29:50 +020011016generate-certificates
11017 This setting is only available when support for OpenSSL was built in. It
11018 enables the dynamic SSL certificates generation. A CA certificate and its
11019 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11020 a transparent forward proxy, SSL requests generate errors because of a common
11021 name mismatch on the certificate presented to the client. With this option
11022 enabled, HAProxy will try to forge a certificate using the SNI hostname
11023 indicated by the client. This is done only if no certificate matches the SNI
11024 hostname (see 'crt-list'). If an error occurs, the default certificate is
11025 used, else the 'strict-sni' option is set.
11026 It can also be used when HAProxy is configured as a reverse proxy to ease the
11027 deployment of an architecture with many backends.
11028
11029 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11030 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011031 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011032 certificate is used many times.
11033
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011034gid <gid>
11035 Sets the group of the UNIX sockets to the designated system gid. It can also
11036 be set by default in the global section's "unix-bind" statement. Note that
11037 some platforms simply ignore this. This setting is equivalent to the "group"
11038 setting except that the group ID is used instead of its name. This setting is
11039 ignored by non UNIX sockets.
11040
11041group <group>
11042 Sets the group of the UNIX sockets to the designated system group. It can
11043 also be set by default in the global section's "unix-bind" statement. Note
11044 that some platforms simply ignore this. This setting is equivalent to the
11045 "gid" setting except that the group name is used instead of its gid. This
11046 setting is ignored by non UNIX sockets.
11047
11048id <id>
11049 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11050 sometimes it is more convenient to fix them to ease monitoring. This value
11051 must be strictly positive and unique within the listener/frontend. This
11052 option can only be used when defining only a single socket.
11053
11054interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011055 Restricts the socket to a specific interface. When specified, only packets
11056 received from that particular interface are processed by the socket. This is
11057 currently only supported on Linux. The interface must be a primary system
11058 interface, not an aliased interface. It is also possible to bind multiple
11059 frontends to the same address if they are bound to different interfaces. Note
11060 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011061 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11062 uses the same interface as inbound traffic, and its associated routing table,
11063 even if there are explicit routes through different interfaces configured.
11064 This can prove useful to address asymmetric routing issues when the same
11065 client IP addresses need to be able to reach frontends hosted on different
11066 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011067
Willy Tarreauabb175f2012-09-24 12:43:26 +020011068level <level>
11069 This setting is used with the stats sockets only to restrict the nature of
11070 the commands that can be issued on the socket. It is ignored by other
11071 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011072 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011073 read, and no change is allowed. It would make sense on systems where it
11074 is not easy to restrict access to the socket.
11075 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011076 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011077 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011078 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011079 all counters).
11080
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011081severity-output <format>
11082 This setting is used with the stats sockets only to configure severity
11083 level output prepended to informational feedback messages. Severity
11084 level of messages can range between 0 and 7, conforming to syslog
11085 rfc5424. Valid and successful socket commands requesting data
11086 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11087 prepended. It is ignored by other sockets. <format> can be one of :
11088 - "none" (default) no severity level is prepended to feedback messages.
11089 - "number" severity level is prepended as a number.
11090 - "string" severity level is prepended as a string following the
11091 rfc5424 convention.
11092
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011093maxconn <maxconn>
11094 Limits the sockets to this number of concurrent connections. Extraneous
11095 connections will remain in the system's backlog until a connection is
11096 released. If unspecified, the limit will be the same as the frontend's
11097 maxconn. Note that in case of port ranges or multiple addresses, the same
11098 value will be applied to each socket. This setting enables different
11099 limitations on expensive sockets, for instance SSL entries which may easily
11100 eat all memory.
11101
11102mode <mode>
11103 Sets the octal mode used to define access permissions on the UNIX socket. It
11104 can also be set by default in the global section's "unix-bind" statement.
11105 Note that some platforms simply ignore this. This setting is ignored by non
11106 UNIX sockets.
11107
11108mss <maxseg>
11109 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11110 connections. This can be used to force a lower MSS for certain specific
11111 ports, for instance for connections passing through a VPN. Note that this
11112 relies on a kernel feature which is theoretically supported under Linux but
11113 was buggy in all versions prior to 2.6.28. It may or may not work on other
11114 operating systems. It may also not change the advertised value but change the
11115 effective size of outgoing segments. The commonly advertised value for TCPv4
11116 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11117 positive, it will be used as the advertised MSS. If it is negative, it will
11118 indicate by how much to reduce the incoming connection's advertised MSS for
11119 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11120
11121name <name>
11122 Sets an optional name for these sockets, which will be reported on the stats
11123 page.
11124
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011125namespace <name>
11126 On Linux, it is possible to specify which network namespace a socket will
11127 belong to. This directive makes it possible to explicitly bind a listener to
11128 a namespace different from the default one. Please refer to your operating
11129 system's documentation to find more details about network namespaces.
11130
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011131nice <nice>
11132 Sets the 'niceness' of connections initiated from the socket. Value must be
11133 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11134 means that such connections are more friendly to others and easily offer
11135 their place in the scheduler. On the opposite, negative values mean that
11136 connections want to run with a higher priority than others. The difference
11137 only happens under high loads when the system is close to saturation.
11138 Negative values are appropriate for low-latency or administration services,
11139 and high values are generally recommended for CPU intensive tasks such as SSL
11140 processing or bulk transfers which are less sensible to latency. For example,
11141 it may make sense to use a positive value for an SMTP socket and a negative
11142 one for an RDP socket.
11143
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011144no-ca-names
11145 This setting is only available when support for OpenSSL was built in. It
11146 prevents from send CA names in server hello message when ca-file is used.
11147
Emeric Brun9b3009b2012-10-05 11:55:06 +020011148no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011149 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011150 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011151 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011152 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011153 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11154 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011155
Emeric Brun90ad8722012-10-02 14:00:59 +020011156no-tls-tickets
11157 This setting is only available when support for OpenSSL was built in. It
11158 disables the stateless session resumption (RFC 5077 TLS Ticket
11159 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011160 session resumption is more expensive in CPU usage. This option is also
11161 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011162
Emeric Brun9b3009b2012-10-05 11:55:06 +020011163no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011164 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011165 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011166 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011167 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011168 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11169 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011170
Emeric Brun9b3009b2012-10-05 11:55:06 +020011171no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011172 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011173 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011174 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011175 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011176 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11177 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011178
Emeric Brun9b3009b2012-10-05 11:55:06 +020011179no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011180 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011181 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011182 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011183 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011184 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11185 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011186
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011187no-tlsv13
11188 This setting is only available when support for OpenSSL was built in. It
11189 disables support for TLSv1.3 on any sockets instantiated from the listener
11190 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11191 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011192 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11193 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011194
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011195npn <protocols>
11196 This enables the NPN TLS extension and advertises the specified protocol list
11197 as supported on top of NPN. The protocol list consists in a comma-delimited
11198 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11199 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011200 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011201 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11202 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11203 version of OpenSSL, NPN might still be used as most clients still support it
11204 at the time of writing this. It is possible to enable both NPN and ALPN
11205 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011206
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011207prefer-client-ciphers
11208 Use the client's preference when selecting the cipher suite, by default
11209 the server's preference is enforced. This option is also available on
11210 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011211 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11212 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11213 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011214
Christopher Fauletc644fa92017-11-23 22:44:11 +010011215process <process-set>[/<thread-set>]
11216 This restricts the list of processes and/or threads on which this listener is
11217 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011218 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011219 between the two is applied. If in the end the listener is not allowed to run
11220 on any remaining process, a warning is emitted, and the listener will either
11221 run on the first process of the listener if a single process was specified,
11222 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011223 set is specified, it limits the threads allowed to process incoming
Christopher Fauletc644fa92017-11-23 22:44:11 +010011224 connections for this listener, for the corresponding process set. For the
11225 unlikely case where several ranges are needed, this directive may be
11226 repeated. <process-set> and <thread-set> must use the format
11227
11228 all | odd | even | number[-[number]]
11229
11230 Ranges can be partially defined. The higher bound can be omitted. In such
11231 case, it is replaced by the corresponding maximum value. The main purpose of
11232 this directive is to be used with the stats sockets and have one different
11233 socket per process. The second purpose is to have multiple bind lines sharing
11234 the same IP:port but not the same process in a listener, so that the system
11235 can distribute the incoming connections into multiple queues and allow a
11236 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11237 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011238
Christopher Fauleta717b992018-04-10 14:43:00 +020011239proto <name>
11240 Forces the multiplexer's protocol to use for the incoming connections. It
11241 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11242 be usable on the frontend side. The list of available protocols is reported
11243 in haproxy -vv.
11244 Idea behind this optipon is to bypass the selection of the best multiplexer's
11245 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011246 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011247 h2" on the bind line.
11248
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011249ssl
11250 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011251 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011252 certificate is necessary (see "crt" above). All contents in the buffers will
11253 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011254 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11255 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011256
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011257ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11258 This option enforces use of <version> or lower on SSL connections instantiated
11259 from this listener. This option is also available on global statement
11260 "ssl-default-bind-options". See also "ssl-min-ver".
11261
11262ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11263 This option enforces use of <version> or upper on SSL connections instantiated
11264 from this listener. This option is also available on global statement
11265 "ssl-default-bind-options". See also "ssl-max-ver".
11266
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011267strict-sni
11268 This setting is only available when support for OpenSSL was built in. The
11269 SSL/TLS negotiation is allow only if the client provided an SNI which match
11270 a certificate. The default certificate is not used.
11271 See the "crt" option for more information.
11272
Willy Tarreau2af207a2015-02-04 00:45:58 +010011273tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011274 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011275 listening socket. This option is available on Linux since version 2.6.37. It
11276 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011277 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011278 useful on long-lived connections experiencing long idle periods such as
11279 remote terminals or database connection pools, where the client and server
11280 timeouts must remain high to allow a long period of idle, but where it is
11281 important to detect that the client has disappeared in order to release all
11282 resources associated with its connection (and the server's session). The
11283 argument is a delay expressed in milliseconds by default. This only works
11284 for regular TCP connections, and is ignored for other protocols.
11285
Willy Tarreau1c862c52012-10-05 16:21:00 +020011286tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011287 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011288 enables TCP Fast Open on the listening socket, which means that clients which
11289 support this feature will be able to send a request and receive a response
11290 during the 3-way handshake starting from second connection, thus saving one
11291 round-trip after the first connection. This only makes sense with protocols
11292 that use high connection rates and where each round trip matters. This can
11293 possibly cause issues with many firewalls which do not accept data on SYN
11294 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011295 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11296 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11297 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011298
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011299tls-ticket-keys <keyfile>
11300 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
11301 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
11302 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
11303 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
11304 used for decryption and the penultimate one for encryption. This enables easy
11305 key rotation by just appending new key to the file and reloading the process.
11306 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
11307 is compromised. It is also a good idea to keep the keys off any permanent
11308 storage such as hard drives (hint: use tmpfs and don't swap those files).
11309 Lifetime hint can be changed using tune.ssl.timeout.
11310
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011311transparent
11312 Is an optional keyword which is supported only on certain Linux kernels. It
11313 indicates that the addresses will be bound even if they do not belong to the
11314 local machine, and that packets targeting any of these addresses will be
11315 intercepted just as if the addresses were locally configured. This normally
11316 requires that IP forwarding is enabled. Caution! do not use this with the
11317 default address '*', as it would redirect any traffic for the specified port.
11318 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11319 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11320 kernel version. Some distribution kernels include backports of the feature,
11321 so check for support with your vendor.
11322
Willy Tarreau77e3af92012-11-24 15:07:23 +010011323v4v6
11324 Is an optional keyword which is supported only on most recent systems
11325 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11326 and IPv6 when it uses the default address. Doing so is sometimes necessary
11327 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011328 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011329
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011330v6only
11331 Is an optional keyword which is supported only on most recent systems
11332 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11333 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011334 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11335 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011336
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011337uid <uid>
11338 Sets the owner of the UNIX sockets to the designated system uid. It can also
11339 be set by default in the global section's "unix-bind" statement. Note that
11340 some platforms simply ignore this. This setting is equivalent to the "user"
11341 setting except that the user numeric ID is used instead of its name. This
11342 setting is ignored by non UNIX sockets.
11343
11344user <user>
11345 Sets the owner of the UNIX sockets to the designated system user. It can also
11346 be set by default in the global section's "unix-bind" statement. Note that
11347 some platforms simply ignore this. This setting is equivalent to the "uid"
11348 setting except that the user name is used instead of its uid. This setting is
11349 ignored by non UNIX sockets.
11350
Emeric Brun1a073b42012-09-28 17:07:34 +020011351verify [none|optional|required]
11352 This setting is only available when support for OpenSSL was built in. If set
11353 to 'none', client certificate is not requested. This is the default. In other
11354 cases, a client certificate is requested. If the client does not provide a
11355 certificate after the request and if 'verify' is set to 'required', then the
11356 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011357 certificate provided by the client is always verified using CAs from
11358 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11359 is aborted, regardless of the 'verify' option, unless the error code exactly
11360 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011361
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200113625.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011363------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011364
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011365The "server" and "default-server" keywords support a certain number of settings
11366which are all passed as arguments on the server line. The order in which those
11367arguments appear does not count, and they are all optional. Some of those
11368settings are single words (booleans) while others expect one or several values
11369after them. In this case, the values must immediately follow the setting name.
11370Except default-server, all those settings must be specified after the server's
11371address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011372
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011373 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011374 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011375
Frédéric Lécailled2376272017-03-21 18:52:12 +010011376Note that all these settings are supported both by "server" and "default-server"
11377keywords, except "id" which is only supported by "server".
11378
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011379The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011380
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011381addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011382 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011383 to send health-checks or to probe the agent-check. On some servers, it may be
11384 desirable to dedicate an IP address to specific component able to perform
11385 complex tests which are more suitable to health-checks than the application.
11386 This parameter is ignored if the "check" parameter is not set. See also the
11387 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011388
Simon Hormand60d6912013-11-25 10:46:36 +090011389agent-check
11390 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011391 health check. An agent health check is performed by making a TCP connection
11392 to the port set by the "agent-port" parameter and reading an ASCII string.
11393 The string is made of a series of words delimited by spaces, tabs or commas
11394 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011395
Willy Tarreau81f5d942013-12-09 20:51:51 +010011396 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011397 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011398 weight of a server as configured when haproxy starts. Note that a zero
11399 weight is reported on the stats page as "DRAIN" since it has the same
11400 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011401
Davor Ocelice9ed2812017-12-25 17:49:28 +010011402 - The string "maxconn:" followed by an integer (no space between). Values
11403 in this format will set the maxconn of a server. The maximum number of
11404 connections advertised needs to be multiplied by the number of load
11405 balancers and different backends that use this health check to get the
11406 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011407
Willy Tarreau81f5d942013-12-09 20:51:51 +010011408 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011409 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011410
Willy Tarreau81f5d942013-12-09 20:51:51 +010011411 - The word "drain". This will turn the server's administrative state to the
11412 DRAIN mode, thus it will not accept any new connections other than those
11413 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011414
Willy Tarreau81f5d942013-12-09 20:51:51 +010011415 - The word "maint". This will turn the server's administrative state to the
11416 MAINT mode, thus it will not accept any new connections at all, and health
11417 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011418
Willy Tarreau81f5d942013-12-09 20:51:51 +010011419 - The words "down", "failed", or "stopped", optionally followed by a
11420 description string after a sharp ('#'). All of these mark the server's
11421 operating state as DOWN, but since the word itself is reported on the stats
11422 page, the difference allows an administrator to know if the situation was
11423 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011424 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011425 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011426
Willy Tarreau81f5d942013-12-09 20:51:51 +010011427 - The word "up" sets back the server's operating state as UP if health checks
11428 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011429
Willy Tarreau81f5d942013-12-09 20:51:51 +010011430 Parameters which are not advertised by the agent are not changed. For
11431 example, an agent might be designed to monitor CPU usage and only report a
11432 relative weight and never interact with the operating status. Similarly, an
11433 agent could be designed as an end-user interface with 3 radio buttons
11434 allowing an administrator to change only the administrative state. However,
11435 it is important to consider that only the agent may revert its own actions,
11436 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11437 agent must implement the other equivalent actions to bring the service into
11438 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011439
Simon Horman2f1f9552013-11-25 10:46:37 +090011440 Failure to connect to the agent is not considered an error as connectivity
11441 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011442 parameter. Warning though, it is not a good idea to stop an agent after it
11443 reports "down", since only an agent reporting "up" will be able to turn the
11444 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011445 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011446
Willy Tarreau81f5d942013-12-09 20:51:51 +010011447 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011448 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011449
James Brown55f9ff12015-10-21 18:19:05 -070011450agent-send <string>
11451 If this option is specified, haproxy will send the given string (verbatim)
11452 to the agent server upon connection. You could, for example, encode
11453 the backend name into this string, which would enable your agent to send
11454 different responses based on the backend. Make sure to include a '\n' if
11455 you want to terminate your request with a newline.
11456
Simon Hormand60d6912013-11-25 10:46:36 +090011457agent-inter <delay>
11458 The "agent-inter" parameter sets the interval between two agent checks
11459 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11460
11461 Just as with every other time-based parameter, it may be entered in any
11462 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11463 parameter also serves as a timeout for agent checks "timeout check" is
11464 not set. In order to reduce "resonance" effects when multiple servers are
11465 hosted on the same hardware, the agent and health checks of all servers
11466 are started with a small time offset between them. It is also possible to
11467 add some random noise in the agent and health checks interval using the
11468 global "spread-checks" keyword. This makes sense for instance when a lot
11469 of backends use the same servers.
11470
11471 See also the "agent-check" and "agent-port" parameters.
11472
Misiek768d8602017-01-09 09:52:43 +010011473agent-addr <addr>
11474 The "agent-addr" parameter sets address for agent check.
11475
11476 You can offload agent-check to another target, so you can make single place
11477 managing status and weights of servers defined in haproxy in case you can't
11478 make self-aware and self-managing services. You can specify both IP or
11479 hostname, it will be resolved.
11480
Simon Hormand60d6912013-11-25 10:46:36 +090011481agent-port <port>
11482 The "agent-port" parameter sets the TCP port used for agent checks.
11483
11484 See also the "agent-check" and "agent-inter" parameters.
11485
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011486backup
11487 When "backup" is present on a server line, the server is only used in load
11488 balancing when all other non-backup servers are unavailable. Requests coming
11489 with a persistence cookie referencing the server will always be served
11490 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011491 the "allbackups" option is set in the backend. See also the "no-backup" and
11492 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011493
Emeric Brunef42d922012-10-11 16:11:36 +020011494ca-file <cafile>
11495 This setting is only available when support for OpenSSL was built in. It
11496 designates a PEM file from which to load CA certificates used to verify
11497 server's certificate.
11498
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011499check
11500 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011501 always considered available. If "check" is set, the server is available when
11502 accepting periodic TCP connections, to ensure that it is really able to serve
11503 requests. The default address and port to send the tests to are those of the
11504 server, and the default source is the same as the one defined in the
11505 backend. It is possible to change the address using the "addr" parameter, the
11506 port using the "port" parameter, the source address using the "source"
11507 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011508 parameters. The request method is define in the backend using the "httpchk",
11509 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011510 refer to those options and parameters for more information. See also
11511 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011512
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011513check-send-proxy
11514 This option forces emission of a PROXY protocol line with outgoing health
11515 checks, regardless of whether the server uses send-proxy or not for the
11516 normal traffic. By default, the PROXY protocol is enabled for health checks
11517 if it is already enabled for normal traffic and if no "port" nor "addr"
11518 directive is present. However, if such a directive is present, the
11519 "check-send-proxy" option needs to be used to force the use of the
11520 protocol. See also the "send-proxy" option for more information.
11521
Olivier Houchard9130a962017-10-17 17:33:43 +020011522check-sni
11523 This option allows you to specify the SNI to be used when doing health checks
11524 over SSL.
11525
Willy Tarreau763a95b2012-10-04 23:15:39 +020011526check-ssl
11527 This option forces encryption of all health checks over SSL, regardless of
11528 whether the server uses SSL or not for the normal traffic. This is generally
11529 used when an explicit "port" or "addr" directive is specified and SSL health
11530 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011531 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011532 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11533 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011534 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011535 See the "ssl" option for more information and "no-check-ssl" to disable
11536 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011537
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011538ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011539 This setting is only available when support for OpenSSL was built in. This
11540 option sets the string describing the list of cipher algorithms that is
11541 negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011542 string is defined in "man 1 ciphers". When SSL is used to communicate with
11543 servers on the local network, it is common to see a weaker set of algorithms
11544 than what is used over the internet. Doing so reduces CPU usage on both the
11545 server and haproxy while still keeping it compatible with deployed software.
11546 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
11547 is needed and just connectivity, using DES can be appropriate.
11548
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011549ciphersuites <ciphersuites>
11550 This setting is only available when support for OpenSSL was built in and
11551 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11552 describing the list of cipher algorithms that is negotiated during the TLS
11553 1.3 handshake with the server. The format of the string is defined in
11554 "man 1 ciphers" under the "ciphersuites" section.
11555
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011556cookie <value>
11557 The "cookie" parameter sets the cookie value assigned to the server to
11558 <value>. This value will be checked in incoming requests, and the first
11559 operational server possessing the same value will be selected. In return, in
11560 cookie insertion or rewrite modes, this value will be assigned to the cookie
11561 sent to the client. There is nothing wrong in having several servers sharing
11562 the same cookie value, and it is in fact somewhat common between normal and
11563 backup servers. See also the "cookie" keyword in backend section.
11564
Emeric Brunef42d922012-10-11 16:11:36 +020011565crl-file <crlfile>
11566 This setting is only available when support for OpenSSL was built in. It
11567 designates a PEM file from which to load certificate revocation list used
11568 to verify server's certificate.
11569
Emeric Bruna7aa3092012-10-26 12:58:00 +020011570crt <cert>
11571 This setting is only available when support for OpenSSL was built in.
11572 It designates a PEM file from which to load both a certificate and the
11573 associated private key. This file can be built by concatenating both PEM
11574 files into one. This certificate will be sent if the server send a client
11575 certificate request.
11576
Willy Tarreau96839092010-03-29 10:02:24 +020011577disabled
11578 The "disabled" keyword starts the server in the "disabled" state. That means
11579 that it is marked down in maintenance mode, and no connection other than the
11580 ones allowed by persist mode will reach it. It is very well suited to setup
11581 new servers, because normal traffic will never reach them, while it is still
11582 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011583 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011584
Frédéric Lécailled2376272017-03-21 18:52:12 +010011585enabled
11586 This option may be used as 'server' setting to reset any 'disabled'
11587 setting which would have been inherited from 'default-server' directive as
11588 default value.
11589 It may also be used as 'default-server' setting to reset any previous
11590 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011591
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011592error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011593 If health observing is enabled, the "error-limit" parameter specifies the
11594 number of consecutive errors that triggers event selected by the "on-error"
11595 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011596
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011597 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011598
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011599fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011600 The "fall" parameter states that a server will be considered as dead after
11601 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11602 unspecified. See also the "check", "inter" and "rise" parameters.
11603
Emeric Brun8694b9a2012-10-05 14:39:07 +020011604force-sslv3
11605 This option enforces use of SSLv3 only when SSL is used to communicate with
11606 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011607 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011608 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011609
11610force-tlsv10
11611 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011612 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011613 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011614
11615force-tlsv11
11616 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011617 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011618 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011619
11620force-tlsv12
11621 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011622 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011623 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011624
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011625force-tlsv13
11626 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11627 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011628 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011629
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011630id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011631 Set a persistent ID for the server. This ID must be positive and unique for
11632 the proxy. An unused ID will automatically be assigned if unset. The first
11633 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011634
Willy Tarreau6a031d12016-11-07 19:42:35 +010011635init-addr {last | libc | none | <ip>},[...]*
11636 Indicate in what order the server's address should be resolved upon startup
11637 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011638 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011639 method which succeeds is used. If the end of the list is reached without
11640 finding a working method, an error is thrown. Method "last" suggests to pick
11641 the address which appears in the state file (see "server-state-file"). Method
11642 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11643 depending on the operating system and build options). Method "none"
11644 specifically indicates that the server should start without any valid IP
11645 address in a down state. It can be useful to ignore some DNS issues upon
11646 startup, waiting for the situation to get fixed later. Finally, an IP address
11647 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011648 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011649 server used to catch old sessions and present them with a decent error
11650 message for example. When the "first" load balancing algorithm is used, this
11651 IP address could point to a fake server used to trigger the creation of new
11652 instances on the fly. This option defaults to "last,libc" indicating that the
11653 previous address found in the state file (if any) is used first, otherwise
11654 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011655 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011656
11657 Example:
11658 defaults
11659 # never fail on address resolution
11660 default-server init-addr last,libc,none
11661
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011662inter <delay>
11663fastinter <delay>
11664downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011665 The "inter" parameter sets the interval between two consecutive health checks
11666 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11667 It is also possible to use "fastinter" and "downinter" to optimize delays
11668 between checks depending on the server state :
11669
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011670 Server state | Interval used
11671 ----------------------------------------+----------------------------------
11672 UP 100% (non-transitional) | "inter"
11673 ----------------------------------------+----------------------------------
11674 Transitionally UP (going down "fall"), | "fastinter" if set,
11675 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11676 or yet unchecked. |
11677 ----------------------------------------+----------------------------------
11678 DOWN 100% (non-transitional) | "downinter" if set,
11679 | "inter" otherwise.
11680 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011681
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011682 Just as with every other time-based parameter, they can be entered in any
11683 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11684 serves as a timeout for health checks sent to servers if "timeout check" is
11685 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011686 hosted on the same hardware, the agent and health checks of all servers
11687 are started with a small time offset between them. It is also possible to
11688 add some random noise in the agent and health checks interval using the
11689 global "spread-checks" keyword. This makes sense for instance when a lot
11690 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011691
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011692maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011693 The "maxconn" parameter specifies the maximal number of concurrent
11694 connections that will be sent to this server. If the number of incoming
11695 concurrent requests goes higher than this value, they will be queued, waiting
11696 for a connection to be released. This parameter is very important as it can
11697 save fragile servers from going down under extreme loads. If a "minconn"
11698 parameter is specified, the limit becomes dynamic. The default value is "0"
11699 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11700 the backend's "fullconn" keyword.
11701
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011702maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011703 The "maxqueue" parameter specifies the maximal number of connections which
11704 will wait in the queue for this server. If this limit is reached, next
11705 requests will be redispatched to other servers instead of indefinitely
11706 waiting to be served. This will break persistence but may allow people to
11707 quickly re-log in when the server they try to connect to is dying. The
11708 default value is "0" which means the queue is unlimited. See also the
11709 "maxconn" and "minconn" parameters.
11710
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011711minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011712 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11713 limit following the backend's load. The server will always accept at least
11714 <minconn> connections, never more than <maxconn>, and the limit will be on
11715 the ramp between both values when the backend has less than <fullconn>
11716 concurrent connections. This makes it possible to limit the load on the
11717 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011718 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011719 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011720
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011721namespace <name>
11722 On Linux, it is possible to specify which network namespace a socket will
11723 belong to. This directive makes it possible to explicitly bind a server to
11724 a namespace different from the default one. Please refer to your operating
11725 system's documentation to find more details about network namespaces.
11726
Frédéric Lécailled2376272017-03-21 18:52:12 +010011727no-agent-check
11728 This option may be used as "server" setting to reset any "agent-check"
11729 setting which would have been inherited from "default-server" directive as
11730 default value.
11731 It may also be used as "default-server" setting to reset any previous
11732 "default-server" "agent-check" setting.
11733
11734no-backup
11735 This option may be used as "server" setting to reset any "backup"
11736 setting which would have been inherited from "default-server" directive as
11737 default value.
11738 It may also be used as "default-server" setting to reset any previous
11739 "default-server" "backup" setting.
11740
11741no-check
11742 This option may be used as "server" setting to reset any "check"
11743 setting which would have been inherited from "default-server" directive as
11744 default value.
11745 It may also be used as "default-server" setting to reset any previous
11746 "default-server" "check" setting.
11747
11748no-check-ssl
11749 This option may be used as "server" setting to reset any "check-ssl"
11750 setting which would have been inherited from "default-server" directive as
11751 default value.
11752 It may also be used as "default-server" setting to reset any previous
11753 "default-server" "check-ssl" setting.
11754
Frédéric Lécailled2376272017-03-21 18:52:12 +010011755no-send-proxy
11756 This option may be used as "server" setting to reset any "send-proxy"
11757 setting which would have been inherited from "default-server" directive as
11758 default value.
11759 It may also be used as "default-server" setting to reset any previous
11760 "default-server" "send-proxy" setting.
11761
11762no-send-proxy-v2
11763 This option may be used as "server" setting to reset any "send-proxy-v2"
11764 setting which would have been inherited from "default-server" directive as
11765 default value.
11766 It may also be used as "default-server" setting to reset any previous
11767 "default-server" "send-proxy-v2" setting.
11768
11769no-send-proxy-v2-ssl
11770 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11771 setting which would have been inherited from "default-server" directive as
11772 default value.
11773 It may also be used as "default-server" setting to reset any previous
11774 "default-server" "send-proxy-v2-ssl" setting.
11775
11776no-send-proxy-v2-ssl-cn
11777 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11778 setting which would have been inherited from "default-server" directive as
11779 default value.
11780 It may also be used as "default-server" setting to reset any previous
11781 "default-server" "send-proxy-v2-ssl-cn" setting.
11782
11783no-ssl
11784 This option may be used as "server" setting to reset any "ssl"
11785 setting which would have been inherited from "default-server" directive as
11786 default value.
11787 It may also be used as "default-server" setting to reset any previous
11788 "default-server" "ssl" setting.
11789
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011790no-ssl-reuse
11791 This option disables SSL session reuse when SSL is used to communicate with
11792 the server. It will force the server to perform a full handshake for every
11793 new connection. It's probably only useful for benchmarking, troubleshooting,
11794 and for paranoid users.
11795
Emeric Brun9b3009b2012-10-05 11:55:06 +020011796no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011797 This option disables support for SSLv3 when SSL is used to communicate with
11798 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011799 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011800
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011801 Supported in default-server: No
11802
Emeric Brunf9c5c472012-10-11 15:28:34 +020011803no-tls-tickets
11804 This setting is only available when support for OpenSSL was built in. It
11805 disables the stateless session resumption (RFC 5077 TLS Ticket
11806 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011807 session resumption is more expensive in CPU usage for servers. This option
11808 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011809 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011810
Emeric Brun9b3009b2012-10-05 11:55:06 +020011811no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011812 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011813 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11814 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011815 often makes sense to disable it when communicating with local servers. This
11816 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011817 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011818
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011819 Supported in default-server: No
11820
Emeric Brun9b3009b2012-10-05 11:55:06 +020011821no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011822 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011823 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11824 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011825 often makes sense to disable it when communicating with local servers. This
11826 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011827 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011828
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011829 Supported in default-server: No
11830
Emeric Brun9b3009b2012-10-05 11:55:06 +020011831no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011832 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011833 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11834 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011835 often makes sense to disable it when communicating with local servers. This
11836 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011837 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011838
11839 Supported in default-server: No
11840
11841no-tlsv13
11842 This option disables support for TLSv1.3 when SSL is used to communicate with
11843 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11844 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11845 often makes sense to disable it when communicating with local servers. This
11846 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011847 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011848
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011849 Supported in default-server: No
11850
Frédéric Lécailled2376272017-03-21 18:52:12 +010011851no-verifyhost
11852 This option may be used as "server" setting to reset any "verifyhost"
11853 setting which would have been inherited from "default-server" directive as
11854 default value.
11855 It may also be used as "default-server" setting to reset any previous
11856 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011857
Simon Hormanfa461682011-06-25 09:39:49 +090011858non-stick
11859 Never add connections allocated to this sever to a stick-table.
11860 This may be used in conjunction with backup to ensure that
11861 stick-table persistence is disabled for backup servers.
