blob: 6c22ab2603d0a5c784545189d195e2142d005a4f [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 Tarreau01fbe742018-10-21 20:28:30 +02007 2018/10/21
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
Robert Tsai81ae1952007-12-05 10:47:29 +0100840 - A filesystem path to a UNIX domain socket, keeping in mind
841 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
William Lallemandb2f07452015-05-12 14:27:13 +0200845 You may want to reference some environment variables in the address
846 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100847
Willy Tarreau18324f52014-06-27 18:10:07 +0200848 <length> is an optional maximum line length. Log lines larger than this value
849 will be truncated before being sent. The reason is that syslog
850 servers act differently on log line length. All servers support the
851 default value of 1024, but some servers simply drop larger lines
852 while others do log them. If a server supports long lines, it may
853 make sense to set this value here in order to avoid truncating long
854 lines. Similarly, if a server drops long lines, it is preferable to
855 truncate them before sending them. Accepted values are 80 to 65535
856 inclusive. The default value of 1024 is generally fine for all
857 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100858 JSON-formatted logs may require larger values. You may also need to
859 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200860
Dragan Dosen7ad31542015-09-28 17:16:47 +0200861 <format> is the log format used when generating syslog messages. It may be
862 one of the following :
863
864 rfc3164 The RFC3164 syslog message format. This is the default.
865 (https://tools.ietf.org/html/rfc3164)
866
867 rfc5424 The RFC5424 syslog message format.
868 (https://tools.ietf.org/html/rfc5424)
869
Robert Tsai81ae1952007-12-05 10:47:29 +0100870 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200871
872 kern user mail daemon auth syslog lpr news
873 uucp cron auth2 ftp ntp audit alert cron2
874 local0 local1 local2 local3 local4 local5 local6 local7
875
876 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200877 all messages are sent. If a maximum level is specified, only messages with a
878 severity at least as important as this level will be sent. An optional minimum
879 level can be specified. If it is set, logs emitted with a more severe level
880 than this one will be capped to this level. This is used to avoid sending
881 "emerg" messages on all terminals on some default syslog configurations.
882 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200883
Cyril Bontédc4d9032012-04-08 21:57:39 +0200884 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200885
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100886log-send-hostname [<string>]
887 Sets the hostname field in the syslog header. If optional "string" parameter
888 is set the header is set to the string contents, otherwise uses the hostname
889 of the system. Generally used if one is not relaying logs through an
890 intermediate syslog server or for simply customizing the hostname printed in
891 the logs.
892
Kevinm48936af2010-12-22 16:08:21 +0000893log-tag <string>
894 Sets the tag field in the syslog header to this string. It defaults to the
895 program name as launched from the command line, which usually is "haproxy".
896 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100897 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000898
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100899lua-load <file>
900 This global directive loads and executes a Lua file. This directive can be
901 used multiple times.
902
William Lallemand4cfede82017-11-24 22:02:34 +0100903master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200904 Master-worker mode. It is equivalent to the command line "-W" argument.
905 This mode will launch a "master" which will monitor the "workers". Using
906 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100907 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200908 or daemon mode. It is recommended to use this mode with multiprocess and
909 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100910 By default, if a worker exits with a bad return code, in the case of a
911 segfault for example, all workers will be killed, and the master will leave.
912 It is convenient to combine this behavior with Restart=on-failure in a
913 systemd unit file in order to relaunch the whole process. If you don't want
914 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200915
William Lallemand4cfede82017-11-24 22:02:34 +0100916 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200917
Willy Tarreau6a06a402007-07-15 20:15:28 +0200918nbproc <number>
919 Creates <number> processes when going daemon. This requires the "daemon"
920 mode. By default, only one process is created, which is the recommended mode
921 of operation. For systems limited to small sets of file descriptors per
922 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
923 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
924
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200925nbthread <number>
926 This setting is only available when support for threads was built in. It
927 creates <number> threads for each created processes. It means if HAProxy is
928 started in foreground, it only creates <number> threads for the first
929 process. FOR NOW, THREADS SUPPORT IN HAPROXY IS HIGHLY EXPERIMENTAL AND IT
930 MUST BE ENABLED WITH CAUTION AND AT YOUR OWN RISK. See also "nbproc".
931
Willy Tarreau6a06a402007-07-15 20:15:28 +0200932pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100933 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200934 the "-p" command line argument. The file must be accessible to the user
935 starting the process. See also "daemon".
936
Willy Tarreau1d549722016-02-16 12:41:57 +0100937presetenv <name> <value>
938 Sets environment variable <name> to value <value>. If the variable exists, it
939 is NOT overwritten. The changes immediately take effect so that the next line
940 in the configuration file sees the new value. See also "setenv", "resetenv",
941 and "unsetenv".
942
943resetenv [<name> ...]
944 Removes all environment variables except the ones specified in argument. It
945 allows to use a clean controlled environment before setting new values with
946 setenv or unsetenv. Please note that some internal functions may make use of
947 some environment variables, such as time manipulation functions, but also
948 OpenSSL or even external checks. This must be used with extreme care and only
949 after complete validation. The changes immediately take effect so that the
950 next line in the configuration file sees the new environment. See also
951 "setenv", "presetenv", and "unsetenv".
952
Christopher Fauletff4121f2017-11-22 16:38:49 +0100953stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200954 Limits the stats socket to a certain set of processes numbers. By default the
955 stats socket is bound to all processes, causing a warning to be emitted when
956 nbproc is greater than 1 because there is no way to select the target process
957 when connecting. However, by using this setting, it becomes possible to pin
958 the stats socket to a specific set of processes, typically the first one. The
959 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100960 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +0100961 word size (32 or 64). Ranges can be partially defined. The higher bound can
962 be omitted. In such case, it is replaced by the corresponding maximum
963 value. A better option consists in using the "process" setting of the "stats
964 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +0200965
Baptiste Assmann5626f482015-08-23 10:00:10 +0200966server-state-base <directory>
967 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200968 file names which do not start with a '/'. See also "server-state-file",
969 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200970
971server-state-file <file>
972 Specifies the path to the file containing state of servers. If the path starts
973 with a slash ('/'), it is considered absolute, otherwise it is considered
974 relative to the directory specified using "server-state-base" (if set) or to
975 the current directory. Before reloading HAProxy, it is possible to save the
976 servers' current state using the stats command "show servers state". The
977 output of this command must be written in the file pointed by <file>. When
978 starting up, before handling traffic, HAProxy will read, load and apply state
979 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200980 configuration. See also "server-state-base" and "show servers state",
981 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +0200982
Willy Tarreau1d549722016-02-16 12:41:57 +0100983setenv <name> <value>
984 Sets environment variable <name> to value <value>. If the variable exists, it
985 is overwritten. The changes immediately take effect so that the next line in
986 the configuration file sees the new value. See also "presetenv", "resetenv",
987 and "unsetenv".
988
Willy Tarreau610f04b2014-02-13 11:36:41 +0100989ssl-default-bind-ciphers <ciphers>
990 This setting is only available when support for OpenSSL was built in. It sets
991 the default string describing the list of cipher algorithms ("cipher suite")
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200992 that are negotiated during the SSL/TLS handshake except for TLSv1.3 for all
993 "bind" lines which do not explicitly define theirs. The format of the string
994 is defined in "man 1 ciphers" from OpenSSL man pages, and can be for instance
995 a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). For
996 TLSv1.3 cipher configuration, please check the "ssl-default-bind-ciphersuites"
997 keyword. Please check the "bind" keyword for more information.
998
999ssl-default-bind-ciphersuites <ciphersuites>
1000 This setting is only available when support for OpenSSL was built in and
1001 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1002 describing the list of cipher algorithms ("cipher suite") that are negotiated
1003 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1004 theirs. The format of the string is defined in
1005 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites", and can
1006 be for instance a string such as
1007 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
1008 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
1009 the "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
1010 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001011
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001012ssl-default-bind-options [<option>]...
1013 This setting is only available when support for OpenSSL was built in. It sets
1014 default ssl-options to force on all "bind" lines. Please check the "bind"
1015 keyword to see available options.
1016
1017 Example:
1018 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001019 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001020
Willy Tarreau610f04b2014-02-13 11:36:41 +01001021ssl-default-server-ciphers <ciphers>
1022 This setting is only available when support for OpenSSL was built in. It
1023 sets the default string describing the list of cipher algorithms that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001024 negotiated during the SSL/TLS handshake except for TLSv1.3 with the server,
1025 for all "server" lines which do not explicitly define theirs. The format of
1026 the string is defined in "man 1 ciphers". For TLSv1.3 cipher configuration,
1027 please check the "ssl-default-server-ciphersuites" keyword. Please check the
1028 "server" keyword for more information.
1029
1030ssl-default-server-ciphersuites <ciphersuites>
1031 This setting is only available when support for OpenSSL was built in and
1032 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1033 string describing the list of cipher algorithms that are negotiated during
1034 the TLSv1.3 handshake with the server, for all "server" lines which do not
1035 explicitly define theirs. The format of the string is defined in
1036 "man 1 ciphers" under the "ciphersuites" section. For cipher configuration for
1037 TLSv1.2 and earlier, please check the "ssl-default-server-ciphers" keyword.
1038 Please check the "server" keyword for more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001039
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001040ssl-default-server-options [<option>]...
1041 This setting is only available when support for OpenSSL was built in. It sets
1042 default ssl-options to force on all "server" lines. Please check the "server"
1043 keyword to see available options.
1044
Remi Gacogne47783ef2015-05-29 15:53:22 +02001045ssl-dh-param-file <file>
1046 This setting is only available when support for OpenSSL was built in. It sets
1047 the default DH parameters that are used during the SSL/TLS handshake when
1048 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001049 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001050 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001051 are not specified either by using ssl-dh-param-file or by setting them
1052 directly in the certificate file, pre-generated DH parameters of the size
1053 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1054 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001055 Custom DH parameters may be generated by using the OpenSSL command
1056 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1057 parameters should not be considered secure anymore.
1058
Emeric Brun850efd52014-01-29 12:24:34 +01001059ssl-server-verify [none|required]
1060 The default behavior for SSL verify on servers side. If specified to 'none',
1061 servers certificates are not verified. The default is 'required' except if
1062 forced using cmdline option '-dV'.
1063
Willy Tarreauabb175f2012-09-24 12:43:26 +02001064stats socket [<address:port>|<path>] [param*]
1065 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1066 Connections to this socket will return various statistics outputs and even
1067 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001068 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001069 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001070
Willy Tarreauabb175f2012-09-24 12:43:26 +02001071 All parameters supported by "bind" lines are supported, for instance to
1072 restrict access to some users or their access rights. Please consult
1073 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001074
1075stats timeout <timeout, in milliseconds>
1076 The default timeout on the stats socket is set to 10 seconds. It is possible
1077 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001078 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001079
1080stats maxconn <connections>
1081 By default, the stats socket is limited to 10 concurrent connections. It is
1082 possible to change this value with "stats maxconn".
1083
Willy Tarreau6a06a402007-07-15 20:15:28 +02001084uid <number>
1085 Changes the process' user ID to <number>. It is recommended that the user ID
1086 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1087 be started with superuser privileges in order to be able to switch to another
1088 one. See also "gid" and "user".
1089
1090ulimit-n <number>
1091 Sets the maximum number of per-process file-descriptors to <number>. By
1092 default, it is automatically computed, so it is recommended not to use this
1093 option.
1094
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001095unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1096 [ group <group> ] [ gid <gid> ]
1097
1098 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1099 This is mainly used to simplify declaration of those UNIX sockets and reduce
1100 the risk of errors, since those settings are most commonly required but are
1101 also process-specific. The <prefix> setting can be used to force all socket
1102 path to be relative to that directory. This might be needed to access another
1103 component's chroot. Note that those paths are resolved before haproxy chroots
1104 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1105 all have the same meaning as their homonyms used by the "bind" statement. If
1106 both are specified, the "bind" statement has priority, meaning that the
1107 "unix-bind" settings may be seen as process-wide default settings.
1108
Willy Tarreau1d549722016-02-16 12:41:57 +01001109unsetenv [<name> ...]
1110 Removes environment variables specified in arguments. This can be useful to
1111 hide some sensitive information that are occasionally inherited from the
1112 user's environment during some operations. Variables which did not exist are
1113 silently ignored so that after the operation, it is certain that none of
1114 these variables remain. The changes immediately take effect so that the next
1115 line in the configuration file will not see these variables. See also
1116 "setenv", "presetenv", and "resetenv".
1117
Willy Tarreau6a06a402007-07-15 20:15:28 +02001118user <user name>
1119 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1120 See also "uid" and "group".
1121
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001122node <name>
1123 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1124
1125 This statement is useful in HA configurations where two or more processes or
1126 servers share the same IP address. By setting a different node-name on all
1127 nodes, it becomes easy to immediately spot what server is handling the
1128 traffic.
1129
1130description <text>
1131 Add a text that describes the instance.
1132
1133 Please note that it is required to escape certain characters (# for example)
1134 and this text is inserted into a html page so you should avoid using
1135 "<" and ">" characters.
1136
Thomas Holmesdb04f192015-05-18 13:21:39 +0100113751degrees-data-file <file path>
1138 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001139 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001140
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001141 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001142 compiled with USE_51DEGREES.
1143
Ben Shillitof25e8e52016-12-02 14:25:37 +0000114451degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001145 A list of 51Degrees property names to be load from the dataset. A full list
1146 of names is available on the 51Degrees website:
1147 https://51degrees.com/resources/property-dictionary
1148
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001149 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001150 compiled with USE_51DEGREES.
1151
Dragan Dosen93b38d92015-06-29 16:43:25 +0200115251degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001153 A char that will be appended to every property value in a response header
1154 containing 51Degrees results. If not set that will be set as ','.
1155
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001156 Please note that this option is only available when haproxy has been
1157 compiled with USE_51DEGREES.
1158
115951degrees-cache-size <number>
1160 Sets the size of the 51Degrees converter cache to <number> entries. This
1161 is an LRU cache which reminds previous device detections and their results.
1162 By default, this cache is disabled.
1163
1164 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001165 compiled with USE_51DEGREES.
1166
scientiamobiled0027ed2016-11-04 10:55:08 +01001167wurfl-data-file <file path>
1168 The path of the WURFL data file to provide device detection services. The
1169 file should be accessible by HAProxy with relevant permissions.
1170
1171 Please note that this option is only available when haproxy has been compiled
1172 with USE_WURFL=1.
1173
1174wurfl-information-list [<capability>]*
1175 A space-delimited list of WURFL capabilities, virtual capabilities, property
1176 names we plan to use in injected headers. A full list of capability and
1177 virtual capability names is available on the Scientiamobile website :
1178
1179 https://www.scientiamobile.com/wurflCapability
1180
1181 Valid WURFL properties are:
1182 - wurfl_id Contains the device ID of the matched device.
1183
1184 - wurfl_root_id Contains the device root ID of the matched
1185 device.
1186
1187 - wurfl_isdevroot Tells if the matched device is a root device.
1188 Possible values are "TRUE" or "FALSE".
1189
1190 - wurfl_useragent The original useragent coming with this
1191 particular web request.
1192
1193 - wurfl_api_version Contains a string representing the currently
1194 used Libwurfl API version.
1195
1196 - wurfl_engine_target Contains a string representing the currently
1197 set WURFL Engine Target. Possible values are
1198 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1199
1200 - wurfl_info A string containing information on the parsed
1201 wurfl.xml and its full path.
1202
1203 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1204 WURFL has been loaded successfully.
1205
1206 - wurfl_normalized_useragent The normalized useragent.
1207
1208 - wurfl_useragent_priority The user agent priority used by WURFL.
1209
1210 Please note that this option is only available when haproxy has been compiled
1211 with USE_WURFL=1.
1212
1213wurfl-information-list-separator <char>
1214 A char that will be used to separate values in a response header containing
1215 WURFL results. If not set that a comma (',') will be used by default.
1216
1217 Please note that this option is only available when haproxy has been compiled
1218 with USE_WURFL=1.
1219
1220wurfl-patch-file [<file path>]
1221 A list of WURFL patch file paths. Note that patches are loaded during startup
1222 thus before the chroot.
1223
1224 Please note that this option is only available when haproxy has been compiled
1225 with USE_WURFL=1.
1226
1227wurfl-engine-mode { accuracy | performance }
1228 Sets the WURFL engine target. You can choose between 'accuracy' or
1229 'performance' targets. In performance mode, desktop web browser detection is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001230 done programmatically without referencing the WURFL data. As a result, most
scientiamobiled0027ed2016-11-04 10:55:08 +01001231 desktop web browsers are returned as generic_web_browser WURFL ID for
Davor Ocelice9ed2812017-12-25 17:49:28 +01001232 performance. If either performance or accuracy are not defined, performance
scientiamobiled0027ed2016-11-04 10:55:08 +01001233 mode is enabled by default.
1234
1235 Please note that this option is only available when haproxy has been compiled
1236 with USE_WURFL=1.
1237
1238wurfl-cache-size <U>[,<D>]
1239 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1240 <D> is the internal device cache size. There are three possibilities here :
1241 - "0" : no cache is used.
1242 - <U> : the Single LRU cache is used, the size is expressed in elements.
1243 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1244 the highest performing option.
1245
1246 Please note that this option is only available when haproxy has been compiled
1247 with USE_WURFL=1.
1248
1249wurfl-useragent-priority { plain | sideloaded_browser }
1250 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1251 over the default sideloaded browser user agent ('sideloaded_browser').
1252
1253 Please note that this option is only available when haproxy has been compiled
1254 with USE_WURFL=1.
1255
Willy Tarreau6a06a402007-07-15 20:15:28 +02001256
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012573.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001258-----------------------
1259
Willy Tarreau1746eec2014-04-25 10:46:47 +02001260max-spread-checks <delay in milliseconds>
1261 By default, haproxy tries to spread the start of health checks across the
1262 smallest health check interval of all the servers in a farm. The principle is
1263 to avoid hammering services running on the same server. But when using large
1264 check intervals (10 seconds or more), the last servers in the farm take some
1265 time before starting to be tested, which can be a problem. This parameter is
1266 used to enforce an upper bound on delay between the first and the last check,
1267 even if the servers' check intervals are larger. When servers run with
1268 shorter intervals, their intervals will be respected though.
1269
Willy Tarreau6a06a402007-07-15 20:15:28 +02001270maxconn <number>
1271 Sets the maximum per-process number of concurrent connections to <number>. It
1272 is equivalent to the command-line argument "-n". Proxies will stop accepting
1273 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001274 automatically adjusted according to this value. See also "ulimit-n". Note:
1275 the "select" poller cannot reliably use more than 1024 file descriptors on
1276 some platforms. If your platform only supports select and reports "select
1277 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001278 below 500 in general). If this value is not set, it will default to the value
1279 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1280 limit is enforced, or will be computed based on the memory limit, the buffer
1281 size, memory allocated to compression, SSL cache size, and use or not of SSL
1282 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001283
Willy Tarreau81c25d02011-09-07 15:17:21 +02001284maxconnrate <number>
1285 Sets the maximum per-process number of connections per second to <number>.
1286 Proxies will stop accepting connections when this limit is reached. It can be
1287 used to limit the global capacity regardless of each frontend capacity. It is
1288 important to note that this can only be used as a service protection measure,
1289 as there will not necessarily be a fair share between frontends when the
1290 limit is reached, so it's a good idea to also limit each frontend to some
1291 value close to its expected share. Also, lowering tune.maxaccept can improve
1292 fairness.
1293
William Lallemandd85f9172012-11-09 17:05:39 +01001294maxcomprate <number>
1295 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001296 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001297 level will be decreased during the session. If the maximum is reached at the
1298 beginning of a session, the session will not compress at all. If the maximum
1299 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001300 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001301 default value.
1302
William Lallemand072a2bf2012-11-20 17:01:01 +01001303maxcompcpuusage <number>
1304 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1305 for new requests or decreasing the compression level of current requests.
1306 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1307 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1308 case of multiple processes (nbproc > 1), each process manages its individual
1309 usage. A value of 100 disable the limit. The default value is 100. Setting
1310 a lower value will prevent the compression work from slowing the whole
1311 process down and from introducing high latencies.
1312
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001313maxpipes <number>
1314 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1315 are only used by kernel-based tcp splicing. Since a pipe contains two file
1316 descriptors, the "ulimit-n" value will be increased accordingly. The default
1317 value is maxconn/4, which seems to be more than enough for most heavy usages.
1318 The splice code dynamically allocates and releases pipes, and can fall back
1319 to standard copy, so setting this value too low may only impact performance.
1320
Willy Tarreau93e7c002013-10-07 18:51:07 +02001321maxsessrate <number>
1322 Sets the maximum per-process number of sessions per second to <number>.
1323 Proxies will stop accepting connections when this limit is reached. It can be
1324 used to limit the global capacity regardless of each frontend capacity. It is
1325 important to note that this can only be used as a service protection measure,
1326 as there will not necessarily be a fair share between frontends when the
1327 limit is reached, so it's a good idea to also limit each frontend to some
1328 value close to its expected share. Also, lowering tune.maxaccept can improve
1329 fairness.
1330
Willy Tarreau403edff2012-09-06 11:58:37 +02001331maxsslconn <number>
1332 Sets the maximum per-process number of concurrent SSL connections to
1333 <number>. By default there is no SSL-specific limit, which means that the
1334 global maxconn setting will apply to all connections. Setting this limit
1335 avoids having openssl use too much memory and crash when malloc returns NULL
1336 (since it unfortunately does not reliably check for such conditions). Note
1337 that the limit applies both to incoming and outgoing connections, so one
1338 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001339 If this value is not set, but a memory limit is enforced, this value will be
1340 automatically computed based on the memory limit, maxconn, the buffer size,
1341 memory allocated to compression, SSL cache size, and use of SSL in either
1342 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1343 when there is a memory limit, haproxy will automatically adjust these values
1344 so that 100% of the connections can be made over SSL with no risk, and will
1345 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001346
Willy Tarreaue43d5322013-10-07 20:01:52 +02001347maxsslrate <number>
1348 Sets the maximum per-process number of SSL sessions per second to <number>.
1349 SSL listeners will stop accepting connections when this limit is reached. It
1350 can be used to limit the global SSL CPU usage regardless of each frontend
1351 capacity. It is important to note that this can only be used as a service
1352 protection measure, as there will not necessarily be a fair share between
1353 frontends when the limit is reached, so it's a good idea to also limit each
1354 frontend to some value close to its expected share. It is also important to
1355 note that the sessions are accounted before they enter the SSL stack and not
1356 after, which also protects the stack against bad handshakes. Also, lowering
1357 tune.maxaccept can improve fairness.
1358
William Lallemand9d5f5482012-11-07 16:12:57 +01001359maxzlibmem <number>
1360 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1361 When the maximum amount is reached, future sessions will not compress as long
1362 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001363 The default value is 0. The value is available in bytes on the UNIX socket
1364 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1365 "ZlibMemUsage" in bytes.
1366
Willy Tarreau6a06a402007-07-15 20:15:28 +02001367noepoll
1368 Disables the use of the "epoll" event polling system on Linux. It is
1369 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001370 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001371
1372nokqueue
1373 Disables the use of the "kqueue" event polling system on BSD. It is
1374 equivalent to the command-line argument "-dk". The next polling system
1375 used will generally be "poll". See also "nopoll".
1376
1377nopoll
1378 Disables the use of the "poll" event polling system. It is equivalent to the
1379 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001380 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001381 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001382
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001383nosplice
1384 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001385 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001386 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001387 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001388 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1389 be used. This option makes it easier to globally disable kernel splicing in
1390 case of doubt. See also "option splice-auto", "option splice-request" and
1391 "option splice-response".
1392
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001393nogetaddrinfo
1394 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1395 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1396
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001397noreuseport
1398 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1399 command line argument "-dR".
1400
Willy Tarreaufe255b72007-10-14 23:09:26 +02001401spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001402 Sometimes it is desirable to avoid sending agent and health checks to
1403 servers at exact intervals, for instance when many logical servers are
1404 located on the same physical server. With the help of this parameter, it
1405 becomes possible to add some randomness in the check interval between 0
1406 and +/- 50%. A value between 2 and 5 seems to show good results. The
1407 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001408
Davor Ocelice9ed2812017-12-25 17:49:28 +01001409ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001410 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001411 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001412 multiple times, it will simply enable multiple crypto engines. Referencing an
1413 unsupported engine will prevent haproxy from starting. Note that many engines
1414 will lead to lower HTTPS performance than pure software with recent
1415 processors. The optional command "algo" sets the default algorithms an ENGINE
1416 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001417 of "ALL" uses the engine for all cryptographic operations. If no list of
1418 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001419 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1420 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1421 openssl configuration file uses:
1422 https://www.openssl.org/docs/man1.0.2/apps/config.html
1423
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001424ssl-mode-async
1425 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001426 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001427 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1428 doesn't support moving read/write buffers and is not compliant with
1429 haproxy's buffer management. So the asynchronous mode is disabled on
1430 read/write operations (it is only enabled during initial and reneg
1431 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001432
Willy Tarreau33cb0652014-12-23 22:52:37 +01001433tune.buffers.limit <number>
1434 Sets a hard limit on the number of buffers which may be allocated per process.
1435 The default value is zero which means unlimited. The minimum non-zero value
1436 will always be greater than "tune.buffers.reserve" and should ideally always
1437 be about twice as large. Forcing this value can be particularly useful to
1438 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001439 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001440 another one to be released by another session. Since buffers are dynamically
1441 allocated and released, the waiting time is very short and not perceptible
1442 provided that limits remain reasonable. In fact sometimes reducing the limit
1443 may even increase performance by increasing the CPU cache's efficiency. Tests
1444 have shown good results on average HTTP traffic with a limit to 1/10 of the
1445 expected global maxconn setting, which also significantly reduces memory
1446 usage. The memory savings come from the fact that a number of connections
1447 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1448 advised to do so by an haproxy core developer.
1449
Willy Tarreau1058ae72014-12-23 22:40:40 +01001450tune.buffers.reserve <number>
1451 Sets the number of buffers which are pre-allocated and reserved for use only
1452 during memory shortage conditions resulting in failed memory allocations. The
1453 minimum value is 2 and is also the default. There is no reason a user would
1454 want to change this value, it's mostly aimed at haproxy core developers.
1455
Willy Tarreau27a674e2009-08-17 07:23:33 +02001456tune.bufsize <number>
1457 Sets the buffer size to this size (in bytes). Lower values allow more
1458 sessions to coexist in the same amount of RAM, and higher values allow some
1459 applications with very large cookies to work. The default value is 16384 and
1460 can be changed at build time. It is strongly recommended not to change this
1461 from the default value, as very low values will break some services such as
1462 statistics, and values larger than default size will increase memory usage,
1463 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001464 parameter should be decreased by the same factor as this one is increased. In
1465 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1466 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001467 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
1468 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +02001469
Willy Tarreau43961d52010-10-04 20:39:20 +02001470tune.chksize <number>
1471 Sets the check buffer size to this size (in bytes). Higher values may help
1472 find string or regex patterns in very large pages, though doing so may imply
1473 more memory and CPU usage. The default value is 16384 and can be changed at
1474 build time. It is not recommended to change this value, but to use better
1475 checks whenever possible.
1476
William Lallemandf3747832012-11-09 12:33:10 +01001477tune.comp.maxlevel <number>
1478 Sets the maximum compression level. The compression level affects CPU
1479 usage during compression. This value affects CPU usage during compression.
1480 Each session using compression initializes the compression algorithm with
1481 this value. The default value is 1.
1482
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001483tune.h2.header-table-size <number>
1484 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1485 cannot be larger than 65536 bytes. A larger value may help certain clients
1486 send more compact requests, depending on their capabilities. This amount of
1487 memory is consumed for each HTTP/2 connection. It is recommended not to
1488 change it.
1489
Willy Tarreaue6baec02017-07-27 11:45:11 +02001490tune.h2.initial-window-size <number>
1491 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001492 can upload before waiting for an acknowledgment from haproxy. This setting
1493 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001494 The default value is 65535, which roughly allows up to 5 Mbps of upload
1495 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1496 over a 1-ms local network. It can make sense to increase this value to allow
1497 faster uploads, or to reduce it to increase fairness when dealing with many
1498 clients. It doesn't affect resource usage.
1499
Willy Tarreau5242ef82017-07-27 11:47:28 +02001500tune.h2.max-concurrent-streams <number>
1501 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1502 number of outstanding requests on a single connection). The default value is
1503 100. A larger one may slightly improve page load time for complex sites when
1504 visited over high latency networks, but increases the amount of resources a
1505 single client may allocate. A value of zero disables the limit so a single
1506 client may create as many streams as allocatable by haproxy. It is highly
1507 recommended not to change this value.
1508
Willy Tarreau193b8c62012-11-22 00:17:38 +01001509tune.http.cookielen <number>
1510 Sets the maximum length of captured cookies. This is the maximum value that
1511 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1512 will automatically be truncated to this one. It is important not to set too
1513 high a value because all cookie captures still allocate this size whatever
1514 their configured value (they share a same pool). This value is per request
1515 per response, so the memory allocated is twice this value per connection.
1516 When not specified, the limit is set to 63 characters. It is recommended not
1517 to change this value.
1518
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001519tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001520 Sets the maximum length of request URI in logs. This prevents truncating long
1521 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001522 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001523 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001524 configuration directives too.
1525 The default value is 1024.
1526
Willy Tarreauac1932d2011-10-24 19:14:41 +02001527tune.http.maxhdr <number>
1528 Sets the maximum number of headers in a request. When a request comes with a
1529 number of headers greater than this value (including the first line), it is
1530 rejected with a "400 Bad Request" status code. Similarly, too large responses
1531 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1532 for all usages, considering that the widely deployed Apache server uses the
1533 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001534 a buggy application to work by the time it gets fixed. The accepted range is
1535 1..32767. Keep in mind that each new header consumes 32bits of memory for
1536 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001537
Willy Tarreau7e312732014-02-12 16:35:14 +01001538tune.idletimer <timeout>
1539 Sets the duration after which haproxy will consider that an empty buffer is
1540 probably associated with an idle stream. This is used to optimally adjust
1541 some packet sizes while forwarding large and small data alternatively. The
1542 decision to use splice() or to send large buffers in SSL is modulated by this
1543 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1544 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001545 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001546 clicking). There should be not reason for changing this value. Please check
1547 tune.ssl.maxrecord below.
1548
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001549tune.lua.forced-yield <number>
1550 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001551 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001552 HAProxy scheduler to process other tasks like accepting connections or
1553 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001554 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001555 lowered. If the Lua code is quite long and its result is absolutely required
1556 to process the data, the <number> can be increased.
1557
Willy Tarreau32f61e22015-03-18 17:54:59 +01001558tune.lua.maxmem
1559 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1560 default it is zero which means unlimited. It is important to set a limit to
1561 ensure that a bug in a script will not result in the system running out of
1562 memory.
1563
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001564tune.lua.session-timeout <timeout>
1565 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001566 preventing infinite loops or spending too much time in Lua. This timeout
1567 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001568 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001569
1570tune.lua.task-timeout <timeout>
1571 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1572 dedicated to the tasks. By default, this timeout isn't set because a task may
1573 remain alive during of the lifetime of HAProxy. For example, a task used to
1574 check servers.
1575
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001576tune.lua.service-timeout <timeout>
1577 This is the execution timeout for the Lua services. This is useful for
1578 preventing infinite loops or spending too much time in Lua. This timeout
1579 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001580 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001581
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001582tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001583 Sets the maximum number of consecutive connections a process may accept in a
1584 row before switching to other work. In single process mode, higher numbers
1585 give better performance at high connection rates. However in multi-process
1586 modes, keeping a bit of fairness between processes generally is better to
1587 increase performance. This value applies individually to each listener, so
1588 that the number of processes a listener is bound to is taken into account.
1589 This value defaults to 64. In multi-process mode, it is divided by twice
1590 the number of processes the listener is bound to. Setting this value to -1
1591 completely disables the limitation. It should normally not be needed to tweak
1592 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001593
1594tune.maxpollevents <number>
1595 Sets the maximum amount of events that can be processed at once in a call to
1596 the polling system. The default value is adapted to the operating system. It
1597 has been noticed that reducing it below 200 tends to slightly decrease
1598 latency at the expense of network bandwidth, and increasing it above 200
1599 tends to trade latency for slightly increased bandwidth.
1600
Willy Tarreau27a674e2009-08-17 07:23:33 +02001601tune.maxrewrite <number>
1602 Sets the reserved buffer space to this size in bytes. The reserved space is
1603 used for header rewriting or appending. The first reads on sockets will never
1604 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1605 bufsize, though that does not make much sense since there are rarely large
1606 numbers of headers to add. Setting it too high prevents processing of large
1607 requests or responses. Setting it too low prevents addition of new headers
1608 to already large requests or to POST requests. It is generally wise to set it
1609 to about 1024. It is automatically readjusted to half of bufsize if it is
1610 larger than that. This means you don't have to worry about it when changing
1611 bufsize.
1612
Willy Tarreauf3045d22015-04-29 16:24:50 +02001613tune.pattern.cache-size <number>
1614 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1615 cache which reminds previous lookups and their results. It is used by ACLs
1616 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1617 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1618 strings. It applies to pattern expressions which means that it will be able
1619 to memorize the result of a lookup among all the patterns specified on a
1620 configuration line (including all those loaded from files). It automatically
1621 invalidates entries which are updated using HTTP actions or on the CLI. The
1622 default cache size is set to 10000 entries, which limits its footprint to
1623 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1624 risk of collision in this cache, which is in the order of the size of the
1625 cache divided by 2^64. Typically, at 10000 requests per second with the
1626 default cache size of 10000 entries, there's 1% chance that a brute force
1627 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1628 This is considered much lower than the risk of a memory corruption caused by
1629 aging components. If this is not acceptable, the cache can be disabled by
1630 setting this parameter to 0.
1631
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001632tune.pipesize <number>
1633 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1634 are the default size for the system. But sometimes when using TCP splicing,
1635 it can improve performance to increase pipe sizes, especially if it is
1636 suspected that pipes are not filled and that many calls to splice() are
1637 performed. This has an impact on the kernel's memory footprint, so this must
1638 not be changed if impacts are not understood.
1639
Willy Tarreaue803de22010-01-21 17:43:04 +01001640tune.rcvbuf.client <number>
1641tune.rcvbuf.server <number>
1642 Forces the kernel socket receive buffer size on the client or the server side
1643 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1644 and backends. It should normally never be set, and the default size (0) lets
1645 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001646 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001647 order to save kernel memory by preventing it from buffering too large amounts
1648 of received data. Lower values will significantly increase CPU usage though.
1649
Willy Tarreaub22fc302015-12-14 12:04:35 +01001650tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001651 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001652 socket buffers. One of them is that a read returns more than <recv_enough>
1653 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1654 may be changed by this setting to better deal with workloads involving lots
1655 of short messages such as telnet or SSH sessions.
1656
Olivier Houchard1599b802018-05-24 18:59:04 +02001657tune.runqueue-depth <number>
1658 Sets the maxinum amount of task that can be processed at once when running
1659 tasks. The default value is 200. Increasing it may incur latency when
1660 dealing with I/Os, making it too small can incur extra overhead.
1661
Willy Tarreaue803de22010-01-21 17:43:04 +01001662tune.sndbuf.client <number>
1663tune.sndbuf.server <number>
1664 Forces the kernel socket send buffer size on the client or the server side to
1665 the specified value in bytes. This value applies to all TCP/HTTP frontends
1666 and backends. It should normally never be set, and the default size (0) lets
1667 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001668 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001669 order to save kernel memory by preventing it from buffering too large amounts
1670 of received data. Lower values will significantly increase CPU usage though.
1671 Another use case is to prevent write timeouts with extremely slow clients due
1672 to the kernel waiting for a large part of the buffer to be read before
1673 notifying haproxy again.
1674
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001675tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001676 Sets the size of the global SSL session cache, in a number of blocks. A block
1677 is large enough to contain an encoded session without peer certificate.