11862
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011863observe <mode>
11864 This option enables health adjusting based on observing communication with
11865 the server. By default this functionality is disabled and enabling it also
11866 requires to enable health checks. There are two supported modes: "layer4" and
11867 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11868 significant. In layer7, which is only allowed for http proxies, responses
11869 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011870 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011871
11872 See also the "check", "on-error" and "error-limit".
11873
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011874on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011875 Select what should happen when enough consecutive errors are detected.
11876 Currently, four modes are available:
11877 - fastinter: force fastinter
11878 - fail-check: simulate a failed check, also forces fastinter (default)
11879 - sudden-death: simulate a pre-fatal failed health check, one more failed
11880 check will mark a server down, forces fastinter
11881 - mark-down: mark the server immediately down and force fastinter
11882
11883 See also the "check", "observe" and "error-limit".
11884
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011885on-marked-down <action>
11886 Modify what occurs when a server is marked down.
11887 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011888 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11889 all connections to the server are immediately terminated when the server
11890 goes down. It might be used if the health check detects more complex cases
11891 than a simple connection status, and long timeouts would cause the service
11892 to remain unresponsive for too long a time. For instance, a health check
11893 might detect that a database is stuck and that there's no chance to reuse
11894 existing connections anymore. Connections killed this way are logged with
11895 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011896
11897 Actions are disabled by default
11898
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011899on-marked-up <action>
11900 Modify what occurs when a server is marked up.
11901 Currently one action is available:
11902 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11903 done only if the server is not in backup state and if it is not disabled
11904 (it must have an effective weight > 0). This can be used sometimes to force
11905 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011906 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11907 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011908 with extreme care. Sessions killed because a server comes up are logged
11909 with an 'U' termination code (for "Up").
11910
11911 Actions are disabled by default
11912
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011913port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011914 Using the "port" parameter, it becomes possible to use a different port to
11915 send health-checks. On some servers, it may be desirable to dedicate a port
11916 to a specific component able to perform complex tests which are more suitable
11917 to health-checks than the application. It is common to run a simple script in
11918 inetd for instance. This parameter is ignored if the "check" parameter is not
11919 set. See also the "addr" parameter.
11920
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011921proto <name>
11922
11923 Forces the multiplexer's protocol to use for the outgoing connections to this
11924 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11925 must also be usable on the backend side. The list of available protocols is
11926 reported in haproxy -vv.
11927 Idea behind this optipon is to bypass the selection of the best multiplexer's
11928 protocol for all connections established to this server.
11929
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011930redir <prefix>
11931 The "redir" parameter enables the redirection mode for all GET and HEAD
11932 requests addressing this server. This means that instead of having HAProxy
11933 forward the request to the server, it will send an "HTTP 302" response with
11934 the "Location" header composed of this prefix immediately followed by the
11935 requested URI beginning at the leading '/' of the path component. That means
11936 that no trailing slash should be used after <prefix>. All invalid requests
11937 will be rejected, and all non-GET or HEAD requests will be normally served by
11938 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011939 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011940 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011941 users to a remote location in case of local disaster. Main use consists in
11942 increasing bandwidth for static servers by having the clients directly
11943 connect to them. Note: never use a relative location here, it would cause a
11944 loop between the client and HAProxy!
11945
11946 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11947
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011948rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011949 The "rise" parameter states that a server will be considered as operational
11950 after <count> consecutive successful health checks. This value defaults to 2
11951 if unspecified. See also the "check", "inter" and "fall" parameters.
11952
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020011953resolve-opts <option>,<option>,...
11954 Comma separated list of options to apply to DNS resolution linked to this
11955 server.
11956
11957 Available options:
11958
11959 * allow-dup-ip
11960 By default, HAProxy prevents IP address duplication in a backend when DNS
11961 resolution at runtime is in operation.
11962 That said, for some cases, it makes sense that two servers (in the same
11963 backend, being resolved by the same FQDN) have the same IP address.
11964 For such case, simply enable this option.
11965 This is the opposite of prevent-dup-ip.
11966
11967 * prevent-dup-ip
11968 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
11969 an IP address already set to a server in the same backend and sharing the
11970 same fqdn.
11971 This is the opposite of allow-dup-ip.
11972
11973 Example:
11974 backend b_myapp
11975 default-server init-addr none resolvers dns
11976 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
11977 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
11978
11979 With the option allow-dup-ip set:
11980 * if the nameserver returns a single IP address, then both servers will use
11981 it
11982 * If the nameserver returns 2 IP addresses, then each server will pick up a
11983 different address
11984
11985 Default value: not set
11986
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011987resolve-prefer <family>
11988 When DNS resolution is enabled for a server and multiple IP addresses from
11989 different families are returned, HAProxy will prefer using an IP address
11990 from the family mentioned in the "resolve-prefer" parameter.
11991 Available families: "ipv4" and "ipv6"
11992
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011993 Default value: ipv6
11994
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011995 Example:
11996
11997 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011998
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011999resolve-net <network>[,<network[,...]]
12000 This options prioritize th choice of an ip address matching a network. This is
12001 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012002 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012003 different datacenters. The latency between datacenter is not negligible, so
12004 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012005 configured network, another address is selected.
12006
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012007 Example:
12008
12009 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012010
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012011resolvers <id>
12012 Points to an existing "resolvers" section to resolve current server's
12013 hostname.
12014
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012015 Example:
12016
12017 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012018
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012019 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012020
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012021send-proxy
12022 The "send-proxy" parameter enforces use of the PROXY protocol over any
12023 connection established to this server. The PROXY protocol informs the other
12024 end about the layer 3/4 addresses of the incoming connection, so that it can
12025 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012026 upper layer protocol. For connections accepted by an "accept-proxy" or
12027 "accept-netscaler-cip" listener, the advertised address will be used. Only
12028 TCPv4 and TCPv6 address families are supported. Other families such as
12029 Unix sockets, will report an UNKNOWN family. Servers using this option can
12030 fully be chained to another instance of haproxy listening with an
12031 "accept-proxy" setting. This setting must not be used if the server isn't
12032 aware of the protocol. When health checks are sent to the server, the PROXY
12033 protocol is automatically used when this option is set, unless there is an
12034 explicit "port" or "addr" directive, in which case an explicit
12035 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012036 See also the "no-send-proxy" option of this section and "accept-proxy" and
12037 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012038
David Safb76832014-05-08 23:42:08 -040012039send-proxy-v2
12040 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12041 over any connection established to this server. The PROXY protocol informs
12042 the other end about the layer 3/4 addresses of the incoming connection, so
12043 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012044 whatever the upper layer protocol. It also send ALPN information if an alpn
12045 have been negotiated. This setting must not be used if the server isn't aware
12046 of this version of the protocol. See also the "no-send-proxy-v2" option of
12047 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012048
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012049proxy-v2-options <option>[,<option>]*
12050 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12051 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012052 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12053 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012054 certificate, "cert-key": key algorithm of the used certificate), "authority":
12055 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012056 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012057
David Safb76832014-05-08 23:42:08 -040012058send-proxy-v2-ssl
12059 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12060 2 over any connection established to this server. The PROXY protocol informs
12061 the other end about the layer 3/4 addresses of the incoming connection, so
12062 that it can know the client's address or the public address it accessed to,
12063 whatever the upper layer protocol. In addition, the SSL information extension
12064 of the PROXY protocol is added to the PROXY protocol header. This setting
12065 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012066 See also the "no-send-proxy-v2-ssl" option of this section and the
12067 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012068
12069send-proxy-v2-ssl-cn
12070 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12071 2 over any connection established to this server. The PROXY protocol informs
12072 the other end about the layer 3/4 addresses of the incoming connection, so
12073 that it can know the client's address or the public address it accessed to,
12074 whatever the upper layer protocol. In addition, the SSL information extension
12075 of the PROXY protocol, along along with the Common Name from the subject of
12076 the client certificate (if any), is added to the PROXY protocol header. This
12077 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012078 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12079 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012080
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012081slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012082 The "slowstart" parameter for a server accepts a value in milliseconds which
12083 indicates after how long a server which has just come back up will run at
12084 full speed. Just as with every other time-based parameter, it can be entered
12085 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12086 linearly from 0 to 100% during this time. The limitation applies to two
12087 parameters :
12088
12089 - maxconn: the number of connections accepted by the server will grow from 1
12090 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12091
12092 - weight: when the backend uses a dynamic weighted algorithm, the weight
12093 grows linearly from 1 to 100%. In this case, the weight is updated at every
12094 health-check. For this reason, it is important that the "inter" parameter
12095 is smaller than the "slowstart", in order to maximize the number of steps.
12096
12097 The slowstart never applies when haproxy starts, otherwise it would cause
12098 trouble to running servers. It only applies when a server has been previously
12099 seen as failed.
12100
Willy Tarreau732eac42015-07-09 11:40:25 +020012101sni <expression>
12102 The "sni" parameter evaluates the sample fetch expression, converts it to a
12103 string and uses the result as the host name sent in the SNI TLS extension to
12104 the server. A typical use case is to send the SNI received from the client in
12105 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012106 expression, though alternatives such as req.hdr(host) can also make sense. If
12107 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012108 name will also be matched against the server certificate's names. See the
12109 "verify" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012110
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012111source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012112source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012113source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012114 The "source" parameter sets the source address which will be used when
12115 connecting to the server. It follows the exact same parameters and principle
12116 as the backend "source" keyword, except that it only applies to the server
12117 referencing it. Please consult the "source" keyword for details.
12118
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012119 Additionally, the "source" statement on a server line allows one to specify a
12120 source port range by indicating the lower and higher bounds delimited by a
12121 dash ('-'). Some operating systems might require a valid IP address when a
12122 source port range is specified. It is permitted to have the same IP/range for
12123 several servers. Doing so makes it possible to bypass the maximum of 64k
12124 total concurrent connections. The limit will then reach 64k connections per
12125 server.
12126
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012127 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12128 specifying the source address without port(s).
12129
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012130ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012131 This option enables SSL ciphering on outgoing connections to the server. It
12132 is critical to verify server certificates using "verify" when using SSL to
12133 connect to servers, otherwise the communication is prone to trivial man in
12134 the-middle attacks rendering SSL useless. When this option is used, health
12135 checks are automatically sent in SSL too unless there is a "port" or an
12136 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012137 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12138 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012139
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012140ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12141 This option enforces use of <version> or lower when SSL is used to communicate
12142 with the server. This option is also available on global statement
12143 "ssl-default-server-options". See also "ssl-min-ver".
12144
12145ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12146 This option enforces use of <version> or upper when SSL is used to communicate
12147 with the server. This option is also available on global statement
12148 "ssl-default-server-options". See also "ssl-max-ver".
12149
Frédéric Lécailled2376272017-03-21 18:52:12 +010012150ssl-reuse
12151 This option may be used as "server" setting to reset any "no-ssl-reuse"
12152 setting which would have been inherited from "default-server" directive as
12153 default value.
12154 It may also be used as "default-server" setting to reset any previous
12155 "default-server" "no-ssl-reuse" setting.
12156
12157stick
12158 This option may be used as "server" setting to reset any "non-stick"
12159 setting which would have been inherited from "default-server" directive as
12160 default value.
12161 It may also be used as "default-server" setting to reset any previous
12162 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012163
Willy Tarreau163d4622015-10-13 16:16:41 +020012164tcp-ut <delay>
12165 Sets the TCP User Timeout for all outgoing connections to this server. This
12166 option is available on Linux since version 2.6.37. It allows haproxy to
12167 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012168 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012169 long-lived connections experiencing long idle periods such as remote
12170 terminals or database connection pools, where the client and server timeouts
12171 must remain high to allow a long period of idle, but where it is important to
12172 detect that the server has disappeared in order to release all resources
12173 associated with its connection (and the client's session). One typical use
12174 case is also to force dead server connections to die when health checks are
12175 too slow or during a soft reload since health checks are then disabled. The
12176 argument is a delay expressed in milliseconds by default. This only works for
12177 regular TCP connections, and is ignored for other protocols.
12178
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012179track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012180 This option enables ability to set the current state of the server by tracking
12181 another one. It is possible to track a server which itself tracks another
12182 server, provided that at the end of the chain, a server has health checks
12183 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012184 used, it has to be enabled on both proxies.
12185
Frédéric Lécailled2376272017-03-21 18:52:12 +010012186tls-tickets
12187 This option may be used as "server" setting to reset any "no-tls-tickets"
12188 setting which would have been inherited from "default-server" directive as
12189 default value.
12190 It may also be used as "default-server" setting to reset any previous
12191 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012192
Emeric Brunef42d922012-10-11 16:11:36 +020012193verify [none|required]
12194 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012195 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012196 certificate provided by the server is verified using CAs from 'ca-file' and
12197 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012198 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012199 the name passed using the "sni" directive, or if not provided, the static
12200 host name passed using the "verifyhost" directive. When no name is found, the
12201 certificate's names are ignored. For this reason, without SNI it's important
12202 to use "verifyhost". On verification failure the handshake is aborted. It is
12203 critically important to verify server certificates when using SSL to connect
12204 to servers, otherwise the communication is prone to trivial man-in-the-middle
12205 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12206 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012207
Evan Broderbe554312013-06-27 00:05:25 -070012208verifyhost <hostname>
12209 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012210 only takes effect if 'verify required' is also specified. This directive sets
12211 a default static hostname to check the server's certificate against when no
12212 SNI was used to connect to the server. If SNI is not used, this is the only
12213 way to enable hostname verification. This static hostname, when set, will
12214 also be used for health checks (which cannot provide an SNI value). If none
12215 of the hostnames in the certificate match the specified hostname, the
12216 handshake is aborted. The hostnames in the server-provided certificate may
12217 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012218
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012219weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012220 The "weight" parameter is used to adjust the server's weight relative to
12221 other servers. All servers will receive a load proportional to their weight
12222 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012223 load. The default weight is 1, and the maximal value is 256. A value of 0
12224 means the server will not participate in load-balancing but will still accept
12225 persistent connections. If this parameter is used to distribute the load
12226 according to server's capacity, it is recommended to start with values which
12227 can both grow and shrink, for instance between 10 and 100 to leave enough
12228 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012229
12230
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122315.3. Server IP address resolution using DNS
12232-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012233
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012234HAProxy allows using a host name on the server line to retrieve its IP address
12235using name servers. By default, HAProxy resolves the name when parsing the
12236configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012237This is not sufficient in some cases, such as in Amazon where a server's IP
12238can change after a reboot or an ELB Virtual IP can change based on current
12239workload.
12240This chapter describes how HAProxy can be configured to process server's name
12241resolution at run time.
12242Whether run time server name resolution has been enable or not, HAProxy will
12243carry on doing the first resolution when parsing the configuration.
12244
12245
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122465.3.1. Global overview
12247----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012248
12249As we've seen in introduction, name resolution in HAProxy occurs at two
12250different steps of the process life:
12251
12252 1. when starting up, HAProxy parses the server line definition and matches a
12253 host name. It uses libc functions to get the host name resolved. This
12254 resolution relies on /etc/resolv.conf file.
12255
Christopher Faulet67957bd2017-09-27 11:00:59 +020012256 2. at run time, HAProxy performs periodically name resolutions for servers
12257 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012258
12259A few other events can trigger a name resolution at run time:
12260 - when a server's health check ends up in a connection timeout: this may be
12261 because the server has a new IP address. So we need to trigger a name
12262 resolution to know this new IP.
12263
Christopher Faulet67957bd2017-09-27 11:00:59 +020012264When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012265HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012266SRV label is specified, then the corresponding SRV records will be retrieved
12267from the DNS server, and the provided hostnames will be used. The SRV label
12268will be checked periodically, and if any server are added or removed, haproxy
12269will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012270
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012271A few things important to notice:
12272 - all the name servers are queried in the mean time. HAProxy will process the
12273 first valid response.
12274
12275 - a resolution is considered as invalid (NX, timeout, refused), when all the
12276 servers return an error.
12277
12278
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122795.3.2. The resolvers section
12280----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012281
12282This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012283HAProxy. There can be as many as resolvers section as needed. Each section can
12284contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012285
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012286When multiple name servers are configured in a resolvers section, then HAProxy
12287uses the first valid response. In case of invalid responses, only the last one
12288is treated. Purpose is to give the chance to a slow server to deliver a valid
12289answer after a fast faulty or outdated server.
12290
12291When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012292used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012293
Christopher Faulet67957bd2017-09-27 11:00:59 +020012294 1. HAProxy retries the same DNS query with a new query type. The A queries are
12295 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12296 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012297
Christopher Faulet67957bd2017-09-27 11:00:59 +020012298 2. When the fallback on the query type was done (or not applicable), HAProxy
12299 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012300
Christopher Faulet67957bd2017-09-27 11:00:59 +020012301 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12302 response is received after that, it stops the DNS resolution and reports
12303 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012304
Christopher Faulet67957bd2017-09-27 11:00:59 +020012305For example, with 2 name servers configured in a resolvers section, the
12306following scenarios are possible:
12307
12308 - First response is valid and is applied directly, second response is
12309 ignored
12310
12311 - First response is invalid and second one is valid, then second response is
12312 applied
12313
12314 - First response is a NX domain and second one a truncated response, then
12315 HAProxy retries the query with a new type
12316
12317 - First response is a NX domain and second one is a timeout, then HAProxy
12318 retries the query with a new type
12319
12320 - Query timed out for both name servers, then HAProxy retries it with the
12321 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012322
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012323As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12324a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012325<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012326
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012327
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012328resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012329 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012330
12331A resolvers section accept the following parameters:
12332
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012333accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012334 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012335 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012336 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12337 by RFC 6891)
12338
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012339 Note: the maximum allowed value is 8192.
12340
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012341nameserver <id> <ip>:<port>
12342 DNS server description:
12343 <id> : label of the server, should be unique
12344 <ip> : IP address of the server
12345 <port> : port where the DNS service actually runs
12346
Ben Draut44e609b2018-05-29 15:40:08 -060012347parse-resolv-conf
12348 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12349 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12350 placed in the resolvers section in place of this directive.
12351
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012352hold <status> <period>
12353 Defines <period> during which the last name resolution should be kept based
12354 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012355 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012356 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012357 <period> : interval between two successive name resolution when the last
12358 answer was in <status>. It follows the HAProxy time format.
12359 <period> is in milliseconds by default.
12360
Baptiste Assmann686408b2017-08-18 10:15:42 +020012361 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012362
Christopher Faulet67957bd2017-09-27 11:00:59 +020012363resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012364 Defines the number of resolutions available in the pool for this resolvers.
12365 If not defines, it defaults to 64. If your configuration requires more than
12366 <nb>, then HAProxy will return an error when parsing the configuration.
12367
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012368resolve_retries <nb>
12369 Defines the number <nb> of queries to send to resolve a server name before
12370 giving up.
12371 Default value: 3
12372
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012373 A retry occurs on name server timeout or when the full sequence of DNS query
12374 type failover is over and we need to start up from the default ANY query
12375 type.
12376
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012377timeout <event> <time>
12378 Defines timeouts related to name resolution
12379 <event> : the event on which the <time> timeout period applies to.
12380 events available are:
Christopher Faulet67957bd2017-09-27 11:00:59 +020012381 - resolve : default time to trigger name resolutions when no
12382 other time applied.
12383 Default value: 1s
12384 - retry : time between two DNS queries, when no valid response
12385 have been received.
12386 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012387 <time> : time related to the event. It follows the HAProxy time format.
12388 <time> is expressed in milliseconds.
12389
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012390 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012391
12392 resolvers mydns
12393 nameserver dns1 10.0.0.1:53
12394 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012395 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012396 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012397 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012398 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012399 hold other 30s
12400 hold refused 30s
12401 hold nx 30s
12402 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012403 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012404 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012405
12406
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124076. HTTP header manipulation
12408---------------------------
12409
12410In HTTP mode, it is possible to rewrite, add or delete some of the request and
12411response headers based on regular expressions. It is also possible to block a
12412request or a response if a particular header matches a regular expression,
12413which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012414against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012415
Willy Tarreau70dffda2014-01-30 03:07:23 +010012416If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12417to process all rsp* rules which can allow, deny, rewrite or delete a header,
12418but it will refuse to add a header to any such messages as this is not
12419HTTP-compliant. The reason for still processing headers in such responses is to
12420stop and/or fix any possible information leak which may happen, for instance
12421because another downstream equipment would unconditionally add a header, or if
12422a server name appears there. When such messages are seen, normal processing
12423still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012424
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012425This section covers common usage of the following keywords, described in detail
12426in section 4.2 :
12427
12428 - reqadd <string>
12429 - reqallow <search>
12430 - reqiallow <search>
12431 - reqdel <search>
12432 - reqidel <search>
12433 - reqdeny <search>
12434 - reqideny <search>
12435 - reqpass <search>
12436 - reqipass <search>
12437 - reqrep <search> <replace>
12438 - reqirep <search> <replace>
12439 - reqtarpit <search>
12440 - reqitarpit <search>
12441 - rspadd <string>
12442 - rspdel <search>
12443 - rspidel <search>
12444 - rspdeny <search>
12445 - rspideny <search>
12446 - rsprep <search> <replace>
12447 - rspirep <search> <replace>
12448
12449With all these keywords, the same conventions are used. The <search> parameter
12450is a POSIX extended regular expression (regex) which supports grouping through
12451parenthesis (without the backslash). Spaces and other delimiters must be
12452prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12453Other characters may be prefixed with a backslash to change their meaning :
12454
12455 \t for a tab
12456 \r for a carriage return (CR)
12457 \n for a new line (LF)
12458 \ to mark a space and differentiate it from a delimiter
12459 \# to mark a sharp and differentiate it from a comment
12460 \\ to use a backslash in a regex
12461 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12462 \xXX to write the ASCII hex code XX as in the C language
12463
12464The <replace> parameter contains the string to be used to replace the largest
12465portion of text matching the regex. It can make use of the special characters
12466above, and can reference a substring which is delimited by parenthesis in the
12467regex, by writing a backslash ('\') immediately followed by one digit from 0 to
124689 indicating the group position (0 designating the entire line). This practice
12469is very common to users of the "sed" program.
12470
12471The <string> parameter represents the string which will systematically be added
12472after the last header line. It can also use special character sequences above.
12473
12474Notes related to these keywords :
12475---------------------------------
12476 - these keywords are not always convenient to allow/deny based on header
12477 contents. It is strongly recommended to use ACLs with the "block" keyword
12478 instead, resulting in far more flexible and manageable rules.
12479
12480 - lines are always considered as a whole. It is not possible to reference
12481 a header name only or a value only. This is important because of the way
12482 headers are written (notably the number of spaces after the colon).
12483
12484 - the first line is always considered as a header, which makes it possible to
12485 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12486 it harder to distinguish between headers and request line. The regex prefix
12487 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12488 ^[^ \t:]*: matches any header name followed by a colon.
12489
12490 - for performances reasons, the number of characters added to a request or to
12491 a response is limited at build time to values between 1 and 4 kB. This
12492 should normally be far more than enough for most usages. If it is too short
12493 on occasional usages, it is possible to gain some space by removing some
12494 useless headers before adding new ones.
12495
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012496 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012497 without the 'i' letter except that they ignore case when matching patterns.
12498
12499 - when a request passes through a frontend then a backend, all req* rules
12500 from the frontend will be evaluated, then all req* rules from the backend
12501 will be evaluated. The reverse path is applied to responses.
12502
12503 - req* statements are applied after "block" statements, so that "block" is
12504 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012505 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012506
12507
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125087. Using ACLs and fetching samples
12509----------------------------------
12510
Davor Ocelice9ed2812017-12-25 17:49:28 +010012511HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012512client or server information, from tables, environmental information etc...
12513The action of extracting such data is called fetching a sample. Once retrieved,
12514these samples may be used for various purposes such as a key to a stick-table,
12515but most common usages consist in matching them against predefined constant
12516data called patterns.
12517
12518
125197.1. ACL basics
12520---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012521
12522The use of Access Control Lists (ACL) provides a flexible solution to perform
12523content switching and generally to take decisions based on content extracted
12524from the request, the response or any environmental status. The principle is
12525simple :
12526
Willy Tarreau74ca5042013-06-11 23:12:07 +020012527 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012528 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012529 - apply one or multiple pattern matching methods on this sample
12530 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012531
Willy Tarreau74ca5042013-06-11 23:12:07 +020012532The actions generally consist in blocking a request, selecting a backend, or
12533adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012534
12535In order to define a test, the "acl" keyword is used. The syntax is :
12536
Willy Tarreau74ca5042013-06-11 23:12:07 +020012537 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012538
12539This creates a new ACL <aclname> or completes an existing one with new tests.
12540Those tests apply to the portion of request/response specified in <criterion>
12541and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012542an operator which may be specified before the set of values. Optionally some
12543conversion operators may be applied to the sample, and they will be specified
12544as a comma-delimited list of keywords just after the first keyword. The values
12545are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012546
12547ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12548'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12549which means that "my_acl" and "My_Acl" are two different ACLs.
12550
12551There is no enforced limit to the number of ACLs. The unused ones do not affect
12552performance, they just consume a small amount of memory.
12553
Willy Tarreau74ca5042013-06-11 23:12:07 +020012554The criterion generally is the name of a sample fetch method, or one of its ACL
12555specific declinations. The default test method is implied by the output type of
12556this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012557methods of a same sample fetch method. The sample fetch methods are the only
12558ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012559
12560Sample fetch methods return data which can be of the following types :
12561 - boolean
12562 - integer (signed or unsigned)
12563 - IPv4 or IPv6 address
12564 - string
12565 - data block
12566
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012567Converters transform any of these data into any of these. For example, some
12568converters might convert a string to a lower-case string while other ones
12569would turn a string to an IPv4 address, or apply a netmask to an IP address.
12570The resulting sample is of the type of the last converter applied to the list,
12571which defaults to the type of the sample fetch method.
12572
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012573Each sample or converter returns data of a specific type, specified with its
12574keyword in this documentation. When an ACL is declared using a standard sample
12575fetch method, certain types automatically involved a default matching method
12576which are summarized in the table below :
12577
12578 +---------------------+-----------------+
12579 | Sample or converter | Default |
12580 | output type | matching method |
12581 +---------------------+-----------------+
12582 | boolean | bool |
12583 +---------------------+-----------------+
12584 | integer | int |
12585 +---------------------+-----------------+
12586 | ip | ip |
12587 +---------------------+-----------------+
12588 | string | str |
12589 +---------------------+-----------------+
12590 | binary | none, use "-m" |
12591 +---------------------+-----------------+
12592
12593Note that in order to match a binary samples, it is mandatory to specify a
12594matching method, see below.
12595
Willy Tarreau74ca5042013-06-11 23:12:07 +020012596The ACL engine can match these types against patterns of the following types :
12597 - boolean
12598 - integer or integer range
12599 - IP address / network
12600 - string (exact, substring, suffix, prefix, subdir, domain)
12601 - regular expression
12602 - hex block
12603
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012604The following ACL flags are currently supported :
12605
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012606 -i : ignore case during matching of all subsequent patterns.
12607 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012608 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012609 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012610 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012611 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012612 -- : force end of flags. Useful when a string looks like one of the flags.
12613
Willy Tarreau74ca5042013-06-11 23:12:07 +020012614The "-f" flag is followed by the name of a file from which all lines will be
12615read as individual values. It is even possible to pass multiple "-f" arguments
12616if the patterns are to be loaded from multiple files. Empty lines as well as
12617lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12618will be stripped. If it is absolutely necessary to insert a valid pattern
12619beginning with a sharp, just prefix it with a space so that it is not taken for
12620a comment. Depending on the data type and match method, haproxy may load the
12621lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12622exact string matching. In this case, duplicates will automatically be removed.
12623
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012624The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12625parsed as two column file. The first column contains the patterns used by the
12626ACL, and the second column contain the samples. The sample can be used later by
12627a map. This can be useful in some rare cases where an ACL would just be used to
12628check for the existence of a pattern in a map before a mapping is applied.
12629
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012630The "-u" flag forces the unique id of the ACL. This unique id is used with the
12631socket interface to identify ACL and dynamically change its values. Note that a
12632file is always identified by its name even if an id is set.
12633
Willy Tarreau74ca5042013-06-11 23:12:07 +020012634Also, note that the "-i" flag applies to subsequent entries and not to entries
12635loaded from files preceding it. For instance :
12636
12637 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12638
12639In this example, each line of "exact-ua.lst" will be exactly matched against
12640the "user-agent" header of the request. Then each line of "generic-ua" will be
12641case-insensitively matched. Then the word "test" will be insensitively matched
12642as well.
12643
12644The "-m" flag is used to select a specific pattern matching method on the input
12645sample. All ACL-specific criteria imply a pattern matching method and generally
12646do not need this flag. However, this flag is useful with generic sample fetch
12647methods to describe how they're going to be matched against the patterns. This
12648is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012649obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012650followed by a pattern matching method name, this method is used instead of the
12651default one for the criterion. This makes it possible to match contents in ways
12652that were not initially planned, or with sample fetch methods which return a
12653string. The matching method also affects the way the patterns are parsed.
12654
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012655The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12656By default, if the parser cannot parse ip address it considers that the parsed
12657string is maybe a domain name and try dns resolution. The flag "-n" disable this
12658resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12659server is not reachable, the haproxy configuration parsing may last many minutes
12660waiting fir the timeout. During this time no error messages are displayed. The
12661flag "-n" disable this behavior. Note also that during the runtime, this
12662function is disabled for the dynamic acl modifications.
12663
Willy Tarreau74ca5042013-06-11 23:12:07 +020012664There are some restrictions however. Not all methods can be used with all
12665sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12666be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012667
12668 - "found" : only check if the requested sample could be found in the stream,
12669 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012670 to pass any pattern to avoid confusion. This matching method is
12671 particularly useful to detect presence of certain contents such
12672 as headers, cookies, etc... even if they are empty and without
12673 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012674
12675 - "bool" : check the value as a boolean. It can only be applied to fetches
12676 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012677 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012678
12679 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012680 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012681
12682 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012683 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012684
Davor Ocelice9ed2812017-12-25 17:49:28 +010012685 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012686 binary sequence. This may be used with binary or string samples.
12687
12688 - "len" : match the sample's length as an integer. This may be used with
12689 binary or string samples.
12690
Willy Tarreau74ca5042013-06-11 23:12:07 +020012691 - "str" : exact match : match the contents against a string. This may be
12692 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012693
Willy Tarreau74ca5042013-06-11 23:12:07 +020012694 - "sub" : substring match : check that the contents contain at least one of
12695 the provided string patterns. This may be used with binary or
12696 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012697
Willy Tarreau74ca5042013-06-11 23:12:07 +020012698 - "reg" : regex match : match the contents against a list of regular
12699 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012700
Willy Tarreau74ca5042013-06-11 23:12:07 +020012701 - "beg" : prefix match : check that the contents begin like the provided
12702 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012703
Willy Tarreau74ca5042013-06-11 23:12:07 +020012704 - "end" : suffix match : check that the contents end like the provided
12705 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012706
Willy Tarreau74ca5042013-06-11 23:12:07 +020012707 - "dir" : subdir match : check that a slash-delimited portion of the
12708 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012709 This may be used with binary or string samples.
12710
Willy Tarreau74ca5042013-06-11 23:12:07 +020012711 - "dom" : domain match : check that a dot-delimited portion of the contents
12712 exactly match one of the provided string patterns. This may be
12713 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012714
12715For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12716request, it is possible to do :
12717
12718 acl jsess_present cook(JSESSIONID) -m found
12719
12720In order to apply a regular expression on the 500 first bytes of data in the
12721buffer, one would use the following acl :
12722
12723 acl script_tag payload(0,500) -m reg -i <script>
12724
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012725On systems where the regex library is much slower when using "-i", it is
12726possible to convert the sample to lowercase before matching, like this :
12727
12728 acl script_tag payload(0,500),lower -m reg <script>
12729
Willy Tarreau74ca5042013-06-11 23:12:07 +020012730All ACL-specific criteria imply a default matching method. Most often, these
12731criteria are composed by concatenating the name of the original sample fetch
12732method and the matching method. For example, "hdr_beg" applies the "beg" match
12733to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12734criteria rely on a sample fetch method, it is always possible instead to use
12735the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012736
Willy Tarreau74ca5042013-06-11 23:12:07 +020012737If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012738the matching method is simply applied to the underlying sample fetch method.
12739For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012740
Willy Tarreau74ca5042013-06-11 23:12:07 +020012741 acl short_form hdr_beg(host) www.
12742 acl alternate1 hdr_beg(host) -m beg www.
12743 acl alternate2 hdr_dom(host) -m beg www.
12744 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012745
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012746
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012747The table below summarizes the compatibility matrix between sample or converter
12748types and the pattern types to fetch against. It indicates for each compatible
12749combination the name of the matching method to be used, surrounded with angle
12750brackets ">" and "<" when the method is the default one and will work by
12751default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012752
Willy Tarreau74ca5042013-06-11 23:12:07 +020012753 +-------------------------------------------------+
12754 | Input sample type |
12755 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012756 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012757 +----------------------+---------+---------+---------+---------+---------+
12758 | none (presence only) | found | found | found | found | found |
12759 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012760 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012761 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012762 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012763 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012764 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012765 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012766 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012767 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012768 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012769 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012770 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012771 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012772 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012773 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012774 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012775 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012776 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012777 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012778 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012779 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012780 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012781 +----------------------+---------+---------+---------+---------+---------+
12782 | hex block | | | | bin | bin |
12783 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012784
12785
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127867.1.1. Matching booleans
12787------------------------
12788
12789In order to match a boolean, no value is needed and all values are ignored.
12790Boolean matching is used by default for all fetch methods of type "boolean".
12791When boolean matching is used, the fetched value is returned as-is, which means
12792that a boolean "true" will always match and a boolean "false" will never match.
12793
12794Boolean matching may also be enforced using "-m bool" on fetch methods which
12795return an integer value. Then, integer value 0 is converted to the boolean
12796"false" and all other values are converted to "true".
12797
Willy Tarreau6a06a402007-07-15 20:15:28 +020012798
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127997.1.2. Matching integers
12800------------------------
12801
12802Integer matching applies by default to integer fetch methods. It can also be
12803enforced on boolean fetches using "-m int". In this case, "false" is converted
12804to the integer 0, and "true" is converted to the integer 1.
12805
12806Integer matching also supports integer ranges and operators. Note that integer
12807matching only applies to positive values. A range is a value expressed with a
12808lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012809
12810For instance, "1024:65535" is a valid range to represent a range of
12811unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12812representation of privileged ports, and ":1023" would also work.
12813
Willy Tarreau62644772008-07-16 18:36:06 +020012814As a special case, some ACL functions support decimal numbers which are in fact
12815two integers separated by a dot. This is used with some version checks for
12816instance. All integer properties apply to those decimal numbers, including
12817ranges and operators.
12818
Willy Tarreau6a06a402007-07-15 20:15:28 +020012819For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012820operators with ranges does not make much sense and is strongly discouraged.
12821Similarly, it does not make much sense to perform order comparisons with a set
12822of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012823
Willy Tarreau0ba27502007-12-24 16:55:16 +010012824Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012825
12826 eq : true if the tested value equals at least one value
12827 ge : true if the tested value is greater than or equal to at least one value
12828 gt : true if the tested value is greater than at least one value
12829 le : true if the tested value is less than or equal to at least one value
12830 lt : true if the tested value is less than at least one value
12831
Willy Tarreau0ba27502007-12-24 16:55:16 +010012832For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012833
12834 acl negative-length hdr_val(content-length) lt 0
12835
Willy Tarreau62644772008-07-16 18:36:06 +020012836This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12837
12838 acl sslv3 req_ssl_ver 3:3.1
12839
Willy Tarreau6a06a402007-07-15 20:15:28 +020012840
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128417.1.3. Matching strings
12842-----------------------
12843
12844String matching applies to string or binary fetch methods, and exists in 6
12845different forms :
12846
12847 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012848 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012849
12850 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012851 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012852
12853 - prefix match (-m beg) : the patterns are compared with the beginning of
12854 the extracted string, and the ACL matches if any of them matches.
12855
12856 - suffix match (-m end) : the patterns are compared with the end of the
12857 extracted string, and the ACL matches if any of them matches.
12858
Baptiste Assmann33db6002016-03-06 23:32:10 +010012859 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012860 string, delimited with slashes ("/"), and the ACL matches if any of them
12861 matches.
12862
12863 - domain match (-m dom) : the patterns are looked up inside the extracted
12864 string, delimited with dots ("."), and the ACL matches if any of them
12865 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012866
12867String matching applies to verbatim strings as they are passed, with the
12868exception of the backslash ("\") which makes it possible to escape some
12869characters such as the space. If the "-i" flag is passed before the first
12870string, then the matching will be performed ignoring the case. In order
12871to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012872before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012873
12874
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128757.1.4. Matching regular expressions (regexes)
12876---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012877
12878Just like with string matching, regex matching applies to verbatim strings as
12879they are passed, with the exception of the backslash ("\") which makes it
12880possible to escape some characters such as the space. If the "-i" flag is
12881passed before the first regex, then the matching will be performed ignoring
12882the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012883the "--" flag before the first string. Same principle applies of course to
12884match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012885
12886
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128877.1.5. Matching arbitrary data blocks
12888-------------------------------------
12889
12890It is possible to match some extracted samples against a binary block which may
12891not safely be represented as a string. For this, the patterns must be passed as
12892a series of hexadecimal digits in an even number, when the match method is set
12893to binary. Each sequence of two digits will represent a byte. The hexadecimal
12894digits may be used upper or lower case.
12895
12896Example :
12897 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12898 acl hello payload(0,6) -m bin 48656c6c6f0a
12899
12900
129017.1.6. Matching IPv4 and IPv6 addresses
12902---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012903
12904IPv4 addresses values can be specified either as plain addresses or with a
12905netmask appended, in which case the IPv4 address matches whenever it is
12906within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012907host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012908difficult to read and debug configurations. If hostnames are used, you should
12909at least ensure that they are present in /etc/hosts so that the configuration
12910does not depend on any random DNS match at the moment the configuration is
12911parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012912
Daniel Schnellereba56342016-04-13 00:26:52 +020012913The dotted IPv4 address notation is supported in both regular as well as the
12914abbreviated form with all-0-octets omitted:
12915
12916 +------------------+------------------+------------------+
12917 | Example 1 | Example 2 | Example 3 |
12918 +------------------+------------------+------------------+
12919 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12920 | 192.168.1 | 10.12 | 127.1 |
12921 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12922 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12923 +------------------+------------------+------------------+
12924
12925Notice that this is different from RFC 4632 CIDR address notation in which
12926192.168.42/24 would be equivalent to 192.168.42.0/24.
12927
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012928IPv6 may be entered in their usual form, with or without a netmask appended.
12929Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12930trouble with randomly resolved IP addresses, host names are never allowed in
12931IPv6 patterns.
12932
12933HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12934following situations :
12935 - tested address is IPv4, pattern address is IPv4, the match applies
12936 in IPv4 using the supplied mask if any.
12937 - tested address is IPv6, pattern address is IPv6, the match applies
12938 in IPv6 using the supplied mask if any.
12939 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12940 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12941 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12942 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12943 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12944 applied in IPv6 using the supplied IPv6 mask.
12945
Willy Tarreau74ca5042013-06-11 23:12:07 +020012946
129477.2. Using ACLs to form conditions
12948----------------------------------
12949
12950Some actions are only performed upon a valid condition. A condition is a
12951combination of ACLs with operators. 3 operators are supported :
12952
12953 - AND (implicit)
12954 - OR (explicit with the "or" keyword or the "||" operator)
12955 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012956
Willy Tarreau74ca5042013-06-11 23:12:07 +020012957A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012958
Willy Tarreau74ca5042013-06-11 23:12:07 +020012959 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012960
Willy Tarreau74ca5042013-06-11 23:12:07 +020012961Such conditions are generally used after an "if" or "unless" statement,
12962indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012963
Willy Tarreau74ca5042013-06-11 23:12:07 +020012964For instance, to block HTTP requests to the "*" URL with methods other than
12965"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12966requests with a content-length greater than 0, and finally every request which
12967is not either GET/HEAD/POST/OPTIONS !
12968
12969 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012970 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12971 http-request deny if METH_GET HTTP_CONTENT
12972 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012973
12974To select a different backend for requests to static contents on the "www" site
12975and to every request on the "img", "video", "download" and "ftp" hosts :
12976
12977 acl url_static path_beg /static /images /img /css
12978 acl url_static path_end .gif .png .jpg .css .js
12979 acl host_www hdr_beg(host) -i www
12980 acl host_static hdr_beg(host) -i img. video. download. ftp.
12981
Davor Ocelice9ed2812017-12-25 17:49:28 +010012982 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020012983 # of host "www". Use backend "www" for the rest.
12984 use_backend static if host_static or host_www url_static
12985 use_backend www if host_www
12986
12987It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
12988expressions that are built on the fly without needing to be declared. They must
12989be enclosed between braces, with a space before and after each brace (because
12990the braces must be seen as independent words). Example :
12991
12992 The following rule :
12993
12994 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012995 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020012996
12997 Can also be written that way :
12998
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012999 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013000
13001It is generally not recommended to use this construct because it's a lot easier
13002to leave errors in the configuration when written that way. However, for very
13003simple rules matching only one source IP address for instance, it can make more
13004sense to use them than to declare ACLs with random names. Another example of
13005good use is the following :
13006
13007 With named ACLs :
13008
13009 acl site_dead nbsrv(dynamic) lt 2
13010 acl site_dead nbsrv(static) lt 2
13011 monitor fail if site_dead
13012
13013 With anonymous ACLs :
13014
13015 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13016
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013017See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13018keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013019
13020
130217.3. Fetching samples
13022---------------------
13023
13024Historically, sample fetch methods were only used to retrieve data to match
13025against patterns using ACLs. With the arrival of stick-tables, a new class of
13026sample fetch methods was created, most often sharing the same syntax as their
13027ACL counterpart. These sample fetch methods are also known as "fetches". As
13028of now, ACLs and fetches have converged. All ACL fetch methods have been made
13029available as fetch methods, and ACLs may use any sample fetch method as well.
13030
13031This section details all available sample fetch methods and their output type.
13032Some sample fetch methods have deprecated aliases that are used to maintain
13033compatibility with existing configurations. They are then explicitly marked as
13034deprecated and should not be used in new setups.
13035
13036The ACL derivatives are also indicated when available, with their respective
13037matching methods. These ones all have a well defined default pattern matching
13038method, so it is never necessary (though allowed) to pass the "-m" option to
13039indicate how the sample will be matched using ACLs.
13040
13041As indicated in the sample type versus matching compatibility matrix above,
13042when using a generic sample fetch method in an ACL, the "-m" option is
13043mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13044the same keyword exists as an ACL keyword and as a standard fetch method, the
13045ACL engine will automatically pick the ACL-only one by default.
13046
13047Some of these keywords support one or multiple mandatory arguments, and one or
13048multiple optional arguments. These arguments are strongly typed and are checked
13049when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013050incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13051are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013052is optional, it will be indicated below between square brackets ('[ ]'). When
13053all arguments are optional, the parenthesis may be omitted.
13054
13055Thus, the syntax of a standard sample fetch method is one of the following :
13056 - name
13057 - name(arg1)
13058 - name(arg1,arg2)
13059
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013060
130617.3.1. Converters
13062-----------------
13063
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013064Sample fetch methods may be combined with transformations to be applied on top
13065of the fetched sample (also called "converters"). These combinations form what
13066is called "sample expressions" and the result is a "sample". Initially this
13067was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013068has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013069unique-id-format, add-header, ...).
13070
13071These transformations are enumerated as a series of specific keywords after the
13072sample fetch method. These keywords may equally be appended immediately after
13073the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013074support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013075
Willy Tarreau97707872015-01-27 15:12:13 +010013076A certain category of converters are bitwise and arithmetic operators which
13077support performing basic operations on integers. Some bitwise operations are
13078supported (and, or, xor, cpl) and some arithmetic operations are supported
13079(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13080bool) which make it possible to report a match without having to write an ACL.
13081
Willy Tarreau74ca5042013-06-11 23:12:07 +020013082The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013083
Ben Shillitof25e8e52016-12-02 14:25:37 +00001308451d.single(<prop>[,<prop>*])
13085 Returns values for the properties requested as a string, where values are
13086 separated by the delimiter specified with "51degrees-property-separator".
13087 The device is identified using the User-Agent header passed to the
13088 converter. The function can be passed up to five property names, and if a
13089 property name can't be found, the value "NoData" is returned.
13090
13091 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013092 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13093 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013094 # User-Agent passed to the converter.
13095 frontend http-in
13096 bind *:8081
13097 default_backend servers
13098 http-request set-header X-51D-DeviceTypeMobileTablet \
13099 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13100
Willy Tarreau97707872015-01-27 15:12:13 +010013101add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013102 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013103 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013104 name. The name of the variable starts with an indication about its scope. The
13105 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013106 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013107 "sess" : the variable is shared with the whole session
13108 "txn" : the variable is shared with the transaction (request and response)
13109 "req" : the variable is shared only during request processing
13110 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013111 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013112 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013113
13114and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013115 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013116 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013117 numeric value or a variable name. The name of the variable starts with an
13118 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013119 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013120 "sess" : the variable is shared with the whole session
13121 "txn" : the variable is shared with the transaction (request and response)
13122 "req" : the variable is shared only during request processing
13123 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013124 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013125 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013126
Holger Just1bfc24b2017-05-06 00:56:53 +020013127b64dec
13128 Converts (decodes) a base64 encoded input string to its binary
13129 representation. It performs the inverse operation of base64().
13130
Emeric Brun53d1a982014-04-30 18:21:37 +020013131base64
13132 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013133 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013134 an SSL ID can be copied in a header).
13135
Willy Tarreau97707872015-01-27 15:12:13 +010013136bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013137 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013138 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013139 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013140 presence of a flag).
13141
Emeric Brun54c4ac82014-11-03 15:32:43 +010013142bytes(<offset>[,<length>])
13143 Extracts some bytes from an input binary sample. The result is a binary
13144 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013145 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013146
Willy Tarreau280f42b2018-02-19 15:34:12 +010013147concat([<start>],[<var>],[<end>])
13148 Concatenates up to 3 fields after the current sample which is then turned to
13149 a string. The first one, <start>, is a constant string, that will be appended
13150 immediately after the existing sample. It may be omitted if not used. The
13151 second one, <var>, is a variable name. The variable will be looked up, its
13152 contents converted to a string, and it will be appended immediately after the
13153 <first> part. If the variable is not found, nothing is appended. It may be
13154 omitted as well. The third field, <end> is a constant string that will be
13155 appended after the variable. It may also be omitted. Together, these elements
13156 allow to concatenate variables with delimiters to an existing set of
13157 variables. This can be used to build new variables made of a succession of
13158 other variables, such as colon-delimited varlues. Note that due to the config
13159 parser, it is not possible to use a comma nor a closing parenthesis as
13160 delimitors.
13161
13162 Example:
13163 tcp-request session set-var(sess.src) src
13164 tcp-request session set-var(sess.dn) ssl_c_s_dn
13165 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13166 http-request set-header x-hap-sig %[var(txn.sig)]
13167
Willy Tarreau97707872015-01-27 15:12:13 +010013168cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013169 Takes the input value of type signed integer, applies a ones-complement
13170 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013171
Willy Tarreau80599772015-01-20 19:35:24 +010013172crc32([<avalanche>])
13173 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13174 hash function. Optionally, it is possible to apply a full avalanche hash
13175 function to the output if the optional <avalanche> argument equals 1. This
13176 converter uses the same functions as used by the various hash-based load
13177 balancing algorithms, so it will provide exactly the same results. It is
13178 provided for compatibility with other software which want a CRC32 to be
13179 computed on some input keys, so it follows the most common implementation as
13180 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13181 but may provide a better or at least less predictable distribution. It must
13182 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013183 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13184
13185crc32c([<avalanche>])
13186 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13187 hash function. Optionally, it is possible to apply a full avalanche hash
13188 function to the output if the optional <avalanche> argument equals 1. This
13189 converter uses the same functions as described in RFC4960, Appendix B [8].
13190 It is provided for compatibility with other software which want a CRC32C to be
13191 computed on some input keys. It is slower than the other algorithms and it must
13192 not be used for security purposes as a 32-bit hash is trivial to break. See
13193 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013194
David Carlier29b3ca32015-09-25 14:09:21 +010013195da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013196 Asks the DeviceAtlas converter to identify the User Agent string passed on
13197 input, and to emit a string made of the concatenation of the properties
13198 enumerated in argument, delimited by the separator defined by the global
13199 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013200 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013201 configuration language.
13202
13203 Example:
13204 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013205 bind *:8881
13206 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013207 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 +020013208
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013209debug
13210 This converter is used as debug tool. It dumps on screen the content and the
13211 type of the input sample. The sample is returned as is on its output. This
13212 converter only exists when haproxy was built with debugging enabled.
13213
Willy Tarreau97707872015-01-27 15:12:13 +010013214div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013215 Divides the input value of type signed integer by <value>, and returns the
13216 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013217 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013218 variable name. The name of the variable starts with an indication about its
13219 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013220 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013221 "sess" : the variable is shared with the whole session
13222 "txn" : the variable is shared with the transaction (request and response)
13223 "req" : the variable is shared only during request processing
13224 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013225 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013226 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013227
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013228djb2([<avalanche>])
13229 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13230 hash function. Optionally, it is possible to apply a full avalanche hash
13231 function to the output if the optional <avalanche> argument equals 1. This
13232 converter uses the same functions as used by the various hash-based load
13233 balancing algorithms, so it will provide exactly the same results. It is
13234 mostly intended for debugging, but can be used as a stick-table entry to
13235 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013236 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13237 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013238
Willy Tarreau97707872015-01-27 15:12:13 +010013239even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013240 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013241 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13242
Marcin Deranek9631a282018-04-16 14:30:46 +020013243field(<index>,<delimiters>[,<count>])
13244 Extracts the substring at the given index counting from the beginning
13245 (positive index) or from the end (negative index) considering given delimiters
13246 from an input string. Indexes start at 1 or -1 and delimiters are a string
13247 formatted list of chars. Optionally you can specify <count> of fields to
13248 extract (default: 1). Value of 0 indicates extraction of all remaining
13249 fields.
13250
13251 Example :
13252 str(f1_f2_f3__f5),field(5,_) # f5
13253 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13254 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13255 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13256 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013257
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013258hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013259 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013260 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013261 in a way that can be reliably transferred (e.g. an SSL ID can be copied in a
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013262 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013263
Dragan Dosen3f957b22017-10-24 09:27:34 +020013264hex2i
13265 Converts a hex string containing two hex digits per input byte to an
13266 integer. If the input value can not be converted, then zero is returned.
13267
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013268http_date([<offset>])
13269 Converts an integer supposed to contain a date since epoch to a string
13270 representing this date in a format suitable for use in HTTP header fields. If
13271 an offset value is specified, then it is a number of seconds that is added to
13272 the date before the conversion is operated. This is particularly useful to
13273 emit Date header fields, Expires values in responses when combined with a
13274 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013275
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013276in_table(<table>)
13277 Uses the string representation of the input sample to perform a look up in
13278 the specified table. If the key is not found in the table, a boolean false
13279 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013280 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013281 or not a source IP address or an Authorization header was already seen).
13282
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013283ipmask(<mask4>, [<mask6>])
13284 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013285 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013286 table entries and as such use the same server. The mask4 can be passed in
13287 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13288 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13289 If no mask6 is given IPv6 addresses will fail to convert for backwards
13290 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013291
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013292json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013293 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013294 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013295 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013296 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13297 of errors:
13298 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13299 bytes, ...)
13300 - invalid range (the decoded value is within a UTF-8 prohibited range),
13301 - code overlong (the value is encoded with more bytes than necessary).
13302
13303 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13304 character is greater than 0xffff because the JSON string escape specification
13305 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13306 in 4 variants designated by a combination of two suffix letters : "p" for
13307 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13308 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013309 - "ascii" : never fails;
13310 - "utf8" : fails on any detected errors;
13311 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013312 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013313 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013314 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13315 characters corresponding to the other errors.
13316
13317 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013318 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013319
13320 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013321 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013322 capture request header user-agent len 150
13323 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013324
13325 Input request from client 127.0.0.1:
13326 GET / HTTP/1.0
13327 User-Agent: Very "Ugly" UA 1/2
13328
13329 Output log:
13330 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13331
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013332language(<value>[,<default>])
13333 Returns the value with the highest q-factor from a list as extracted from the
13334 "accept-language" header using "req.fhdr". Values with no q-factor have a
13335 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13336 belong to the list of semi-colon delimited <values> will be considered. The
13337 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13338 given list and a default value is provided, it is returned. Note that language
13339 names may have a variant after a dash ('-'). If this variant is present in the
13340 list, it will be matched, but if it is not, only the base language is checked.
13341 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013342 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013343 ordering of the values in the request counts, as the first value among
13344 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013345
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013346 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013347
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013348 # this configuration switches to the backend matching a
13349 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013350
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013351 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13352 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13353 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13354 use_backend spanish if es
13355 use_backend french if fr
13356 use_backend english if en
13357 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013358
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013359length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013360 Get the length of the string. This can only be placed after a string
13361 sample fetch function or after a transformation keyword returning a string
13362 type. The result is of type integer.
13363
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013364lower
13365 Convert a string sample to lower case. This can only be placed after a string
13366 sample fetch function or after a transformation keyword returning a string
13367 type. The result is of type string.
13368
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013369ltime(<format>[,<offset>])
13370 Converts an integer supposed to contain a date since epoch to a string
13371 representing this date in local time using a format defined by the <format>
13372 string using strftime(3). The purpose is to allow any date format to be used
13373 in logs. An optional <offset> in seconds may be applied to the input date
13374 (positive or negative). See the strftime() man page for the format supported
13375 by your operating system. See also the utime converter.
13376
13377 Example :
13378
13379 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013380 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013381 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13382
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013383map(<map_file>[,<default_value>])
13384map_<match_type>(<map_file>[,<default_value>])
13385map_<match_type>_<output_type>(<map_file>[,<default_value>])
13386 Search the input value from <map_file> using the <match_type> matching method,
13387 and return the associated value converted to the type <output_type>. If the
13388 input value cannot be found in the <map_file>, the converter returns the
13389 <default_value>. If the <default_value> is not set, the converter fails and
13390 acts as if no input value could be fetched. If the <match_type> is not set, it
13391 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13392 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13393 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013394
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013395 It is important to avoid overlapping between the keys : IP addresses and
13396 strings are stored in trees, so the first of the finest match will be used.
13397 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013398
Tim Düsterhus4896c442016-11-29 02:15:19 +010013399 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013400 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013401
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013402 input type | match method | output type str | output type int | output type ip
13403 -----------+--------------+-----------------+-----------------+---------------
13404 str | str | map_str | map_str_int | map_str_ip
13405 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013406 str | beg | map_beg | map_beg_int | map_end_ip
13407 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013408 str | sub | map_sub | map_sub_int | map_sub_ip
13409 -----------+--------------+-----------------+-----------------+---------------
13410 str | dir | map_dir | map_dir_int | map_dir_ip
13411 -----------+--------------+-----------------+-----------------+---------------
13412 str | dom | map_dom | map_dom_int | map_dom_ip
13413 -----------+--------------+-----------------+-----------------+---------------
13414 str | end | map_end | map_end_int | map_end_ip
13415 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013416 str | reg | map_reg | map_reg_int | map_reg_ip
13417 -----------+--------------+-----------------+-----------------+---------------
13418 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013419 -----------+--------------+-----------------+-----------------+---------------
13420 int | int | map_int | map_int_int | map_int_ip
13421 -----------+--------------+-----------------+-----------------+---------------
13422 ip | ip | map_ip | map_ip_int | map_ip_ip
13423 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013424
Thierry Fournier8feaa662016-02-10 22:55:20 +010013425 The special map called "map_regm" expect matching zone in the regular
13426 expression and modify the output replacing back reference (like "\1") by
13427 the corresponding match text.
13428
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013429 The file contains one key + value per line. Lines which start with '#' are
13430 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13431 is then the first "word" (series of non-space/tabs characters), and the value
13432 is what follows this series of space/tab till the end of the line excluding
13433 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013434
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013435 Example :
13436
13437 # this is a comment and is ignored
13438 2.22.246.0/23 United Kingdom \n
13439 <-><-----------><--><------------><---->
13440 | | | | `- trailing spaces ignored
13441 | | | `---------- value
13442 | | `-------------------- middle spaces ignored
13443 | `---------------------------- key
13444 `------------------------------------ leading spaces ignored
13445
Willy Tarreau97707872015-01-27 15:12:13 +010013446mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013447 Divides the input value of type signed integer by <value>, and returns the
13448 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013449 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013450 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013451 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013452 "sess" : the variable is shared with the whole session
13453 "txn" : the variable is shared with the transaction (request and response)
13454 "req" : the variable is shared only during request processing
13455 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013456 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013457 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013458
13459mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013460 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013461 the product as an signed integer. In case of overflow, the largest possible
13462 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013463 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013464 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013465 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013466 "sess" : the variable is shared with the whole session
13467 "txn" : the variable is shared with the transaction (request and response)
13468 "req" : the variable is shared only during request processing
13469 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013470 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013471 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013472
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013473nbsrv
13474 Takes an input value of type string, interprets it as a backend name and
13475 returns the number of usable servers in that backend. Can be used in places
13476 where we want to look up a backend from a dynamic name, like a result of a
13477 map lookup.
13478
Willy Tarreau97707872015-01-27 15:12:13 +010013479neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013480 Takes the input value of type signed integer, computes the opposite value,
13481 and returns the remainder as an signed integer. 0 is identity. This operator
13482 is provided for reversed subtracts : in order to subtract the input from a
13483 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013484
13485not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013486 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013487 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013488 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013489 absence of a flag).
13490
13491odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013492 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013493 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13494
13495or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013496 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013497 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013498 numeric value or a variable name. The name of the variable starts with an
13499 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013500 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013501 "sess" : the variable is shared with the whole session
13502 "txn" : the variable is shared with the transaction (request and response)
13503 "req" : the variable is shared only during request processing
13504 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013505 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013506 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013507
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013508regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013509 Applies a regex-based substitution to the input string. It does the same
13510 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13511 default it will replace in the input string the first occurrence of the
13512 largest part matching the regular expression <regex> with the substitution
13513 string <subst>. It is possible to replace all occurrences instead by adding
13514 the flag "g" in the third argument <flags>. It is also possible to make the
13515 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13516 string, it is made up from the concatenation of all desired flags. Thus if
13517 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13518 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013519 configuration parser, some characters such as closing parenthesis, closing
13520 square brackets or comma are not possible to use in the arguments. The first
13521 use of this converter is to replace certain characters or sequence of
13522 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013523
13524 Example :
13525
13526 # de-duplicate "/" in header "x-path".
13527 # input: x-path: /////a///b/c/xzxyz/
13528 # output: x-path: /a/b/c/xzxyz/
13529 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13530
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013531capture-req(<id>)
13532 Capture the string entry in the request slot <id> and returns the entry as
13533 is. If the slot doesn't exist, the capture fails silently.
13534
13535 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013536 "http-response capture", "capture.req.hdr" and
13537 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013538
13539capture-res(<id>)
13540 Capture the string entry in the response slot <id> and returns the entry as
13541 is. If the slot doesn't exist, the capture fails silently.
13542
13543 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013544 "http-response capture", "capture.req.hdr" and
13545 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013546
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013547sdbm([<avalanche>])
13548 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13549 hash function. Optionally, it is possible to apply a full avalanche hash
13550 function to the output if the optional <avalanche> argument equals 1. This
13551 converter uses the same functions as used by the various hash-based load
13552 balancing algorithms, so it will provide exactly the same results. It is
13553 mostly intended for debugging, but can be used as a stick-table entry to
13554 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013555 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13556 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013557
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013558set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013559 Sets a variable with the input content and returns the content on the output
13560 as-is. The variable keeps the value and the associated input type. The name of
13561 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013562 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013563 "sess" : the variable is shared with the whole session
13564 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013565 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013566 "req" : the variable is shared only during request processing,
13567 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013568 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013569 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013570
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013571sha1
13572 Converts a binary input sample to a SHA1 digest. The result is a binary
13573 sample with length of 20 bytes.
13574
Tim Duesterhusca097c12018-04-27 21:18:45 +020013575strcmp(<var>)
13576 Compares the contents of <var> with the input value of type string. Returns
13577 the result as a signed integer compatible with strcmp(3): 0 if both strings
13578 are identical. A value less than 0 if the left string is lexicographically
13579 smaller than the right string or if the left string is shorter. A value greater
13580 than 0 otherwise (right string greater than left string or the right string is
13581 shorter).
13582
13583 Example :
13584
13585 http-request set-var(txn.host) hdr(host)
13586 # Check whether the client is attempting domain fronting.
13587 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13588
13589
Willy Tarreau97707872015-01-27 15:12:13 +010013590sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013591 Subtracts <value> from the input value of type signed integer, and returns
13592 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013593 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013594 or a variable name. The name of the variable starts with an indication about
13595 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013596 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013597 "sess" : the variable is shared with the whole session
13598 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013599 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013600 "req" : the variable is shared only during request processing,
13601 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013602 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013603 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013604
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013605table_bytes_in_rate(<table>)
13606 Uses the string representation of the input sample to perform a look up in
13607 the specified table. If the key is not found in the table, integer value zero
13608 is returned. Otherwise the converter returns the average client-to-server
13609 bytes rate associated with the input sample in the designated table, measured
13610 in amount of bytes over the period configured in the table. See also the
13611 sc_bytes_in_rate sample fetch keyword.
13612
13613
13614table_bytes_out_rate(<table>)
13615 Uses the string representation of the input sample to perform a look up in
13616 the specified table. If the key is not found in the table, integer value zero
13617 is returned. Otherwise the converter returns the average server-to-client
13618 bytes rate associated with the input sample in the designated table, measured
13619 in amount of bytes over the period configured in the table. See also the
13620 sc_bytes_out_rate sample fetch keyword.
13621
13622table_conn_cnt(<table>)
13623 Uses the string representation of the input sample to perform a look up in
13624 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013625 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013626 connections associated with the input sample in the designated table. See
13627 also the sc_conn_cnt sample fetch keyword.
13628
13629table_conn_cur(<table>)
13630 Uses the string representation of the input sample to perform a look up in
13631 the specified table. If the key is not found in the table, integer value zero
13632 is returned. Otherwise the converter returns the current amount of concurrent
13633 tracked connections associated with the input sample in the designated table.
13634 See also the sc_conn_cur sample fetch keyword.
13635
13636table_conn_rate(<table>)
13637 Uses the string representation of the input sample to perform a look up in
13638 the specified table. If the key is not found in the table, integer value zero
13639 is returned. Otherwise the converter returns the average incoming connection
13640 rate associated with the input sample in the designated table. See also the
13641 sc_conn_rate sample fetch keyword.
13642
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013643table_gpt0(<table>)
13644 Uses the string representation of the input sample to perform a look up in
13645 the specified table. If the key is not found in the table, boolean value zero
13646 is returned. Otherwise the converter returns the current value of the first
13647 general purpose tag associated with the input sample in the designated table.