1678 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001679 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001680 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001681 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001682 and reassigned. Higher values reduce the occurrence of such a purge, hence
1683 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1684 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001685 and are shared between all processes if "nbproc" is greater than 1. Setting
1686 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001687
Emeric Brun8dc60392014-05-09 13:52:00 +02001688tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001689 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001690 should normally not be used since it will force many renegotiations due to
1691 clients hitting a random process. But it may be required on some operating
1692 systems where none of the SSL cache synchronization method may be used. In
1693 this case, adding a first layer of hash-based load balancing before the SSL
1694 layer might limit the impact of the lack of session sharing.
1695
Emeric Brun4f65bff2012-11-16 15:11:00 +01001696tune.ssl.lifetime <timeout>
1697 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001698 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001699 does not guarantee that sessions will last that long, because if the cache is
1700 full, the longest idle sessions will be purged despite their configured
1701 lifetime. The real usefulness of this setting is to prevent sessions from
1702 being used for too long.
1703
Willy Tarreaubfd59462013-02-21 07:46:09 +01001704tune.ssl.maxrecord <number>
1705 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1706 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1707 data only once it has received a full record. With large records, it means
1708 that clients might have to download up to 16kB of data before starting to
1709 process them. Limiting the value can improve page load times on browsers
1710 located over high latency or low bandwidth networks. It is suggested to find
1711 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1712 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1713 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1714 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001715 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001716 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001717
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001718tune.ssl.default-dh-param <number>
1719 Sets the maximum size of the Diffie-Hellman parameters used for generating
1720 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1721 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1722 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1723 this maximum value. Default value if 1024. Only 1024 or higher values are
1724 allowed. Higher values will increase the CPU load, and values greater than
1725 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001726 used if static Diffie-Hellman parameters are supplied either directly
1727 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001728
Christopher Faulet31af49d2015-06-09 17:29:50 +02001729tune.ssl.ssl-ctx-cache-size <number>
1730 Sets the size of the cache used to store generated certificates to <number>
1731 entries. This is a LRU cache. Because generating a SSL certificate
1732 dynamically is expensive, they are cached. The default cache size is set to
1733 1000 entries.
1734
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001735tune.ssl.capture-cipherlist-size <number>
1736 Sets the maximum size of the buffer used for capturing client-hello cipher
1737 list. If the value is 0 (default value) the capture is disabled, otherwise
1738 a buffer is allocated for each SSL/TLS connection.
1739
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001740tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001741tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001742tune.vars.reqres-max-size <size>
1743tune.vars.sess-max-size <size>
1744tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001745 These five tunes help to manage the maximum amount of memory used by the
1746 variables system. "global" limits the overall amount of memory available for
1747 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1748 memory for the session scope, "txn" for the transaction scope, and "reqres"
1749 limits the memory for each request or response processing.
1750 Memory accounting is hierarchical, meaning more coarse grained limits include
1751 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1752 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001753
Daniel Schneller0b547052016-03-21 20:46:57 +01001754 For example, when "tune.vars.sess-max-size" is limited to 100,
1755 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1756 100 either. If we create a variable "txn.var" that contains 100 bytes,
1757 all available space is consumed.
1758 Notice that exceeding the limits at runtime will not result in an error
1759 message, but values might be cut off or corrupted. So make sure to accurately
1760 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001761
William Lallemanda509e4c2012-11-07 16:54:34 +01001762tune.zlib.memlevel <number>
1763 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001764 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001765 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001766 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001767 between 1 and 9. The default value is 8.
1768
1769tune.zlib.windowsize <number>
1770 Sets the window size (the size of the history buffer) as a parameter of the
1771 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001772 in better compression at the expense of memory usage. Can be a value between
1773 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001774
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017753.3. Debugging
1776--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001777
1778debug
1779 Enables debug mode which dumps to stdout all exchanges, and disables forking
1780 into background. It is the equivalent of the command-line argument "-d". It
1781 should never be used in a production configuration since it may prevent full
1782 system startup.
1783
1784quiet
1785 Do not display any message during startup. It is equivalent to the command-
1786 line argument "-q".
1787
Emeric Brunf099e792010-09-27 12:05:28 +02001788
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010017893.4. Userlists
1790--------------
1791It is possible to control access to frontend/backend/listen sections or to
1792http stats by allowing only authenticated and authorized users. To do this,
1793it is required to create at least one userlist and to define users.
1794
1795userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001796 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001797 used to store authentication & authorization data for independent customers.
1798
1799group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001800 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001801 attach users to this group by using a comma separated list of names
1802 proceeded by "users" keyword.
1803
Cyril Bontéf0c60612010-02-06 14:44:47 +01001804user <username> [password|insecure-password <password>]
1805 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001806 Adds user <username> to the current userlist. Both secure (encrypted) and
1807 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001808 evaluated using the crypt(3) function, so depending on the system's
1809 capabilities, different algorithms are supported. For example, modern Glibc
1810 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1811 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001812
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001813 Attention: Be aware that using encrypted passwords might cause significantly
1814 increased CPU usage, depending on the number of requests, and the algorithm
1815 used. For any of the hashed variants, the password for each request must
1816 be processed through the chosen algorithm, before it can be compared to the
1817 value specified in the config file. Most current algorithms are deliberately
1818 designed to be expensive to compute to achieve resistance against brute
1819 force attacks. They do not simply salt/hash the clear text password once,
1820 but thousands of times. This can quickly become a major factor in haproxy's
1821 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001822
1823 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001824 userlist L1
1825 group G1 users tiger,scott
1826 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001827
Cyril Bontéf0c60612010-02-06 14:44:47 +01001828 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1829 user scott insecure-password elgato
1830 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001831
Cyril Bontéf0c60612010-02-06 14:44:47 +01001832 userlist L2
1833 group G1
1834 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001835
Cyril Bontéf0c60612010-02-06 14:44:47 +01001836 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1837 user scott insecure-password elgato groups G1,G2
1838 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001839
1840 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001841
Emeric Brunf099e792010-09-27 12:05:28 +02001842
18433.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001844----------
Emeric Brun94900952015-06-11 18:25:54 +02001845It is possible to propagate entries of any data-types in stick-tables between
1846several haproxy instances over TCP connections in a multi-master fashion. Each
1847instance pushes its local updates and insertions to remote peers. The pushed
1848values overwrite remote ones without aggregation. Interrupted exchanges are
1849automatically detected and recovered from the last known point.
1850In addition, during a soft restart, the old process connects to the new one
1851using such a TCP connection to push all its entries before the new process
1852tries to connect to other peers. That ensures very fast replication during a
1853reload, it typically takes a fraction of a second even for large tables.
1854Note that Server IDs are used to identify servers remotely, so it is important
1855that configurations look similar or at least that the same IDs are forced on
1856each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001857
1858peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001859 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001860 which is referenced by one or more stick-tables.
1861
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001862disabled
1863 Disables a peers section. It disables both listening and any synchronization
1864 related to this section. This is provided to disable synchronization of stick
1865 tables without having to comment out all "peers" references.
1866
1867enable
1868 This re-enables a disabled peers section which was previously disabled.
1869
Emeric Brunf099e792010-09-27 12:05:28 +02001870peer <peername> <ip>:<port>
1871 Defines a peer inside a peers section.
1872 If <peername> is set to the local peer name (by default hostname, or forced
1873 using "-L" command line option), haproxy will listen for incoming remote peer
1874 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1875 to join the remote peer, and <peername> is used at the protocol level to
1876 identify and validate the remote peer on the server side.
1877
1878 During a soft restart, local peer <ip>:<port> is used by the old instance to
1879 connect the new one and initiate a complete replication (teaching process).
1880
1881 It is strongly recommended to have the exact same peers declaration on all
1882 peers and to only rely on the "-L" command line argument to change the local
1883 peer name. This makes it easier to maintain coherent configuration files
1884 across all peers.
1885
William Lallemandb2f07452015-05-12 14:27:13 +02001886 You may want to reference some environment variables in the address
1887 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001888
Cyril Bontédc4d9032012-04-08 21:57:39 +02001889 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001890 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001891 peer haproxy1 192.168.0.1:1024
1892 peer haproxy2 192.168.0.2:1024
1893 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001894
1895 backend mybackend
1896 mode tcp
1897 balance roundrobin
1898 stick-table type ip size 20k peers mypeers
1899 stick on src
1900
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001901 server srv1 192.168.0.30:80
1902 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001903
1904
Simon Horman51a1cf62015-02-03 13:00:44 +090019053.6. Mailers
1906------------
1907It is possible to send email alerts when the state of servers changes.
1908If configured email alerts are sent to each mailer that is configured
1909in a mailers section. Email is sent to mailers using SMTP.
1910
Pieter Baauw386a1272015-08-16 15:26:24 +02001911mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001912 Creates a new mailer list with the name <mailersect>. It is an
1913 independent section which is referenced by one or more proxies.
1914
1915mailer <mailername> <ip>:<port>
1916 Defines a mailer inside a mailers section.
1917
1918 Example:
1919 mailers mymailers
1920 mailer smtp1 192.168.0.1:587
1921 mailer smtp2 192.168.0.2:587
1922
1923 backend mybackend
1924 mode tcp
1925 balance roundrobin
1926
1927 email-alert mailers mymailers
1928 email-alert from test1@horms.org
1929 email-alert to test2@horms.org
1930
1931 server srv1 192.168.0.30:80
1932 server srv2 192.168.0.31:80
1933
Pieter Baauw235fcfc2016-02-13 15:33:40 +01001934timeout mail <time>
1935 Defines the time available for a mail/connection to be made and send to
1936 the mail-server. If not defined the default value is 10 seconds. To allow
1937 for at least two SYN-ACK packets to be send during initial TCP handshake it
1938 is advised to keep this value above 4 seconds.
1939
1940 Example:
1941 mailers mymailers
1942 timeout mail 20s
1943 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09001944
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019454. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001946----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001947
Willy Tarreau6a06a402007-07-15 20:15:28 +02001948Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02001949 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001950 - frontend <name>
1951 - backend <name>
1952 - listen <name>
1953
1954A "defaults" section sets default parameters for all other sections following
1955its declaration. Those default parameters are reset by the next "defaults"
1956section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001957section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001958
1959A "frontend" section describes a set of listening sockets accepting client
1960connections.
1961
1962A "backend" section describes a set of servers to which the proxy will connect
1963to forward incoming connections.
1964
1965A "listen" section defines a complete proxy with its frontend and backend
1966parts combined in one section. It is generally useful for TCP-only traffic.
1967
Willy Tarreau0ba27502007-12-24 16:55:16 +01001968All proxy names must be formed from upper and lower case letters, digits,
1969'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1970case-sensitive, which means that "www" and "WWW" are two different proxies.
1971
1972Historically, all proxy names could overlap, it just caused troubles in the
1973logs. Since the introduction of content switching, it is mandatory that two
1974proxies with overlapping capabilities (frontend/backend) have different names.
1975However, it is still permitted that a frontend and a backend share the same
1976name, as this configuration seems to be commonly encountered.
1977
1978Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1979and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001980bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001981protocol, and can interact with it by allowing, blocking, switching, adding,
1982modifying, or removing arbitrary contents in requests or responses, based on
1983arbitrary criteria.
1984
Willy Tarreau70dffda2014-01-30 03:07:23 +01001985In HTTP mode, the processing applied to requests and responses flowing over
1986a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02001987the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01001988
1989 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1990 requests and responses are processed, and connections remain open but idle
1991 between responses and new requests.
1992
1993 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1994 1.0 to 1.5-dev21 : only the first request and response are processed, and
1995 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet4212a302018-09-21 10:42:19 +02001996 be used as it creates lots of trouble with logging and HTTP processing. It
1997 is supported only on frontends.
Willy Tarreau70dffda2014-01-30 03:07:23 +01001998
Willy Tarreau70dffda2014-01-30 03:07:23 +01001999 - SCL: server close ("option http-server-close") : the server-facing
2000 connection is closed after the end of the response is received, but the
2001 client-facing connection remains open.
2002
Christopher Faulet315b39c2018-09-21 16:26:19 +02002003 - CLO: close ("option httpclose"): the connection is closed after the end of
2004 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002005
2006The effective mode that will be applied to a connection passing through a
2007frontend and a backend can be determined by both proxy modes according to the
2008following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002009weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002010
Christopher Faulet315b39c2018-09-21 16:26:19 +02002011 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002012
Christopher Faulet315b39c2018-09-21 16:26:19 +02002013 | KAL | SCL | CLO
2014 ----+-----+-----+----
2015 KAL | KAL | SCL | CLO
2016 ----+-----+-----+----
2017 TUN | TUN | SCL | CLO
2018 Frontend ----+-----+-----+----
2019 mode SCL | SCL | SCL | CLO
2020 ----+-----+-----+----
2021 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002022
Willy Tarreau0ba27502007-12-24 16:55:16 +01002023
Willy Tarreau70dffda2014-01-30 03:07:23 +01002024
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020254.1. Proxy keywords matrix
2026--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002027
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002028The following list of keywords is supported. Most of them may only be used in a
2029limited set of section types. Some of them are marked as "deprecated" because
2030they are inherited from an old syntax which may be confusing or functionally
2031limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002032marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002033option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002034and must be disabled for a specific instance. Such options may also be prefixed
2035with "default" in order to restore default settings regardless of what has been
2036specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002037
Willy Tarreau6a06a402007-07-15 20:15:28 +02002038
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002039 keyword defaults frontend listen backend
2040------------------------------------+----------+----------+---------+---------
2041acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002042appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002043backlog X X X -
2044balance X - X X
2045bind - X X -
2046bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002047block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002048capture cookie - X X -
2049capture request header - X X -
2050capture response header - X X -
2051clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002052compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002053contimeout (deprecated) X - X X
2054cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002055declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002056default-server X - X X
2057default_backend X X X -
2058description - X X X
2059disabled X X X X
2060dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002061email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002062email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002063email-alert mailers X X X X
2064email-alert myhostname X X X X
2065email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002066enabled X X X X
2067errorfile X X X X
2068errorloc X X X X
2069errorloc302 X X X X
2070-- keyword -------------------------- defaults - frontend - listen -- backend -
2071errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002072force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002073filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002074fullconn X - X X
2075grace X X X X
2076hash-type X - X X
2077http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002078http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002079http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002080http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002081http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002082http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002083http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002084id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002085ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002086load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002087log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002088log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002089log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002090log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002091max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002092maxconn X X X -
2093mode X X X X
2094monitor fail - X X -
2095monitor-net X X X -
2096monitor-uri X X X -
2097option abortonclose (*) X - X X
2098option accept-invalid-http-request (*) X X X -
2099option accept-invalid-http-response (*) X - X X
2100option allbackups (*) X - X X
2101option checkcache (*) X - X X
2102option clitcpka (*) X X X -
2103option contstats (*) X X X -
2104option dontlog-normal (*) X X X -
2105option dontlognull (*) X X X -
Christopher Faulet315b39c2018-09-21 16:26:19 +02002106option forceclose (deprectated) (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002107-- keyword -------------------------- defaults - frontend - listen -- backend -
2108option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002109option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002110option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002111option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002112option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002113option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002114option http-server-close (*) X X X X
Christopher Faulet4212a302018-09-21 10:42:19 +02002115option http-tunnel (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002116option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002117option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002118option httpchk X - X X
2119option httpclose (*) X X X X
2120option httplog X X X X
2121option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002122option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002123option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002124option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002125option log-health-checks (*) X - X X
2126option log-separate-errors (*) X X X -
2127option logasap (*) X X X -
2128option mysql-check X - X X
2129option nolinger (*) X X X X
2130option originalto X X X X
2131option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002132option pgsql-check X - X X
2133option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002134option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002135option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002136option smtpchk X - X X
2137option socket-stats (*) X X X -
2138option splice-auto (*) X X X X
2139option splice-request (*) X X X X
2140option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002141option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002142option srvtcpka (*) X - X X
2143option ssl-hello-chk X - X X
2144-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002145option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002146option tcp-smart-accept (*) X X X -
2147option tcp-smart-connect (*) X - X X
2148option tcpka X X X X
2149option tcplog X X X X
2150option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002151external-check command X - X X
2152external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002153persist rdp-cookie X - X X
2154rate-limit sessions X X X -
2155redirect - X X X
2156redisp (deprecated) X - X X
2157redispatch (deprecated) X - X X
2158reqadd - X X X
2159reqallow - X X X
2160reqdel - X X X
2161reqdeny - X X X
2162reqiallow - X X X
2163reqidel - X X X
2164reqideny - X X X
2165reqipass - X X X
2166reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002167reqitarpit - X X X
2168reqpass - X X X
2169reqrep - X X X
2170-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002171reqtarpit - X X X
2172retries X - X X
2173rspadd - X X X
2174rspdel - X X X
2175rspdeny - X X X
2176rspidel - X X X
2177rspideny - X X X
2178rspirep - X X X
2179rsprep - X X X
2180server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002181server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002182server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002183source X - X X
2184srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002185stats admin - X X X
2186stats auth X X X X
2187stats enable X X X X
2188stats hide-version X X X X
2189stats http-request - X X X
2190stats realm X X X X
2191stats refresh X X X X
2192stats scope X X X X
2193stats show-desc X X X X
2194stats show-legends X X X X
2195stats show-node X X X X
2196stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002197-- keyword -------------------------- defaults - frontend - listen -- backend -
2198stick match - - X X
2199stick on - - X X
2200stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002201stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002202stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002203tcp-check connect - - X X
2204tcp-check expect - - X X
2205tcp-check send - - X X
2206tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002207tcp-request connection - X X -
2208tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002209tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002210tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002211tcp-response content - - X X
2212tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002213timeout check X - X X
2214timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002215timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002216timeout clitimeout (deprecated) X X X -
2217timeout connect X - X X
2218timeout contimeout (deprecated) X - X X
2219timeout http-keep-alive X X X X
2220timeout http-request X X X X
2221timeout queue X - X X
2222timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002223timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002224timeout srvtimeout (deprecated) X - X X
2225timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002226timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002227transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002228unique-id-format X X X -
2229unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002230use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002231use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002232------------------------------------+----------+----------+---------+---------
2233 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002234
Willy Tarreau0ba27502007-12-24 16:55:16 +01002235
Willy Tarreauc57f0e22009-05-10 13:12:33 +020022364.2. Alphabetically sorted keywords reference
2237---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002238
2239This section provides a description of each keyword and its usage.