13648 See also the sc_get_gpt0 sample fetch keyword.
13649
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013650table_gpc0(<table>)
13651 Uses the string representation of the input sample to perform a look up in
13652 the specified table. If the key is not found in the table, integer value zero
13653 is returned. Otherwise the converter returns the current value of the first
13654 general purpose counter associated with the input sample in the designated
13655 table. See also the sc_get_gpc0 sample fetch keyword.
13656
13657table_gpc0_rate(<table>)
13658 Uses the string representation of the input sample to perform a look up in
13659 the specified table. If the key is not found in the table, integer value zero
13660 is returned. Otherwise the converter returns the frequency which the gpc0
13661 counter was incremented over the configured period in the table, associated
13662 with the input sample in the designated table. See also the sc_get_gpc0_rate
13663 sample fetch keyword.
13664
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013665table_gpc1(<table>)
13666 Uses the string representation of the input sample to perform a look up in
13667 the specified table. If the key is not found in the table, integer value zero
13668 is returned. Otherwise the converter returns the current value of the second
13669 general purpose counter associated with the input sample in the designated
13670 table. See also the sc_get_gpc1 sample fetch keyword.
13671
13672table_gpc1_rate(<table>)
13673 Uses the string representation of the input sample to perform a look up in
13674 the specified table. If the key is not found in the table, integer value zero
13675 is returned. Otherwise the converter returns the frequency which the gpc1
13676 counter was incremented over the configured period in the table, associated
13677 with the input sample in the designated table. See also the sc_get_gpc1_rate
13678 sample fetch keyword.
13679
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013680table_http_err_cnt(<table>)
13681 Uses the string representation of the input sample to perform a look up in
13682 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013683 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013684 errors associated with the input sample in the designated table. See also the
13685 sc_http_err_cnt sample fetch keyword.
13686
13687table_http_err_rate(<table>)
13688 Uses the string representation of the input sample to perform a look up in
13689 the specified table. If the key is not found in the table, integer value zero
13690 is returned. Otherwise the average rate of HTTP errors associated with the
13691 input sample in the designated table, measured in amount of errors over the
13692 period configured in the table. See also the sc_http_err_rate sample fetch
13693 keyword.
13694
13695table_http_req_cnt(<table>)
13696 Uses the string representation of the input sample to perform a look up in
13697 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013698 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013699 requests associated with the input sample in the designated table. See also
13700 the sc_http_req_cnt sample fetch keyword.
13701
13702table_http_req_rate(<table>)
13703 Uses the string representation of the input sample to perform a look up in
13704 the specified table. If the key is not found in the table, integer value zero
13705 is returned. Otherwise the average rate of HTTP requests associated with the
13706 input sample in the designated table, measured in amount of requests over the
13707 period configured in the table. See also the sc_http_req_rate sample fetch
13708 keyword.
13709
13710table_kbytes_in(<table>)
13711 Uses the string representation of the input sample to perform a look up in
13712 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013713 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013714 to-server data associated with the input sample in the designated table,
13715 measured in kilobytes. The test is currently performed on 32-bit integers,
13716 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13717 keyword.
13718
13719table_kbytes_out(<table>)
13720 Uses the string representation of the input sample to perform a look up in
13721 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013722 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013723 to-client data associated with the input sample in the designated table,
13724 measured in kilobytes. The test is currently performed on 32-bit integers,
13725 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13726 keyword.
13727
13728table_server_id(<table>)
13729 Uses the string representation of the input sample to perform a look up in
13730 the specified table. If the key is not found in the table, integer value zero
13731 is returned. Otherwise the converter returns the server ID associated with
13732 the input sample in the designated table. A server ID is associated to a
13733 sample by a "stick" rule when a connection to a server succeeds. A server ID
13734 zero means that no server is associated with this key.
13735
13736table_sess_cnt(<table>)
13737 Uses the string representation of the input sample to perform a look up in
13738 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013739 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013740 sessions associated with the input sample in the designated table. Note that
13741 a session here refers to an incoming connection being accepted by the
13742 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13743 keyword.
13744
13745table_sess_rate(<table>)
13746 Uses the string representation of the input sample to perform a look up in
13747 the specified table. If the key is not found in the table, integer value zero
13748 is returned. Otherwise the converter returns the average incoming session
13749 rate associated with the input sample in the designated table. Note that a
13750 session here refers to an incoming connection being accepted by the
13751 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13752 keyword.
13753
13754table_trackers(<table>)
13755 Uses the string representation of the input sample to perform a look up in
13756 the specified table. If the key is not found in the table, integer value zero
13757 is returned. Otherwise the converter returns the current amount of concurrent
13758 connections tracking the same key as the input sample in the designated
13759 table. It differs from table_conn_cur in that it does not rely on any stored
13760 information but on the table's reference count (the "use" value which is
13761 returned by "show table" on the CLI). This may sometimes be more suited for
13762 layer7 tracking. It can be used to tell a server how many concurrent
13763 connections there are from a given address for example. See also the
13764 sc_trackers sample fetch keyword.
13765
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013766upper
13767 Convert a string sample to upper case. This can only be placed after a string
13768 sample fetch function or after a transformation keyword returning a string
13769 type. The result is of type string.
13770
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013771url_dec
13772 Takes an url-encoded string provided as input and returns the decoded
13773 version as output. The input and the output are of type string.
13774
Christopher Faulet85d79c92016-11-09 16:54:56 +010013775unset-var(<var name>)
13776 Unsets a variable if the input content is defined. The name of the variable
13777 starts with an indication about its scope. The scopes allowed are:
13778 "proc" : the variable is shared with the whole process
13779 "sess" : the variable is shared with the whole session
13780 "txn" : the variable is shared with the transaction (request and
13781 response),
13782 "req" : the variable is shared only during request processing,
13783 "res" : the variable is shared only during response processing.
13784 This prefix is followed by a name. The separator is a '.'. The name may only
13785 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13786
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013787utime(<format>[,<offset>])
13788 Converts an integer supposed to contain a date since epoch to a string
13789 representing this date in UTC time using a format defined by the <format>
13790 string using strftime(3). The purpose is to allow any date format to be used
13791 in logs. An optional <offset> in seconds may be applied to the input date
13792 (positive or negative). See the strftime() man page for the format supported
13793 by your operating system. See also the ltime converter.
13794
13795 Example :
13796
13797 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013798 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013799 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13800
Marcin Deranek9631a282018-04-16 14:30:46 +020013801word(<index>,<delimiters>[,<count>])
13802 Extracts the nth word counting from the beginning (positive index) or from
13803 the end (negative index) considering given delimiters from an input string.
13804 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13805 Optionally you can specify <count> of words to extract (default: 1).
13806 Value of 0 indicates extraction of all remaining words.
13807
13808 Example :
13809 str(f1_f2_f3__f5),word(4,_) # f5
13810 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13811 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13812 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13813 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013814
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013815wt6([<avalanche>])
13816 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13817 hash function. Optionally, it is possible to apply a full avalanche hash
13818 function to the output if the optional <avalanche> argument equals 1. This
13819 converter uses the same functions as used by the various hash-based load
13820 balancing algorithms, so it will provide exactly the same results. It is
13821 mostly intended for debugging, but can be used as a stick-table entry to
13822 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013823 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13824 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013825
Willy Tarreau97707872015-01-27 15:12:13 +010013826xor(<value>)
13827 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013828 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013829 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013830 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013831 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013832 "sess" : the variable is shared with the whole session
13833 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013834 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013835 "req" : the variable is shared only during request processing,
13836 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013837 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013838 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013839
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013840xxh32([<seed>])
13841 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13842 variant of the XXHash hash function. This hash supports a seed which defaults
13843 to zero but a different value maybe passed as the <seed> argument. This hash
13844 is known to be very good and very fast so it can be used to hash URLs and/or
13845 URL parameters for use as stick-table keys to collect statistics with a low
13846 collision rate, though care must be taken as the algorithm is not considered
13847 as cryptographically secure.
13848
13849xxh64([<seed>])
13850 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13851 variant of the XXHash hash function. This hash supports a seed which defaults
13852 to zero but a different value maybe passed as the <seed> argument. This hash
13853 is known to be very good and very fast so it can be used to hash URLs and/or
13854 URL parameters for use as stick-table keys to collect statistics with a low
13855 collision rate, though care must be taken as the algorithm is not considered
13856 as cryptographically secure.
13857
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013858
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200138597.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013860--------------------------------------------
13861
13862A first set of sample fetch methods applies to internal information which does
13863not even relate to any client information. These ones are sometimes used with
13864"monitor-fail" directives to report an internal status to external watchers.
13865The sample fetch methods described in this section are usable anywhere.
13866
13867always_false : boolean
13868 Always returns the boolean "false" value. It may be used with ACLs as a
13869 temporary replacement for another one when adjusting configurations.
13870
13871always_true : boolean
13872 Always returns the boolean "true" value. It may be used with ACLs as a
13873 temporary replacement for another one when adjusting configurations.
13874
13875avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013876 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013877 divided by the number of active servers. The current backend is used if no
13878 backend is specified. This is very similar to "queue" except that the size of
13879 the farm is considered, in order to give a more accurate measurement of the
13880 time it may take for a new connection to be processed. The main usage is with
13881 ACL to return a sorry page to new users when it becomes certain they will get
13882 a degraded service, or to pass to the backend servers in a header so that
13883 they decide to work in degraded mode or to disable some functions to speed up
13884 the processing a bit. Note that in the event there would not be any active
13885 server anymore, twice the number of queued connections would be considered as
13886 the measured value. This is a fair estimate, as we expect one server to get
13887 back soon anyway, but we still prefer to send new traffic to another backend
13888 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13889 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013890
Willy Tarreau74ca5042013-06-11 23:12:07 +020013891be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013892 Applies to the number of currently established connections on the backend,
13893 possibly including the connection being evaluated. If no backend name is
13894 specified, the current one is used. But it is also possible to check another
13895 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013896 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
13897
13898be_conn_free([<backend>]) : integer
13899 Returns an integer value corresponding to the number of available connections
13900 across available servers in the backend. Queue slots are not included. Backup
13901 servers are also not included, unless all other servers are down. If no
13902 backend name is specified, the current one is used. But it is also possible
13903 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013904 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
13905 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013906
13907 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
13908 (meaning unlimited), then this fetch clearly does not make sense, in which
13909 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013910
Willy Tarreau74ca5042013-06-11 23:12:07 +020013911be_sess_rate([<backend>]) : integer
13912 Returns an integer value corresponding to the sessions creation rate on the
13913 backend, in number of new sessions per second. This is used with ACLs to
13914 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013915 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013916 online dictionary). It can also be useful to add this element to logs using a
13917 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013918
13919 Example :
13920 # Redirect to an error page if the dictionary is requested too often
13921 backend dynamic
13922 mode http
13923 acl being_scanned be_sess_rate gt 100
13924 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013925
Davor Ocelice9ed2812017-12-25 17:49:28 +010013926bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013927 Returns a binary chain. The input is the hexadecimal representation
13928 of the string.
13929
13930bool(<bool>) : bool
13931 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
13932 'false' and '0' are the same. 'true' and '1' are the same.
13933
Willy Tarreau74ca5042013-06-11 23:12:07 +020013934connslots([<backend>]) : integer
13935 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013936 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020013937 connections on all servers and the maximum queue size. This is probably only
13938 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050013939
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013940 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020013941 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013942 usage; see "use_backend" keyword) can be redirected to a different backend.
13943
Willy Tarreau55165fe2009-05-10 12:02:55 +020013944 'connslots' = number of available server connection slots, + number of
13945 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013946
Willy Tarreaua36af912009-10-10 12:02:45 +020013947 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020013948 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020013949 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020013950 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010013951 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020013952 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020013953 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013954
Willy Tarreau55165fe2009-05-10 12:02:55 +020013955 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
13956 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020013957 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020013958 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013959
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013960date([<offset>]) : integer
13961 Returns the current date as the epoch (number of seconds since 01/01/1970).
13962 If an offset value is specified, then it is a number of seconds that is added
13963 to the current date before returning the value. This is particularly useful
13964 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020013965 It is useful combined with the http_date converter.
13966
13967 Example :
13968
13969 # set an expires header to now+1 hour in every response
13970 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013971
Etienne Carrierea792a0a2018-01-17 13:43:24 +010013972date_us : integer
13973 Return the microseconds part of the date (the "second" part is returned by
13974 date sample). This sample is coherent with the date sample as it is comes
13975 from the same timeval structure.
13976
Willy Tarreaud716f9b2017-10-13 11:03:15 +020013977distcc_body(<token>[,<occ>]) : binary
13978 Parses a distcc message and returns the body associated to occurrence #<occ>
13979 of the token <token>. Occurrences start at 1, and when unspecified, any may
13980 match though in practice only the first one is checked for now. This can be
13981 used to extract file names or arguments in files built using distcc through
13982 haproxy. Please refer to distcc's protocol documentation for the complete
13983 list of supported tokens.
13984
13985distcc_param(<token>[,<occ>]) : integer
13986 Parses a distcc message and returns the parameter associated to occurrence
13987 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
13988 any may match though in practice only the first one is checked for now. This
13989 can be used to extract certain information such as the protocol version, the
13990 file size or the argument in files built using distcc through haproxy.
13991 Another use case consists in waiting for the start of the preprocessed file
13992 contents before connecting to the server to avoid keeping idle connections.
13993 Please refer to distcc's protocol documentation for the complete list of
13994 supported tokens.
13995
13996 Example :
13997 # wait up to 20s for the pre-processed file to be uploaded
13998 tcp-request inspect-delay 20s
13999 tcp-request content accept if { distcc_param(DOTI) -m found }
14000 # send large files to the big farm
14001 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14002
Willy Tarreau595ec542013-06-12 21:34:28 +020014003env(<name>) : string
14004 Returns a string containing the value of environment variable <name>. As a
14005 reminder, environment variables are per-process and are sampled when the
14006 process starts. This can be useful to pass some information to a next hop
14007 server, or with ACLs to take specific action when the process is started a
14008 certain way.
14009
14010 Examples :
14011 # Pass the Via header to next hop with the local hostname in it
14012 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14013
14014 # reject cookie-less requests when the STOP environment variable is set
14015 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14016
Willy Tarreau74ca5042013-06-11 23:12:07 +020014017fe_conn([<frontend>]) : integer
14018 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014019 possibly including the connection being evaluated. If no frontend name is
14020 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014021 frontend. It can be used to return a sorry page before hard-blocking, or to
14022 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014023 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014024 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14025 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014026
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014027fe_req_rate([<frontend>]) : integer
14028 Returns an integer value corresponding to the number of HTTP requests per
14029 second sent to a frontend. This number can differ from "fe_sess_rate" in
14030 situations where client-side keep-alive is enabled.
14031
Willy Tarreau74ca5042013-06-11 23:12:07 +020014032fe_sess_rate([<frontend>]) : integer
14033 Returns an integer value corresponding to the sessions creation rate on the
14034 frontend, in number of new sessions per second. This is used with ACLs to
14035 limit the incoming session rate to an acceptable range in order to prevent
14036 abuse of service at the earliest moment, for example when combined with other
14037 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14038 down below the limit. It can also be useful to add this element to logs using
14039 a log-format directive. See also the "rate-limit sessions" directive for use
14040 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014041
14042 Example :
14043 # This frontend limits incoming mails to 10/s with a max of 100
14044 # concurrent connections. We accept any connection below 10/s, and
14045 # force excess clients to wait for 100 ms. Since clients are limited to
14046 # 100 max, there cannot be more than 10 incoming mails per second.
14047 frontend mail
14048 bind :25
14049 mode tcp
14050 maxconn 100
14051 acl too_fast fe_sess_rate ge 10
14052 tcp-request inspect-delay 100ms
14053 tcp-request content accept if ! too_fast
14054 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014055
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014056hostname : string
14057 Returns the system hostname.
14058
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014059int(<integer>) : signed integer
14060 Returns a signed integer.
14061
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014062ipv4(<ipv4>) : ipv4
14063 Returns an ipv4.
14064
14065ipv6(<ipv6>) : ipv6
14066 Returns an ipv6.
14067
14068meth(<method>) : method
14069 Returns a method.
14070
Willy Tarreau0f30d262014-11-24 16:02:05 +010014071nbproc : integer
14072 Returns an integer value corresponding to the number of processes that were
14073 started (it equals the global "nbproc" setting). This is useful for logging
14074 and debugging purposes.
14075
Willy Tarreau74ca5042013-06-11 23:12:07 +020014076nbsrv([<backend>]) : integer
14077 Returns an integer value corresponding to the number of usable servers of
14078 either the current backend or the named backend. This is mostly used with
14079 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014080 switch to an alternate backend when the number of servers is too low to
14081 to handle some load. It is useful to report a failure when combined with
14082 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014083
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014084prio_class : integer
14085 Returns the priority class of the current session for http mode or connection
14086 for tcp mode. The value will be that set by the last call to "http-request
14087 set-priority-class" or "tcp-request content set-priority-class".
14088
14089prio_offset : integer
14090 Returns the priority offset of the current session for http mode or
14091 connection for tcp mode. The value will be that set by the last call to
14092 "http-request set-priority-offset" or "tcp-request content
14093 set-priority-offset".
14094
Willy Tarreau0f30d262014-11-24 16:02:05 +010014095proc : integer
14096 Returns an integer value corresponding to the position of the process calling
14097 the function, between 1 and global.nbproc. This is useful for logging and
14098 debugging purposes.
14099
Willy Tarreau74ca5042013-06-11 23:12:07 +020014100queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014101 Returns the total number of queued connections of the designated backend,
14102 including all the connections in server queues. If no backend name is
14103 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014104 one. This is useful with ACLs or to pass statistics to backend servers. This
14105 can be used to take actions when queuing goes above a known level, generally
14106 indicating a surge of traffic or a massive slowdown on the servers. One
14107 possible action could be to reject new users but still accept old ones. See
14108 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14109
Willy Tarreau84310e22014-02-14 11:59:04 +010014110rand([<range>]) : integer
14111 Returns a random integer value within a range of <range> possible values,
14112 starting at zero. If the range is not specified, it defaults to 2^32, which
14113 gives numbers between 0 and 4294967295. It can be useful to pass some values
14114 needed to take some routing decisions for example, or just for debugging
14115 purposes. This random must not be used for security purposes.
14116
Willy Tarreau74ca5042013-06-11 23:12:07 +020014117srv_conn([<backend>/]<server>) : integer
14118 Returns an integer value corresponding to the number of currently established
14119 connections on the designated server, possibly including the connection being
14120 evaluated. If <backend> is omitted, then the server is looked up in the
14121 current backend. It can be used to use a specific farm when one server is
14122 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014123 connections with it. See also the "fe_conn", "be_conn", "queue", and
14124 "srv_conn_free" fetch methods.
14125
14126srv_conn_free([<backend>/]<server>) : integer
14127 Returns an integer value corresponding to the number of available connections
14128 on the designated server, possibly including the connection being evaluated.
14129 The value does not include queue slots. If <backend> is omitted, then the
14130 server is looked up in the current backend. It can be used to use a specific
14131 farm when one server is full, or to inform the server about our view of the
14132 number of active connections with it. See also the "be_conn_free" and
14133 "srv_conn" fetch methods.
14134
14135 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14136 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014137
14138srv_is_up([<backend>/]<server>) : boolean
14139 Returns true when the designated server is UP, and false when it is either
14140 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14141 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014142 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014143 availability). Another possible use which is more of a hack consists in
14144 using dummy servers as boolean variables that can be enabled or disabled from
14145 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14146
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014147srv_queue([<backend>/]<server>) : integer
14148 Returns an integer value corresponding to the number of connections currently
14149 pending in the designated server's queue. If <backend> is omitted, then the
14150 server is looked up in the current backend. It can sometimes be used together
14151 with the "use-server" directive to force to use a known faster server when it
14152 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14153 fetch methods.
14154
Willy Tarreau74ca5042013-06-11 23:12:07 +020014155srv_sess_rate([<backend>/]<server>) : integer
14156 Returns an integer corresponding to the sessions creation rate on the
14157 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014158 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014159 used with ACLs but can make sense with logs too. This is used to switch to an
14160 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014161 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014162 overloading servers).
14163
14164 Example :
14165 # Redirect to a separate back
14166 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14167 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14168 use_backend be2 if srv1_full or srv2_full
14169
Willy Tarreau0f30d262014-11-24 16:02:05 +010014170stopping : boolean
14171 Returns TRUE if the process calling the function is currently stopping. This
14172 can be useful for logging, or for relaxing certain checks or helping close
14173 certain connections upon graceful shutdown.
14174
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014175str(<string>) : string
14176 Returns a string.
14177
Willy Tarreau74ca5042013-06-11 23:12:07 +020014178table_avl([<table>]) : integer
14179 Returns the total number of available entries in the current proxy's
14180 stick-table or in the designated stick-table. See also table_cnt.
14181
14182table_cnt([<table>]) : integer
14183 Returns the total number of entries currently in use in the current proxy's
14184 stick-table or in the designated stick-table. See also src_conn_cnt and
14185 table_avl for other entry counting methods.
14186
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014187thread : integer
14188 Returns an integer value corresponding to the position of the thread calling
14189 the function, between 0 and (global.nbthread-1). This is useful for logging
14190 and debugging purposes.
14191
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014192var(<var-name>) : undefined
14193 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014194 sample fetch fails. The name of the variable starts with an indication
14195 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014196 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014197 "sess" : the variable is shared with the whole session
14198 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014199 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014200 "req" : the variable is shared only during request processing,
14201 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014202 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014203 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014204
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200142057.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014206----------------------------------
14207
14208The layer 4 usually describes just the transport layer which in haproxy is
14209closest to the connection, where no content is yet made available. The fetch
14210methods described here are usable as low as the "tcp-request connection" rule
14211sets unless they require some future information. Those generally include
14212TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014213the incoming connection. For retrieving a value from a sticky counters, the
14214counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014215"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14216used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14217can be specified as the first integer argument when using the "sc_" prefix.
14218Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14219table may be specified with the "sc*" form, in which case the currently
14220tracked key will be looked up into this alternate table instead of the table
14221currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014222
14223be_id : integer
14224 Returns an integer containing the current backend's id. It can be used in
14225 frontends with responses to check which backend processed the request.
14226
Marcin Deranekd2471c22016-12-12 14:08:05 +010014227be_name : string
14228 Returns a string containing the current backend's name. It can be used in
14229 frontends with responses to check which backend processed the request.
14230
Willy Tarreau74ca5042013-06-11 23:12:07 +020014231dst : ip
14232 This is the destination IPv4 address of the connection on the client side,
14233 which is the address the client connected to. It can be useful when running
14234 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14235 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
14236 RFC 4291.
14237
14238dst_conn : integer
14239 Returns an integer value corresponding to the number of currently established
14240 connections on the same socket including the one being evaluated. It is
14241 normally used with ACLs but can as well be used to pass the information to
14242 servers in an HTTP header or in logs. It can be used to either return a sorry
14243 page before hard-blocking, or to use a specific backend to drain new requests
14244 when the socket is considered saturated. This offers the ability to assign
14245 different limits to different listening ports or addresses. See also the
14246 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014247
Willy Tarreau16e01562016-08-09 16:46:18 +020014248dst_is_local : boolean
14249 Returns true if the destination address of the incoming connection is local
14250 to the system, or false if the address doesn't exist on the system, meaning
14251 that it was intercepted in transparent mode. It can be useful to apply
14252 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014253 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014254 be delivered only on this address, or SSH access could be locally redirected.
14255 Please note that the check involves a few system calls, so it's better to do
14256 it only once per connection.
14257
Willy Tarreau74ca5042013-06-11 23:12:07 +020014258dst_port : integer
14259 Returns an integer value corresponding to the destination TCP port of the
14260 connection on the client side, which is the port the client connected to.
14261 This might be used when running in transparent mode, when assigning dynamic
14262 ports to some clients for a whole application session, to stick all users to
14263 a same server, or to pass the destination port information to a server using
14264 an HTTP header.
14265
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014266fc_http_major : integer
14267 Reports the front connection's HTTP major version encoding, which may be 1
14268 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14269 encoding and not on the version present in the request header.
14270
Emeric Brun4f603012017-01-05 15:11:44 +010014271fc_rcvd_proxy : boolean
14272 Returns true if the client initiated the connection with a PROXY protocol
14273 header.
14274
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014275fc_rtt(<unit>) : integer
14276 Returns the Round Trip Time (RTT) measured by the kernel for the client
14277 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14278 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14279 connection is not established, if the connection is not TCP or if the
14280 operating system does not support TCP_INFO, for example Linux kernels before
14281 2.4, the sample fetch fails.
14282
14283fc_rttvar(<unit>) : integer
14284 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14285 client connection. <unit> is facultative, by default the unit is milliseconds.
14286 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14287 server connection is not established, if the connection is not TCP or if the
14288 operating system does not support TCP_INFO, for example Linux kernels before
14289 2.4, the sample fetch fails.
14290
Joe Williams30fcd392016-08-10 07:06:44 -070014291fc_unacked(<unit>) : integer
14292 Returns the unacked counter measured by the kernel for the client connection.
14293 If the server connection is not established, if the connection is not TCP or
14294 if the operating system does not support TCP_INFO, for example Linux kernels
14295 before 2.4, the sample fetch fails.
14296
14297fc_sacked(<unit>) : integer
14298 Returns the sacked counter measured by the kernel for the client connection.
14299 If the server connection is not established, if the connection is not TCP or
14300 if the operating system does not support TCP_INFO, for example Linux kernels
14301 before 2.4, the sample fetch fails.
14302
14303fc_retrans(<unit>) : integer
14304 Returns the retransmits counter measured by the kernel for the client
14305 connection. If the server connection is not established, if the connection is
14306 not TCP or if the operating system does not support TCP_INFO, for example
14307 Linux kernels before 2.4, the sample fetch fails.
14308
14309fc_fackets(<unit>) : integer
14310 Returns the fack counter measured by the kernel for the client
14311 connection. If the server connection is not established, if the connection is
14312 not TCP or if the operating system does not support TCP_INFO, for example
14313 Linux kernels before 2.4, the sample fetch fails.
14314
14315fc_lost(<unit>) : integer
14316 Returns the lost counter measured by the kernel for the client
14317 connection. If the server connection is not established, if the connection is
14318 not TCP or if the operating system does not support TCP_INFO, for example
14319 Linux kernels before 2.4, the sample fetch fails.
14320
14321fc_reordering(<unit>) : integer
14322 Returns the reordering counter measured by the kernel for the client
14323 connection. If the server connection is not established, if the connection is
14324 not TCP or if the operating system does not support TCP_INFO, for example
14325 Linux kernels before 2.4, the sample fetch fails.
14326
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014327fe_defbe : string
14328 Returns a string containing the frontend's default backend name. It can be
14329 used in frontends to check which backend will handle requests by default.
14330
Willy Tarreau74ca5042013-06-11 23:12:07 +020014331fe_id : integer
14332 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014333 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014334 coming via a same frontend to the same server.
14335
Marcin Deranekd2471c22016-12-12 14:08:05 +010014336fe_name : string
14337 Returns a string containing the current frontend's name. It can be used in
14338 backends to check from which frontend it was called, or to stick all users
14339 coming via a same frontend to the same server.
14340
Cyril Bonté62ba8702014-04-22 23:52:25 +020014341sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014342sc0_bytes_in_rate([<table>]) : integer
14343sc1_bytes_in_rate([<table>]) : integer
14344sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014345 Returns the average client-to-server bytes rate from the currently tracked
14346 counters, measured in amount of bytes over the period configured in the
14347 table. See also src_bytes_in_rate.
14348
Cyril Bonté62ba8702014-04-22 23:52:25 +020014349sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014350sc0_bytes_out_rate([<table>]) : integer
14351sc1_bytes_out_rate([<table>]) : integer
14352sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014353 Returns the average server-to-client bytes rate from the currently tracked
14354 counters, measured in amount of bytes over the period configured in the
14355 table. See also src_bytes_out_rate.
14356
Cyril Bonté62ba8702014-04-22 23:52:25 +020014357sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014358sc0_clr_gpc0([<table>]) : integer
14359sc1_clr_gpc0([<table>]) : integer
14360sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014361 Clears the first General Purpose Counter associated to the currently tracked
14362 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014363 stored value is zero, so first invocation will always return zero. This is
14364 typically used as a second ACL in an expression in order to mark a connection
14365 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014366
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014367 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014368 # block if 5 consecutive requests continue to come faster than 10 sess
14369 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014370 acl abuse sc0_http_req_rate gt 10
14371 acl kill sc0_inc_gpc0 gt 5
14372 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014373 tcp-request connection accept if !abuse save
14374 tcp-request connection reject if abuse kill
14375
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014376sc_clr_gpc1(<ctr>[,<table>]) : integer
14377sc0_clr_gpc1([<table>]) : integer
14378sc1_clr_gpc1([<table>]) : integer
14379sc2_clr_gpc1([<table>]) : integer
14380 Clears the second General Purpose Counter associated to the currently tracked
14381 counters, and returns its previous value. Before the first invocation, the
14382 stored value is zero, so first invocation will always return zero. This is
14383 typically used as a second ACL in an expression in order to mark a connection
14384 when a first ACL was verified.
14385
Cyril Bonté62ba8702014-04-22 23:52:25 +020014386sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014387sc0_conn_cnt([<table>]) : integer
14388sc1_conn_cnt([<table>]) : integer
14389sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014390 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014391 counters. See also src_conn_cnt.
14392
Cyril Bonté62ba8702014-04-22 23:52:25 +020014393sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014394sc0_conn_cur([<table>]) : integer
14395sc1_conn_cur([<table>]) : integer
14396sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014397 Returns the current amount of concurrent connections tracking the same
14398 tracked counters. This number is automatically incremented when tracking
14399 begins and decremented when tracking stops. See also src_conn_cur.
14400
Cyril Bonté62ba8702014-04-22 23:52:25 +020014401sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014402sc0_conn_rate([<table>]) : integer
14403sc1_conn_rate([<table>]) : integer
14404sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014405 Returns the average connection rate from the currently tracked counters,
14406 measured in amount of connections over the period configured in the table.
14407 See also src_conn_rate.
14408
Cyril Bonté62ba8702014-04-22 23:52:25 +020014409sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014410sc0_get_gpc0([<table>]) : integer
14411sc1_get_gpc0([<table>]) : integer
14412sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014413 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014414 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014415
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014416sc_get_gpc1(<ctr>[,<table>]) : integer
14417sc0_get_gpc1([<table>]) : integer
14418sc1_get_gpc1([<table>]) : integer
14419sc2_get_gpc1([<table>]) : integer
14420 Returns the value of the second General Purpose Counter associated to the
14421 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14422
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014423sc_get_gpt0(<ctr>[,<table>]) : integer
14424sc0_get_gpt0([<table>]) : integer
14425sc1_get_gpt0([<table>]) : integer
14426sc2_get_gpt0([<table>]) : integer
14427 Returns the value of the first General Purpose Tag associated to the
14428 currently tracked counters. See also src_get_gpt0.
14429
Cyril Bonté62ba8702014-04-22 23:52:25 +020014430sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014431sc0_gpc0_rate([<table>]) : integer
14432sc1_gpc0_rate([<table>]) : integer
14433sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014434 Returns the average increment rate of the first General Purpose Counter
14435 associated to the currently tracked counters. It reports the frequency
14436 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014437 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14438 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14439 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014440
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014441sc_gpc1_rate(<ctr>[,<table>]) : integer
14442sc0_gpc1_rate([<table>]) : integer
14443sc1_gpc1_rate([<table>]) : integer
14444sc2_gpc1_rate([<table>]) : integer
14445 Returns the average increment rate of the second General Purpose Counter
14446 associated to the currently tracked counters. It reports the frequency
14447 which the gpc1 counter was incremented over the configured period. See also
14448 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14449 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14450 be returned, as "gpc1" only holds the event count.
14451
Cyril Bonté62ba8702014-04-22 23:52:25 +020014452sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014453sc0_http_err_cnt([<table>]) : integer
14454sc1_http_err_cnt([<table>]) : integer
14455sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014456 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014457 counters. This includes the both request errors and 4xx error responses.
14458 See also src_http_err_cnt.
14459
Cyril Bonté62ba8702014-04-22 23:52:25 +020014460sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014461sc0_http_err_rate([<table>]) : integer
14462sc1_http_err_rate([<table>]) : integer
14463sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014464 Returns the average rate of HTTP errors from the currently tracked counters,
14465 measured in amount of errors over the period configured in the table. This
14466 includes the both request errors and 4xx error responses. See also
14467 src_http_err_rate.
14468
Cyril Bonté62ba8702014-04-22 23:52:25 +020014469sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014470sc0_http_req_cnt([<table>]) : integer
14471sc1_http_req_cnt([<table>]) : integer
14472sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014473 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014474 counters. This includes every started request, valid or not. See also
14475 src_http_req_cnt.
14476
Cyril Bonté62ba8702014-04-22 23:52:25 +020014477sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014478sc0_http_req_rate([<table>]) : integer
14479sc1_http_req_rate([<table>]) : integer
14480sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014481 Returns the average rate of HTTP requests from the currently tracked
14482 counters, measured in amount of requests over the period configured in
14483 the table. This includes every started request, valid or not. See also
14484 src_http_req_rate.
14485
Cyril Bonté62ba8702014-04-22 23:52:25 +020014486sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014487sc0_inc_gpc0([<table>]) : integer
14488sc1_inc_gpc0([<table>]) : integer
14489sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014490 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014491 tracked counters, and returns its new value. Before the first invocation,
14492 the stored value is zero, so first invocation will increase it to 1 and will
14493 return 1. This is typically used as a second ACL in an expression in order
14494 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014495
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014496 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014497 acl abuse sc0_http_req_rate gt 10
14498 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014499 tcp-request connection reject if abuse kill
14500
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014501sc_inc_gpc1(<ctr>[,<table>]) : integer
14502sc0_inc_gpc1([<table>]) : integer
14503sc1_inc_gpc1([<table>]) : integer
14504sc2_inc_gpc1([<table>]) : integer
14505 Increments the second General Purpose Counter associated to the currently
14506 tracked counters, and returns its new value. Before the first invocation,
14507 the stored value is zero, so first invocation will increase it to 1 and will
14508 return 1. This is typically used as a second ACL in an expression in order
14509 to mark a connection when a first ACL was verified.
14510
Cyril Bonté62ba8702014-04-22 23:52:25 +020014511sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014512sc0_kbytes_in([<table>]) : integer
14513sc1_kbytes_in([<table>]) : integer
14514sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014515 Returns the total amount of client-to-server data from the currently tracked
14516 counters, measured in kilobytes. The test is currently performed on 32-bit
14517 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014518
Cyril Bonté62ba8702014-04-22 23:52:25 +020014519sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014520sc0_kbytes_out([<table>]) : integer
14521sc1_kbytes_out([<table>]) : integer
14522sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014523 Returns the total amount of server-to-client data from the currently tracked
14524 counters, measured in kilobytes. The test is currently performed on 32-bit
14525 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014526
Cyril Bonté62ba8702014-04-22 23:52:25 +020014527sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014528sc0_sess_cnt([<table>]) : integer
14529sc1_sess_cnt([<table>]) : integer
14530sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014531 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014532 into sessions, which means that they were accepted by a "tcp-request
14533 connection" rule, from the currently tracked counters. A backend may count
14534 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014535 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014536 with the client. See also src_sess_cnt.
14537
Cyril Bonté62ba8702014-04-22 23:52:25 +020014538sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014539sc0_sess_rate([<table>]) : integer
14540sc1_sess_rate([<table>]) : integer
14541sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014542 Returns the average session rate from the currently tracked counters,
14543 measured in amount of sessions over the period configured in the table. A
14544 session is a connection that got past the early "tcp-request connection"
14545 rules. A backend may count more sessions than connections because each
14546 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014547 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014548
Cyril Bonté62ba8702014-04-22 23:52:25 +020014549sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014550sc0_tracked([<table>]) : boolean
14551sc1_tracked([<table>]) : boolean
14552sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014553 Returns true if the designated session counter is currently being tracked by
14554 the current session. This can be useful when deciding whether or not we want
14555 to set some values in a header passed to the server.
14556
Cyril Bonté62ba8702014-04-22 23:52:25 +020014557sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014558sc0_trackers([<table>]) : integer
14559sc1_trackers([<table>]) : integer
14560sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014561 Returns the current amount of concurrent connections tracking the same
14562 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014563 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014564 that it does not rely on any stored information but on the table's reference
14565 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014566 may sometimes be more suited for layer7 tracking. It can be used to tell a
14567 server how many concurrent connections there are from a given address for
14568 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014569
Willy Tarreau74ca5042013-06-11 23:12:07 +020014570so_id : integer
14571 Returns an integer containing the current listening socket's id. It is useful
14572 in frontends involving many "bind" lines, or to stick all users coming via a
14573 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014574
Willy Tarreau74ca5042013-06-11 23:12:07 +020014575src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014576 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014577 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14578 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14579 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014580 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14581 directive is used, it can be the address of a client behind another
14582 PROXY-protocol compatible component for all rule sets except
14583 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014584
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014585 Example:
14586 # add an HTTP header in requests with the originating address' country
14587 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14588
Willy Tarreau74ca5042013-06-11 23:12:07 +020014589src_bytes_in_rate([<table>]) : integer
14590 Returns the average bytes rate from the incoming connection's source address
14591 in the current proxy's stick-table or in the designated stick-table, measured
14592 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014593 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014594
Willy Tarreau74ca5042013-06-11 23:12:07 +020014595src_bytes_out_rate([<table>]) : integer
14596 Returns the average bytes rate to the incoming connection's source address in
14597 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014598 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014599 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014600
Willy Tarreau74ca5042013-06-11 23:12:07 +020014601src_clr_gpc0([<table>]) : integer
14602 Clears the first General Purpose Counter associated to the incoming
14603 connection's source address in the current proxy's stick-table or in the
14604 designated stick-table, and returns its previous value. If the address is not
14605 found, an entry is created and 0 is returned. This is typically used as a
14606 second ACL in an expression in order to mark a connection when a first ACL
14607 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014608
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014609 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014610 # block if 5 consecutive requests continue to come faster than 10 sess
14611 # per second, and reset the counter as soon as the traffic slows down.
14612 acl abuse src_http_req_rate gt 10
14613 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014614 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014615 tcp-request connection accept if !abuse save
14616 tcp-request connection reject if abuse kill
14617
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014618src_clr_gpc1([<table>]) : integer
14619 Clears the second General Purpose Counter associated to the incoming
14620 connection's source address in the current proxy's stick-table or in the
14621 designated stick-table, and returns its previous value. If the address is not
14622 found, an entry is created and 0 is returned. This is typically used as a
14623 second ACL in an expression in order to mark a connection when a first ACL
14624 was verified.
14625
Willy Tarreau74ca5042013-06-11 23:12:07 +020014626src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014627 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014628 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014629 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014630 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014631
Willy Tarreau74ca5042013-06-11 23:12:07 +020014632src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014633 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014634 current incoming connection's source address in the current proxy's
14635 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014636 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014637
Willy Tarreau74ca5042013-06-11 23:12:07 +020014638src_conn_rate([<table>]) : integer
14639 Returns the average connection rate from the incoming connection's source
14640 address in the current proxy's stick-table or in the designated stick-table,
14641 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014642 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014643
Willy Tarreau74ca5042013-06-11 23:12:07 +020014644src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014645 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014646 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014647 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014648 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014649
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014650src_get_gpc1([<table>]) : integer
14651 Returns the value of the second General Purpose Counter associated to the
14652 incoming connection's source address in the current proxy's stick-table or in
14653 the designated stick-table. If the address is not found, zero is returned.
14654 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14655
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014656src_get_gpt0([<table>]) : integer
14657 Returns the value of the first General Purpose Tag associated to the
14658 incoming connection's source address in the current proxy's stick-table or in
14659 the designated stick-table. If the address is not found, zero is returned.
14660 See also sc/sc0/sc1/sc2_get_gpt0.
14661
Willy Tarreau74ca5042013-06-11 23:12:07 +020014662src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014663 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014664 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014665 stick-table or in the designated stick-table. It reports the frequency
14666 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014667 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14668 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14669 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014670
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014671src_gpc1_rate([<table>]) : integer
14672 Returns the average increment rate of the second General Purpose Counter
14673 associated to the incoming connection's source address in the current proxy's
14674 stick-table or in the designated stick-table. It reports the frequency
14675 which the gpc1 counter was incremented over the configured period. See also
14676 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14677 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14678 be returned, as "gpc1" only holds the event count.
14679
Willy Tarreau74ca5042013-06-11 23:12:07 +020014680src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014681 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014682 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014683 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014684 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014685 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014686
Willy Tarreau74ca5042013-06-11 23:12:07 +020014687src_http_err_rate([<table>]) : integer
14688 Returns the average rate of HTTP errors from the incoming connection's source
14689 address in the current proxy's stick-table or in the designated stick-table,
14690 measured in amount of errors over the period configured in the table. This
14691 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014692 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014693
Willy Tarreau74ca5042013-06-11 23:12:07 +020014694src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014695 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014696 source address in the current proxy's stick-table or in the designated stick-
14697 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014698 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014699
Willy Tarreau74ca5042013-06-11 23:12:07 +020014700src_http_req_rate([<table>]) : integer
14701 Returns the average rate of HTTP requests from the incoming connection's
14702 source address in the current proxy's stick-table or in the designated stick-
14703 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014704 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014705 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014706
Willy Tarreau74ca5042013-06-11 23:12:07 +020014707src_inc_gpc0([<table>]) : integer
14708 Increments the first General Purpose Counter associated to the incoming
14709 connection's source address in the current proxy's stick-table or in the
14710 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014711 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014712 This is typically used as a second ACL in an expression in order to mark a
14713 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014714
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014715 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014716 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014717 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014718 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014719
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014720src_inc_gpc1([<table>]) : integer
14721 Increments the second General Purpose Counter associated to the incoming
14722 connection's source address in the current proxy's stick-table or in the
14723 designated stick-table, and returns its new value. If the address is not
14724 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14725 This is typically used as a second ACL in an expression in order to mark a
14726 connection when a first ACL was verified.
14727
Willy Tarreau16e01562016-08-09 16:46:18 +020014728src_is_local : boolean
14729 Returns true if the source address of the incoming connection is local to the
14730 system, or false if the address doesn't exist on the system, meaning that it
14731 comes from a remote machine. Note that UNIX addresses are considered local.
14732 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014733 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014734 note that the check involves a few system calls, so it's better to do it only
14735 once per connection.
14736
Willy Tarreau74ca5042013-06-11 23:12:07 +020014737src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014738 Returns the total amount of data received from the incoming connection's
14739 source address in the current proxy's stick-table or in the designated
14740 stick-table, measured in kilobytes. If the address is not found, zero is
14741 returned. The test is currently performed on 32-bit integers, which limits
14742 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014743
Willy Tarreau74ca5042013-06-11 23:12:07 +020014744src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014745 Returns the total amount of data sent to the incoming connection's source
14746 address in the current proxy's stick-table or in the designated stick-table,
14747 measured in kilobytes. If the address is not found, zero is returned. The
14748 test is currently performed on 32-bit integers, which limits values to 4
14749 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014750
Willy Tarreau74ca5042013-06-11 23:12:07 +020014751src_port : integer
14752 Returns an integer value corresponding to the TCP source port of the
14753 connection on the client side, which is the port the client connected from.
14754 Usage of this function is very limited as modern protocols do not care much
14755 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014756
Willy Tarreau74ca5042013-06-11 23:12:07 +020014757src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014758 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014759 connection's source IPv4 address in the current proxy's stick-table or in the
14760 designated stick-table, that were transformed into sessions, which means that
14761 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014762 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014763
Willy Tarreau74ca5042013-06-11 23:12:07 +020014764src_sess_rate([<table>]) : integer
14765 Returns the average session rate from the incoming connection's source
14766 address in the current proxy's stick-table or in the designated stick-table,
14767 measured in amount of sessions over the period configured in the table. A
14768 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014769 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014770
Willy Tarreau74ca5042013-06-11 23:12:07 +020014771src_updt_conn_cnt([<table>]) : integer
14772 Creates or updates the entry associated to the incoming connection's source
14773 address in the current proxy's stick-table or in the designated stick-table.
14774 This table must be configured to store the "conn_cnt" data type, otherwise
14775 the match will be ignored. The current count is incremented by one, and the
14776 expiration timer refreshed. The updated count is returned, so this match
14777 can't return zero. This was used to reject service abusers based on their
14778 source address. Note: it is recommended to use the more complete "track-sc*"
14779 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014780
14781 Example :
14782 # This frontend limits incoming SSH connections to 3 per 10 second for
14783 # each source address, and rejects excess connections until a 10 second
14784 # silence is observed. At most 20 addresses are tracked.
14785 listen ssh
14786 bind :22
14787 mode tcp
14788 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014789 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014790 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014791 server local 127.0.0.1:22
14792
Willy Tarreau74ca5042013-06-11 23:12:07 +020014793srv_id : integer
14794 Returns an integer containing the server's id when processing the response.
14795 While it's almost only used with ACLs, it may be used for logging or
14796 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014797
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200147987.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014799----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801The layer 5 usually describes just the session layer which in haproxy is
14802closest to the session once all the connection handshakes are finished, but
14803when no content is yet made available. The fetch methods described here are
14804usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014805future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014806
Ben Shillitof25e8e52016-12-02 14:25:37 +00001480751d.all(<prop>[,<prop>*]) : string
14808 Returns values for the properties requested as a string, where values are
14809 separated by the delimiter specified with "51degrees-property-separator".
14810 The device is identified using all the important HTTP headers from the
14811 request. The function can be passed up to five property names, and if a
14812 property name can't be found, the value "NoData" is returned.
14813
14814 Example :
14815 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14816 # containing the three properties requested using all relevant headers from
14817 # the request.
14818 frontend http-in
14819 bind *:8081
14820 default_backend servers
14821 http-request set-header X-51D-DeviceTypeMobileTablet \
14822 %[51d.all(DeviceType,IsMobile,IsTablet)]
14823
Emeric Brun645ae792014-04-30 14:21:06 +020014824ssl_bc : boolean
14825 Returns true when the back connection was made via an SSL/TLS transport
14826 layer and is locally deciphered. This means the outgoing connection was made
14827 other a server with the "ssl" option.
14828
14829ssl_bc_alg_keysize : integer
14830 Returns the symmetric cipher key size supported in bits when the outgoing
14831 connection was made over an SSL/TLS transport layer.
14832
14833ssl_bc_cipher : string
14834 Returns the name of the used cipher when the outgoing connection was made
14835 over an SSL/TLS transport layer.
14836
Emeric Brun74f7ffa2018-02-19 16:14:12 +010014837ssl_bc_is_resumed : boolean
14838 Returns true when the back connection was made over an SSL/TLS transport
14839 layer and the newly created SSL session was resumed using a cached
14840 session or a TLS ticket.
14841
Emeric Brun645ae792014-04-30 14:21:06 +020014842ssl_bc_protocol : string
14843 Returns the name of the used protocol when the outgoing connection was made
14844 over an SSL/TLS transport layer.
14845
Emeric Brunb73a9b02014-04-30 18:49:19 +020014846ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020014847 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014848 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14849 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020014850
14851ssl_bc_session_id : binary
14852 Returns the SSL ID of the back connection when the outgoing connection was
14853 made over an SSL/TLS transport layer. It is useful to log if we want to know
14854 if session was reused or not.
14855
Patrick Hemmere0275472018-04-28 19:15:51 -040014856ssl_bc_session_key : binary
14857 Returns the SSL session master key of the back connection when the outgoing
14858 connection was made over an SSL/TLS transport layer. It is useful to decrypt
14859 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
14860 BoringSSL.
14861
Emeric Brun645ae792014-04-30 14:21:06 +020014862ssl_bc_use_keysize : integer
14863 Returns the symmetric cipher key size used in bits when the outgoing
14864 connection was made over an SSL/TLS transport layer.
14865
Willy Tarreau74ca5042013-06-11 23:12:07 +020014866ssl_c_ca_err : integer
14867 When the incoming connection was made over an SSL/TLS transport layer,
14868 returns the ID of the first error detected during verification of the client
14869 certificate at depth > 0, or 0 if no error was encountered during this
14870 verification process. Please refer to your SSL library's documentation to
14871 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020014872
Willy Tarreau74ca5042013-06-11 23:12:07 +020014873ssl_c_ca_err_depth : integer
14874 When the incoming connection was made over an SSL/TLS transport layer,
14875 returns the depth in the CA chain of the first error detected during the
14876 verification of the client certificate. If no error is encountered, 0 is
14877 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014878
Emeric Brun43e79582014-10-29 19:03:26 +010014879ssl_c_der : binary
14880 Returns the DER formatted certificate presented by the client when the
14881 incoming connection was made over an SSL/TLS transport layer. When used for
14882 an ACL, the value(s) to match against can be passed in hexadecimal form.
14883
Willy Tarreau74ca5042013-06-11 23:12:07 +020014884ssl_c_err : integer
14885 When the incoming connection was made over an SSL/TLS transport layer,
14886 returns the ID of the first error detected during verification at depth 0, or
14887 0 if no error was encountered during this verification process. Please refer
14888 to your SSL library's documentation to find the exhaustive list of error
14889 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020014890
Willy Tarreau74ca5042013-06-11 23:12:07 +020014891ssl_c_i_dn([<entry>[,<occ>]]) : string
14892 When the incoming connection was made over an SSL/TLS transport layer,
14893 returns the full distinguished name of the issuer of the certificate
14894 presented by the client when no <entry> is specified, or the value of the
14895 first given entry found from the beginning of the DN. If a positive/negative
14896 occurrence number is specified as the optional second argument, it returns
14897 the value of the nth given entry value from the beginning/end of the DN.
14898 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
14899 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020014900
Willy Tarreau74ca5042013-06-11 23:12:07 +020014901ssl_c_key_alg : string
14902 Returns the name of the algorithm used to generate the key of the certificate
14903 presented by the client when the incoming connection was made over an SSL/TLS
14904 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020014905
Willy Tarreau74ca5042013-06-11 23:12:07 +020014906ssl_c_notafter : string
14907 Returns the end date presented by the client as a formatted string
14908 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14909 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020014910
Willy Tarreau74ca5042013-06-11 23:12:07 +020014911ssl_c_notbefore : string
14912 Returns the start date presented by the client as a formatted string
14913 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14914 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014915
Willy Tarreau74ca5042013-06-11 23:12:07 +020014916ssl_c_s_dn([<entry>[,<occ>]]) : string
14917 When the incoming connection was made over an SSL/TLS transport layer,
14918 returns the full distinguished name of the subject of the certificate
14919 presented by the client when no <entry> is specified, or the value of the
14920 first given entry found from the beginning of the DN. If a positive/negative
14921 occurrence number is specified as the optional second argument, it returns
14922 the value of the nth given entry value from the beginning/end of the DN.
14923 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
14924 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014925
Willy Tarreau74ca5042013-06-11 23:12:07 +020014926ssl_c_serial : binary
14927 Returns the serial of the certificate presented by the client when the
14928 incoming connection was made over an SSL/TLS transport layer. When used for
14929 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014930
Willy Tarreau74ca5042013-06-11 23:12:07 +020014931ssl_c_sha1 : binary
14932 Returns the SHA-1 fingerprint of the certificate presented by the client when
14933 the incoming connection was made over an SSL/TLS transport layer. This can be
14934 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014935 Note that the output is binary, so if you want to pass that signature to the
14936 server, you need to encode it in hex or base64, such as in the example below:
14937
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014938 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014939 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020014940
Willy Tarreau74ca5042013-06-11 23:12:07 +020014941ssl_c_sig_alg : string
14942 Returns the name of the algorithm used to sign the certificate presented by
14943 the client when the incoming connection was made over an SSL/TLS transport
14944 layer.
Emeric Brun87855892012-10-17 17:39:35 +020014945
Willy Tarreau74ca5042013-06-11 23:12:07 +020014946ssl_c_used : boolean
14947 Returns true if current SSL session uses a client certificate even if current
14948 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020014949
Willy Tarreau74ca5042013-06-11 23:12:07 +020014950ssl_c_verify : integer
14951 Returns the verify result error ID when the incoming connection was made over
14952 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
14953 refer to your SSL library's documentation for an exhaustive list of error
14954 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020014955
Willy Tarreau74ca5042013-06-11 23:12:07 +020014956ssl_c_version : integer
14957 Returns the version of the certificate presented by the client when the
14958 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020014959
Emeric Brun43e79582014-10-29 19:03:26 +010014960ssl_f_der : binary
14961 Returns the DER formatted certificate presented by the frontend when the
14962 incoming connection was made over an SSL/TLS transport layer. When used for
14963 an ACL, the value(s) to match against can be passed in hexadecimal form.
14964
Willy Tarreau74ca5042013-06-11 23:12:07 +020014965ssl_f_i_dn([<entry>[,<occ>]]) : string
14966 When the incoming connection was made over an SSL/TLS transport layer,
14967 returns the full distinguished name of the issuer of the certificate
14968 presented by the frontend when no <entry> is specified, or the value of the
14969 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020014970 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020014971 the value of the nth given entry value from the beginning/end of the DN.
14972 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
14973 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020014974
Willy Tarreau74ca5042013-06-11 23:12:07 +020014975ssl_f_key_alg : string
14976 Returns the name of the algorithm used to generate the key of the certificate
14977 presented by the frontend when the incoming connection was made over an
14978 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014979
Willy Tarreau74ca5042013-06-11 23:12:07 +020014980ssl_f_notafter : string
14981 Returns the end date presented by the frontend as a formatted string
14982 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14983 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014984
Willy Tarreau74ca5042013-06-11 23:12:07 +020014985ssl_f_notbefore : string
14986 Returns the start date presented by the frontend as a formatted string
14987 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14988 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020014989
Willy Tarreau74ca5042013-06-11 23:12:07 +020014990ssl_f_s_dn([<entry>[,<occ>]]) : string
14991 When the incoming connection was made over an SSL/TLS transport layer,
14992 returns the full distinguished name of the subject of the certificate
14993 presented by the frontend when no <entry> is specified, or the value of the
14994 first given entry found from the beginning of the DN. If a positive/negative
14995 occurrence number is specified as the optional second argument, it returns
14996 the value of the nth given entry value from the beginning/end of the DN.
14997 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
14998 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020014999
Willy Tarreau74ca5042013-06-11 23:12:07 +020015000ssl_f_serial : binary
15001 Returns the serial of the certificate presented by the frontend when the
15002 incoming connection was made over an SSL/TLS transport layer. When used for
15003 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015004
Emeric Brun55f4fa82014-04-30 17:11:25 +020015005ssl_f_sha1 : binary
15006 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15007 when the incoming connection was made over an SSL/TLS transport layer. This
15008 can be used to know which certificate was chosen using SNI.
15009
Willy Tarreau74ca5042013-06-11 23:12:07 +020015010ssl_f_sig_alg : string
15011 Returns the name of the algorithm used to sign the certificate presented by
15012 the frontend when the incoming connection was made over an SSL/TLS transport
15013 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015014
Willy Tarreau74ca5042013-06-11 23:12:07 +020015015ssl_f_version : integer
15016 Returns the version of the certificate presented by the frontend when the
15017 incoming connection was made over an SSL/TLS transport layer.
15018
15019ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015020 Returns true when the front connection was made via an SSL/TLS transport
15021 layer and is locally deciphered. This means it has matched a socket declared
15022 with a "bind" line having the "ssl" option.
15023
Willy Tarreau74ca5042013-06-11 23:12:07 +020015024 Example :
15025 # This passes "X-Proto: https" to servers when client connects over SSL
15026 listen http-https
15027 bind :80
15028 bind :443 ssl crt /etc/haproxy.pem
15029 http-request add-header X-Proto https if { ssl_fc }
15030
15031ssl_fc_alg_keysize : integer
15032 Returns the symmetric cipher key size supported in bits when the incoming
15033 connection was made over an SSL/TLS transport layer.
15034
15035ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015036 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015037 incoming connection made via a TLS transport layer and locally deciphered by
15038 haproxy. The result is a string containing the protocol name advertised by
15039 the client. The SSL library must have been built with support for TLS
15040 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15041 not advertised unless the "alpn" keyword on the "bind" line specifies a
15042 protocol list. Also, nothing forces the client to pick a protocol from this
15043 list, any other one may be requested. The TLS ALPN extension is meant to
15044 replace the TLS NPN extension. See also "ssl_fc_npn".
15045
Willy Tarreau74ca5042013-06-11 23:12:07 +020015046ssl_fc_cipher : string
15047 Returns the name of the used cipher when the incoming connection was made
15048 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015049
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015050ssl_fc_cipherlist_bin : binary
15051 Returns the binary form of the client hello cipher list. The maximum returned
15052 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015053 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015054
15055ssl_fc_cipherlist_hex : string
15056 Returns the binary form of the client hello cipher list encoded as
15057 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015058 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015059
15060ssl_fc_cipherlist_str : string
15061 Returns the decoded text form of the client hello cipher list. The maximum
15062 number of ciphers returned is according with the value of
15063 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015064 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015065 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015066
15067ssl_fc_cipherlist_xxh : integer
15068 Returns a xxh64 of the cipher list. This hash can be return only is the value
15069 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015070 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015071
Willy Tarreau74ca5042013-06-11 23:12:07 +020015072ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015073 Returns true if a client certificate is present in an incoming connection over
15074 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015075 Note: on SSL session resumption with Session ID or TLS ticket, client
15076 certificate is not present in the current connection but may be retrieved
15077 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15078 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015079
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015080ssl_fc_has_early : boolean
15081 Returns true if early data were sent, and the handshake didn't happen yet. As
15082 it has security implications, it is useful to be able to refuse those, or
15083 wait until the handshake happened.
15084
Willy Tarreau74ca5042013-06-11 23:12:07 +020015085ssl_fc_has_sni : boolean
15086 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015087 in an incoming connection was made over an SSL/TLS transport layer. Returns
15088 true when the incoming connection presents a TLS SNI field. This requires
15089 that the SSL library is build with support for TLS extensions enabled (check
15090 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015091
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015092ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015093 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015094 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15095 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015096
Willy Tarreau74ca5042013-06-11 23:12:07 +020015097ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015098 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015099 made via a TLS transport layer and locally deciphered by haproxy. The result
15100 is a string containing the protocol name advertised by the client. The SSL
15101 library must have been built with support for TLS extensions enabled (check
15102 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15103 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15104 forces the client to pick a protocol from this list, any other one may be
15105 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015106
Willy Tarreau74ca5042013-06-11 23:12:07 +020015107ssl_fc_protocol : string
15108 Returns the name of the used protocol when the incoming connection was made
15109 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015110
Emeric Brunb73a9b02014-04-30 18:49:19 +020015111ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015112 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015113 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15114 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015115
Willy Tarreau74ca5042013-06-11 23:12:07 +020015116ssl_fc_session_id : binary
15117 Returns the SSL ID of the front connection when the incoming connection was
15118 made over an SSL/TLS transport layer. It is useful to stick a given client to
15119 a server. It is important to note that some browsers refresh their session ID
15120 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015121
Patrick Hemmere0275472018-04-28 19:15:51 -040015122ssl_fc_session_key : binary
15123 Returns the SSL session master key of the front connection when the incoming
15124 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15125 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15126 BoringSSL.
15127
15128
Willy Tarreau74ca5042013-06-11 23:12:07 +020015129ssl_fc_sni : string
15130 This extracts the Server Name Indication TLS extension (SNI) field from an
15131 incoming connection made via an SSL/TLS transport layer and locally
15132 deciphered by haproxy. The result (when present) typically is a string
15133 matching the HTTPS host name (253 chars or less). The SSL library must have
15134 been built with support for TLS extensions enabled (check haproxy -vv).
15135
15136 This fetch is different from "req_ssl_sni" above in that it applies to the
15137 connection being deciphered by haproxy and not to SSL contents being blindly
15138 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015139 requires that the SSL library is build with support for TLS extensions
15140 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015141
Willy Tarreau74ca5042013-06-11 23:12:07 +020015142 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015143 ssl_fc_sni_end : suffix match
15144 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015145
Willy Tarreau74ca5042013-06-11 23:12:07 +020015146ssl_fc_use_keysize : integer
15147 Returns the symmetric cipher key size used in bits when the incoming
15148 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015149
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015150
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200151517.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015152------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015153
Willy Tarreau74ca5042013-06-11 23:12:07 +020015154Fetching samples from buffer contents is a bit different from the previous
15155sample fetches above because the sampled data are ephemeral. These data can
15156only be used when they're available and will be lost when they're forwarded.
15157For this reason, samples fetched from buffer contents during a request cannot
15158be used in a response for example. Even while the data are being fetched, they
15159can change. Sometimes it is necessary to set some delays or combine multiple
15160sample fetch methods to ensure that the expected data are complete and usable,
15161for example through TCP request content inspection. Please see the "tcp-request
15162content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015163
Willy Tarreau74ca5042013-06-11 23:12:07 +020015164payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015165 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015166 "stick on", "stick match"), and for "res.payload" when used in the context of
15167 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015168
Willy Tarreau74ca5042013-06-11 23:12:07 +020015169payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15170 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015171 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015172 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015173
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015174req.hdrs : string
15175 Returns the current request headers as string including the last empty line
15176 separating headers from the request body. The last empty line can be used to
15177 detect a truncated header block. This sample fetch is useful for some SPOE
15178 headers analyzers and for advanced logging.
15179
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015180req.hdrs_bin : binary
15181 Returns the current request headers contained in preparsed binary form. This
15182 is useful for offloading some processing with SPOE. Each string is described
15183 by a length followed by the number of bytes indicated in the length. The
15184 length is represented using the variable integer encoding detailed in the
15185 SPOE documentation. The end of the list is marked by a couple of empty header
15186 names and values (length of 0 for both).