2240
2241
2242acl <aclname> <criterion> [flags] [operator] <value> ...
2243 Declare or complete an access list.
2244 May be used in sections : defaults | frontend | listen | backend
2245 no | yes | yes | yes
2246 Example:
2247 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2248 acl invalid_src src_port 0:1023
2249 acl local_dst hdr(host) -i localhost
2250
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002251 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002252
2253
Cyril Bontéb21570a2009-11-29 20:04:48 +01002254appsession <cookie> len <length> timeout <holdtime>
2255 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002256 Define session stickiness on an existing application cookie.
2257 May be used in sections : defaults | frontend | listen | backend
2258 no | no | yes | yes
2259 Arguments :
2260 <cookie> this is the name of the cookie used by the application and which
2261 HAProxy will have to learn for each new session.
2262
Cyril Bontéb21570a2009-11-29 20:04:48 +01002263 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002264 checked in each cookie value.
2265
2266 <holdtime> this is the time after which the cookie will be removed from
2267 memory if unused. If no unit is specified, this time is in
2268 milliseconds.
2269
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002270 request-learn
2271 If this option is specified, then haproxy will be able to learn
2272 the cookie found in the request in case the server does not
2273 specify any in response. This is typically what happens with
2274 PHPSESSID cookies, or when haproxy's session expires before
2275 the application's session and the correct server is selected.
2276 It is recommended to specify this option to improve reliability.
2277
Cyril Bontéb21570a2009-11-29 20:04:48 +01002278 prefix When this option is specified, haproxy will match on the cookie
2279 prefix (or URL parameter prefix). The appsession value is the
2280 data following this prefix.
2281
2282 Example :
2283 appsession ASPSESSIONID len 64 timeout 3h prefix
2284
Davor Ocelice9ed2812017-12-25 17:49:28 +01002285 This will match the cookie ASPSESSIONIDXXX=XXXX,
2286 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002287
2288 mode This option allows to change the URL parser mode.
2289 2 modes are currently supported :
2290 - path-parameters :
2291 The parser looks for the appsession in the path parameters
2292 part (each parameter is separated by a semi-colon), which is
2293 convenient for JSESSIONID for example.
2294 This is the default mode if the option is not set.
2295 - query-string :
2296 In this mode, the parser will look for the appsession in the
2297 query string.
2298
Willy Tarreau294d0f02015-08-10 19:40:12 +02002299 As of version 1.6, appsessions was removed. It is more flexible and more
2300 convenient to use stick-tables instead, and stick-tables support multi-master
2301 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002302
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002303 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2304 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002305
2306
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002307backlog <conns>
2308 Give hints to the system about the approximate listen backlog desired size
2309 May be used in sections : defaults | frontend | listen | backend
2310 yes | yes | yes | no
2311 Arguments :
2312 <conns> is the number of pending connections. Depending on the operating
2313 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002314 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002315
2316 In order to protect against SYN flood attacks, one solution is to increase
2317 the system's SYN backlog size. Depending on the system, sometimes it is just
2318 tunable via a system parameter, sometimes it is not adjustable at all, and
2319 sometimes the system relies on hints given by the application at the time of
2320 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2321 to the listen() syscall. On systems which can make use of this value, it can
2322 sometimes be useful to be able to specify a different value, hence this
2323 backlog parameter.
2324
2325 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2326 used as a hint and the system accepts up to the smallest greater power of
2327 two, and never more than some limits (usually 32768).
2328
2329 See also : "maxconn" and the target operating system's tuning guide.
2330
2331
Willy Tarreau0ba27502007-12-24 16:55:16 +01002332balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002333balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002334 Define the load balancing algorithm to be used in a backend.
2335 May be used in sections : defaults | frontend | listen | backend
2336 yes | no | yes | yes
2337 Arguments :
2338 <algorithm> is the algorithm used to select a server when doing load
2339 balancing. This only applies when no persistence information
2340 is available, or when a connection is redispatched to another
2341 server. <algorithm> may be one of the following :
2342
2343 roundrobin Each server is used in turns, according to their weights.
2344 This is the smoothest and fairest algorithm when the server's
2345 processing time remains equally distributed. This algorithm
2346 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002347 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002348 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002349 large farms, when a server becomes up after having been down
2350 for a very short time, it may sometimes take a few hundreds
2351 requests for it to be re-integrated into the farm and start
2352 receiving traffic. This is normal, though very rare. It is
2353 indicated here in case you would have the chance to observe
2354 it, so that you don't worry.
2355
2356 static-rr Each server is used in turns, according to their weights.
2357 This algorithm is as similar to roundrobin except that it is
2358 static, which means that changing a server's weight on the
2359 fly will have no effect. On the other hand, it has no design
2360 limitation on the number of servers, and when a server goes
2361 up, it is always immediately reintroduced into the farm, once
2362 the full map is recomputed. It also uses slightly less CPU to
2363 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002364
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002365 leastconn The server with the lowest number of connections receives the
2366 connection. Round-robin is performed within groups of servers
2367 of the same load to ensure that all servers will be used. Use
2368 of this algorithm is recommended where very long sessions are
2369 expected, such as LDAP, SQL, TSE, etc... but is not very well
2370 suited for protocols using short sessions such as HTTP. This
2371 algorithm is dynamic, which means that server weights may be
2372 adjusted on the fly for slow starts for instance.
2373
Willy Tarreauf09c6602012-02-13 17:12:08 +01002374 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002375 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002376 identifier to the highest (see server parameter "id"), which
2377 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002378 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002379 not make sense to use this algorithm without setting maxconn.
2380 The purpose of this algorithm is to always use the smallest
2381 number of servers so that extra servers can be powered off
2382 during non-intensive hours. This algorithm ignores the server
2383 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002384 or IMAP than HTTP, though it can be useful there too. In
2385 order to use this algorithm efficiently, it is recommended
2386 that a cloud controller regularly checks server usage to turn
2387 them off when unused, and regularly checks backend queue to
2388 turn new servers on when the queue inflates. Alternatively,
2389 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002390
Willy Tarreau0ba27502007-12-24 16:55:16 +01002391 source The source IP address is hashed and divided by the total
2392 weight of the running servers to designate which server will
2393 receive the request. This ensures that the same client IP
2394 address will always reach the same server as long as no
2395 server goes down or up. If the hash result changes due to the
2396 number of running servers changing, many clients will be
2397 directed to a different server. This algorithm is generally
2398 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002399 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002400 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002401 static by default, which means that changing a server's
2402 weight on the fly will have no effect, but this can be
2403 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002404
Oskar Stolc8dc41842012-05-19 10:19:54 +01002405 uri This algorithm hashes either the left part of the URI (before
2406 the question mark) or the whole URI (if the "whole" parameter
2407 is present) and divides the hash value by the total weight of
2408 the running servers. The result designates which server will
2409 receive the request. This ensures that the same URI will
2410 always be directed to the same server as long as no server
2411 goes up or down. This is used with proxy caches and
2412 anti-virus proxies in order to maximize the cache hit rate.
2413 Note that this algorithm may only be used in an HTTP backend.
2414 This algorithm is static by default, which means that
2415 changing a server's weight on the fly will have no effect,
2416 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002417
Oskar Stolc8dc41842012-05-19 10:19:54 +01002418 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002419 "depth", both followed by a positive integer number. These
2420 options may be helpful when it is needed to balance servers
2421 based on the beginning of the URI only. The "len" parameter
2422 indicates that the algorithm should only consider that many
2423 characters at the beginning of the URI to compute the hash.
2424 Note that having "len" set to 1 rarely makes sense since most
2425 URIs start with a leading "/".
2426
2427 The "depth" parameter indicates the maximum directory depth
2428 to be used to compute the hash. One level is counted for each
2429 slash in the request. If both parameters are specified, the
2430 evaluation stops when either is reached.
2431
Willy Tarreau0ba27502007-12-24 16:55:16 +01002432 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002433 the query string of each HTTP GET request.
2434
2435 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002436 request entity will be searched for the parameter argument,
2437 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002438 ('?') in the URL. The message body will only start to be
2439 analyzed once either the advertised amount of data has been
2440 received or the request buffer is full. In the unlikely event
2441 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002442 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002443 be randomly balanced if at all. This keyword used to support
2444 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002445
2446 If the parameter is found followed by an equal sign ('=') and
2447 a value, then the value is hashed and divided by the total
2448 weight of the running servers. The result designates which
2449 server will receive the request.
2450
2451 This is used to track user identifiers in requests and ensure
2452 that a same user ID will always be sent to the same server as
2453 long as no server goes up or down. If no value is found or if
2454 the parameter is not found, then a round robin algorithm is
2455 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002456 backend. This algorithm is static by default, which means
2457 that changing a server's weight on the fly will have no
2458 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002459
Cyril Bontédc4d9032012-04-08 21:57:39 +02002460 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2461 request. Just as with the equivalent ACL 'hdr()' function,
2462 the header name in parenthesis is not case sensitive. If the
2463 header is absent or if it does not contain any value, the
2464 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002465
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002466 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002467 reducing the hash algorithm to the main domain part with some
2468 specific headers such as 'Host'. For instance, in the Host
2469 value "haproxy.1wt.eu", only "1wt" will be considered.
2470
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002471 This algorithm is static by default, which means that
2472 changing a server's weight on the fly will have no effect,
2473 but this can be changed using "hash-type".
2474
Willy Tarreau760e81d2018-05-03 07:20:40 +02002475 random A random number will be used as the key for the consistent
2476 hashing function. This means that the servers' weights are
2477 respected, dynamic weight changes immediately take effect, as
2478 well as new server additions. Random load balancing can be
2479 useful with large farms or when servers are frequently added
2480 or removed. The hash-balance-factor directive can be used to
2481 further improve fairness of the load balancing, especially
2482 in situations where servers show highly variable response
2483 times.
2484
Emeric Brun736aa232009-06-30 17:56:00 +02002485 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002486 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002487 The RDP cookie <name> (or "mstshash" if omitted) will be
2488 looked up and hashed for each incoming TCP request. Just as
2489 with the equivalent ACL 'req_rdp_cookie()' function, the name
2490 is not case-sensitive. This mechanism is useful as a degraded
2491 persistence mode, as it makes it possible to always send the
2492 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002493 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002494 used instead.
2495
2496 Note that for this to work, the frontend must ensure that an
2497 RDP cookie is already present in the request buffer. For this
2498 you must use 'tcp-request content accept' rule combined with
2499 a 'req_rdp_cookie_cnt' ACL.
2500
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002501 This algorithm is static by default, which means that
2502 changing a server's weight on the fly will have no effect,
2503 but this can be changed using "hash-type".
2504
Cyril Bontédc4d9032012-04-08 21:57:39 +02002505 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002506
Willy Tarreau0ba27502007-12-24 16:55:16 +01002507 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002508 algorithms. Right now, only "url_param" and "uri" support an
2509 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002510
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002511 The load balancing algorithm of a backend is set to roundrobin when no other
2512 algorithm, mode nor option have been set. The algorithm may only be set once
2513 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002514
Lukas Tribus80512b12018-10-27 20:07:40 +02002515 With authentication schemes that require the same connection like NTLM, URI
2516 based alghoritms must not be used, as they would cause subsequent requests
2517 to be routed to different backend servers, breaking the invalid assumptions
2518 NTLM relies on.
2519
Willy Tarreau0ba27502007-12-24 16:55:16 +01002520 Examples :
2521 balance roundrobin
2522 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002523 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002524 balance hdr(User-Agent)
2525 balance hdr(host)
2526 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002527
2528 Note: the following caveats and limitations on using the "check_post"
2529 extension with "url_param" must be considered :
2530
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002531 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002532 to determine if the parameters will be found in the body or entity which
2533 may contain binary data. Therefore another method may be required to
2534 restrict consideration of POST requests that have no URL parameters in
2535 the body. (see acl reqideny http_end)
2536
2537 - using a <max_wait> value larger than the request buffer size does not
2538 make sense and is useless. The buffer size is set at build time, and
2539 defaults to 16 kB.
2540
2541 - Content-Encoding is not supported, the parameter search will probably
2542 fail; and load balancing will fall back to Round Robin.
2543
2544 - Expect: 100-continue is not supported, load balancing will fall back to
2545 Round Robin.
2546
Lukas Tribus23953682017-04-28 13:24:30 +00002547 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002548 If the entire parameter value is not present in the first chunk, the
2549 selection of server is undefined (actually, defined by how little
2550 actually appeared in the first chunk).
2551
2552 - This feature does not support generation of a 100, 411 or 501 response.
2553
2554 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002555 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002556 white space or control characters are found, indicating the end of what
2557 might be a URL parameter list. This is probably not a concern with SGML
2558 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002559
Willy Tarreau294d0f02015-08-10 19:40:12 +02002560 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002561
2562
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002563bind [<address>]:<port_range> [, ...] [param*]
2564bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002565 Define one or several listening addresses and/or ports in a frontend.
2566 May be used in sections : defaults | frontend | listen | backend
2567 no | yes | yes | no
2568 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002569 <address> is optional and can be a host name, an IPv4 address, an IPv6
2570 address, or '*'. It designates the address the frontend will
2571 listen on. If unset, all IPv4 addresses of the system will be
2572 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002573 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002574 Optionally, an address family prefix may be used before the
2575 address to force the family regardless of the address format,
2576 which can be useful to specify a path to a unix socket with
2577 no slash ('/'). Currently supported prefixes are :
2578 - 'ipv4@' -> address is always IPv4
2579 - 'ipv6@' -> address is always IPv6
2580 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002581 - 'abns@' -> address is in abstract namespace (Linux only).
2582 Note: since abstract sockets are not "rebindable", they
2583 do not cope well with multi-process mode during
2584 soft-restart, so it is better to avoid them if
2585 nbproc is greater than 1. The effect is that if the
2586 new process fails to start, only one of the old ones
2587 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002588 - 'fd@<n>' -> use file descriptor <n> inherited from the
2589 parent. The fd must be bound and may or may not already
2590 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002591 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2592 connected unix socket or of a socketpair. The bind waits
2593 to receive a FD over the unix socket and uses it as if it
2594 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002595 You may want to reference some environment variables in the
2596 address parameter, see section 2.3 about environment
2597 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002598
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002599 <port_range> is either a unique TCP port, or a port range for which the
2600 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002601 above. The port is mandatory for TCP listeners. Note that in
2602 the case of an IPv6 address, the port is always the number
2603 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002604 - a numerical port (ex: '80')
2605 - a dash-delimited ports range explicitly stating the lower
2606 and upper bounds (ex: '2000-2100') which are included in
2607 the range.
2608
2609 Particular care must be taken against port ranges, because
2610 every <address:port> couple consumes one socket (= a file
2611 descriptor), so it's easy to consume lots of descriptors
2612 with a simple range, and to run out of sockets. Also, each
2613 <address:port> couple must be used only once among all
2614 instances running on a same system. Please note that binding
2615 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002616 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002617 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002618
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002619 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002620 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002621 receive UNIX connections on the socket located at this place.
2622 The path must begin with a slash and by default is absolute.
2623 It can be relative to the prefix defined by "unix-bind" in
2624 the global section. Note that the total length of the prefix
2625 followed by the socket path cannot exceed some system limits
2626 for UNIX sockets, which commonly are set to 107 characters.
2627
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002628 <param*> is a list of parameters common to all sockets declared on the
2629 same line. These numerous parameters depend on OS and build
2630 options and have a complete section dedicated to them. Please
2631 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002632
Willy Tarreau0ba27502007-12-24 16:55:16 +01002633 It is possible to specify a list of address:port combinations delimited by
2634 commas. The frontend will then listen on all of these addresses. There is no
2635 fixed limit to the number of addresses and ports which can be listened on in
2636 a frontend, as well as there is no limit to the number of "bind" statements
2637 in a frontend.
2638
2639 Example :
2640 listen http_proxy
2641 bind :80,:443
2642 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002643 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002644
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002645 listen http_https_proxy
2646 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002647 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002648
Willy Tarreau24709282013-03-10 21:32:12 +01002649 listen http_https_proxy_explicit
2650 bind ipv6@:80
2651 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2652 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2653
Willy Tarreaudad36a32013-03-11 01:20:04 +01002654 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002655 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002656
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002657 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2658 sun_path length is used for the address length. Some other programs
2659 such as socat use the string length only by default. Pass the option
2660 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2661 make it compatible with HAProxy's.
2662
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002663 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002664 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002665
2666
Christopher Fauletff4121f2017-11-22 16:38:49 +01002667bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002668 Limit visibility of an instance to a certain set of processes numbers.
2669 May be used in sections : defaults | frontend | listen | backend
2670 yes | yes | yes | yes
2671 Arguments :
2672 all All process will see this instance. This is the default. It
2673 may be used to override a default value.
2674
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002675 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002676 option may be combined with other numbers.
2677
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002678 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002679 option may be combined with other numbers. Do not use it
2680 with less than 2 processes otherwise some instances might be
2681 missing from all processes.
2682
Christopher Fauletff4121f2017-11-22 16:38:49 +01002683 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002684 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002685 the machine's word size. Ranges can be partially defined. The
2686 higher bound can be omitted. In such case, it is replaced by
2687 the corresponding maximum value. If a proxy is bound to
2688 process numbers greater than the configured global.nbproc, it
2689 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002690 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002691
2692 This keyword limits binding of certain instances to certain processes. This
2693 is useful in order not to have too many processes listening to the same
2694 ports. For instance, on a dual-core machine, it might make sense to set
2695 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2696 and 'even' instances.
2697
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002698 At the moment, it is not possible to reference more than 32 or 64 processes
2699 using this keyword, but this should be more than enough for most setups.
2700 Please note that 'all' really means all processes regardless of the machine's
2701 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002702
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002703 Each "bind" line may further be limited to a subset of the proxy's processes,
2704 please consult the "process" bind keyword in section 5.1.