15187
15188 *(<str:header-name><str:header-value>)<empty string><empty string>
15189
15190 int: refer to the SPOE documentation for the encoding
15191 str: <int:length><bytes>
15192
Willy Tarreau74ca5042013-06-11 23:12:07 +020015193req.len : integer
15194req_len : integer (deprecated)
15195 Returns an integer value corresponding to the number of bytes present in the
15196 request buffer. This is mostly used in ACL. It is important to understand
15197 that this test does not return false as long as the buffer is changing. This
15198 means that a check with equality to zero will almost always immediately match
15199 at the beginning of the session, while a test for more data will wait for
15200 that data to come in and return false only when haproxy is certain that no
15201 more data will come in. This test was designed to be used with TCP request
15202 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015203
Willy Tarreau74ca5042013-06-11 23:12:07 +020015204req.payload(<offset>,<length>) : binary
15205 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015206 in the request buffer. As a special case, if the <length> argument is zero,
15207 the the whole buffer from <offset> to the end is extracted. This can be used
15208 with ACLs in order to check for the presence of some content in a buffer at
15209 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015210
Willy Tarreau74ca5042013-06-11 23:12:07 +020015211 ACL alternatives :
15212 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015213
Willy Tarreau74ca5042013-06-11 23:12:07 +020015214req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15215 This extracts a binary block whose size is specified at <offset1> for <length>
15216 bytes, and which starts at <offset2> if specified or just after the length in
15217 the request buffer. The <offset2> parameter also supports relative offsets if
15218 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015219
Willy Tarreau74ca5042013-06-11 23:12:07 +020015220 ACL alternatives :
15221 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015222
Willy Tarreau74ca5042013-06-11 23:12:07 +020015223 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015224
Willy Tarreau74ca5042013-06-11 23:12:07 +020015225req.proto_http : boolean
15226req_proto_http : boolean (deprecated)
15227 Returns true when data in the request buffer look like HTTP and correctly
15228 parses as such. It is the same parser as the common HTTP request parser which
15229 is used so there should be no surprises. The test does not match until the
15230 request is complete, failed or timed out. This test may be used to report the
15231 protocol in TCP logs, but the biggest use is to block TCP request analysis
15232 until a complete HTTP request is present in the buffer, for example to track
15233 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015234
Willy Tarreau74ca5042013-06-11 23:12:07 +020015235 Example:
15236 # track request counts per "base" (concatenation of Host+URL)
15237 tcp-request inspect-delay 10s
15238 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015239 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015240
Willy Tarreau74ca5042013-06-11 23:12:07 +020015241req.rdp_cookie([<name>]) : string
15242rdp_cookie([<name>]) : string (deprecated)
15243 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15244 <name>, or any cookie if unspecified. The parser only checks for the first
15245 cookie, as illustrated in the RDP protocol specification. The cookie name is
15246 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15247 contain the user name of the client connecting to the server if properly
15248 configured on the client. The "MSTSHASH" cookie is often used as well for
15249 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015250
Willy Tarreau74ca5042013-06-11 23:12:07 +020015251 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15252 used and thus the distribution of clients to backend servers is not linked to
15253 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15254 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15255 distribution of clients to backend servers than the hash used by "balance
15256 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015257
Willy Tarreau74ca5042013-06-11 23:12:07 +020015258 ACL derivatives :
15259 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015260
Willy Tarreau74ca5042013-06-11 23:12:07 +020015261 Example :
15262 listen tse-farm
15263 bind 0.0.0.0:3389
15264 # wait up to 5s for an RDP cookie in the request
15265 tcp-request inspect-delay 5s
15266 tcp-request content accept if RDP_COOKIE
15267 # apply RDP cookie persistence
15268 persist rdp-cookie
15269 # Persist based on the mstshash cookie
15270 # This is only useful makes sense if
15271 # balance rdp-cookie is not used
15272 stick-table type string size 204800
15273 stick on req.rdp_cookie(mstshash)
15274 server srv1 1.1.1.1:3389
15275 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015276
Willy Tarreau74ca5042013-06-11 23:12:07 +020015277 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15278 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015279
Willy Tarreau74ca5042013-06-11 23:12:07 +020015280req.rdp_cookie_cnt([name]) : integer
15281rdp_cookie_cnt([name]) : integer (deprecated)
15282 Tries to parse the request buffer as RDP protocol, then returns an integer
15283 corresponding to the number of RDP cookies found. If an optional cookie name
15284 is passed, only cookies matching this name are considered. This is mostly
15285 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015286
Willy Tarreau74ca5042013-06-11 23:12:07 +020015287 ACL derivatives :
15288 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015289
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015290req.ssl_ec_ext : boolean
15291 Returns a boolean identifying if client sent the Supported Elliptic Curves
15292 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015293 message. This can be used to present ECC compatible clients with EC
15294 certificate and to use RSA for all others, on the same IP address. Note that
15295 this only applies to raw contents found in the request buffer and not to
15296 contents deciphered via an SSL data layer, so this will not work with "bind"
15297 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015298
Willy Tarreau74ca5042013-06-11 23:12:07 +020015299req.ssl_hello_type : integer
15300req_ssl_hello_type : integer (deprecated)
15301 Returns an integer value containing the type of the SSL hello message found
15302 in the request buffer if the buffer contains data that parse as a complete
15303 SSL (v3 or superior) client hello message. Note that this only applies to raw
15304 contents found in the request buffer and not to contents deciphered via an
15305 SSL data layer, so this will not work with "bind" lines having the "ssl"
15306 option. This is mostly used in ACL to detect presence of an SSL hello message
15307 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015308
Willy Tarreau74ca5042013-06-11 23:12:07 +020015309req.ssl_sni : string
15310req_ssl_sni : string (deprecated)
15311 Returns a string containing the value of the Server Name TLS extension sent
15312 by a client in a TLS stream passing through the request buffer if the buffer
15313 contains data that parse as a complete SSL (v3 or superior) client hello
15314 message. Note that this only applies to raw contents found in the request
15315 buffer and not to contents deciphered via an SSL data layer, so this will not
15316 work with "bind" lines having the "ssl" option. SNI normally contains the
15317 name of the host the client tries to connect to (for recent browsers). SNI is
15318 useful for allowing or denying access to certain hosts when SSL/TLS is used
15319 by the client. This test was designed to be used with TCP request content
15320 inspection. If content switching is needed, it is recommended to first wait
15321 for a complete client hello (type 1), like in the example below. See also
15322 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015323
Willy Tarreau74ca5042013-06-11 23:12:07 +020015324 ACL derivatives :
15325 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015326
Willy Tarreau74ca5042013-06-11 23:12:07 +020015327 Examples :
15328 # Wait for a client hello for at most 5 seconds
15329 tcp-request inspect-delay 5s
15330 tcp-request content accept if { req_ssl_hello_type 1 }
15331 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15332 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015333
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015334req.ssl_st_ext : integer
15335 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15336 Returns 1 if the client sent SessionTicket TLS Extension
15337 Returns 2 if the client also sent non-zero length TLS SessionTicket
15338 Note that this only applies to raw contents found in the request buffer and
15339 not to contents deciphered via an SSL data layer, so this will not work with
15340 "bind" lines having the "ssl" option. This can for example be used to detect
15341 whether the client sent a SessionTicket or not and stick it accordingly, if
15342 no SessionTicket then stick on SessionID or don't stick as there's no server
15343 side state is there when SessionTickets are in use.
15344
Willy Tarreau74ca5042013-06-11 23:12:07 +020015345req.ssl_ver : integer
15346req_ssl_ver : integer (deprecated)
15347 Returns an integer value containing the version of the SSL/TLS protocol of a
15348 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15349 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15350 composed of the major version multiplied by 65536, added to the minor
15351 version. Note that this only applies to raw contents found in the request
15352 buffer and not to contents deciphered via an SSL data layer, so this will not
15353 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015354 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015355 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015356
Willy Tarreau74ca5042013-06-11 23:12:07 +020015357 ACL derivatives :
15358 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015359
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015360res.len : integer
15361 Returns an integer value corresponding to the number of bytes present in the
15362 response buffer. This is mostly used in ACL. It is important to understand
15363 that this test does not return false as long as the buffer is changing. This
15364 means that a check with equality to zero will almost always immediately match
15365 at the beginning of the session, while a test for more data will wait for
15366 that data to come in and return false only when haproxy is certain that no
15367 more data will come in. This test was designed to be used with TCP response
15368 content inspection.
15369
Willy Tarreau74ca5042013-06-11 23:12:07 +020015370res.payload(<offset>,<length>) : binary
15371 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015372 in the response buffer. As a special case, if the <length> argument is zero,
15373 the the whole buffer from <offset> to the end is extracted. This can be used
15374 with ACLs in order to check for the presence of some content in a buffer at
15375 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015376
Willy Tarreau74ca5042013-06-11 23:12:07 +020015377res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15378 This extracts a binary block whose size is specified at <offset1> for <length>
15379 bytes, and which starts at <offset2> if specified or just after the length in
15380 the response buffer. The <offset2> parameter also supports relative offsets
15381 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015382
Willy Tarreau74ca5042013-06-11 23:12:07 +020015383 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015384
Willy Tarreau971f7b62015-09-29 14:06:59 +020015385res.ssl_hello_type : integer
15386rep_ssl_hello_type : integer (deprecated)
15387 Returns an integer value containing the type of the SSL hello message found
15388 in the response buffer if the buffer contains data that parses as a complete
15389 SSL (v3 or superior) hello message. Note that this only applies to raw
15390 contents found in the response buffer and not to contents deciphered via an
15391 SSL data layer, so this will not work with "server" lines having the "ssl"
15392 option. This is mostly used in ACL to detect presence of an SSL hello message
15393 that is supposed to contain an SSL session ID usable for stickiness.
15394
Willy Tarreau74ca5042013-06-11 23:12:07 +020015395wait_end : boolean
15396 This fetch either returns true when the inspection period is over, or does
15397 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015398 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015399 actions, such as a delayed reject for some special addresses. Since it either
15400 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015401 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015402 "WAIT_END" is always usable without prior declaration. This test was designed
15403 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015404
Willy Tarreau74ca5042013-06-11 23:12:07 +020015405 Examples :
15406 # delay every incoming request by 2 seconds
15407 tcp-request inspect-delay 2s
15408 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015409
Willy Tarreau74ca5042013-06-11 23:12:07 +020015410 # don't immediately tell bad guys they are rejected
15411 tcp-request inspect-delay 10s
15412 acl goodguys src 10.0.0.0/24
15413 acl badguys src 10.0.1.0/24
15414 tcp-request content accept if goodguys
15415 tcp-request content reject if badguys WAIT_END
15416 tcp-request content reject
15417
15418
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200154197.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015420--------------------------------------
15421
15422It is possible to fetch samples from HTTP contents, requests and responses.
15423This application layer is also called layer 7. It is only possible to fetch the
15424data in this section when a full HTTP request or response has been parsed from
15425its respective request or response buffer. This is always the case with all
15426HTTP specific rules and for sections running with "mode http". When using TCP
15427content inspection, it may be necessary to support an inspection delay in order
15428to let the request or response come in first. These fetches may require a bit
15429more CPU resources than the layer 4 ones, but not much since the request and
15430response are indexed.
15431
15432base : string
15433 This returns the concatenation of the first Host header and the path part of
15434 the request, which starts at the first slash and ends before the question
15435 mark. It can be useful in virtual hosted environments to detect URL abuses as
15436 well as to improve shared caches efficiency. Using this with a limited size
15437 stick table also allows one to collect statistics about most commonly
15438 requested objects by host/path. With ACLs it can allow simple content
15439 switching rules involving the host and the path at the same time, such as
15440 "www.example.com/favicon.ico". See also "path" and "uri".
15441
15442 ACL derivatives :
15443 base : exact string match
15444 base_beg : prefix match
15445 base_dir : subdir match
15446 base_dom : domain match
15447 base_end : suffix match
15448 base_len : length match
15449 base_reg : regex match
15450 base_sub : substring match
15451
15452base32 : integer
15453 This returns a 32-bit hash of the value returned by the "base" fetch method
15454 above. This is useful to track per-URL activity on high traffic sites without
15455 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015456 memory. The output type is an unsigned integer. The hash function used is
15457 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15458 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015459
15460base32+src : binary
15461 This returns the concatenation of the base32 fetch above and the src fetch
15462 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15463 depending on the source address family. This can be used to track per-IP,
15464 per-URL counters.
15465
William Lallemand65ad6e12014-01-31 15:08:02 +010015466capture.req.hdr(<idx>) : string
15467 This extracts the content of the header captured by the "capture request
15468 header", idx is the position of the capture keyword in the configuration.
15469 The first entry is an index of 0. See also: "capture request header".
15470
15471capture.req.method : string
15472 This extracts the METHOD of an HTTP request. It can be used in both request
15473 and response. Unlike "method", it can be used in both request and response
15474 because it's allocated.
15475
15476capture.req.uri : string
15477 This extracts the request's URI, which starts at the first slash and ends
15478 before the first space in the request (without the host part). Unlike "path"
15479 and "url", it can be used in both request and response because it's
15480 allocated.
15481
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015482capture.req.ver : string
15483 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15484 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15485 logs because it relies on a persistent flag.
15486
William Lallemand65ad6e12014-01-31 15:08:02 +010015487capture.res.hdr(<idx>) : string
15488 This extracts the content of the header captured by the "capture response
15489 header", idx is the position of the capture keyword in the configuration.
15490 The first entry is an index of 0.
15491 See also: "capture response header"
15492
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015493capture.res.ver : string
15494 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15495 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15496 persistent flag.
15497
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015498req.body : binary
15499 This returns the HTTP request's available body as a block of data. It
15500 requires that the request body has been buffered made available using
15501 "option http-buffer-request". In case of chunked-encoded body, currently only
15502 the first chunk is analyzed.
15503
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015504req.body_param([<name>) : string
15505 This fetch assumes that the body of the POST request is url-encoded. The user
15506 can check if the "content-type" contains the value
15507 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15508 parameter <name> in the body, which ends before '&'. The parameter name is
15509 case-sensitive. If no name is given, any parameter will match, and the first
15510 one will be returned. The result is a string corresponding to the value of the
15511 parameter <name> as presented in the request body (no URL decoding is
15512 performed). Note that the ACL version of this fetch iterates over multiple
15513 parameters and will iteratively report all parameters values if no name is
15514 given.
15515
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015516req.body_len : integer
15517 This returns the length of the HTTP request's available body in bytes. It may
15518 be lower than the advertised length if the body is larger than the buffer. It
15519 requires that the request body has been buffered made available using
15520 "option http-buffer-request".
15521
15522req.body_size : integer
15523 This returns the advertised length of the HTTP request's body in bytes. It
15524 will represent the advertised Content-Length header, or the size of the first
15525 chunk in case of chunked encoding. In order to parse the chunks, it requires
15526 that the request body has been buffered made available using
15527 "option http-buffer-request".
15528
Willy Tarreau74ca5042013-06-11 23:12:07 +020015529req.cook([<name>]) : string
15530cook([<name>]) : string (deprecated)
15531 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15532 header line from the request, and returns its value as string. If no name is
15533 specified, the first cookie value is returned. When used with ACLs, all
15534 matching cookies are evaluated. Spaces around the name and the value are
15535 ignored as requested by the Cookie header specification (RFC6265). The cookie
15536 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15537 well return an empty value if it is present. Use the "found" match to detect
15538 presence. Use the res.cook() variant for response cookies sent by the server.
15539
15540 ACL derivatives :
15541 cook([<name>]) : exact string match
15542 cook_beg([<name>]) : prefix match
15543 cook_dir([<name>]) : subdir match
15544 cook_dom([<name>]) : domain match
15545 cook_end([<name>]) : suffix match
15546 cook_len([<name>]) : length match
15547 cook_reg([<name>]) : regex match
15548 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015549
Willy Tarreau74ca5042013-06-11 23:12:07 +020015550req.cook_cnt([<name>]) : integer
15551cook_cnt([<name>]) : integer (deprecated)
15552 Returns an integer value representing the number of occurrences of the cookie
15553 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015554
Willy Tarreau74ca5042013-06-11 23:12:07 +020015555req.cook_val([<name>]) : integer
15556cook_val([<name>]) : integer (deprecated)
15557 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15558 header line from the request, and converts its value to an integer which is
15559 returned. If no name is specified, the first cookie value is returned. When
15560 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015561
Willy Tarreau74ca5042013-06-11 23:12:07 +020015562cookie([<name>]) : string (deprecated)
15563 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15564 header line from the request, or a "Set-Cookie" header from the response, and
15565 returns its value as a string. A typical use is to get multiple clients
15566 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015567 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015568 multi-peer synchronization and state keeping across restarts. If no name is
15569 specified, the first cookie value is returned. This fetch should not be used
15570 anymore and should be replaced by req.cook() or res.cook() instead as it
15571 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015572
Willy Tarreau74ca5042013-06-11 23:12:07 +020015573hdr([<name>[,<occ>]]) : string
15574 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15575 used on responses. Please refer to these respective fetches for more details.
15576 In case of doubt about the fetch direction, please use the explicit ones.
15577 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015578 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015579
Willy Tarreau74ca5042013-06-11 23:12:07 +020015580req.fhdr(<name>[,<occ>]) : string
15581 This extracts the last occurrence of header <name> in an HTTP request. When
15582 used from an ACL, all occurrences are iterated over until a match is found.
15583 Optionally, a specific occurrence might be specified as a position number.
15584 Positive values indicate a position from the first occurrence, with 1 being
15585 the first one. Negative values indicate positions relative to the last one,
15586 with -1 being the last one. It differs from req.hdr() in that any commas
15587 present in the value are returned and are not used as delimiters. This is
15588 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015589
Willy Tarreau74ca5042013-06-11 23:12:07 +020015590req.fhdr_cnt([<name>]) : integer
15591 Returns an integer value representing the number of occurrences of request
15592 header field name <name>, or the total number of header fields if <name> is
15593 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15594 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015595
Willy Tarreau74ca5042013-06-11 23:12:07 +020015596req.hdr([<name>[,<occ>]]) : string
15597 This extracts the last occurrence of header <name> in an HTTP request. When
15598 used from an ACL, all occurrences are iterated over until a match is found.
15599 Optionally, a specific occurrence might be specified as a position number.
15600 Positive values indicate a position from the first occurrence, with 1 being
15601 the first one. Negative values indicate positions relative to the last one,
15602 with -1 being the last one. A typical use is with the X-Forwarded-For header
15603 once converted to IP, associated with an IP stick-table. The function
15604 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015605 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015606 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015607 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015608
Willy Tarreau74ca5042013-06-11 23:12:07 +020015609 ACL derivatives :
15610 hdr([<name>[,<occ>]]) : exact string match
15611 hdr_beg([<name>[,<occ>]]) : prefix match
15612 hdr_dir([<name>[,<occ>]]) : subdir match
15613 hdr_dom([<name>[,<occ>]]) : domain match
15614 hdr_end([<name>[,<occ>]]) : suffix match
15615 hdr_len([<name>[,<occ>]]) : length match
15616 hdr_reg([<name>[,<occ>]]) : regex match
15617 hdr_sub([<name>[,<occ>]]) : substring match
15618
15619req.hdr_cnt([<name>]) : integer
15620hdr_cnt([<header>]) : integer (deprecated)
15621 Returns an integer value representing the number of occurrences of request
15622 header field name <name>, or the total number of header field values if
15623 <name> is not specified. It is important to remember that one header line may
15624 count as several headers if it has several values. The function considers any
15625 comma as a delimiter for distinct values. If full-line headers are desired
15626 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15627 detect presence, absence or abuse of a specific header, as well as to block
15628 request smuggling attacks by rejecting requests which contain more than one
15629 of certain headers. See "req.hdr" for more information on header matching.
15630
15631req.hdr_ip([<name>[,<occ>]]) : ip
15632hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15633 This extracts the last occurrence of header <name> in an HTTP request,
15634 converts it to an IPv4 or IPv6 address and returns this address. When used
15635 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15636 of every header is checked. Optionally, a specific occurrence might be
15637 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015638 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015639 positions relative to the last one, with -1 being the last one. A typical use
15640 is with the X-Forwarded-For and X-Client-IP headers.
15641
15642req.hdr_val([<name>[,<occ>]]) : integer
15643hdr_val([<name>[,<occ>]]) : integer (deprecated)
15644 This extracts the last occurrence of header <name> in an HTTP request, and
15645 converts it to an integer value. When used with ACLs, all occurrences are
15646 checked, and if <name> is omitted, every value of every header is checked.
15647 Optionally, a specific occurrence might be specified as a position number.
15648 Positive values indicate a position from the first occurrence, with 1 being
15649 the first one. Negative values indicate positions relative to the last one,
15650 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15651
15652http_auth(<userlist>) : boolean
15653 Returns a boolean indicating whether the authentication data received from
15654 the client match a username & password stored in the specified userlist. This
15655 fetch function is not really useful outside of ACLs. Currently only http
15656 basic auth is supported.
15657
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015658http_auth_group(<userlist>) : string
15659 Returns a string corresponding to the user name found in the authentication
15660 data received from the client if both the user name and password are valid
15661 according to the specified userlist. The main purpose is to use it in ACLs
15662 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015663 This fetch function is not really useful outside of ACLs. Currently only http
15664 basic auth is supported.
15665
15666 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015667 http_auth_group(<userlist>) : group ...
15668 Returns true when the user extracted from the request and whose password is
15669 valid according to the specified userlist belongs to at least one of the
15670 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015671
15672http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015673 Returns true when the request being processed is the first one of the
15674 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015675 from some requests when a request is not the first one, or to help grouping
15676 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015677
Willy Tarreau74ca5042013-06-11 23:12:07 +020015678method : integer + string
15679 Returns an integer value corresponding to the method in the HTTP request. For
15680 example, "GET" equals 1 (check sources to establish the matching). Value 9
15681 means "other method" and may be converted to a string extracted from the
15682 stream. This should not be used directly as a sample, this is only meant to
15683 be used from ACLs, which transparently convert methods from patterns to these
15684 integer + string values. Some predefined ACL already check for most common
15685 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015686
Willy Tarreau74ca5042013-06-11 23:12:07 +020015687 ACL derivatives :
15688 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015689
Willy Tarreau74ca5042013-06-11 23:12:07 +020015690 Example :
15691 # only accept GET and HEAD requests
15692 acl valid_method method GET HEAD
15693 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015694
Willy Tarreau74ca5042013-06-11 23:12:07 +020015695path : string
15696 This extracts the request's URL path, which starts at the first slash and
15697 ends before the question mark (without the host part). A typical use is with
15698 prefetch-capable caches, and with portals which need to aggregate multiple
15699 information from databases and keep them in caches. Note that with outgoing
15700 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015701 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015702 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015703
Willy Tarreau74ca5042013-06-11 23:12:07 +020015704 ACL derivatives :
15705 path : exact string match
15706 path_beg : prefix match
15707 path_dir : subdir match
15708 path_dom : domain match
15709 path_end : suffix match
15710 path_len : length match
15711 path_reg : regex match
15712 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015713
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015714query : string
15715 This extracts the request's query string, which starts after the first
15716 question mark. If no question mark is present, this fetch returns nothing. If
15717 a question mark is present but nothing follows, it returns an empty string.
15718 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015719 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015720 which stops before the question mark.
15721
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015722req.hdr_names([<delim>]) : string
15723 This builds a string made from the concatenation of all header names as they
15724 appear in the request when the rule is evaluated. The default delimiter is
15725 the comma (',') but it may be overridden as an optional argument <delim>. In
15726 this case, only the first character of <delim> is considered.
15727
Willy Tarreau74ca5042013-06-11 23:12:07 +020015728req.ver : string
15729req_ver : string (deprecated)
15730 Returns the version string from the HTTP request, for example "1.1". This can
15731 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15732 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015733
Willy Tarreau74ca5042013-06-11 23:12:07 +020015734 ACL derivatives :
15735 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015736
Willy Tarreau74ca5042013-06-11 23:12:07 +020015737res.comp : boolean
15738 Returns the boolean "true" value if the response has been compressed by
15739 HAProxy, otherwise returns boolean "false". This may be used to add
15740 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015741
Willy Tarreau74ca5042013-06-11 23:12:07 +020015742res.comp_algo : string
15743 Returns a string containing the name of the algorithm used if the response
15744 was compressed by HAProxy, for example : "deflate". This may be used to add
15745 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015746
Willy Tarreau74ca5042013-06-11 23:12:07 +020015747res.cook([<name>]) : string
15748scook([<name>]) : string (deprecated)
15749 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15750 header line from the response, and returns its value as string. If no name is
15751 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015752
Willy Tarreau74ca5042013-06-11 23:12:07 +020015753 ACL derivatives :
15754 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015755
Willy Tarreau74ca5042013-06-11 23:12:07 +020015756res.cook_cnt([<name>]) : integer
15757scook_cnt([<name>]) : integer (deprecated)
15758 Returns an integer value representing the number of occurrences of the cookie
15759 <name> in the response, or all cookies if <name> is not specified. This is
15760 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015761
Willy Tarreau74ca5042013-06-11 23:12:07 +020015762res.cook_val([<name>]) : integer
15763scook_val([<name>]) : integer (deprecated)
15764 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15765 header line from the response, and converts its value to an integer which is
15766 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015767
Willy Tarreau74ca5042013-06-11 23:12:07 +020015768res.fhdr([<name>[,<occ>]]) : string
15769 This extracts the last occurrence of header <name> in an HTTP response, or of
15770 the last header if no <name> is specified. Optionally, a specific occurrence
15771 might be specified as a position number. Positive values indicate a position
15772 from the first occurrence, with 1 being the first one. Negative values
15773 indicate positions relative to the last one, with -1 being the last one. It
15774 differs from res.hdr() in that any commas present in the value are returned
15775 and are not used as delimiters. If this is not desired, the res.hdr() fetch
15776 should be used instead. This is sometimes useful with headers such as Date or
15777 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015778
Willy Tarreau74ca5042013-06-11 23:12:07 +020015779res.fhdr_cnt([<name>]) : integer
15780 Returns an integer value representing the number of occurrences of response
15781 header field name <name>, or the total number of header fields if <name> is
15782 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
15783 the number of full line headers and does not stop on commas. If this is not
15784 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015785
Willy Tarreau74ca5042013-06-11 23:12:07 +020015786res.hdr([<name>[,<occ>]]) : string
15787shdr([<name>[,<occ>]]) : string (deprecated)
15788 This extracts the last occurrence of header <name> in an HTTP response, or of
15789 the last header if no <name> is specified. Optionally, a specific occurrence
15790 might be specified as a position number. Positive values indicate a position
15791 from the first occurrence, with 1 being the first one. Negative values
15792 indicate positions relative to the last one, with -1 being the last one. This
15793 can be useful to learn some data into a stick-table. The function considers
15794 any comma as a delimiter for distinct values. If this is not desired, the
15795 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015796
Willy Tarreau74ca5042013-06-11 23:12:07 +020015797 ACL derivatives :
15798 shdr([<name>[,<occ>]]) : exact string match
15799 shdr_beg([<name>[,<occ>]]) : prefix match
15800 shdr_dir([<name>[,<occ>]]) : subdir match
15801 shdr_dom([<name>[,<occ>]]) : domain match
15802 shdr_end([<name>[,<occ>]]) : suffix match
15803 shdr_len([<name>[,<occ>]]) : length match
15804 shdr_reg([<name>[,<occ>]]) : regex match
15805 shdr_sub([<name>[,<occ>]]) : substring match
15806
15807res.hdr_cnt([<name>]) : integer
15808shdr_cnt([<name>]) : integer (deprecated)
15809 Returns an integer value representing the number of occurrences of response
15810 header field name <name>, or the total number of header fields if <name> is
15811 not specified. The function considers any comma as a delimiter for distinct
15812 values. If this is not desired, the res.fhdr_cnt() fetch should be used
15813 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015814
Willy Tarreau74ca5042013-06-11 23:12:07 +020015815res.hdr_ip([<name>[,<occ>]]) : ip
15816shdr_ip([<name>[,<occ>]]) : ip (deprecated)
15817 This extracts the last occurrence of header <name> in an HTTP response,
15818 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
15819 specific occurrence might be specified as a position number. Positive values
15820 indicate a position from the first occurrence, with 1 being the first one.
15821 Negative values indicate positions relative to the last one, with -1 being
15822 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015823
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015824res.hdr_names([<delim>]) : string
15825 This builds a string made from the concatenation of all header names as they
15826 appear in the response when the rule is evaluated. The default delimiter is
15827 the comma (',') but it may be overridden as an optional argument <delim>. In
15828 this case, only the first character of <delim> is considered.
15829
Willy Tarreau74ca5042013-06-11 23:12:07 +020015830res.hdr_val([<name>[,<occ>]]) : integer
15831shdr_val([<name>[,<occ>]]) : integer (deprecated)
15832 This extracts the last occurrence of header <name> in an HTTP response, and
15833 converts it to an integer value. Optionally, a specific occurrence might be
15834 specified as a position number. Positive values indicate a position from the
15835 first occurrence, with 1 being the first one. Negative values indicate
15836 positions relative to the last one, with -1 being the last one. This can be
15837 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015838
Willy Tarreau74ca5042013-06-11 23:12:07 +020015839res.ver : string
15840resp_ver : string (deprecated)
15841 Returns the version string from the HTTP response, for example "1.1". This
15842 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020015843
Willy Tarreau74ca5042013-06-11 23:12:07 +020015844 ACL derivatives :
15845 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015846
Willy Tarreau74ca5042013-06-11 23:12:07 +020015847set-cookie([<name>]) : string (deprecated)
15848 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15849 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020015850 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020015851 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015852
Willy Tarreau74ca5042013-06-11 23:12:07 +020015853 This fetch function is deprecated and has been superseded by the "res.cook"
15854 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015855
Willy Tarreau74ca5042013-06-11 23:12:07 +020015856status : integer
15857 Returns an integer containing the HTTP status code in the HTTP response, for
15858 example, 302. It is mostly used within ACLs and integer ranges, for example,
15859 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015860
Thierry Fournier0e00dca2016-04-07 15:47:40 +020015861unique-id : string
15862 Returns the unique-id attached to the request. The directive
15863 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
15864 fails. Note that the unique-id is usually used with HTTP requests, however this
15865 sample fetch can be used with other protocols. Obviously, if it is used with
15866 other protocols than HTTP, the unique-id-format directive must not contain
15867 HTTP parts. See: unique-id-format and unique-id-header
15868
Willy Tarreau74ca5042013-06-11 23:12:07 +020015869url : string
15870 This extracts the request's URL as presented in the request. A typical use is
15871 with prefetch-capable caches, and with portals which need to aggregate
15872 multiple information from databases and keep them in caches. With ACLs, using
15873 "path" is preferred over using "url", because clients may send a full URL as
15874 is normally done with proxies. The only real use is to match "*" which does
15875 not match in "path", and for which there is already a predefined ACL. See
15876 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015877
Willy Tarreau74ca5042013-06-11 23:12:07 +020015878 ACL derivatives :
15879 url : exact string match
15880 url_beg : prefix match
15881 url_dir : subdir match
15882 url_dom : domain match
15883 url_end : suffix match
15884 url_len : length match
15885 url_reg : regex match
15886 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015887
Willy Tarreau74ca5042013-06-11 23:12:07 +020015888url_ip : ip
15889 This extracts the IP address from the request's URL when the host part is
15890 presented as an IP address. Its use is very limited. For instance, a
15891 monitoring system might use this field as an alternative for the source IP in
15892 order to test what path a given source address would follow, or to force an
15893 entry in a table for a given source address. With ACLs it can be used to
15894 restrict access to certain systems through a proxy, for example when combined
15895 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015896
Willy Tarreau74ca5042013-06-11 23:12:07 +020015897url_port : integer
15898 This extracts the port part from the request's URL. Note that if the port is
15899 not specified in the request, port 80 is assumed. With ACLs it can be used to
15900 restrict access to certain systems through a proxy, for example when combined
15901 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015902
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015903urlp([<name>[,<delim>]]) : string
15904url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015905 This extracts the first occurrence of the parameter <name> in the query
15906 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015907 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
15908 any parameter will match, and the first one will be returned. The result is
15909 a string corresponding to the value of the parameter <name> as presented in
15910 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020015911 stickiness based on a client ID, to extract an application cookie passed as a
15912 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015913 this fetch iterates over multiple parameters and will iteratively report all
15914 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015915
Willy Tarreau74ca5042013-06-11 23:12:07 +020015916 ACL derivatives :
15917 urlp(<name>[,<delim>]) : exact string match
15918 urlp_beg(<name>[,<delim>]) : prefix match
15919 urlp_dir(<name>[,<delim>]) : subdir match
15920 urlp_dom(<name>[,<delim>]) : domain match
15921 urlp_end(<name>[,<delim>]) : suffix match
15922 urlp_len(<name>[,<delim>]) : length match
15923 urlp_reg(<name>[,<delim>]) : regex match
15924 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015925
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015926
Willy Tarreau74ca5042013-06-11 23:12:07 +020015927 Example :
15928 # match http://example.com/foo?PHPSESSIONID=some_id
15929 stick on urlp(PHPSESSIONID)
15930 # match http://example.com/foo;JSESSIONID=some_id
15931 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015932
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015933urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020015934 See "urlp" above. This one extracts the URL parameter <name> in the request
15935 and converts it to an integer value. This can be used for session stickiness
15936 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020015937
Dragan Dosen0070cd52016-06-16 12:19:49 +020015938url32 : integer
15939 This returns a 32-bit hash of the value obtained by concatenating the first
15940 Host header and the whole URL including parameters (not only the path part of
15941 the request, as in the "base32" fetch above). This is useful to track per-URL
15942 activity. A shorter hash is stored, saving a lot of memory. The output type
15943 is an unsigned integer.
15944
15945url32+src : binary
15946 This returns the concatenation of the "url32" fetch and the "src" fetch. The
15947 resulting type is of type binary, with a size of 8 or 20 bytes depending on
15948 the source address family. This can be used to track per-IP, per-URL counters.
15949
Willy Tarreau198a7442008-01-17 12:05:32 +010015950
Willy Tarreau74ca5042013-06-11 23:12:07 +0200159517.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015952---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015953
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015954Some predefined ACLs are hard-coded so that they do not have to be declared in
15955every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020015956order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010015957
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015958ACL name Equivalent to Usage
15959---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015960FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020015961HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015962HTTP_1.0 req_ver 1.0 match HTTP version 1.0
15963HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010015964HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
15965HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
15966HTTP_URL_SLASH url_beg / match URL beginning with "/"
15967HTTP_URL_STAR url * match URL equal to "*"
15968LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015969METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015970METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015971METH_GET method GET HEAD match HTTP GET or HEAD method
15972METH_HEAD method HEAD match HTTP HEAD method
15973METH_OPTIONS method OPTIONS match HTTP OPTIONS method
15974METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015975METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015976METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020015977RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015978REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015979TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015980WAIT_END wait_end wait for end of content analysis
15981---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015982
Willy Tarreaub937b7e2010-01-12 15:27:54 +010015983
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200159848. Logging
15985----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010015986
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015987One of HAProxy's strong points certainly lies is its precise logs. It probably
15988provides the finest level of information available for such a product, which is
15989very important for troubleshooting complex environments. Standard information
15990provided in logs include client ports, TCP/HTTP state timers, precise session
15991state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010015992to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015993headers.
15994
15995In order to improve administrators reactivity, it offers a great transparency
15996about encountered problems, both internal and external, and it is possible to
15997send logs to different sources at the same time with different level filters :
15998
15999 - global process-level logs (system errors, start/stop, etc..)
16000 - per-instance system and internal errors (lack of resource, bugs, ...)
16001 - per-instance external troubles (servers up/down, max connections)
16002 - per-instance activity (client connections), either at the establishment or
16003 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016004 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016005 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016006
16007The ability to distribute different levels of logs to different log servers
16008allow several production teams to interact and to fix their problems as soon
16009as possible. For example, the system team might monitor system-wide errors,
16010while the application team might be monitoring the up/down for their servers in
16011real time, and the security team might analyze the activity logs with one hour
16012delay.
16013
16014
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160158.1. Log levels
16016---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016017
Simon Hormandf791f52011-05-29 15:01:10 +090016018TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016019source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016020HTTP request, HTTP return code, number of bytes transmitted, conditions
16021in which the session ended, and even exchanged cookies values. For example
16022track a particular user's problems. All messages may be sent to up to two
16023syslog servers. Check the "log" keyword in section 4.2 for more information
16024about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016025
16026
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160278.2. Log formats
16028----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016029
William Lallemand48940402012-01-30 16:47:22 +010016030HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016031and will be detailed in the following sections. A few of them may vary
16032slightly with the configuration, due to indicators specific to certain
16033options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016034
16035 - the default format, which is very basic and very rarely used. It only
16036 provides very basic information about the incoming connection at the moment
16037 it is accepted : source IP:port, destination IP:port, and frontend-name.
16038 This mode will eventually disappear so it will not be described to great
16039 extents.
16040
16041 - the TCP format, which is more advanced. This format is enabled when "option
16042 tcplog" is set on the frontend. HAProxy will then usually wait for the
16043 connection to terminate before logging. This format provides much richer
16044 information, such as timers, connection counts, queue size, etc... This
16045 format is recommended for pure TCP proxies.
16046
16047 - the HTTP format, which is the most advanced for HTTP proxying. This format
16048 is enabled when "option httplog" is set on the frontend. It provides the
16049 same information as the TCP format with some HTTP-specific fields such as
16050 the request, the status code, and captures of headers and cookies. This
16051 format is recommended for HTTP proxies.
16052
Emeric Brun3a058f32009-06-30 18:26:00 +020016053 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16054 fields arranged in the same order as the CLF format. In this mode, all
16055 timers, captures, flags, etc... appear one per field after the end of the
16056 common fields, in the same order they appear in the standard HTTP format.
16057
William Lallemand48940402012-01-30 16:47:22 +010016058 - the custom log format, allows you to make your own log line.
16059
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016060Next sections will go deeper into details for each of these formats. Format
16061specification will be performed on a "field" basis. Unless stated otherwise, a
16062field is a portion of text delimited by any number of spaces. Since syslog
16063servers are susceptible of inserting fields at the beginning of a line, it is
16064always assumed that the first field is the one containing the process name and
16065identifier.
16066
16067Note : Since log lines may be quite long, the log examples in sections below
16068 might be broken into multiple lines. The example log lines will be
16069 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16070 broken into multiple lines, each non-final line will end with a
16071 backslash ('\') and the next line will start indented by two characters.
16072
16073
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160748.2.1. Default log format
16075-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016076
16077This format is used when no specific option is set. The log is emitted as soon
16078as the connection is accepted. One should note that this currently is the only
16079format which logs the request's destination IP and ports.
16080
16081 Example :
16082 listen www
16083 mode http
16084 log global
16085 server srv1 127.0.0.1:8000
16086
16087 >>> Feb 6 12:12:09 localhost \
16088 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16089 (www/HTTP)
16090
16091 Field Format Extract from the example above
16092 1 process_name '[' pid ']:' haproxy[14385]:
16093 2 'Connect from' Connect from
16094 3 source_ip ':' source_port 10.0.1.2:33312
16095 4 'to' to
16096 5 destination_ip ':' destination_port 10.0.3.31:8012
16097 6 '(' frontend_name '/' mode ')' (www/HTTP)
16098
16099Detailed fields description :
16100 - "source_ip" is the IP address of the client which initiated the connection.
16101 - "source_port" is the TCP port of the client which initiated the connection.
16102 - "destination_ip" is the IP address the client connected to.
16103 - "destination_port" is the TCP port the client connected to.
16104 - "frontend_name" is the name of the frontend (or listener) which received
16105 and processed the connection.
16106 - "mode is the mode the frontend is operating (TCP or HTTP).
16107
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016108In case of a UNIX socket, the source and destination addresses are marked as
16109"unix:" and the ports reflect the internal ID of the socket which accepted the
16110connection (the same ID as reported in the stats).
16111
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016112It is advised not to use this deprecated format for newer installations as it
16113will eventually disappear.
16114
16115
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161168.2.2. TCP log format
16117---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016118
16119The TCP format is used when "option tcplog" is specified in the frontend, and
16120is the recommended format for pure TCP proxies. It provides a lot of precious
16121information for troubleshooting. Since this format includes timers and byte
16122counts, the log is normally emitted at the end of the session. It can be
16123emitted earlier if "option logasap" is specified, which makes sense in most
16124environments with long sessions such as remote terminals. Sessions which match
16125the "monitor" rules are never logged. It is also possible not to emit logs for
16126sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016127specifying "option dontlognull" in the frontend. Successful connections will
16128not be logged if "option dontlog-normal" is specified in the frontend. A few
16129fields may slightly vary depending on some configuration options, those are
16130marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016131
16132 Example :
16133 frontend fnt
16134 mode tcp
16135 option tcplog
16136 log global
16137 default_backend bck
16138
16139 backend bck
16140 server srv1 127.0.0.1:8000
16141
16142 >>> Feb 6 12:12:56 localhost \
16143 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16144 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16145
16146 Field Format Extract from the example above
16147 1 process_name '[' pid ']:' haproxy[14387]:
16148 2 client_ip ':' client_port 10.0.1.2:33313
16149 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16150 4 frontend_name fnt
16151 5 backend_name '/' server_name bck/srv1
16152 6 Tw '/' Tc '/' Tt* 0/0/5007
16153 7 bytes_read* 212
16154 8 termination_state --
16155 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16156 10 srv_queue '/' backend_queue 0/0
16157
16158Detailed fields description :
16159 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016160 connection to haproxy. If the connection was accepted on a UNIX socket
16161 instead, the IP address would be replaced with the word "unix". Note that
16162 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016163 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016164 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016165 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016166
16167 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016168 If the connection was accepted on a UNIX socket instead, the port would be
16169 replaced with the ID of the accepting socket, which is also reported in the
16170 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016171
16172 - "accept_date" is the exact date when the connection was received by haproxy
16173 (which might be very slightly different from the date observed on the
16174 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016175 the same date which may appear in any upstream firewall's log. When used in
16176 HTTP mode, the accept_date field will be reset to the first moment the
16177 connection is ready to receive a new request (end of previous response for
16178 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016179
16180 - "frontend_name" is the name of the frontend (or listener) which received
16181 and processed the connection.
16182
16183 - "backend_name" is the name of the backend (or listener) which was selected
16184 to manage the connection to the server. This will be the same as the
16185 frontend if no switching rule has been applied, which is common for TCP
16186 applications.
16187
16188 - "server_name" is the name of the last server to which the connection was
16189 sent, which might differ from the first one if there were connection errors
16190 and a redispatch occurred. Note that this server belongs to the backend
16191 which processed the request. If the connection was aborted before reaching
16192 a server, "<NOSRV>" is indicated instead of a server name.
16193
16194 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16195 It can be "-1" if the connection was aborted before reaching the queue.
16196 See "Timers" below for more details.
16197
16198 - "Tc" is the total time in milliseconds spent waiting for the connection to
16199 establish to the final server, including retries. It can be "-1" if the
16200 connection was aborted before a connection could be established. See
16201 "Timers" below for more details.
16202
16203 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016204 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016205 "option logasap" was specified, then the time counting stops at the moment
16206 the log is emitted. In this case, a '+' sign is prepended before the value,
16207 indicating that the final one will be larger. See "Timers" below for more
16208 details.
16209
16210 - "bytes_read" is the total number of bytes transmitted from the server to
16211 the client when the log is emitted. If "option logasap" is specified, the
16212 this value will be prefixed with a '+' sign indicating that the final one
16213 may be larger. Please note that this value is a 64-bit counter, so log
16214 analysis tools must be able to handle it without overflowing.
16215
16216 - "termination_state" is the condition the session was in when the session
16217 ended. This indicates the session state, which side caused the end of
16218 session to happen, and for what reason (timeout, error, ...). The normal
16219 flags should be "--", indicating the session was closed by either end with
16220 no data remaining in buffers. See below "Session state at disconnection"
16221 for more details.
16222
16223 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016224 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016225 limits have been reached. For instance, if actconn is close to 512 when
16226 multiple connection errors occur, chances are high that the system limits
16227 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016228 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016229
16230 - "feconn" is the total number of concurrent connections on the frontend when
16231 the session was logged. It is useful to estimate the amount of resource
16232 required to sustain high loads, and to detect when the frontend's "maxconn"
16233 has been reached. Most often when this value increases by huge jumps, it is
16234 because there is congestion on the backend servers, but sometimes it can be
16235 caused by a denial of service attack.
16236
16237 - "beconn" is the total number of concurrent connections handled by the
16238 backend when the session was logged. It includes the total number of
16239 concurrent connections active on servers as well as the number of
16240 connections pending in queues. It is useful to estimate the amount of
16241 additional servers needed to support high loads for a given application.
16242 Most often when this value increases by huge jumps, it is because there is
16243 congestion on the backend servers, but sometimes it can be caused by a
16244 denial of service attack.
16245
16246 - "srv_conn" is the total number of concurrent connections still active on
16247 the server when the session was logged. It can never exceed the server's
16248 configured "maxconn" parameter. If this value is very often close or equal
16249 to the server's "maxconn", it means that traffic regulation is involved a
16250 lot, meaning that either the server's maxconn value is too low, or that
16251 there aren't enough servers to process the load with an optimal response
16252 time. When only one of the server's "srv_conn" is high, it usually means
16253 that this server has some trouble causing the connections to take longer to
16254 be processed than on other servers.
16255
16256 - "retries" is the number of connection retries experienced by this session
16257 when trying to connect to the server. It must normally be zero, unless a
16258 server is being stopped at the same moment the connection was attempted.
16259 Frequent retries generally indicate either a network problem between
16260 haproxy and the server, or a misconfigured system backlog on the server
16261 preventing new connections from being queued. This field may optionally be
16262 prefixed with a '+' sign, indicating that the session has experienced a
16263 redispatch after the maximal retry count has been reached on the initial
16264 server. In this case, the server name appearing in the log is the one the
16265 connection was redispatched to, and not the first one, though both may
16266 sometimes be the same in case of hashing for instance. So as a general rule
16267 of thumb, when a '+' is present in front of the retry count, this count
16268 should not be attributed to the logged server.
16269
16270 - "srv_queue" is the total number of requests which were processed before
16271 this one in the server queue. It is zero when the request has not gone
16272 through the server queue. It makes it possible to estimate the approximate
16273 server's response time by dividing the time spent in queue by the number of
16274 requests in the queue. It is worth noting that if a session experiences a
16275 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016276 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016277 backend queue unless a redispatch occurs.
16278
16279 - "backend_queue" is the total number of requests which were processed before
16280 this one in the backend's global queue. It is zero when the request has not
16281 gone through the global queue. It makes it possible to estimate the average
16282 queue length, which easily translates into a number of missing servers when
16283 divided by a server's "maxconn" parameter. It is worth noting that if a
16284 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016285 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016286 through both the server queue and the backend queue unless a redispatch
16287 occurs.
16288
16289
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162908.2.3. HTTP log format
16291----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016292
16293The HTTP format is the most complete and the best suited for HTTP proxies. It
16294is enabled by when "option httplog" is specified in the frontend. It provides
16295the same level of information as the TCP format with additional features which
16296are specific to the HTTP protocol. Just like the TCP format, the log is usually
16297emitted at the end of the session, unless "option logasap" is specified, which
16298generally only makes sense for download sites. A session which matches the
16299"monitor" rules will never logged. It is also possible not to log sessions for
16300which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016301frontend. Successful connections will not be logged if "option dontlog-normal"
16302is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016303
16304Most fields are shared with the TCP log, some being different. A few fields may
16305slightly vary depending on some configuration options. Those ones are marked
16306with a star ('*') after the field name below.
16307
16308 Example :
16309 frontend http-in
16310 mode http
16311 option httplog
16312 log global
16313 default_backend bck
16314
16315 backend static
16316 server srv1 127.0.0.1:8000
16317
16318 >>> Feb 6 12:14:14 localhost \
16319 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16320 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 +010016321 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016322
16323 Field Format Extract from the example above
16324 1 process_name '[' pid ']:' haproxy[14389]:
16325 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016326 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016327 4 frontend_name http-in
16328 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016329 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016330 7 status_code 200
16331 8 bytes_read* 2750
16332 9 captured_request_cookie -
16333 10 captured_response_cookie -
16334 11 termination_state ----
16335 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16336 13 srv_queue '/' backend_queue 0/0
16337 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16338 15 '{' captured_response_headers* '}' {}
16339 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016340
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016341Detailed fields description :
16342 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016343 connection to haproxy. If the connection was accepted on a UNIX socket
16344 instead, the IP address would be replaced with the word "unix". Note that
16345 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016346 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016347 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016348 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016349
16350 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016351 If the connection was accepted on a UNIX socket instead, the port would be
16352 replaced with the ID of the accepting socket, which is also reported in the
16353 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016354
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016355 - "request_date" is the exact date when the first byte of the HTTP request
16356 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016357
16358 - "frontend_name" is the name of the frontend (or listener) which received
16359 and processed the connection.
16360
16361 - "backend_name" is the name of the backend (or listener) which was selected
16362 to manage the connection to the server. This will be the same as the
16363 frontend if no switching rule has been applied.
16364
16365 - "server_name" is the name of the last server to which the connection was
16366 sent, which might differ from the first one if there were connection errors
16367 and a redispatch occurred. Note that this server belongs to the backend
16368 which processed the request. If the request was aborted before reaching a
16369 server, "<NOSRV>" is indicated instead of a server name. If the request was
16370 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16371
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016372 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16373 request from the client (not counting body) after the first byte was
16374 received. It can be "-1" if the connection was aborted before a complete
16375 request could be received or the a bad request was received. It should
16376 always be very small because a request generally fits in one single packet.
16377 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016378 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16379 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016380
16381 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16382 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016383 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016384
16385 - "Tc" is the total time in milliseconds spent waiting for the connection to
16386 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016387 request was aborted before a connection could be established. See section
16388 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016389
16390 - "Tr" is the total time in milliseconds spent waiting for the server to send
16391 a full HTTP response, not counting data. It can be "-1" if the request was
16392 aborted before a complete response could be received. It generally matches
16393 the server's processing time for the request, though it may be altered by
16394 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016395 "GET" requests generally indicate an overloaded server. See section 8.4
16396 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016397
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016398 - "Ta" is the time the request remained active in haproxy, which is the total
16399 time in milliseconds elapsed between the first byte of the request was
16400 received and the last byte of response was sent. It covers all possible
16401 processing except the handshake (see Th) and idle time (see Ti). There is
16402 one exception, if "option logasap" was specified, then the time counting
16403 stops at the moment the log is emitted. In this case, a '+' sign is
16404 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016405 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016406
16407 - "status_code" is the HTTP status code returned to the client. This status
16408 is generally set by the server, but it might also be set by haproxy when
16409 the server cannot be reached or when its response is blocked by haproxy.
16410
16411 - "bytes_read" is the total number of bytes transmitted to the client when
16412 the log is emitted. This does include HTTP headers. If "option logasap" is
16413 specified, the this value will be prefixed with a '+' sign indicating that
16414 the final one may be larger. Please note that this value is a 64-bit
16415 counter, so log analysis tools must be able to handle it without
16416 overflowing.
16417
16418 - "captured_request_cookie" is an optional "name=value" entry indicating that
16419 the client had this cookie in the request. The cookie name and its maximum
16420 length are defined by the "capture cookie" statement in the frontend
16421 configuration. The field is a single dash ('-') when the option is not
16422 set. Only one cookie may be captured, it is generally used to track session
16423 ID exchanges between a client and a server to detect session crossing
16424 between clients due to application bugs. For more details, please consult
16425 the section "Capturing HTTP headers and cookies" below.
16426
16427 - "captured_response_cookie" is an optional "name=value" entry indicating
16428 that the server has returned a cookie with its response. The cookie name
16429 and its maximum length are defined by the "capture cookie" statement in the
16430 frontend configuration. The field is a single dash ('-') when the option is
16431 not set. Only one cookie may be captured, it is generally used to track
16432 session ID exchanges between a client and a server to detect session
16433 crossing between clients due to application bugs. For more details, please
16434 consult the section "Capturing HTTP headers and cookies" below.
16435
16436 - "termination_state" is the condition the session was in when the session
16437 ended. This indicates the session state, which side caused the end of
16438 session to happen, for what reason (timeout, error, ...), just like in TCP
16439 logs, and information about persistence operations on cookies in the last
16440 two characters. The normal flags should begin with "--", indicating the
16441 session was closed by either end with no data remaining in buffers. See
16442 below "Session state at disconnection" for more details.
16443
16444 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016445 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016446 limits have been reached. For instance, if actconn is close to 512 or 1024
16447 when multiple connection errors occur, chances are high that the system
16448 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016449 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016450 system.
16451
16452 - "feconn" is the total number of concurrent connections on the frontend when
16453 the session was logged. It is useful to estimate the amount of resource
16454 required to sustain high loads, and to detect when the frontend's "maxconn"
16455 has been reached. Most often when this value increases by huge jumps, it is
16456 because there is congestion on the backend servers, but sometimes it can be
16457 caused by a denial of service attack.
16458
16459 - "beconn" is the total number of concurrent connections handled by the
16460 backend when the session was logged. It includes the total number of
16461 concurrent connections active on servers as well as the number of
16462 connections pending in queues. It is useful to estimate the amount of
16463 additional servers needed to support high loads for a given application.
16464 Most often when this value increases by huge jumps, it is because there is
16465 congestion on the backend servers, but sometimes it can be caused by a
16466 denial of service attack.
16467
16468 - "srv_conn" is the total number of concurrent connections still active on
16469 the server when the session was logged. It can never exceed the server's
16470 configured "maxconn" parameter. If this value is very often close or equal
16471 to the server's "maxconn", it means that traffic regulation is involved a
16472 lot, meaning that either the server's maxconn value is too low, or that
16473 there aren't enough servers to process the load with an optimal response
16474 time. When only one of the server's "srv_conn" is high, it usually means
16475 that this server has some trouble causing the requests to take longer to be
16476 processed than on other servers.
16477
16478 - "retries" is the number of connection retries experienced by this session
16479 when trying to connect to the server. It must normally be zero, unless a
16480 server is being stopped at the same moment the connection was attempted.
16481 Frequent retries generally indicate either a network problem between
16482 haproxy and the server, or a misconfigured system backlog on the server
16483 preventing new connections from being queued. This field may optionally be
16484 prefixed with a '+' sign, indicating that the session has experienced a
16485 redispatch after the maximal retry count has been reached on the initial
16486 server. In this case, the server name appearing in the log is the one the
16487 connection was redispatched to, and not the first one, though both may
16488 sometimes be the same in case of hashing for instance. So as a general rule
16489 of thumb, when a '+' is present in front of the retry count, this count
16490 should not be attributed to the logged server.
16491
16492 - "srv_queue" is the total number of requests which were processed before
16493 this one in the server queue. It is zero when the request has not gone
16494 through the server queue. It makes it possible to estimate the approximate
16495 server's response time by dividing the time spent in queue by the number of
16496 requests in the queue. It is worth noting that if a session experiences a
16497 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016498 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016499 backend queue unless a redispatch occurs.
16500
16501 - "backend_queue" is the total number of requests which were processed before
16502 this one in the backend's global queue. It is zero when the request has not
16503 gone through the global queue. It makes it possible to estimate the average
16504 queue length, which easily translates into a number of missing servers when
16505 divided by a server's "maxconn" parameter. It is worth noting that if a
16506 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016507 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016508 through both the server queue and the backend queue unless a redispatch
16509 occurs.
16510
16511 - "captured_request_headers" is a list of headers captured in the request due
16512 to the presence of the "capture request header" statement in the frontend.
16513 Multiple headers can be captured, they will be delimited by a vertical bar
16514 ('|'). When no capture is enabled, the braces do not appear, causing a
16515 shift of remaining fields. It is important to note that this field may
16516 contain spaces, and that using it requires a smarter log parser than when
16517 it's not used. Please consult the section "Capturing HTTP headers and
16518 cookies" below for more details.
16519
16520 - "captured_response_headers" is a list of headers captured in the response
16521 due to the presence of the "capture response header" statement in the
16522 frontend. Multiple headers can be captured, they will be delimited by a
16523 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16524 causing a shift of remaining fields. It is important to note that this
16525 field may contain spaces, and that using it requires a smarter log parser
16526 than when it's not used. Please consult the section "Capturing HTTP headers
16527 and cookies" below for more details.
16528
16529 - "http_request" is the complete HTTP request line, including the method,
16530 request and HTTP version string. Non-printable characters are encoded (see
16531 below the section "Non-printable characters"). This is always the last
16532 field, and it is always delimited by quotes and is the only one which can
16533 contain quotes. If new fields are added to the log format, they will be
16534 added before this field. This field might be truncated if the request is
16535 huge and does not fit in the standard syslog buffer (1024 characters). This
16536 is the reason why this field must always remain the last one.
16537
16538
Cyril Bontédc4d9032012-04-08 21:57:39 +0200165398.2.4. Custom log format
16540------------------------
William Lallemand48940402012-01-30 16:47:22 +010016541
Willy Tarreau2beef582012-12-20 17:22:52 +010016542The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016543mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016544
Davor Ocelice9ed2812017-12-25 17:49:28 +010016545HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016546Variables can take arguments using braces ('{}'), and multiple arguments are
16547separated by commas within the braces. Flags may be added or removed by
16548prefixing them with a '+' or '-' sign.
16549
16550Special variable "%o" may be used to propagate its flags to all other
16551variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016552("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016553
Willy Tarreauc8368452012-12-21 00:09:23 +010016554If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016555as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016556less common information such as the client's SSL certificate's DN, or to log
16557the key that would be used to store an entry into a stick table.
16558
William Lallemand48940402012-01-30 16:47:22 +010016559Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016560In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016561in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016562
Dragan Dosen835b9212016-02-12 13:23:03 +010016563Note: when using the RFC5424 syslog message format, the characters '"',
16564'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16565https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16566such cases, the use of the flag "E" should be considered.
16567
William Lallemand48940402012-01-30 16:47:22 +010016568Flags are :
16569 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016570 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016571 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16572 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016573
16574 Example:
16575
16576 log-format %T\ %t\ Some\ Text
16577 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16578
Dragan Dosen835b9212016-02-12 13:23:03 +010016579 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16580
William Lallemand48940402012-01-30 16:47:22 +010016581At the moment, the default HTTP format is defined this way :
16582
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016583 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16584 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016585
William Lallemandbddd4fd2012-02-27 11:23:10 +010016586the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016587
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016588 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16589 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16590 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016591
William Lallemandbddd4fd2012-02-27 11:23:10 +010016592and the default TCP format is defined this way :
16593
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016594 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16595 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016596
William Lallemand48940402012-01-30 16:47:22 +010016597Please refer to the table below for currently defined variables :
16598
William Lallemandbddd4fd2012-02-27 11:23:10 +010016599 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016600 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016601 +---+------+-----------------------------------------------+-------------+
16602 | | %o | special variable, apply flags on all next var | |
16603 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016604 | | %B | bytes_read (from server to client) | numeric |
16605 | H | %CC | captured_request_cookie | string |
16606 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016607 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016608 | H | %HM | HTTP method (ex: POST) | string |
16609 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016610 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016611 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16612 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016613 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016614 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016615 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016616 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016617 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016618 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016619 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016620 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16621 | H | %Ti | idle time before the HTTP request | numeric |
16622 | H | %Tq | Th + Ti + TR | numeric |
16623 | H | %TR | time to receive the full request from 1st byte| numeric |
16624 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016625 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016626 | | %Tt | Tt | numeric |
16627 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016628 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016629 | | %ac | actconn | numeric |
16630 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016631 | | %bc | beconn (backend concurrent connections) | numeric |
16632 | | %bi | backend_source_ip (connecting address) | IP |
16633 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016634 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016635 | | %ci | client_ip (accepted address) | IP |
16636 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016637 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016638 | | %fc | feconn (frontend concurrent connections) | numeric |
16639 | | %fi | frontend_ip (accepting address) | IP |
16640 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016641 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016642 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016643 | | %hr | captured_request_headers default style | string |
16644 | | %hrl | captured_request_headers CLF style | string list |
16645 | | %hs | captured_response_headers default style | string |
16646 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016647 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016648 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016649 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016650 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016651 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016652 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016653 | | %sc | srv_conn (server concurrent connections) | numeric |
16654 | | %si | server_IP (target address) | IP |
16655 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016656 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016657 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16658 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016659 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016660 | H | %tr | date_time of HTTP request | date |
16661 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016662 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016663 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016664 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016665 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016666
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016667 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016668
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016669
166708.2.5. Error log format
16671-----------------------
16672
16673When an incoming connection fails due to an SSL handshake or an invalid PROXY
16674protocol header, haproxy will log the event using a shorter, fixed line format.
16675By default, logs are emitted at the LOG_INFO level, unless the option
16676"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016677will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016678logged if the "dontlognull" option is set.
16679
16680The format looks like this :
16681
16682 >>> Dec 3 18:27:14 localhost \
16683 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16684 Connection error during SSL handshake
16685
16686 Field Format Extract from the example above
16687 1 process_name '[' pid ']:' haproxy[6103]:
16688 2 client_ip ':' client_port 127.0.0.1:56059
16689 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16690 4 frontend_name "/" bind_name ":" frt/f1:
16691 5 message Connection error during SSL handshake
16692
16693These fields just provide minimal information to help debugging connection
16694failures.
16695
16696
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166978.3. Advanced logging options
16698-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016699
16700Some advanced logging options are often looked for but are not easy to find out
16701just by looking at the various options. Here is an entry point for the few
16702options which can enable better logging. Please refer to the keywords reference
16703for more information about their usage.
16704
16705
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167068.3.1. Disabling logging of external tests
16707------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016708
16709It is quite common to have some monitoring tools perform health checks on
16710haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16711commercial load-balancer, and sometimes it will simply be a more complete
16712monitoring system such as Nagios. When the tests are very frequent, users often
16713ask how to disable logging for those checks. There are three possibilities :
16714
16715 - if connections come from everywhere and are just TCP probes, it is often
16716 desired to simply disable logging of connections without data exchange, by
16717 setting "option dontlognull" in the frontend. It also disables logging of
16718 port scans, which may or may not be desired.
16719
16720 - if the connection come from a known source network, use "monitor-net" to
16721 declare this network as monitoring only. Any host in this network will then
16722 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016723 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016724 such as other load-balancers.
16725
16726 - if the tests are performed on a known URI, use "monitor-uri" to declare
16727 this URI as dedicated to monitoring. Any host sending this request will
16728 only get the result of a health-check, and the request will not be logged.
16729
16730
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167318.3.2. Logging before waiting for the session to terminate
16732----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016733
16734The problem with logging at end of connection is that you have no clue about
16735what is happening during very long sessions, such as remote terminal sessions
16736or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016737"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016738just before data transfer begins. This means that in case of TCP, it will still
16739log the connection status to the server, and in case of HTTP, it will log just
16740after processing the server headers. In this case, the number of bytes reported
16741is the number of header bytes sent to the client. In order to avoid confusion
16742with normal logs, the total time field and the number of bytes are prefixed
16743with a '+' sign which means that real numbers are certainly larger.
16744
16745
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167468.3.3. Raising log level upon errors
16747------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016748
16749Sometimes it is more convenient to separate normal traffic from errors logs,
16750for instance in order to ease error monitoring from log files. When the option
16751"log-separate-errors" is used, connections which experience errors, timeouts,
16752retries, redispatches or HTTP status codes 5xx will see their syslog level
16753raised from "info" to "err". This will help a syslog daemon store the log in
16754a separate file. It is very important to keep the errors in the normal traffic
16755file too, so that log ordering is not altered. You should also be careful if
16756you already have configured your syslog daemon to store all logs higher than
16757"notice" in an "admin" file, because the "err" level is higher than "notice".
16758
16759
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167608.3.4. Disabling logging of successful connections
16761--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016762
16763Although this may sound strange at first, some large sites have to deal with
16764multiple thousands of logs per second and are experiencing difficulties keeping
16765them intact for a long time or detecting errors within them. If the option
16766"dontlog-normal" is set on the frontend, all normal connections will not be
16767logged. In this regard, a normal connection is defined as one without any
16768error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
16769and a response with a status 5xx is not considered normal and will be logged
16770too. Of course, doing is is really discouraged as it will remove most of the
16771useful information from the logs. Do this only if you have no other
16772alternative.
16773
16774
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200167758.4. Timing events
16776------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016777
16778Timers provide a great help in troubleshooting network problems. All values are
16779reported in milliseconds (ms). These timers should be used in conjunction with
16780the session termination flags. In TCP mode with "option tcplog" set on the
16781frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016782mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
16783addition, three other measures are provided, "Th", "Ti", and "Tq".
16784
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010016785Timings events in HTTP mode:
16786
16787 first request 2nd request
16788 |<-------------------------------->|<-------------- ...
16789 t tr t tr ...
16790 ---|----|----|----|----|----|----|----|----|--
16791 : Th Ti TR Tw Tc Tr Td : Ti ...