2705
Willy Tarreaub369a042014-09-16 13:21:03 +02002706 When a frontend has no explicit "bind-process" line, it tries to bind to all
2707 the processes referenced by its "bind" lines. That means that frontends can
2708 easily adapt to their listeners' processes.
2709
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002710 If some backends are referenced by frontends bound to other processes, the
2711 backend automatically inherits the frontend's processes.
2712
2713 Example :
2714 listen app_ip1
2715 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002716 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002717
2718 listen app_ip2
2719 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002720 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002721
2722 listen management
2723 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002724 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002725
Willy Tarreau110ecc12012-11-15 17:50:01 +01002726 listen management
2727 bind 10.0.0.4:80
2728 bind-process 1-4
2729
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002730 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002731
2732
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002733block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002734 Block a layer 7 request if/unless a condition is matched
2735 May be used in sections : defaults | frontend | listen | backend
2736 no | yes | yes | yes
2737
2738 The HTTP request will be blocked very early in the layer 7 processing
2739 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002740 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002741 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002742 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002743 "block" statements per instance. To block connections at layer 4 (without
2744 sending a 403 error) see "tcp-request connection reject" and
2745 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002746
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002747 This form is deprecated, do not use it in any new configuration, use the new
2748 "http-request deny" instead.
2749
Willy Tarreau0ba27502007-12-24 16:55:16 +01002750 Example:
2751 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2752 acl invalid_src src_port 0:1023
2753 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002754 # block is deprecated. Use http-request deny instead:
2755 #block if invalid_src || local_dst
2756 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002757
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002758 See also : section 7 about ACL usage, "http-request deny",
2759 "http-response deny", "tcp-request connection reject" and
2760 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002761
2762capture cookie <name> len <length>
2763 Capture and log a cookie in the request and in the response.
2764 May be used in sections : defaults | frontend | listen | backend
2765 no | yes | yes | no
2766 Arguments :
2767 <name> is the beginning of the name of the cookie to capture. In order
2768 to match the exact name, simply suffix the name with an equal
2769 sign ('='). The full name will appear in the logs, which is
2770 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002771 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002772
2773 <length> is the maximum number of characters to report in the logs, which
2774 include the cookie name, the equal sign and the value, all in the
2775 standard "name=value" form. The string will be truncated on the
2776 right if it exceeds <length>.
2777
2778 Only the first cookie is captured. Both the "cookie" request headers and the
2779 "set-cookie" response headers are monitored. This is particularly useful to
2780 check for application bugs causing session crossing or stealing between
2781 users, because generally the user's cookies can only change on a login page.
2782
2783 When the cookie was not presented by the client, the associated log column
2784 will report "-". When a request does not cause a cookie to be assigned by the
2785 server, a "-" is reported in the response column.
2786
2787 The capture is performed in the frontend only because it is necessary that
2788 the log format does not change for a given frontend depending on the
2789 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002790 "capture cookie" statement in a frontend. The maximum capture length is set
2791 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2792 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002793
2794 Example:
2795 capture cookie ASPSESSION len 32
2796
2797 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002798 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002799
2800
2801capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002802 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002803 May be used in sections : defaults | frontend | listen | backend
2804 no | yes | yes | no
2805 Arguments :
2806 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002807 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002808 appear in the requests, with the first letter of each word in
2809 upper case. The header name will not appear in the logs, only the
2810 value is reported, but the position in the logs is respected.
2811
2812 <length> is the maximum number of characters to extract from the value and
2813 report in the logs. The string will be truncated on the right if
2814 it exceeds <length>.
2815
Willy Tarreau4460d032012-11-21 23:37:37 +01002816 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002817 value will be added to the logs between braces ('{}'). If multiple headers
2818 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002819 in the same order they were declared in the configuration. Non-existent
2820 headers will be logged just as an empty string. Common uses for request
2821 header captures include the "Host" field in virtual hosting environments, the
2822 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002823 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002824 environments to find where the request came from.
2825
2826 Note that when capturing headers such as "User-agent", some spaces may be
2827 logged, making the log analysis more difficult. Thus be careful about what
2828 you log if you know your log parser is not smart enough to rely on the
2829 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002830
Willy Tarreau0900abb2012-11-22 00:21:46 +01002831 There is no limit to the number of captured request headers nor to their
2832 length, though it is wise to keep them low to limit memory usage per session.
2833 In order to keep log format consistent for a same frontend, header captures
2834 can only be declared in a frontend. It is not possible to specify a capture
2835 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002836
2837 Example:
2838 capture request header Host len 15
2839 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002840 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002841
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002842 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002843 about logging.
2844
2845
2846capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002847 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002848 May be used in sections : defaults | frontend | listen | backend
2849 no | yes | yes | no
2850 Arguments :
2851 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002852 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002853 appear in the response, with the first letter of each word in
2854 upper case. The header name will not appear in the logs, only the
2855 value is reported, but the position in the logs is respected.
2856
2857 <length> is the maximum number of characters to extract from the value and
2858 report in the logs. The string will be truncated on the right if
2859 it exceeds <length>.
2860
Willy Tarreau4460d032012-11-21 23:37:37 +01002861 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002862 result will be added to the logs between braces ('{}') after the captured
2863 request headers. If multiple headers are captured, they will be delimited by
2864 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002865 the configuration. Non-existent headers will be logged just as an empty
2866 string. Common uses for response header captures include the "Content-length"
2867 header which indicates how many bytes are expected to be returned, the
2868 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002869
Willy Tarreau0900abb2012-11-22 00:21:46 +01002870 There is no limit to the number of captured response headers nor to their
2871 length, though it is wise to keep them low to limit memory usage per session.
2872 In order to keep log format consistent for a same frontend, header captures
2873 can only be declared in a frontend. It is not possible to specify a capture
2874 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002875
2876 Example:
2877 capture response header Content-length len 9
2878 capture response header Location len 15
2879
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002880 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002881 about logging.
2882
2883
Cyril Bontéf0c60612010-02-06 14:44:47 +01002884clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002885 Set the maximum inactivity time on the client side.
2886 May be used in sections : defaults | frontend | listen | backend
2887 yes | yes | yes | no
2888 Arguments :
2889 <timeout> is the timeout value is specified in milliseconds by default, but
2890 can be in any other unit if the number is suffixed by the unit,
2891 as explained at the top of this document.
2892
2893 The inactivity timeout applies when the client is expected to acknowledge or
2894 send data. In HTTP mode, this timeout is particularly important to consider
2895 during the first phase, when the client sends the request, and during the
2896 response while it is reading data sent by the server. The value is specified
2897 in milliseconds by default, but can be in any other unit if the number is
2898 suffixed by the unit, as specified at the top of this document. In TCP mode
2899 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2900 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002901 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002902 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002903 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002904
2905 This parameter is specific to frontends, but can be specified once for all in
2906 "defaults" sections. This is in fact one of the easiest solutions not to
2907 forget about it. An unspecified timeout results in an infinite timeout, which
2908 is not recommended. Such a usage is accepted and works but reports a warning
2909 during startup because it may results in accumulation of expired sessions in
2910 the system if the system's timeouts are not configured either.
2911
2912 This parameter is provided for compatibility but is currently deprecated.
2913 Please use "timeout client" instead.
2914
Willy Tarreau036fae02008-01-06 13:24:40 +01002915 See also : "timeout client", "timeout http-request", "timeout server", and
2916 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002917
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002918compression algo <algorithm> ...
2919compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002920compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002921 Enable HTTP compression.
2922 May be used in sections : defaults | frontend | listen | backend
2923 yes | yes | yes | yes
2924 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002925 algo is followed by the list of supported compression algorithms.
2926 type is followed by the list of MIME types that will be compressed.
2927 offload makes haproxy work as a compression offloader only (see notes).
2928
2929 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002930 identity this is mostly for debugging, and it was useful for developing
2931 the compression feature. Identity does not apply any change on
2932 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002933
Willy Tarreauc91840a2015-03-28 17:00:39 +01002934 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01002935 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01002936
2937 deflate same as "gzip", but with deflate algorithm and zlib format.
2938 Note that this algorithm has ambiguous support on many
2939 browsers and no support at all from recent ones. It is
2940 strongly recommended not to use it for anything else than
2941 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01002942 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002943
Willy Tarreauc91840a2015-03-28 17:00:39 +01002944 raw-deflate same as "deflate" without the zlib wrapper, and used as an
2945 alternative when the browser wants "deflate". All major
2946 browsers understand it and despite violating the standards,
2947 it is known to work better than "deflate", at least on MSIE
2948 and some versions of Safari. Do not use it in conjunction
2949 with "deflate", use either one or the other since both react
2950 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01002951 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002952
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002953 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002954 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002955 If backend servers support HTTP compression, these directives
2956 will be no-op: haproxy will see the compressed response and will not
2957 compress again. If backend servers do not support HTTP compression and
2958 there is Accept-Encoding header in request, haproxy will compress the
2959 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002960
2961 The "offload" setting makes haproxy remove the Accept-Encoding header to
2962 prevent backend servers from compressing responses. It is strongly
2963 recommended not to do this because this means that all the compression work
2964 will be done on the single point where haproxy is located. However in some
2965 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002966 with broken HTTP compression implementation which can't be turned off.
2967 In that case haproxy can be used to prevent that gateway from emitting
2968 invalid payloads. In this case, simply removing the header in the
2969 configuration does not work because it applies before the header is parsed,
2970 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02002971 then be used for such scenarios. Note: for now, the "offload" setting is
2972 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002973
William Lallemand05097442012-11-20 12:14:28 +01002974 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002975 * the request does not advertise a supported compression algorithm in the
2976 "Accept-Encoding" header
2977 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002978 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002979 * response header "Transfer-Encoding" contains "chunked" (Temporary
2980 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002981 * response contain neither a "Content-Length" header nor a
2982 "Transfer-Encoding" whose last value is "chunked"
2983 * response contains a "Content-Type" header whose first value starts with
2984 "multipart"
2985 * the response contains the "no-transform" value in the "Cache-control"
2986 header
2987 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2988 and later
2989 * The response contains a "Content-Encoding" header, indicating that the
2990 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002991
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002992 Note: The compression does not rewrite Etag headers, and does not emit the
2993 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002994
William Lallemand82fe75c2012-10-23 10:25:10 +02002995 Examples :
2996 compression algo gzip
2997 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002998
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002999
Cyril Bontéf0c60612010-02-06 14:44:47 +01003000contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003001 Set the maximum time to wait for a connection attempt to a server to succeed.
3002 May be used in sections : defaults | frontend | listen | backend
3003 yes | no | yes | yes
3004 Arguments :
3005 <timeout> is the timeout value is specified in milliseconds by default, but
3006 can be in any other unit if the number is suffixed by the unit,
3007 as explained at the top of this document.
3008
3009 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003010 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003011 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003012 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003013 connect timeout also presets the queue timeout to the same value if this one
3014 has not been specified. Historically, the contimeout was also used to set the
3015 tarpit timeout in a listen section, which is not possible in a pure frontend.
3016
3017 This parameter is specific to backends, but can be specified once for all in
3018 "defaults" sections. This is in fact one of the easiest solutions not to
3019 forget about it. An unspecified timeout results in an infinite timeout, which
3020 is not recommended. Such a usage is accepted and works but reports a warning
3021 during startup because it may results in accumulation of failed sessions in
3022 the system if the system's timeouts are not configured either.
3023
3024 This parameter is provided for backwards compatibility but is currently
3025 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3026 instead.
3027
3028 See also : "timeout connect", "timeout queue", "timeout tarpit",
3029 "timeout server", "contimeout".
3030
3031
Willy Tarreau55165fe2009-05-10 12:02:55 +02003032cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003033 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3034 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003035 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003036 Enable cookie-based persistence in a backend.
3037 May be used in sections : defaults | frontend | listen | backend
3038 yes | no | yes | yes
3039 Arguments :
3040 <name> is the name of the cookie which will be monitored, modified or
3041 inserted in order to bring persistence. This cookie is sent to
3042 the client via a "Set-Cookie" header in the response, and is
3043 brought back by the client in a "Cookie" header in all requests.
3044 Special care should be taken to choose a name which does not
3045 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003046 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003047 HTTP/HTTPS), care should be taken to use different cookie names
3048 between all backends if persistence between them is not desired.
3049
3050 rewrite This keyword indicates that the cookie will be provided by the
3051 server and that haproxy will have to modify its value to set the
3052 server's identifier in it. This mode is handy when the management
3053 of complex combinations of "Set-cookie" and "Cache-control"
3054 headers is left to the application. The application can then
3055 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003056 cookie. Since all responses should be monitored, this mode
3057 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003058 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003059 start with this mode for new deployments. This keyword is
3060 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003061
3062 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003063 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003064
Willy Tarreaua79094d2010-08-31 22:54:15 +02003065 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003066 server. When used without the "preserve" option, if the server
3067 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003068 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003069 existing configurations running in the "rewrite" mode. The cookie
3070 will only be a session cookie and will not be stored on the
3071 client's disk. By default, unless the "indirect" option is added,
3072 the server will see the cookies emitted by the client. Due to
3073 caching effects, it is generally wise to add the "nocache" or
3074 "postonly" keywords (see below). The "insert" keyword is not
3075 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003076
3077 prefix This keyword indicates that instead of relying on a dedicated
3078 cookie for the persistence, an existing one will be completed.
3079 This may be needed in some specific environments where the client
3080 does not support more than one single cookie and the application
3081 already needs it. In this case, whenever the server sets a cookie
3082 named <name>, it will be prefixed with the server's identifier
3083 and a delimiter. The prefix will be removed from all client
3084 requests so that the server still finds the cookie it emitted.
3085 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003086 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003087 not compatible with "rewrite" and "insert". Note: it is highly
3088 recommended not to use "indirect" with "prefix", otherwise server
3089 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003090
Willy Tarreaua79094d2010-08-31 22:54:15 +02003091 indirect When this option is specified, no cookie will be emitted to a
3092 client which already has a valid one for the server which has
3093 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003094 it will be removed, unless the "preserve" option is also set. In
3095 "insert" mode, this will additionally remove cookies from the
3096 requests transmitted to the server, making the persistence
3097 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003098 Note: it is highly recommended not to use "indirect" with
3099 "prefix", otherwise server cookie updates would not be sent to
3100 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003101
3102 nocache This option is recommended in conjunction with the insert mode
3103 when there is a cache between the client and HAProxy, as it
3104 ensures that a cacheable response will be tagged non-cacheable if
3105 a cookie needs to be inserted. This is important because if all
3106 persistence cookies are added on a cacheable home page for
3107 instance, then all customers will then fetch the page from an
3108 outer cache and will all share the same persistence cookie,
3109 leading to one server receiving much more traffic than others.
3110 See also the "insert" and "postonly" options.
3111
3112 postonly This option ensures that cookie insertion will only be performed
3113 on responses to POST requests. It is an alternative to the
3114 "nocache" option, because POST responses are not cacheable, so
3115 this ensures that the persistence cookie will never get cached.
3116 Since most sites do not need any sort of persistence before the
3117 first POST which generally is a login request, this is a very
3118 efficient method to optimize caching without risking to find a
3119 persistence cookie in the cache.
3120 See also the "insert" and "nocache" options.
3121
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003122 preserve This option may only be used with "insert" and/or "indirect". It
3123 allows the server to emit the persistence cookie itself. In this
3124 case, if a cookie is found in the response, haproxy will leave it
3125 untouched. This is useful in order to end persistence after a
3126 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003127 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003128 the past. By combining this mechanism with the "disable-on-404"
3129 check option, it is possible to perform a completely graceful
3130 shutdown because users will definitely leave the server after
3131 they logout.
3132
Willy Tarreau4992dd22012-05-31 21:02:17 +02003133 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3134 when a cookie is inserted. This attribute is used so that a
3135 user agent doesn't share the cookie with non-HTTP components.
3136 Please check RFC6265 for more information on this attribute.
3137
3138 secure This option tells haproxy to add a "Secure" cookie attribute when
3139 a cookie is inserted. This attribute is used so that a user agent
3140 never emits this cookie over non-secure channels, which means
3141 that a cookie learned with this flag will be presented only over
3142 SSL/TLS connections. Please check RFC6265 for more information on
3143 this attribute.
3144
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003145 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003146 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003147 name. If the domain begins with a dot, the browser is allowed to
3148 use it for any host ending with that name. It is also possible to
3149 specify several domain names by invoking this option multiple
3150 times. Some browsers might have small limits on the number of
3151 domains, so be careful when doing that. For the record, sending
3152 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003153
Willy Tarreau996a92c2010-10-13 19:30:47 +02003154 maxidle This option allows inserted cookies to be ignored after some idle
3155 time. It only works with insert-mode cookies. When a cookie is
3156 sent to the client, the date this cookie was emitted is sent too.
3157 Upon further presentations of this cookie, if the date is older
3158 than the delay indicated by the parameter (in seconds), it will
3159 be ignored. Otherwise, it will be refreshed if needed when the
3160 response is sent to the client. This is particularly useful to
3161 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003162 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003163 this option is set and a cookie has no date, it is always
3164 accepted, but gets refreshed in the response. This maintains the
3165 ability for admins to access their sites. Cookies that have a
3166 date in the future further than 24 hours are ignored. Doing so
3167 lets admins fix timezone issues without risking kicking users off
3168 the site.
3169
3170 maxlife This option allows inserted cookies to be ignored after some life
3171 time, whether they're in use or not. It only works with insert
3172 mode cookies. When a cookie is first sent to the client, the date
3173 this cookie was emitted is sent too. Upon further presentations
3174 of this cookie, if the date is older than the delay indicated by
3175 the parameter (in seconds), it will be ignored. If the cookie in
3176 the request has no date, it is accepted and a date will be set.
3177 Cookies that have a date in the future further than 24 hours are
3178 ignored. Doing so lets admins fix timezone issues without risking
3179 kicking users off the site. Contrary to maxidle, this value is
3180 not refreshed, only the first visit date counts. Both maxidle and
3181 maxlife may be used at the time. This is particularly useful to
3182 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003183 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003184 is stronger than the maxidle method in that it forces a
3185 redispatch after some absolute delay.