16792 :<---- Tq ---->: :
16793 :<-------------- Tt -------------->:
16794 :<--------- Ta --------->:
16795
16796Timings events in TCP mode:
16797
16798 TCP session
16799 |<----------------->|
16800 t t
16801 ---|----|----|----|----|---
16802 | Th Tw Tc Td |
16803 |<------ Tt ------->|
16804
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016805 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016806 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016807 only happen once during the whole connection's lifetime. A large time here
16808 may indicate that the client only pre-established the connection without
16809 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010016810 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020016811 SSL handshake was very expensive to compute. Please note that this time is
16812 reported only before the first request, so it is safe to average it over
16813 all request to calculate the amortized value. The second and subsequent
16814 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016815
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016816 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
16817 counts between the end of the handshakes and the first byte of the HTTP
16818 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020016819 to count after the end of the transmission the previous response. When a
16820 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
16821 after the previous request. Some browsers pre-establish connections to a
16822 server in order to reduce the latency of a future request, and keep them
16823 pending until they need it. This delay will be reported as the idle time. A
16824 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016825
16826 - TR: total time to get the client request (HTTP mode only). It's the time
16827 elapsed between the first bytes received and the moment the proxy received
16828 the empty line marking the end of the HTTP headers. The value "-1"
16829 indicates that the end of headers has never been seen. This happens when
16830 the client closes prematurely or times out. This time is usually very short
16831 since most requests fit in a single packet. A large time may indicate a
16832 request typed by hand during a test.
16833
16834 - Tq: total time to get the client request from the accept date or since the
16835 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010016836 exactly equal to Th + Ti + TR unless any of them is -1, in which case it
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016837 returns -1 as well. This timer used to be very useful before the arrival of
16838 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
16839 it in favor of TR nowadays, as the idle time adds a lot of noise to the
16840 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016841
16842 - Tw: total time spent in the queues waiting for a connection slot. It
16843 accounts for backend queue as well as the server queues, and depends on the
16844 queue size, and the time needed for the server to complete previous
16845 requests. The value "-1" means that the request was killed before reaching
16846 the queue, which is generally what happens with invalid or denied requests.
16847
16848 - Tc: total time to establish the TCP connection to the server. It's the time
16849 elapsed between the moment the proxy sent the connection request, and the
16850 moment it was acknowledged by the server, or between the TCP SYN packet and
16851 the matching SYN/ACK packet in return. The value "-1" means that the
16852 connection never established.
16853
16854 - Tr: server response time (HTTP mode only). It's the time elapsed between
16855 the moment the TCP connection was established to the server and the moment
16856 the server sent its complete response headers. It purely shows its request
16857 processing time, without the network overhead due to the data transmission.
16858 It is worth noting that when the client has data to send to the server, for
16859 instance during a POST request, the time already runs, and this can distort
16860 apparent response time. For this reason, it's generally wise not to trust
16861 too much this field for POST requests initiated from clients behind an
16862 untrusted network. A value of "-1" here means that the last the response
16863 header (empty line) was never seen, most likely because the server timeout
16864 stroke before the server managed to process the request.
16865
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016866 - Ta: total active time for the HTTP request, between the moment the proxy
16867 received the first byte of the request header and the emission of the last
16868 byte of the response body. The exception is when the "logasap" option is
16869 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
16870 a '+' sign. From this field, we can deduce "Td", the data transmission time,
16871 by subtracting other timers when valid :
16872
16873 Td = Ta - (TR + Tw + Tc + Tr)
16874
16875 Timers with "-1" values have to be excluded from this equation. Note that
16876 "Ta" can never be negative.
16877
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016878 - Tt: total session duration time, between the moment the proxy accepted it
16879 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016880 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
16881 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016882 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016883
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016884 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016885
16886 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016887 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
16888 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016889
16890These timers provide precious indications on trouble causes. Since the TCP
16891protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
16892that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016893due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
16894"Tt" is close to a timeout value specified in the configuration, it often means
16895that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016896
16897Most common cases :
16898
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016899 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
16900 the client and the proxy. This is very rare on local networks but might
16901 happen when clients are on far remote networks and send large requests. It
16902 may happen that values larger than usual appear here without any network
16903 cause. Sometimes, during an attack or just after a resource starvation has
16904 ended, haproxy may accept thousands of connections in a few milliseconds.
16905 The time spent accepting these connections will inevitably slightly delay
16906 processing of other connections, and it can happen that request times in the
16907 order of a few tens of milliseconds are measured after a few thousands of
16908 new connections have been accepted at once. Using one of the keep-alive
16909 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020016910 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016911
16912 - If "Tc" is close to 3000, a packet has probably been lost between the
16913 server and the proxy during the server connection phase. This value should
16914 always be very low, such as 1 ms on local networks and less than a few tens
16915 of ms on remote networks.
16916
Willy Tarreau55165fe2009-05-10 12:02:55 +020016917 - If "Tr" is nearly always lower than 3000 except some rare values which seem
16918 to be the average majored by 3000, there are probably some packets lost
16919 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016920
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016921 - If "Ta" is large even for small byte counts, it generally is because
16922 neither the client nor the server decides to close the connection while
16923 haproxy is running in tunnel mode and both have agreed on a keep-alive
16924 connection mode. In order to solve this issue, it will be needed to specify
16925 one of the HTTP options to manipulate keep-alive or close options on either
16926 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
16927 important when connection regulation is used with the "maxconn" option on
16928 the servers, since no new connection will be sent to the server until
16929 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016930
16931Other noticeable HTTP log cases ('xx' means any value to be ignored) :
16932
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016933 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016934 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016935 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016936
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016937 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016938 or it aborted too early. Check the session termination flags
16939 then "timeout http-request" and "timeout client" settings.
16940
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016941 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016942 servers were out of order, because the request was invalid
16943 or forbidden by ACL rules. Check the session termination
16944 flags.
16945
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016946 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
16947 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016948 Check the session termination flags, then check the
16949 "timeout connect" setting. Note that the tarpit action might
16950 return similar-looking patterns, with "Tw" equal to the time
16951 the client connection was maintained open.
16952
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016953 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016954 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016955 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016956 termination flags, then check the "timeout server" setting.
16957
16958
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169598.5. Session state at disconnection
16960-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016961
16962TCP and HTTP logs provide a session termination indicator in the
16963"termination_state" field, just before the number of active connections. It is
169642-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
16965each of which has a special meaning :
16966
16967 - On the first character, a code reporting the first event which caused the
16968 session to terminate :
16969
16970 C : the TCP session was unexpectedly aborted by the client.
16971
16972 S : the TCP session was unexpectedly aborted by the server, or the
16973 server explicitly refused it.
16974
16975 P : the session was prematurely aborted by the proxy, because of a
16976 connection limit enforcement, because a DENY filter was matched,
16977 because of a security check which detected and blocked a dangerous
16978 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010016979 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020016980
16981 L : the session was locally processed by haproxy and was not passed to
16982 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016983
16984 R : a resource on the proxy has been exhausted (memory, sockets, source
16985 ports, ...). Usually, this appears during the connection phase, and
16986 system logs should contain a copy of the precise error. If this
16987 happens, it must be considered as a very serious anomaly which
16988 should be fixed as soon as possible by any means.
16989
16990 I : an internal error was identified by the proxy during a self-check.
16991 This should NEVER happen, and you are encouraged to report any log
16992 containing this, because this would almost certainly be a bug. It
16993 would be wise to preventively restart the process after such an
16994 event too, in case it would be caused by memory corruption.
16995
Simon Horman752dc4a2011-06-21 14:34:59 +090016996 D : the session was killed by haproxy because the server was detected
16997 as down and was configured to kill all connections when going down.
16998
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070016999 U : the session was killed by haproxy on this backup server because an
17000 active server was detected as up and was configured to kill all
17001 backup connections when going up.
17002
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017003 K : the session was actively killed by an admin operating on haproxy.
17004
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017005 c : the client-side timeout expired while waiting for the client to
17006 send or receive data.
17007
17008 s : the server-side timeout expired while waiting for the server to
17009 send or receive data.
17010
17011 - : normal session completion, both the client and the server closed
17012 with nothing left in the buffers.
17013
17014 - on the second character, the TCP or HTTP session state when it was closed :
17015
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017016 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017017 (HTTP mode only). Nothing was sent to any server.
17018
17019 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17020 only happen when servers have a 'maxconn' parameter set. It can
17021 also happen in the global queue after a redispatch consecutive to
17022 a failed attempt to connect to a dying server. If no redispatch is
17023 reported, then no connection attempt was made to any server.
17024
17025 C : the proxy was waiting for the CONNECTION to establish on the
17026 server. The server might at most have noticed a connection attempt.
17027
17028 H : the proxy was waiting for complete, valid response HEADERS from the
17029 server (HTTP only).
17030
17031 D : the session was in the DATA phase.
17032
17033 L : the proxy was still transmitting LAST data to the client while the
17034 server had already finished. This one is very rare as it can only
17035 happen when the client dies while receiving the last packets.
17036
17037 T : the request was tarpitted. It has been held open with the client
17038 during the whole "timeout tarpit" duration or until the client
17039 closed, both of which will be reported in the "Tw" timer.
17040
17041 - : normal session completion after end of data transfer.
17042
17043 - the third character tells whether the persistence cookie was provided by
17044 the client (only in HTTP mode) :
17045
17046 N : the client provided NO cookie. This is usually the case for new
17047 visitors, so counting the number of occurrences of this flag in the
17048 logs generally indicate a valid trend for the site frequentation.
17049
17050 I : the client provided an INVALID cookie matching no known server.
17051 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017052 cookies between HTTP/HTTPS sites, persistence conditionally
17053 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017054
17055 D : the client provided a cookie designating a server which was DOWN,
17056 so either "option persist" was used and the client was sent to
17057 this server, or it was not set and the client was redispatched to
17058 another server.
17059
Willy Tarreau996a92c2010-10-13 19:30:47 +020017060 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017061 server.
17062
Willy Tarreau996a92c2010-10-13 19:30:47 +020017063 E : the client provided a valid cookie, but with a last date which was
17064 older than what is allowed by the "maxidle" cookie parameter, so
17065 the cookie is consider EXPIRED and is ignored. The request will be
17066 redispatched just as if there was no cookie.
17067
17068 O : the client provided a valid cookie, but with a first date which was
17069 older than what is allowed by the "maxlife" cookie parameter, so
17070 the cookie is consider too OLD and is ignored. The request will be
17071 redispatched just as if there was no cookie.
17072
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017073 U : a cookie was present but was not used to select the server because
17074 some other server selection mechanism was used instead (typically a
17075 "use-server" rule).
17076
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017077 - : does not apply (no cookie set in configuration).
17078
17079 - the last character reports what operations were performed on the persistence
17080 cookie returned by the server (only in HTTP mode) :
17081
17082 N : NO cookie was provided by the server, and none was inserted either.
17083
17084 I : no cookie was provided by the server, and the proxy INSERTED one.
17085 Note that in "cookie insert" mode, if the server provides a cookie,
17086 it will still be overwritten and reported as "I" here.
17087
Willy Tarreau996a92c2010-10-13 19:30:47 +020017088 U : the proxy UPDATED the last date in the cookie that was presented by
17089 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017090 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017091 date indicated in the cookie. If any other change happens, such as
17092 a redispatch, then the cookie will be marked as inserted instead.
17093
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017094 P : a cookie was PROVIDED by the server and transmitted as-is.
17095
17096 R : the cookie provided by the server was REWRITTEN by the proxy, which
17097 happens in "cookie rewrite" or "cookie prefix" modes.
17098
17099 D : the cookie provided by the server was DELETED by the proxy.
17100
17101 - : does not apply (no cookie set in configuration).
17102
Willy Tarreau996a92c2010-10-13 19:30:47 +020017103The combination of the two first flags gives a lot of information about what
17104was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017105helpful to detect server saturation, network troubles, local system resource
17106starvation, attacks, etc...
17107
17108The most common termination flags combinations are indicated below. They are
17109alphabetically sorted, with the lowercase set just after the upper case for
17110easier finding and understanding.
17111
17112 Flags Reason
17113
17114 -- Normal termination.
17115
17116 CC The client aborted before the connection could be established to the
17117 server. This can happen when haproxy tries to connect to a recently
17118 dead (or unchecked) server, and the client aborts while haproxy is
17119 waiting for the server to respond or for "timeout connect" to expire.
17120
17121 CD The client unexpectedly aborted during data transfer. This can be
17122 caused by a browser crash, by an intermediate equipment between the
17123 client and haproxy which decided to actively break the connection,
17124 by network routing issues between the client and haproxy, or by a
17125 keep-alive session between the server and the client terminated first
17126 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017127
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017128 cD The client did not send nor acknowledge any data for as long as the
17129 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017130 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017131
17132 CH The client aborted while waiting for the server to start responding.
17133 It might be the server taking too long to respond or the client
17134 clicking the 'Stop' button too fast.
17135
17136 cH The "timeout client" stroke while waiting for client data during a
17137 POST request. This is sometimes caused by too large TCP MSS values
17138 for PPPoE networks which cannot transport full-sized packets. It can
17139 also happen when client timeout is smaller than server timeout and
17140 the server takes too long to respond.
17141
17142 CQ The client aborted while its session was queued, waiting for a server
17143 with enough empty slots to accept it. It might be that either all the
17144 servers were saturated or that the assigned server was taking too
17145 long a time to respond.
17146
17147 CR The client aborted before sending a full HTTP request. Most likely
17148 the request was typed by hand using a telnet client, and aborted
17149 too early. The HTTP status code is likely a 400 here. Sometimes this
17150 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017151 and the client. "option http-ignore-probes" can be used to ignore
17152 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017153
17154 cR The "timeout http-request" stroke before the client sent a full HTTP
17155 request. This is sometimes caused by too large TCP MSS values on the
17156 client side for PPPoE networks which cannot transport full-sized
17157 packets, or by clients sending requests by hand and not typing fast
17158 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017159 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017160 some browsers started to implement a "pre-connect" feature consisting
17161 in speculatively connecting to some recently visited web sites just
17162 in case the user would like to visit them. This results in many
17163 connections being established to web sites, which end up in 408
17164 Request Timeout if the timeout strikes first, or 400 Bad Request when
17165 the browser decides to close them first. These ones pollute the log
17166 and feed the error counters. Some versions of some browsers have even
17167 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017168 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017169 http-ignore-probes" in the frontend, resulting in connections with
17170 zero data transfer to be totally ignored. This will definitely hide
17171 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017172
17173 CT The client aborted while its session was tarpitted. It is important to
17174 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017175 wrong tarpit rules have been written. If a lot of them happen, it
17176 might make sense to lower the "timeout tarpit" value to something
17177 closer to the average reported "Tw" timer, in order not to consume
17178 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017179
Willy Tarreau570f2212013-06-10 16:42:09 +020017180 LR The request was intercepted and locally handled by haproxy. Generally
17181 it means that this was a redirect or a stats request.
17182
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017183 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017184 the TCP connection (the proxy received a TCP RST or an ICMP message
17185 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017186 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017187 or no ARP response on local network). When this happens in HTTP mode,
17188 the status code is likely a 502 or 503 here.
17189
17190 sC The "timeout connect" stroke before a connection to the server could
17191 complete. When this happens in HTTP mode, the status code is likely a
17192 503 or 504 here.
17193
17194 SD The connection to the server died with an error during the data
17195 transfer. This usually means that haproxy has received an RST from
17196 the server or an ICMP message from an intermediate equipment while
17197 exchanging data with the server. This can be caused by a server crash
17198 or by a network issue on an intermediate equipment.
17199
17200 sD The server did not send nor acknowledge any data for as long as the
17201 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017202 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017203 load-balancers, ...), as well as keep-alive sessions maintained
17204 between the client and the server expiring first on haproxy.
17205
17206 SH The server aborted before sending its full HTTP response headers, or
17207 it crashed while processing the request. Since a server aborting at
17208 this moment is very rare, it would be wise to inspect its logs to
17209 control whether it crashed and why. The logged request may indicate a
17210 small set of faulty requests, demonstrating bugs in the application.
17211 Sometimes this might also be caused by an IDS killing the connection
17212 between haproxy and the server.
17213
17214 sH The "timeout server" stroke before the server could return its
17215 response headers. This is the most common anomaly, indicating too
17216 long transactions, probably caused by server or database saturation.
17217 The immediate workaround consists in increasing the "timeout server"
17218 setting, but it is important to keep in mind that the user experience
17219 will suffer from these long response times. The only long term
17220 solution is to fix the application.
17221
17222 sQ The session spent too much time in queue and has been expired. See
17223 the "timeout queue" and "timeout connect" settings to find out how to
17224 fix this if it happens too often. If it often happens massively in
17225 short periods, it may indicate general problems on the affected
17226 servers due to I/O or database congestion, or saturation caused by
17227 external attacks.
17228
17229 PC The proxy refused to establish a connection to the server because the
17230 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017231 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017232 so that it does not happen anymore. This status is very rare and
17233 might happen when the global "ulimit-n" parameter is forced by hand.
17234
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017235 PD The proxy blocked an incorrectly formatted chunked encoded message in
17236 a request or a response, after the server has emitted its headers. In
17237 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017238 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017239 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017240
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017241 PH The proxy blocked the server's response, because it was invalid,
17242 incomplete, dangerous (cache control), or matched a security filter.
17243 In any case, an HTTP 502 error is sent to the client. One possible
17244 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017245 containing unauthorized characters. It is also possible but quite
17246 rare, that the proxy blocked a chunked-encoding request from the
17247 client due to an invalid syntax, before the server responded. In this
17248 case, an HTTP 400 error is sent to the client and reported in the
17249 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017250
17251 PR The proxy blocked the client's HTTP request, either because of an
17252 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17253 the client, or because a deny filter matched, in which case it
17254 returned an HTTP 403 error.
17255
17256 PT The proxy blocked the client's request and has tarpitted its
17257 connection before returning it a 500 server error. Nothing was sent
17258 to the server. The connection was maintained open for as long as
17259 reported by the "Tw" timer field.
17260
17261 RC A local resource has been exhausted (memory, sockets, source ports)
17262 preventing the connection to the server from establishing. The error
17263 logs will tell precisely what was missing. This is very rare and can
17264 only be solved by proper system tuning.
17265
Willy Tarreau996a92c2010-10-13 19:30:47 +020017266The combination of the two last flags gives a lot of information about how
17267persistence was handled by the client, the server and by haproxy. This is very
17268important to troubleshoot disconnections, when users complain they have to
17269re-authenticate. The commonly encountered flags are :
17270
17271 -- Persistence cookie is not enabled.
17272
17273 NN No cookie was provided by the client, none was inserted in the
17274 response. For instance, this can be in insert mode with "postonly"
17275 set on a GET request.
17276
17277 II A cookie designating an invalid server was provided by the client,
17278 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017279 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017280 value can be presented by a client when no other server knows it.
17281
17282 NI No cookie was provided by the client, one was inserted in the
17283 response. This typically happens for first requests from every user
17284 in "insert" mode, which makes it an easy way to count real users.
17285
17286 VN A cookie was provided by the client, none was inserted in the
17287 response. This happens for most responses for which the client has
17288 already got a cookie.
17289
17290 VU A cookie was provided by the client, with a last visit date which is
17291 not completely up-to-date, so an updated cookie was provided in
17292 response. This can also happen if there was no date at all, or if
17293 there was a date but the "maxidle" parameter was not set, so that the
17294 cookie can be switched to unlimited time.
17295
17296 EI A cookie was provided by the client, with a last visit date which is
17297 too old for the "maxidle" parameter, so the cookie was ignored and a
17298 new cookie was inserted in the response.
17299
17300 OI A cookie was provided by the client, with a first visit date which is
17301 too old for the "maxlife" parameter, so the cookie was ignored and a
17302 new cookie was inserted in the response.
17303
17304 DI The server designated by the cookie was down, a new server was
17305 selected and a new cookie was emitted in the response.
17306
17307 VI The server designated by the cookie was not marked dead but could not
17308 be reached. A redispatch happened and selected another one, which was
17309 then advertised in the response.
17310
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017311
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173128.6. Non-printable characters
17313-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017314
17315In order not to cause trouble to log analysis tools or terminals during log
17316consulting, non-printable characters are not sent as-is into log files, but are
17317converted to the two-digits hexadecimal representation of their ASCII code,
17318prefixed by the character '#'. The only characters that can be logged without
17319being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17320escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17321is the same for the character '"' which becomes "#22", as well as '{', '|' and
17322'}' when logging headers.
17323
17324Note that the space character (' ') is not encoded in headers, which can cause
17325issues for tools relying on space count to locate fields. A typical header
17326containing spaces is "User-Agent".
17327
17328Last, it has been observed that some syslog daemons such as syslog-ng escape
17329the quote ('"') with a backslash ('\'). The reverse operation can safely be
17330performed since no quote may appear anywhere else in the logs.
17331
17332
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173338.7. Capturing HTTP cookies
17334---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017335
17336Cookie capture simplifies the tracking a complete user session. This can be
17337achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017338section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017339cookie will simultaneously be checked in the request ("Cookie:" header) and in
17340the response ("Set-Cookie:" header). The respective values will be reported in
17341the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017342locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017343not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17344user switches to a new session for example, because the server will reassign it
17345a new cookie. It is also possible to detect if a server unexpectedly sets a
17346wrong cookie to a client, leading to session crossing.
17347
17348 Examples :
17349 # capture the first cookie whose name starts with "ASPSESSION"
17350 capture cookie ASPSESSION len 32
17351
17352 # capture the first cookie whose name is exactly "vgnvisitor"
17353 capture cookie vgnvisitor= len 32
17354
17355
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173568.8. Capturing HTTP headers
17357---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017358
17359Header captures are useful to track unique request identifiers set by an upper
17360proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17361the response, one can search for information about the response length, how the
17362server asked the cache to behave, or an object location during a redirection.
17363
17364Header captures are performed using the "capture request header" and "capture
17365response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017366section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017367
17368It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017369time. Non-existent headers are logged as empty strings, and if one header
17370appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017371are grouped within braces '{' and '}' in the same order as they were declared,
17372and delimited with a vertical bar '|' without any space. Response headers
17373follow the same representation, but are displayed after a space following the
17374request headers block. These blocks are displayed just before the HTTP request
17375in the logs.
17376
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017377As a special case, it is possible to specify an HTTP header capture in a TCP
17378frontend. The purpose is to enable logging of headers which will be parsed in
17379an HTTP backend if the request is then switched to this HTTP backend.
17380
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017381 Example :
17382 # This instance chains to the outgoing proxy
17383 listen proxy-out
17384 mode http
17385 option httplog
17386 option logasap
17387 log global
17388 server cache1 192.168.1.1:3128
17389
17390 # log the name of the virtual server
17391 capture request header Host len 20
17392
17393 # log the amount of data uploaded during a POST
17394 capture request header Content-Length len 10
17395
17396 # log the beginning of the referrer
17397 capture request header Referer len 20
17398
17399 # server name (useful for outgoing proxies only)
17400 capture response header Server len 20
17401
17402 # logging the content-length is useful with "option logasap"
17403 capture response header Content-Length len 10
17404
Davor Ocelice9ed2812017-12-25 17:49:28 +010017405 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017406 capture response header Cache-Control len 8
17407
17408 # the Via header will report the next proxy's name
17409 capture response header Via len 20
17410
17411 # log the URL location during a redirection
17412 capture response header Location len 20
17413
17414 >>> Aug 9 20:26:09 localhost \
17415 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17416 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17417 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17418 "GET http://fr.adserver.yahoo.com/"
17419
17420 >>> Aug 9 20:30:46 localhost \
17421 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17422 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17423 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017424 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017425
17426 >>> Aug 9 20:30:46 localhost \
17427 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17428 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17429 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17430 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017431 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017432
17433
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174348.9. Examples of logs
17435---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017436
17437These are real-world examples of logs accompanied with an explanation. Some of
17438them have been made up by hand. The syslog part has been removed for better
17439reading. Their sole purpose is to explain how to decipher them.
17440
17441 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17442 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17443 "HEAD / HTTP/1.0"
17444
17445 => long request (6.5s) entered by hand through 'telnet'. The server replied
17446 in 147 ms, and the session ended normally ('----')
17447
17448 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17449 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17450 0/9 "HEAD / HTTP/1.0"
17451
17452 => Idem, but the request was queued in the global queue behind 9 other
17453 requests, and waited there for 1230 ms.
17454
17455 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17456 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17457 "GET /image.iso HTTP/1.0"
17458
17459 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017460 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017461 14 ms, 243 bytes of headers were sent to the client, and total time from
17462 accept to first data byte is 30 ms.
17463
17464 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17465 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17466 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17467
17468 => the proxy blocked a server response either because of an "rspdeny" or
17469 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017470 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017471 risked being cached. In this case, the response is replaced with a "502
17472 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17473 to return the 502 and not the server.
17474
17475 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017476 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 +010017477
17478 => the client never completed its request and aborted itself ("C---") after
17479 8.5s, while the proxy was waiting for the request headers ("-R--").
17480 Nothing was sent to any server.
17481
17482 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17483 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17484
17485 => The client never completed its request, which was aborted by the
17486 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017487 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017488 send a 408 return code to the client.
17489
17490 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17491 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17492
17493 => This log was produced with "option tcplog". The client timed out after
17494 5 seconds ("c----").
17495
17496 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17497 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017498 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017499
17500 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017501 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017502 (config says 'retries 3'), and no redispatch (otherwise we would have
17503 seen "/+3"). Status code 503 was returned to the client. There were 115
17504 connections on this server, 202 connections on this proxy, and 205 on
17505 the global process. It is possible that the server refused the
17506 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017507
Willy Tarreau52b2d222011-09-07 23:48:48 +020017508
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200175099. Supported filters
17510--------------------
17511
17512Here are listed officially supported filters with the list of parameters they
17513accept. Depending on compile options, some of these filters might be
17514unavailable. The list of available filters is reported in haproxy -vv.
17515
17516See also : "filter"
17517
175189.1. Trace
17519----------
17520
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017521filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017522
17523 Arguments:
17524 <name> is an arbitrary name that will be reported in
17525 messages. If no name is provided, "TRACE" is used.
17526
17527 <random-parsing> enables the random parsing of data exchanged between
17528 the client and the server. By default, this filter
17529 parses all available data. With this parameter, it
17530 only parses a random amount of the available data.
17531
Davor Ocelice9ed2812017-12-25 17:49:28 +010017532 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017533 default, this filter forwards all previously parsed
17534 data. With this parameter, it only forwards a random
17535 amount of the parsed data.
17536
Davor Ocelice9ed2812017-12-25 17:49:28 +010017537 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017538
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017539This filter can be used as a base to develop new filters. It defines all
17540callbacks and print a message on the standard error stream (stderr) with useful
17541information for all of them. It may be useful to debug the activity of other
17542filters or, quite simply, HAProxy's activity.
17543
17544Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17545tests the behavior of a filter that parses data exchanged between a client and
17546a server by adding some latencies in the processing.
17547
17548
175499.2. HTTP compression
17550---------------------
17551
17552filter compression
17553
17554The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17555keyword must still be used to enable and configure the HTTP compression. And
17556when no other filter is used, it is enough. But it is mandatory to explicitly
17557use a filter line to enable the HTTP compression when two or more filters are
17558used for the same listener/frontend/backend. This is important to know the
17559filters evaluation order.
17560
17561See also : "compression"
17562
17563
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200175649.3. Stream Processing Offload Engine (SPOE)
17565--------------------------------------------
17566
17567filter spoe [engine <name>] config <file>
17568
17569 Arguments :
17570
17571 <name> is the engine name that will be used to find the right scope in
17572 the configuration file. If not provided, all the file will be
17573 parsed.
17574
17575 <file> is the path of the engine configuration file. This file can
17576 contain configuration of several engines. In this case, each
17577 part must be placed in its own scope.
17578
17579The Stream Processing Offload Engine (SPOE) is a filter communicating with
17580external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017581streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017582exchanged with them are configured in dedicated files, for the main part. It
17583also requires dedicated backends, defined in HAProxy configuration.
17584
17585SPOE communicates with external components using an in-house binary protocol,
17586the Stream Processing Offload Protocol (SPOP).
17587
Tim Düsterhus4896c442016-11-29 02:15:19 +010017588For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017589"doc/SPOE.txt".
17590
17591Important note:
17592 The SPOE filter is highly experimental for now and was not heavily
17593 tested. It is really not production ready. So use it carefully.
17594
William Lallemand86d0df02017-11-24 21:36:45 +01001759510. Cache
17596---------
17597
17598HAProxy provides a cache, which was designed to perform cache on small objects
17599(favicon, css...). This is a minimalist low-maintenance cache which runs in
17600RAM.
17601
17602The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017603this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017604
17605If an object is not used anymore, it can be deleted to store a new object
17606independently of its expiration date. The oldest objects are deleted first
17607when we try to allocate a new one.
17608
Cyril Bonté7b888f12017-11-26 22:24:31 +010017609The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017610
17611It's possible to view the status of a cache using the Unix socket command
17612"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17613for more details.
17614
17615When an object is delivered from the cache, the server name in the log is
17616replaced by "<CACHE>".
17617
Cyril Bonté7b888f12017-11-26 22:24:31 +01001761810.1. Limitation
17619----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017620
17621The cache won't store and won't deliver objects in these cases:
17622
17623- If the response is not a 200
17624- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017625- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010017626- If the response is not cacheable
17627
17628- If the request is not a GET
17629- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020017630- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010017631
17632Caution!: Due to the current limitation of the filters, it is not recommended
17633to use the cache with other filters. Using them can cause undefined behavior
Cyril Bonté7b888f12017-11-26 22:24:31 +010017634if they modify the response (compression for example).
William Lallemand86d0df02017-11-24 21:36:45 +010017635
Cyril Bonté7b888f12017-11-26 22:24:31 +01001763610.2. Setup
17637-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017638
17639To setup a cache, you must define a cache section and use it in a proxy with
17640the corresponding http-request and response actions.
17641
Cyril Bonté7b888f12017-11-26 22:24:31 +01001764210.2.1. Cache section
17643---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017644
17645cache <name>
17646 Declare a cache section, allocate a shared cache memory named <name>, the
17647 size of cache is mandatory.
17648
17649total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017650 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017651 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010017652
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017653max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017654 Define the maximum size of the objects to be cached. Must not be greater than
17655 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
17656 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017657
William Lallemand86d0df02017-11-24 21:36:45 +010017658max-age <seconds>
17659 Define the maximum expiration duration. The expiration is set has the lowest
17660 value between the s-maxage or max-age (in this order) directive in the
17661 Cache-Control response header and this value. The default value is 60
17662 seconds, which means that you can't cache an object more than 60 seconds by
17663 default.
17664
Cyril Bonté7b888f12017-11-26 22:24:31 +01001766510.2.2. Proxy section
17666---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017667
17668http-request cache-use <name>
17669 Try to deliver a cached object from the cache <name>. This directive is also
17670 mandatory to store the cache as it calculates the cache hash. If you want to
17671 use a condition for both storage and delivering that's a good idea to put it
17672 after this one.
17673
17674http-response cache-store <name>
17675 Store an http-response within the cache. The storage of the response headers
17676 is done at this step, which means you can use others http-response actions
17677 to modify headers before or after the storage of the response. This action
17678 is responsible for the setup of the cache storage filter.
17679
17680
17681Example:
17682
17683 backend bck1
17684 mode http
17685
17686 http-request cache-use foobar
17687 http-response cache-store foobar
17688 server srv1 127.0.0.1:80
17689
17690 cache foobar
17691 total-max-size 4
17692 max-age 240
17693
Willy Tarreau0ba27502007-12-24 16:55:16 +010017694/*
17695 * Local variables:
17696 * fill-column: 79
17697 * End:
17698 */