3186
Olivier Houchard4e694042017-03-14 20:01:29 +01003187 dynamic Activate dynamic cookies. When used, a session cookie is
3188 dynamically created for each server, based on the IP and port
3189 of the server, and a secret key, specified in the
3190 "dynamic-cookie-key" backend directive.
3191 The cookie will be regenerated each time the IP address change,
3192 and is only generated for IPv4/IPv6.
3193
Willy Tarreau0ba27502007-12-24 16:55:16 +01003194 There can be only one persistence cookie per HTTP backend, and it can be
3195 declared in a defaults section. The value of the cookie will be the value
3196 indicated after the "cookie" keyword in a "server" statement. If no cookie
3197 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003198
Willy Tarreau0ba27502007-12-24 16:55:16 +01003199 Examples :
3200 cookie JSESSIONID prefix
3201 cookie SRV insert indirect nocache
3202 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003203 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003204
Willy Tarreau294d0f02015-08-10 19:40:12 +02003205 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003206
Willy Tarreau983e01e2010-01-11 18:42:06 +01003207
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003208declare capture [ request | response ] len <length>
3209 Declares a capture slot.
3210 May be used in sections : defaults | frontend | listen | backend
3211 no | yes | yes | no
3212 Arguments:
3213 <length> is the length allowed for the capture.
3214
3215 This declaration is only available in the frontend or listen section, but the
3216 reserved slot can be used in the backends. The "request" keyword allocates a
3217 capture slot for use in the request, and "response" allocates a capture slot
3218 for use in the response.
3219
3220 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003221 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003222 "http-request capture" and "http-response capture".
3223
3224
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003225default-server [param*]
3226 Change default options for a server in a backend
3227 May be used in sections : defaults | frontend | listen | backend
3228 yes | no | yes | yes
3229 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003230 <param*> is a list of parameters for this server. The "default-server"
3231 keyword accepts an important number of options and has a complete
3232 section dedicated to it. Please refer to section 5 for more
3233 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003234
Willy Tarreau983e01e2010-01-11 18:42:06 +01003235 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003236 default-server inter 1000 weight 13
3237
3238 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003239
Willy Tarreau983e01e2010-01-11 18:42:06 +01003240
Willy Tarreau0ba27502007-12-24 16:55:16 +01003241default_backend <backend>
3242 Specify the backend to use when no "use_backend" rule has been matched.
3243 May be used in sections : defaults | frontend | listen | backend
3244 yes | yes | yes | no
3245 Arguments :
3246 <backend> is the name of the backend to use.
3247
3248 When doing content-switching between frontend and backends using the
3249 "use_backend" keyword, it is often useful to indicate which backend will be
3250 used when no rule has matched. It generally is the dynamic backend which
3251 will catch all undetermined requests.
3252
Willy Tarreau0ba27502007-12-24 16:55:16 +01003253 Example :
3254
3255 use_backend dynamic if url_dyn
3256 use_backend static if url_css url_img extension_img
3257 default_backend dynamic
3258
Willy Tarreau98d04852015-05-26 12:18:29 +02003259 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003260
Willy Tarreau0ba27502007-12-24 16:55:16 +01003261
Baptiste Assmann27f51342013-10-09 06:51:49 +02003262description <string>
3263 Describe a listen, frontend or backend.
3264 May be used in sections : defaults | frontend | listen | backend
3265 no | yes | yes | yes
3266 Arguments : string
3267
3268 Allows to add a sentence to describe the related object in the HAProxy HTML
3269 stats page. The description will be printed on the right of the object name
3270 it describes.
3271 No need to backslash spaces in the <string> arguments.
3272
3273
Willy Tarreau0ba27502007-12-24 16:55:16 +01003274disabled
3275 Disable a proxy, frontend or backend.
3276 May be used in sections : defaults | frontend | listen | backend
3277 yes | yes | yes | yes
3278 Arguments : none
3279
3280 The "disabled" keyword is used to disable an instance, mainly in order to
3281 liberate a listening port or to temporarily disable a service. The instance
3282 will still be created and its configuration will be checked, but it will be
3283 created in the "stopped" state and will appear as such in the statistics. It
3284 will not receive any traffic nor will it send any health-checks or logs. It
3285 is possible to disable many instances at once by adding the "disabled"
3286 keyword in a "defaults" section.
3287
3288 See also : "enabled"
3289
3290
Willy Tarreau5ce94572010-06-07 14:35:41 +02003291dispatch <address>:<port>
3292 Set a default server address
3293 May be used in sections : defaults | frontend | listen | backend
3294 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003295 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003296
3297 <address> is the IPv4 address of the default server. Alternatively, a
3298 resolvable hostname is supported, but this name will be resolved
3299 during start-up.
3300
3301 <ports> is a mandatory port specification. All connections will be sent
3302 to this port, and it is not permitted to use port offsets as is
3303 possible with normal servers.
3304
Willy Tarreau787aed52011-04-15 06:45:37 +02003305 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003306 server can take the connection. In the past it was used to forward non
3307 persistent connections to an auxiliary load balancer. Due to its simple
3308 syntax, it has also been used for simple TCP relays. It is recommended not to
3309 use it for more clarity, and to use the "server" directive instead.
3310
3311 See also : "server"
3312
Olivier Houchard4e694042017-03-14 20:01:29 +01003313
3314dynamic-cookie-key <string>
3315 Set the dynamic cookie secret key for a backend.
3316 May be used in sections : defaults | frontend | listen | backend
3317 yes | no | yes | yes
3318 Arguments : The secret key to be used.
3319
3320 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003321 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003322 specified on the "server" line), using a hash of the IP address of the
3323 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003324 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003325 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003326
Willy Tarreau0ba27502007-12-24 16:55:16 +01003327enabled
3328 Enable a proxy, frontend or backend.
3329 May be used in sections : defaults | frontend | listen | backend
3330 yes | yes | yes | yes
3331 Arguments : none
3332
3333 The "enabled" keyword is used to explicitly enable an instance, when the
3334 defaults has been set to "disabled". This is very rarely used.
3335
3336 See also : "disabled"
3337
3338
3339errorfile <code> <file>
3340 Return a file contents instead of errors generated by HAProxy
3341 May be used in sections : defaults | frontend | listen | backend
3342 yes | yes | yes | yes
3343 Arguments :
3344 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003345 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3346 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003347
3348 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003349 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003350 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003351 error pages, and to use absolute paths, since files are read
3352 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003353
3354 It is important to understand that this keyword is not meant to rewrite
3355 errors returned by the server, but errors detected and returned by HAProxy.
3356 This is why the list of supported errors is limited to a small set.
3357
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003358 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3359
Willy Tarreau0ba27502007-12-24 16:55:16 +01003360 The files are returned verbatim on the TCP socket. This allows any trick such
3361 as redirections to another URL or site, as well as tricks to clean cookies,
3362 force enable or disable caching, etc... The package provides default error
3363 files returning the same contents as default errors.
3364
Willy Tarreau59140a22009-02-22 12:02:30 +01003365 The files should not exceed the configured buffer size (BUFSIZE), which
3366 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003367 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003368 loops between the client and HAProxy when all servers are down, causing an
3369 error to be returned instead of an image. For better HTTP compliance, it is
3370 recommended that all header lines end with CR-LF and not LF alone.
3371
Willy Tarreau0ba27502007-12-24 16:55:16 +01003372 The files are read at the same time as the configuration and kept in memory.
3373 For this reason, the errors continue to be returned even when the process is
3374 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003375 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003376 403 status code and interrogating a blocked URL.
3377
3378 See also : "errorloc", "errorloc302", "errorloc303"
3379
Willy Tarreau59140a22009-02-22 12:02:30 +01003380 Example :
3381 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003382 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003383 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3384 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3385
Willy Tarreau2769aa02007-12-27 18:26:09 +01003386
3387errorloc <code> <url>
3388errorloc302 <code> <url>
3389 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3390 May be used in sections : defaults | frontend | listen | backend
3391 yes | yes | yes | yes
3392 Arguments :
3393 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003394 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3395 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003396
3397 <url> it is the exact contents of the "Location" header. It may contain
3398 either a relative URI to an error page hosted on the same site,
3399 or an absolute URI designating an error page on another site.
3400 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003401 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003402
3403 It is important to understand that this keyword is not meant to rewrite
3404 errors returned by the server, but errors detected and returned by HAProxy.
3405 This is why the list of supported errors is limited to a small set.
3406
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003407 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3408
Willy Tarreau2769aa02007-12-27 18:26:09 +01003409 Note that both keyword return the HTTP 302 status code, which tells the
3410 client to fetch the designated URL using the same HTTP method. This can be
3411 quite problematic in case of non-GET methods such as POST, because the URL
3412 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003413 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003414 status code, indicating to the client that the URL must be fetched with a GET
3415 request.
3416
3417 See also : "errorfile", "errorloc303"
3418
3419
3420errorloc303 <code> <url>
3421 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3422 May be used in sections : defaults | frontend | listen | backend
3423 yes | yes | yes | yes
3424 Arguments :
3425 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003426 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3427 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003428
3429 <url> it is the exact contents of the "Location" header. It may contain
3430 either a relative URI to an error page hosted on the same site,
3431 or an absolute URI designating an error page on another site.
3432 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003433 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003434
3435 It is important to understand that this keyword is not meant to rewrite
3436 errors returned by the server, but errors detected and returned by HAProxy.
3437 This is why the list of supported errors is limited to a small set.
3438
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003439 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3440
Willy Tarreau2769aa02007-12-27 18:26:09 +01003441 Note that both keyword return the HTTP 303 status code, which tells the
3442 client to fetch the designated URL using the same HTTP GET method. This
3443 solves the usual problems associated with "errorloc" and the 302 code. It is
3444 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003445 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003446
3447 See also : "errorfile", "errorloc", "errorloc302"
3448
3449
Simon Horman51a1cf62015-02-03 13:00:44 +09003450email-alert from <emailaddr>
3451 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003452 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003453 May be used in sections: defaults | frontend | listen | backend
3454 yes | yes | yes | yes
3455
3456 Arguments :
3457
3458 <emailaddr> is the from email address to use when sending email alerts
3459
3460 Also requires "email-alert mailers" and "email-alert to" to be set
3461 and if so sending email alerts is enabled for the proxy.
3462
Simon Horman64e34162015-02-06 11:11:57 +09003463 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003464 "email-alert myhostname", "email-alert to", section 3.6 about
3465 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003466
3467
3468email-alert level <level>
3469 Declare the maximum log level of messages for which email alerts will be
3470 sent. This acts as a filter on the sending of email alerts.
3471 May be used in sections: defaults | frontend | listen | backend
3472 yes | yes | yes | yes
3473
3474 Arguments :
3475
3476 <level> One of the 8 syslog levels:
3477 emerg alert crit err warning notice info debug
3478 The above syslog levels are ordered from lowest to highest.
3479
3480 By default level is alert
3481
3482 Also requires "email-alert from", "email-alert mailers" and
3483 "email-alert to" to be set and if so sending email alerts is enabled
3484 for the proxy.
3485
Simon Horman1421e212015-04-30 13:10:35 +09003486 Alerts are sent when :
3487
3488 * An un-paused server is marked as down and <level> is alert or lower
3489 * A paused server is marked as down and <level> is notice or lower
3490 * A server is marked as up or enters the drain state and <level>
3491 is notice or lower
3492 * "option log-health-checks" is enabled, <level> is info or lower,
3493 and a health check status update occurs
3494
Simon Horman64e34162015-02-06 11:11:57 +09003495 See also : "email-alert from", "email-alert mailers",
3496 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003497 section 3.6 about mailers.
3498
3499
3500email-alert mailers <mailersect>
3501 Declare the mailers to be used when sending email alerts
3502 May be used in sections: defaults | frontend | listen | backend
3503 yes | yes | yes | yes
3504
3505 Arguments :
3506
3507 <mailersect> is the name of the mailers section to send email alerts.
3508
3509 Also requires "email-alert from" and "email-alert to" to be set
3510 and if so sending email alerts is enabled for the proxy.
3511
Simon Horman64e34162015-02-06 11:11:57 +09003512 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3513 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003514
3515
3516email-alert myhostname <hostname>
3517 Declare the to hostname address to be used when communicating with
3518 mailers.
3519 May be used in sections: defaults | frontend | listen | backend
3520 yes | yes | yes | yes
3521
3522 Arguments :
3523
Baptiste Assmann738bad92015-12-21 15:27:53 +01003524 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003525
3526 By default the systems hostname is used.
3527
3528 Also requires "email-alert from", "email-alert mailers" and
3529 "email-alert to" to be set and if so sending email alerts is enabled
3530 for the proxy.
3531
Simon Horman64e34162015-02-06 11:11:57 +09003532 See also : "email-alert from", "email-alert level", "email-alert mailers",
3533 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003534
3535
3536email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003537 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003538 header of email alerts. This is the address that email alerts are sent to.
3539 May be used in sections: defaults | frontend | listen | backend
3540 yes | yes | yes | yes
3541
3542 Arguments :
3543
3544 <emailaddr> is the to email address to use when sending email alerts
3545
3546 Also requires "email-alert mailers" and "email-alert to" to be set
3547 and if so sending email alerts is enabled for the proxy.
3548
Simon Horman64e34162015-02-06 11:11:57 +09003549 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003550 "email-alert myhostname", section 3.6 about mailers.
3551
3552
Willy Tarreau4de91492010-01-22 19:10:05 +01003553force-persist { if | unless } <condition>
3554 Declare a condition to force persistence on down servers
3555 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003556 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003557
3558 By default, requests are not dispatched to down servers. It is possible to
3559 force this using "option persist", but it is unconditional and redispatches
3560 to a valid server if "option redispatch" is set. That leaves with very little
3561 possibilities to force some requests to reach a server which is artificially
3562 marked down for maintenance operations.
3563
3564 The "force-persist" statement allows one to declare various ACL-based
3565 conditions which, when met, will cause a request to ignore the down status of
3566 a server and still try to connect to it. That makes it possible to start a
3567 server, still replying an error to the health checks, and run a specially
3568 configured browser to test the service. Among the handy methods, one could
3569 use a specific source IP address, or a specific cookie. The cookie also has
3570 the advantage that it can easily be added/removed on the browser from a test
3571 page. Once the service is validated, it is then possible to open the service
3572 to the world by returning a valid response to health checks.
3573
3574 The forced persistence is enabled when an "if" condition is met, or unless an
3575 "unless" condition is met. The final redispatch is always disabled when this
3576 is used.
3577
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003578 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003579 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003580
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003581
3582filter <name> [param*]
3583 Add the filter <name> in the filter list attached to the proxy.
3584 May be used in sections : defaults | frontend | listen | backend
3585 no | yes | yes | yes
3586 Arguments :
3587 <name> is the name of the filter. Officially supported filters are
3588 referenced in section 9.
3589
Tim Düsterhus4896c442016-11-29 02:15:19 +01003590 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003591 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003592 filter. Please refer to the documentation of the corresponding
3593 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003594
3595 Multiple occurrences of the filter line can be used for the same proxy. The
3596 same filter can be referenced many times if needed.
3597
3598 Example:
3599 listen
3600 bind *:80
3601
3602 filter trace name BEFORE-HTTP-COMP
3603 filter compression
3604 filter trace name AFTER-HTTP-COMP
3605
3606 compression algo gzip
3607 compression offload
3608
3609 server srv1 192.168.0.1:80
3610
3611 See also : section 9.
3612
Willy Tarreau4de91492010-01-22 19:10:05 +01003613
Willy Tarreau2769aa02007-12-27 18:26:09 +01003614fullconn <conns>
3615 Specify at what backend load the servers will reach their maxconn
3616 May be used in sections : defaults | frontend | listen | backend
3617 yes | no | yes | yes
3618 Arguments :
3619 <conns> is the number of connections on the backend which will make the
3620 servers use the maximal number of connections.
3621
Willy Tarreau198a7442008-01-17 12:05:32 +01003622 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003623 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003624 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003625 load. The server will then always accept at least <minconn> connections,
3626 never more than <maxconn>, and the limit will be on the ramp between both
3627 values when the backend has less than <conns> concurrent connections. This
3628 makes it possible to limit the load on the servers during normal loads, but
3629 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003630 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003631
Willy Tarreaufbb78422011-06-05 15:38:35 +02003632 Since it's hard to get this value right, haproxy automatically sets it to
3633 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003634 backend (based on "use_backend" and "default_backend" rules). That way it's
3635 safe to leave it unset. However, "use_backend" involving dynamic names are
3636 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003637
Willy Tarreau2769aa02007-12-27 18:26:09 +01003638 Example :
3639 # The servers will accept between 100 and 1000 concurrent connections each
3640 # and the maximum of 1000 will be reached when the backend reaches 10000
3641 # connections.
3642 backend dynamic
3643 fullconn 10000
3644 server srv1 dyn1:80 minconn 100 maxconn 1000
3645 server srv2 dyn2:80 minconn 100 maxconn 1000
3646
3647 See also : "maxconn", "server"
3648
3649
3650grace <time>
3651 Maintain a proxy operational for some time after a soft stop
3652 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003653 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003654 Arguments :
3655 <time> is the time (by default in milliseconds) for which the instance
3656 will remain operational with the frontend sockets still listening
3657 when a soft-stop is received via the SIGUSR1 signal.
3658
3659 This may be used to ensure that the services disappear in a certain order.
3660 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003661 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003662 needed by the equipment to detect the failure.
3663
3664 Note that currently, there is very little benefit in using this parameter,
3665 and it may in fact complicate the soft-reconfiguration process more than
3666 simplify it.
3667
Willy Tarreau0ba27502007-12-24 16:55:16 +01003668
Andrew Rodland17be45e2016-10-25 17:04:12 -04003669hash-balance-factor <factor>
3670 Specify the balancing factor for bounded-load consistent hashing
3671 May be used in sections : defaults | frontend | listen | backend
3672 yes | no | no | yes
3673 Arguments :
3674 <factor> is the control for the maximum number of concurrent requests to
3675 send to a server, expressed as a percentage of the average number
3676 of concurrent requests across all of the active servers.
3677
3678 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3679 enables an algorithm that prevents any one server from getting too many
3680 requests at once, even if some hash buckets receive many more requests than
3681 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3682 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3683 then no server will be allowed to have a load more than 1.5 times the average.
3684 If server weights are used, they will be respected.
3685
3686 If the first-choice server is disqualified, the algorithm will choose another
3687 server based on the request hash, until a server with additional capacity is
3688 found. A higher <factor> allows more imbalance between the servers, while a
3689 lower <factor> means that more servers will be checked on average, affecting
3690 performance. Reasonable values are from 125 to 200.
3691
Willy Tarreau760e81d2018-05-03 07:20:40 +02003692 This setting is also used by "balance random" which internally relies on the
3693 consistent hashing mechanism.
3694
Andrew Rodland17be45e2016-10-25 17:04:12 -04003695 See also : "balance" and "hash-type".
3696
3697
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003698hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003699 Specify a method to use for mapping hashes to servers
3700 May be used in sections : defaults | frontend | listen | backend
3701 yes | no | yes | yes
3702 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003703 <method> is the method used to select a server from the hash computed by
3704 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003705
Bhaskar98634f02013-10-29 23:30:51 -04003706 map-based the hash table is a static array containing all alive servers.
3707 The hashes will be very smooth, will consider weights, but
3708 will be static in that weight changes while a server is up
3709 will be ignored. This means that there will be no slow start.
3710 Also, since a server is selected by its position in the array,
3711 most mappings are changed when the server count changes. This
3712 means that when a server goes up or down, or when a server is
3713 added to a farm, most connections will be redistributed to
3714 different servers. This can be inconvenient with caches for
3715 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003716
Bhaskar98634f02013-10-29 23:30:51 -04003717 consistent the hash table is a tree filled with many occurrences of each
3718 server. The hash key is looked up in the tree and the closest
3719 server is chosen. This hash is dynamic, it supports changing
3720 weights while the servers are up, so it is compatible with the
3721 slow start feature. It has the advantage that when a server
3722 goes up or down, only its associations are moved. When a
3723 server is added to the farm, only a few part of the mappings
3724 are redistributed, making it an ideal method for caches.
3725 However, due to its principle, the distribution will never be
3726 very smooth and it may sometimes be necessary to adjust a
3727 server's weight or its ID to get a more balanced distribution.
3728 In order to get the same distribution on multiple load
3729 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003730 same IDs. Note: consistent hash uses sdbm and avalanche if no
3731 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003732
3733 <function> is the hash function to be used :
3734
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003735 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003736 reimplementation of ndbm) database library. It was found to do
3737 well in scrambling bits, causing better distribution of the keys
3738 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003739 function with good distribution, unless the total server weight
3740 is a multiple of 64, in which case applying the avalanche
3741 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003742
3743 djb2 this function was first proposed by Dan Bernstein many years ago
3744 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003745 function provides a better distribution than sdbm. It generally
3746 works well with text-based inputs though it can perform extremely
3747 poorly with numeric-only input or when the total server weight is
3748 a multiple of 33, unless the avalanche modifier is also used.
3749
Willy Tarreaua0f42712013-11-14 14:30:35 +01003750 wt6 this function was designed for haproxy while testing other
3751 functions in the past. It is not as smooth as the other ones, but
3752 is much less sensible to the input data set or to the number of
3753 servers. It can make sense as an alternative to sdbm+avalanche or
3754 djb2+avalanche for consistent hashing or when hashing on numeric
3755 data such as a source IP address or a visitor identifier in a URL
3756 parameter.
3757
Willy Tarreau324f07f2015-01-20 19:44:50 +01003758 crc32 this is the most common CRC32 implementation as used in Ethernet,
3759 gzip, PNG, etc. It is slower than the other ones but may provide
3760 a better distribution or less predictable results especially when
3761 used on strings.
3762
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003763 <modifier> indicates an optional method applied after hashing the key :
3764
3765 avalanche This directive indicates that the result from the hash
3766 function above should not be used in its raw form but that
3767 a 4-byte full avalanche hash must be applied first. The
3768 purpose of this step is to mix the resulting bits from the
3769 previous hash in order to avoid any undesired effect when
3770 the input contains some limited values or when the number of
3771 servers is a multiple of one of the hash's components (64
3772 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3773 result less predictable, but it's also not as smooth as when
3774 using the original function. Some testing might be needed
3775 with some workloads. This hash is one of the many proposed
3776 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003777
Bhaskar98634f02013-10-29 23:30:51 -04003778 The default hash type is "map-based" and is recommended for most usages. The
3779 default function is "sdbm", the selection of a function should be based on
3780 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003781
Andrew Rodland17be45e2016-10-25 17:04:12 -04003782 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003783
3784
Willy Tarreau0ba27502007-12-24 16:55:16 +01003785http-check disable-on-404
3786 Enable a maintenance mode upon HTTP/404 response to health-checks
3787 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003788 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003789 Arguments : none
3790
3791 When this option is set, a server which returns an HTTP code 404 will be
3792 excluded from further load-balancing, but will still receive persistent
3793 connections. This provides a very convenient method for Web administrators
3794 to perform a graceful shutdown of their servers. It is also important to note
3795 that a server which is detected as failed while it was in this mode will not
3796 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3797 will immediately be reinserted into the farm. The status on the stats page
3798 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003799 option only works in conjunction with the "httpchk" option. If this option
3800 is used with "http-check expect", then it has precedence over it so that 404
3801 responses will still be considered as soft-stop.
3802
3803 See also : "option httpchk", "http-check expect"
3804
3805
3806http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003807 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003808 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003809 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003810 Arguments :
3811 <match> is a keyword indicating how to look for a specific pattern in the
3812 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003813 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003814 exclamation mark ("!") to negate the match. Spaces are allowed
3815 between the exclamation mark and the keyword. See below for more
3816 details on the supported keywords.
3817
3818 <pattern> is the pattern to look for. It may be a string or a regular
3819 expression. If the pattern contains spaces, they must be escaped
3820 with the usual backslash ('\').
3821
3822 By default, "option httpchk" considers that response statuses 2xx and 3xx
3823 are valid, and that others are invalid. When "http-check expect" is used,
3824 it defines what is considered valid or invalid. Only one "http-check"
3825 statement is supported in a backend. If a server fails to respond or times
3826 out, the check obviously fails. The available matches are :
3827
3828 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003829 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003830 response's status code is exactly this string. If the
3831 "status" keyword is prefixed with "!", then the response
3832 will be considered invalid if the status code matches.
3833
3834 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003835 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003836 response's status code matches the expression. If the
3837 "rstatus" keyword is prefixed with "!", then the response
3838 will be considered invalid if the status code matches.
3839 This is mostly used to check for multiple codes.
3840
3841 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003842 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003843 response's body contains this exact string. If the
3844 "string" keyword is prefixed with "!", then the response
3845 will be considered invalid if the body contains this
3846 string. This can be used to look for a mandatory word at
3847 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003848 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003849 trace).
3850
3851 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003852 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003853 response's body matches this expression. If the "rstring"
3854 keyword is prefixed with "!", then the response will be
3855 considered invalid if the body matches the expression.
3856 This can be used to look for a mandatory word at the end
3857 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003858 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003859
3860 It is important to note that the responses will be limited to a certain size
3861 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3862 Thus, too large responses may not contain the mandatory pattern when using
3863 "string" or "rstring". If a large response is absolutely required, it is
3864 possible to change the default max size by setting the global variable.
3865 However, it is worth keeping in mind that parsing very large responses can
3866 waste some CPU cycles, especially when regular expressions are used, and that
3867 it is always better to focus the checks on smaller resources.
3868
Cyril Bonté32602d22015-01-30 00:07:07 +01003869 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3870 will automatically append a "Connection: close" header, meaning that this
3871 header should not be present in the request provided by "option httpchk".
3872
Willy Tarreaubd741542010-03-16 18:46:54 +01003873 Last, if "http-check expect" is combined with "http-check disable-on-404",
3874 then this last one has precedence when the server responds with 404.
3875
3876 Examples :
3877 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003878 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003879
3880 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003881 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003882
3883 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003884 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003885
3886 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003887 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003888
Willy Tarreaubd741542010-03-16 18:46:54 +01003889 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003890
3891
Willy Tarreauef781042010-01-27 11:53:01 +01003892http-check send-state
3893 Enable emission of a state header with HTTP health checks
3894 May be used in sections : defaults | frontend | listen | backend
3895 yes | no | yes | yes
3896 Arguments : none
3897
3898 When this option is set, haproxy will systematically send a special header
3899 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3900 how they are seen by haproxy. This can be used for instance when a server is
3901 manipulated without access to haproxy and the operator needs to know whether
3902 haproxy still sees it up or not, or if the server is the last one in a farm.
3903
3904 The header is composed of fields delimited by semi-colons, the first of which
3905 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3906 checks on the total number before transition, just as appears in the stats
3907 interface. Next headers are in the form "<variable>=<value>", indicating in
3908 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003909 - a variable "address", containing the address of the backend server.
3910 This corresponds to the <address> field in the server declaration. For
3911 unix domain sockets, it will read "unix".
3912
3913 - a variable "port", containing the port of the backend server. This
3914 corresponds to the <port> field in the server declaration. For unix
3915 domain sockets, it will read "unix".
3916
Willy Tarreauef781042010-01-27 11:53:01 +01003917 - a variable "name", containing the name of the backend followed by a slash
3918 ("/") then the name of the server. This can be used when a server is
3919 checked in multiple backends.
3920
3921 - a variable "node" containing the name of the haproxy node, as set in the
3922 global "node" variable, otherwise the system's hostname if unspecified.
3923
3924 - a variable "weight" indicating the weight of the server, a slash ("/")
3925 and the total weight of the farm (just counting usable servers). This
3926 helps to know if other servers are available to handle the load when this
3927 one fails.
3928
3929 - a variable "scur" indicating the current number of concurrent connections
3930 on the server, followed by a slash ("/") then the total number of
3931 connections on all servers of the same backend.
3932
3933 - a variable "qcur" indicating the current number of requests in the
3934 server's queue.
3935
3936 Example of a header received by the application server :
3937 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
3938 scur=13/22; qcur=0
3939
3940 See also : "option httpchk", "http-check disable-on-404"
3941
Willy Tarreau53275e82017-11-24 07:52:01 +01003942http-request { allow | auth [realm <realm>] | redirect <rule> | reject |
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003943 tarpit [deny_status <status>] | deny [deny_status <status>] |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003944 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003945 capture <sample> [ len <length> | id <id> ] |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003946 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003947 replace-header <name> <match-regex> <replace-fmt> |
3948 replace-value <name> <match-regex> <replace-fmt> |
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01003949 set-method <fmt> | set-path <fmt> | set-query <fmt> |
3950 set-uri <fmt> | set-tos <tos> | set-mark <mark> |
Patrick Hemmer268a7072018-05-11 12:52:31 -04003951 set-priority-class <expr> | set-priority-offset <expr>
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003952 add-acl(<file name>) <key fmt> |
3953 del-acl(<file name>) <key fmt> |
3954 del-map(<file name>) <key fmt> |
Baptiste Assmannbb7e86a2014-09-03 18:29:47 +02003955 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02003956 set-var(<var name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01003957 unset-var(<var name>) |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01003958 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02003959 sc-inc-gpc0(<sc-id>) |
Frédéric Lécaille6778b272018-01-29 15:22:53 +01003960 sc-inc-gpc1(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02003961 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02003962 silent-drop |
William Lallemand86d0df02017-11-24 21:36:45 +01003963 send-spoe-group |
3964 cache-use
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003965 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01003966 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003967 Access control for Layer 7 requests
3968
3969 May be used in sections: defaults | frontend | listen | backend
3970 no | yes | yes | yes
3971
Willy Tarreau20b0de52012-12-24 15:45:22 +01003972 The http-request statement defines a set of rules which apply to layer 7
3973 processing. The rules are evaluated in their declaration order when they are
3974 met in a frontend, listen or backend section. Any rule may optionally be
3975 followed by an ACL-based condition, in which case it will only be evaluated
3976 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003977
Willy Tarreau20b0de52012-12-24 15:45:22 +01003978 The first keyword is the rule's action. Currently supported actions include :
3979 - "allow" : this stops the evaluation of the rules and lets the request
3980 pass the check. No further "http-request" rules are evaluated.
3981
3982 - "deny" : this stops the evaluation of the rules and immediately rejects
Willy Tarreaube1d34d2016-06-26 19:37:59 +02003983 the request and emits an HTTP 403 error, or optionally the status code
3984 specified as an argument to "deny_status". The list of permitted status
3985 codes is limited to those that can be overridden by the "errorfile"
3986 directive. No further "http-request" rules are evaluated.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003987
Willy Tarreau53275e82017-11-24 07:52:01 +01003988 - "reject" : this stops the evaluation of the rules and immediately closes
3989 the connection without sending any response. It acts similarly to the
3990 "tcp-request content reject" rules. It can be useful to force an
3991 immediate connection closure on HTTP/2 connections.
3992
Willy Tarreauccbcc372012-12-27 12:37:57 +01003993 - "tarpit" : this stops the evaluation of the rules and immediately blocks
3994 the request without responding for a delay specified by "timeout tarpit"
3995 or "timeout connect" if the former is not set. After that delay, if the
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003996 client is still connected, an HTTP error 500 (or optionally the status
3997 code specified as an argument to "deny_status") is returned so that the
Willy Tarreauccbcc372012-12-27 12:37:57 +01003998 client does not suspect it has been tarpitted. Logs will report the flags
3999 "PT". The goal of the tarpit rule is to slow down robots during an attack
4000 when they're limited on the number of concurrent requests. It can be very
4001 efficient against very dumb robots, and will significantly reduce the
4002 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004003 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreau2d392c22015-08-24 01:43:45 +02004004 front firewall to support insane number of concurrent connections. See
4005 also the "silent-drop" action below.
Willy Tarreauccbcc372012-12-27 12:37:57 +01004006
Willy Tarreau20b0de52012-12-24 15:45:22 +01004007 - "auth" : this stops the evaluation of the rules and immediately responds
4008 with an HTTP 401 or 407 error code to invite the user to present a valid
4009 user name and password. No further "http-request" rules are evaluated. An
4010 optional "realm" parameter is supported, it sets the authentication realm
4011 that is returned with the response (typically the application's name).
4012
Willy Tarreau81499eb2012-12-27 12:19:02 +01004013 - "redirect" : this performs an HTTP redirection based on a redirect rule.
4014 This is exactly the same as the "redirect" statement except that it
4015 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01004016 "http-request" rules and that these rules use the "log-format" strings.
4017 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01004018
Willy Tarreau20b0de52012-12-24 15:45:22 +01004019 - "add-header" appends an HTTP header field whose name is specified in
4020 <name> and whose value is defined by <fmt> which follows the log-format
4021 rules (see Custom Log Format in section 8.2.4). This is particularly
Davor Ocelice9ed2812017-12-25 17:49:28 +01004022 useful to pass connection-specific information to the server (e.g. the
Willy Tarreau20b0de52012-12-24 15:45:22 +01004023 client's SSL certificate), or to combine several headers into one. This
4024 rule is not final, so it is possible to add other similar rules. Note
4025 that header addition is performed immediately, so one rule might reuse
4026 the resulting header from a previous rule.
4027
4028 - "set-header" does the same as "add-header" except that the header name
4029 is first removed if it existed. This is useful when passing security
4030 information to the server, where the header must not be manipulated by
Willy Tarreau85603282015-01-21 20:39:27 +01004031 external users. Note that the new value is computed before the removal so
4032 it is possible to concatenate a value to an existing header.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004033
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004034 - "del-header" removes all HTTP header fields whose name is specified in
4035 <name>.
4036
Sasha Pachev218f0642014-06-16 12:05:59 -06004037 - "replace-header" matches the regular expression in all occurrences of
4038 header field <name> according to <match-regex>, and replaces them with
4039 the <replace-fmt> argument. Format characters are allowed in replace-fmt
4040 and work like in <fmt> arguments in "add-header". The match is only
4041 case-sensitive. It is important to understand that this action only
4042 considers whole header lines, regardless of the number of values they
4043 may contain. This usage is suited to headers naturally containing commas
4044 in their value, such as If-Modified-Since and so on.
4045
4046 Example:
4047
4048 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
4049
4050 applied to:
4051
4052 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4053
4054 outputs:
4055
4056 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4057
4058 assuming the backend IP is 192.168.1.20
4059
4060 - "replace-value" works like "replace-header" except that it matches the
4061 regex against every comma-delimited value of the header field <name>
4062 instead of the entire header. This is suited for all headers which are
4063 allowed to carry more than one value. An example could be the Accept
4064 header.
4065
4066 Example:
4067
4068 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
4069
4070 applied to:
4071
4072 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
4073
4074 outputs:
4075
4076 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
4077
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004078 - "set-method" rewrites the request method with the result of the
4079 evaluation of format string <fmt>. There should be very few valid reasons
4080 for having to do so as this is more likely to break something than to fix
4081 it.
4082
4083 - "set-path" rewrites the request path with the result of the evaluation of
4084 format string <fmt>. The query string, if any, is left intact. If a
4085 scheme and authority is found before the path, they are left intact as
4086 well. If the request doesn't have a path ("*"), this one is replaced with
4087 the format. This can be used to prepend a directory component in front of
4088 a path for example. See also "set-query" and "set-uri".
4089
4090 Example :
4091 # prepend the host name before the path
4092 http-request set-path /%[hdr(host)]%[path]
4093
4094 - "set-query" rewrites the request's query string which appears after the
4095 first question mark ("?") with the result of the evaluation of format
4096 string <fmt>. The part prior to the question mark is left intact. If the
4097 request doesn't contain a question mark and the new value is not empty,
4098 then one is added at the end of the URI, followed by the new value. If
4099 a question mark was present, it will never be removed even if the value
4100 is empty. This can be used to add or remove parameters from the query
4101 string. See also "set-query" and "set-uri".
4102
4103 Example :
4104 # replace "%3D" with "=" in the query string
4105 http-request set-query %[query,regsub(%3D,=,g)]
4106
4107 - "set-uri" rewrites the request URI with the result of the evaluation of
4108 format string <fmt>. The scheme, authority, path and query string are all
4109 replaced at once. This can be used to rewrite hosts in front of proxies,
4110 or to perform complex modifications to the URI such as moving parts
4111 between the path and the query string. See also "set-path" and
4112 "set-query".
4113
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004114 - "set-nice" sets the "nice" factor of the current request being processed.
4115 It only has effect against the other requests being processed at the same
4116 time. The default value is 0, unless altered by the "nice" setting on the
4117 "bind" line. The accepted range is -1024..1024. The higher the value, the
4118 nicest the request will be. Lower values will make the request more
4119 important than other ones. This can be useful to improve the speed of
4120 some requests, or lower the priority of non-important requests. Using
4121 this setting without prior experimentation can cause some major slowdown.
4122
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004123 - "set-log-level" is used to change the log level of the current request
4124 when a certain condition is met. Valid levels are the 8 syslog levels
4125 (see the "log" keyword) plus the special level "silent" which disables
4126 logging for this request. This rule is not final so the last matching
4127 rule wins. This rule can be useful to disable health checks coming from
4128 another equipment.
4129
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004130 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
4131 the client to the value passed in <tos> on platforms which support this.
4132 This value represents the whole 8 bits of the IP TOS field, and can be
4133 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
4134 that only the 6 higher bits are used in DSCP or TOS, and the two lower
Davor Ocelice9ed2812017-12-25 17:49:28 +01004135 bits are always 0. This can be used to adjust some routing behavior on
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004136 border routers based on some information from the request. See RFC 2474,
4137 2597, 3260 and 4594 for more information.
4138
Willy Tarreau51347ed2013-06-11 19:34:13 +02004139 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
4140 client to the value passed in <mark> on platforms which support it. This
4141 value is an unsigned 32 bit value which can be matched by netfilter and
4142 by the routing table. It can be expressed both in decimal or hexadecimal
4143 format (prefixed by "0x"). This can be useful to force certain packets to
4144 take a different route (for example a cheaper network path for bulk
4145 downloads). This works on Linux kernels 2.6.32 and above and requires
4146 admin privileges.
4147
Patrick Hemmer268a7072018-05-11 12:52:31 -04004148 - "set-priority-class" is used to set the queue priority class of the
4149 current request. The value must be a sample expression which converts to
4150 an integer in the range -2047..2047. Results outside this range will be
4151 truncated. The priority class determines the order in which queued
4152 requests are processed. Lower values have higher priority.
4153
4154 - "set-priority-offset" is used to set the queue priority timestamp offset
4155 of the current request. The value must be a sample expression which
4156 converts to an integer in the range -524287..524287. Results outside this
4157 range will be truncated. When a request is queued, it is ordered first by
4158 the priority class, then by the current timestamp adjusted by the given
4159 offset in milliseconds. Lower values have higher priority.
4160 Note that the resulting timestamp is is only tracked with enough precision
4161 for 524,287ms (8m44s287ms). If the request is queued long enough to where
4162 the adjusted timestamp exceeds this value, it will be misidentified as
4163 highest priority. Thus it is important to set "timeout queue" to a value,
4164 where when combined with the offset, does not exceed this limit.
4165
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004166 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
4167 from a file (even a dummy empty file). The file name of the ACL to be
4168 updated is passed between parentheses. It takes one argument: <key fmt>,
4169 which follows log-format rules, to collect content of the new entry. It
4170 performs a lookup in the ACL before insertion, to avoid duplicated (or
4171 more) values. This lookup is done by a linear search and can be expensive
4172 with large lists! It is the equivalent of the "add acl" command from the
4173 stats socket, but can be triggered by an HTTP request.
4174
4175 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
4176 from a file (even a dummy empty file). The file name of the ACL to be
4177 updated is passed between parentheses. It takes one argument: <key fmt>,
4178 which follows log-format rules, to collect content of the entry to delete.
4179 It is the equivalent of the "del acl" command from the stats socket, but
4180 can be triggered by an HTTP request.
4181
4182 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
4183 from a file (even a dummy empty file). The file name of the MAP to be
4184 updated is passed between parentheses. It takes one argument: <key fmt>,
4185 which follows log-format rules, to collect content of the entry to delete.
4186 It takes one argument: "file name" It is the equivalent of the "del map"
4187 command from the stats socket, but can be triggered by an HTTP request.
4188
4189 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
4190 from a file (even a dummy empty file). The file name of the MAP to be
4191 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
4192 which follows log-format rules, used to collect MAP key, and <value fmt>,
4193 which follows log-format rules, used to collect content for the new entry.
4194 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4195 more) values. This lookup is done by a linear search and can be expensive
4196 with large lists! It is the equivalent of the "set map" command from the
4197 stats socket, but can be triggered by an HTTP request.
4198
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004199 - capture <sample> [ len <length> | id <id> ] :
Willy Tarreaua9083d02015-05-08 15:27:59 +02004200 captures sample expression <sample> from the request buffer, and converts
4201 it to a string of at most <len> characters. The resulting string is
4202 stored into the next request "capture" slot, so it will possibly appear
4203 next to some captured HTTP headers. It will then automatically appear in
4204 the logs, and it will be possible to extract it using sample fetch rules
4205 to feed it into headers or anything. The length should be limited given
4206 that this size will be allocated for each capture during the whole
4207 session life. Please check section 7.3 (Fetching samples) and "capture
4208 request header" for more information.
4209
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004210 If the keyword "id" is used instead of "len", the action tries to store
4211 the captured string in a previously declared capture slot. This is useful
4212 to run captures in backends. The slot id can be declared by a previous
4213 directive "http-request capture" or with the "declare capture" keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01004214 If the slot <id> doesn't exist, then HAProxy fails parsing the
4215 configuration to prevent unexpected behavior at run time.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004216
William Lallemand86d0df02017-11-24 21:36:45 +01004217 - cache-use <name> :
4218 See section 10.2 about cache setup.
4219
Willy Tarreau09448f72014-06-25 18:12:15 +02004220 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4221 enables tracking of sticky counters from current request. These rules
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02004222 do not stop evaluation and do not change default action. The number of
4223 counters that may be simultaneously tracked by the same connection is set
4224 in MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults
4225 to 3, so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The
4226 first "track-sc0" rule executed enables tracking of the counters of the
Willy Tarreau09448f72014-06-25 18:12:15 +02004227 specified table as the first set. The first "track-sc1" rule executed
4228 enables tracking of the counters of the specified table as the second
4229 set. The first "track-sc2" rule executed enables tracking of the
4230 counters of the specified table as the third set. It is a recommended
4231 practice to use the first set of counters for the per-frontend counters
4232 and the second set for the per-backend ones. But this is just a
4233 guideline, all may be used everywhere.
4234
4235 These actions take one or two arguments :
4236 <key> is mandatory, and is a sample expression rule as described
4237 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01004238 request or connection will be analyzed, extracted, combined,
Willy Tarreau09448f72014-06-25 18:12:15 +02004239 and used to select which table entry to update the counters.
4240
4241 <table> is an optional table to be used instead of the default one,
4242 which is the stick-table declared in the current proxy. All
4243 the counters for the matches and updates for the key will
4244 then be performed in that table until the session ends.
4245
4246 Once a "track-sc*" rule is executed, the key is looked up in the table
4247 and if it is not found, an entry is allocated for it. Then a pointer to
4248 that entry is kept during all the session's life, and this entry's
4249 counters are updated as often as possible, every time the session's
4250 counters are updated, and also systematically when the session ends.
4251 Counters are only updated for events that happen after the tracking has
4252 been started. As an exception, connection counters and request counters
4253 are systematically updated so that they reflect useful information.
4254
4255 If the entry tracks concurrent connection counters, one connection is
4256 counted for as long as the entry is tracked, and the entry will not
4257 expire during that time. Tracking counters also provides a performance
4258 advantage over just checking the keys, because only one table lookup is
4259 performed for all ACL checks that make use of it.
4260
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004261 - sc-set-gpt0(<sc-id>) <int> :
4262 This action sets the GPT0 tag according to the sticky counter designated
4263 by <sc-id> and the value of <int>. The expected result is a boolean. If
4264 an error occurs, this action silently fails and the actions evaluation
4265 continues.
4266
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004267 - sc-inc-gpc0(<sc-id>):
4268 This action increments the GPC0 counter according with the sticky counter
4269 designated by <sc-id>. If an error occurs, this action silently fails and
4270 the actions evaluation continues.
4271
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004272 - sc-inc-gpc1(<sc-id>):
4273 Same as "sc-inc-gpc0" action above but for GPC1 counter.
4274
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004275 - set-var(<var-name>) <expr> :
4276 Is used to set the contents of a variable. The variable is declared
4277 inline.
4278
Daniel Schneller0b547052016-03-21 20:46:57 +01004279 <var-name> The name of the variable starts with an indication about
4280 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004281 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004282 "sess" : the variable is shared with the whole session
4283 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004284 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004285 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004286 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004287 "res" : the variable is shared only during response
4288 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004289 This prefix is followed by a name. The separator is a '.'.
Daniel Schneller0b547052016-03-21 20:46:57 +01004290 The name may only contain characters 'a-z', 'A-Z', '0-9'
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004291 and '_'.
4292
4293 <expr> Is a standard HAProxy expression formed by a sample-fetch
4294 followed by some converters.
4295
4296 Example:
4297
4298 http-request set-var(req.my_var) req.fhdr(user-agent),lower
4299
Christopher Faulet85d79c92016-11-09 16:54:56 +01004300 - unset-var(<var-name>) :
4301 Is used to unset a variable. See above for details about <var-name>.
4302
4303 Example:
4304
4305 http-request unset-var(req.my_var)
4306
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004307 - set-src <expr> :
4308 Is used to set the source IP address to the value of specified
4309 expression. Useful when a proxy in front of HAProxy rewrites source IP,
4310 but provides the correct IP in a HTTP header; or you want to mask
4311 source IP for privacy.
4312
4313 <expr> Is a standard HAProxy expression formed by a sample-fetch
4314 followed by some converters.
4315
4316 Example:
4317
4318 http-request set-src hdr(x-forwarded-for)
4319 http-request set-src src,ipmask(24)
4320
Willy Tarreau00005ce2016-10-21 15:07:45 +02004321 When possible, set-src preserves the original source port as long as the
4322 address family allows it, otherwise the source port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004323
William Lallemand44be6402016-05-25 01:51:35 +02004324 - set-src-port <expr> :
4325 Is used to set the source port address to the value of specified
4326 expression.
4327
4328 <expr> Is a standard HAProxy expression formed by a sample-fetch
4329 followed by some converters.
4330
4331 Example:
4332
4333 http-request set-src-port hdr(x-port)
4334 http-request set-src-port int(4000)
4335
Willy Tarreau00005ce2016-10-21 15:07:45 +02004336 When possible, set-src-port preserves the original source address as long
4337 as the address family supports a port, otherwise it forces the source
4338 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004339
William Lallemand13e9b0c2016-05-25 02:34:07 +02004340 - set-dst <expr> :
4341 Is used to set the destination IP address to the value of specified
4342 expression. Useful when a proxy in front of HAProxy rewrites destination
4343 IP, but provides the correct IP in a HTTP header; or you want to mask
4344 the IP for privacy. If you want to connect to the new address/port, use
4345 '0.0.0.0:0' as a server address in the backend.
4346
4347 <expr> Is a standard HAProxy expression formed by a sample-fetch
4348 followed by some converters.
4349
4350 Example:
4351
4352 http-request set-dst hdr(x-dst)
4353 http-request set-dst dst,ipmask(24)
4354
Willy Tarreau00005ce2016-10-21 15:07:45 +02004355 When possible, set-dst preserves the original destination port as long as
4356 the address family allows it, otherwise the destination port is set to 0.
4357
William Lallemand13e9b0c2016-05-25 02:34:07 +02004358 - set-dst-port <expr> :
4359 Is used to set the destination port address to the value of specified
4360 expression. If you want to connect to the new address/port, use
4361 '0.0.0.0:0' as a server address in the backend.
4362
4363 <expr> Is a standard HAProxy expression formed by a sample-fetch
4364 followed by some converters.
4365
4366 Example:
4367
4368 http-request set-dst-port hdr(x-port)
4369 http-request set-dst-port int(4000)
4370
Willy Tarreau00005ce2016-10-21 15:07:45 +02004371 When possible, set-dst-port preserves the original destination address as
4372 long as the address family supports a port, otherwise it forces the
4373 destination address to IPv4 "0.0.0.0" before rewriting the port.
4374
Willy Tarreau2d392c22015-08-24 01:43:45 +02004375 - "silent-drop" : this stops the evaluation of the rules and makes the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004376 client-facing connection suddenly disappear using a system-dependent way
Willy Tarreau2d392c22015-08-24 01:43:45 +02004377 that tries to prevent the client from being notified. The effect it then
4378 that the client still sees an established connection while there's none
4379 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4380 except that it doesn't use any local resource at all on the machine
4381 running HAProxy. It can resist much higher loads than "tarpit", and slow
Davor Ocelice9ed2812017-12-25 17:49:28 +01004382 down stronger attackers. It is important to understand the impact of using
4383 this mechanism. All stateful equipment placed between the client and
Willy Tarreau2d392c22015-08-24 01:43:45 +02004384 HAProxy (firewalls, proxies, load balancers) will also keep the
4385 established connection for a long time and may suffer from this action.
4386 On modern Linux systems running with enough privileges, the TCP_REPAIR
4387 socket option is used to block the emission of a TCP reset. On other
4388 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4389 pass the first router, though it's still delivered to local networks. Do
4390 not use it unless you fully understand how it works.
4391
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004392
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004393 - "wait-for-handshake" : this will delay the processing of the request
4394 until the SSL handshake happened. This is mostly useful to delay
4395 processing early data until we're sure they are valid.
4396
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004397 - send-spoe-group <engine-name> <group-name> :
4398 This action is used to trigger sending of a group of SPOE messages. To do
4399 so, the SPOE engine used to send messages must be defined, as well as the
4400 SPOE group to send. Of course, the SPOE engine must refer to an existing
4401 SPOE filter. If not engine name is provided on the SPOE filter line, the
4402 SPOE agent name must be used.
4403
4404 <engine-name> The SPOE engine name.
4405
4406 <group-name> The SPOE group name as specified in the engine
4407 configuration.
4408
Willy Tarreau20b0de52012-12-24 15:45:22 +01004409 There is no limit to the number of http-request statements per instance.
4410
4411 It is important to know that http-request rules are processed very early in
4412 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004413 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4414 visible by almost all further ACL rules.
4415
4416 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4417 in newer versions (>= 1.5). But if you need to use regular expression to
4418 delete headers, you can still use "reqdel". Also please use
4419 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004420
4421 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01004422 acl nagios src 192.168.129.3
4423 acl local_net src 192.168.0.0/16
4424 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004425
Cyril Bonté78caf842010-03-10 22:41:43 +01004426 http-request allow if nagios
4427 http-request allow if local_net auth_ok
4428 http-request auth realm Gimme if local_net auth_ok
4429 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004430
Cyril Bonté78caf842010-03-10 22:41:43 +01004431 Example:
4432 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01004433 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004434
Willy Tarreau20b0de52012-12-24 15:45:22 +01004435 Example:
4436 http-request set-header X-Haproxy-Current-Date %T
4437 http-request set-header X-SSL %[ssl_fc]
Willy Tarreaufca42612015-08-27 17:15:05 +02004438 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004439 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4440 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4441 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4442 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4443 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4444 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
4445
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004446 Example:
4447 acl key req.hdr(X-Add-Acl-Key) -m found
4448 acl add path /addacl
4449 acl del path /delacl
4450
4451 acl myhost hdr(Host) -f myhost.lst
4452
4453 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4454 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
4455
4456 Example:
4457 acl value req.hdr(X-Value) -m found
4458 acl setmap path /setmap
4459 acl delmap path /delmap
4460
4461 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4462
4463 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4464 http-request del-map(map.lst) %[src] if delmap
4465
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02004466 See also : "stats http-request", section 3.4 about userlists and section 7
4467 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01004468
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004469http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Willy Tarreau51d861a2015-05-22 17:30:48 +02004470 capture <sample> id <id> | redirect <rule> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004471 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06004472 replace-header <name> <regex-match> <replace-fmt> |
4473 replace-value <name> <regex-match> <replace-fmt> |
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004474 set-status <status> [reason <str>] |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004475 set-log-level <level> | set-mark <mark> | set-tos <tos> |
4476 add-acl(<file name>) <key fmt> |
4477 del-acl(<file name>) <key fmt> |
4478 del-map(<file name>) <key fmt> |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01004479 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004480 set-var(<var-name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01004481 unset-var(<var-name>) |
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004482 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004483 sc-inc-gpc0(<sc-id>) |
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004484 sc-inc-gpc1(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004485 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02004486 silent-drop |
William Lallemand86d0df02017-11-24 21:36:45 +01004487 send-spoe-group |
4488 cache-store
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004489 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02004490 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004491 Access control for Layer 7 responses
4492
4493 May be used in sections: defaults | frontend | listen | backend
4494 no | yes | yes | yes
4495
4496 The http-response statement defines a set of rules which apply to layer 7
4497 processing. The rules are evaluated in their declaration order when they are
4498 met in a frontend, listen or backend section. Any rule may optionally be
4499 followed by an ACL-based condition, in which case it will only be evaluated
4500 if the condition is true. Since these rules apply on responses, the backend
4501 rules are applied first, followed by the frontend's rules.
4502
4503 The first keyword is the rule's action. Currently supported actions include :
4504 - "allow" : this stops the evaluation of the rules and lets the response
4505 pass the check. No further "http-response" rules are evaluated for the
4506 current section.
4507
4508 - "deny" : this stops the evaluation of the rules and immediately rejects
4509 the response and emits an HTTP 502 error. No further "http-response"
4510 rules are evaluated.
4511
4512 - "add-header" appends an HTTP header field whose name is specified in
4513 <name> and whose value is defined by <fmt> which follows the log-format
4514 rules (see Custom Log Format in section 8.2.4). This may be used to send
4515 a cookie to a client for example, or to pass some internal information.
4516 This rule is not final, so it is possible to add other similar rules.
4517 Note that header addition is performed immediately, so one rule might
4518 reuse the resulting header from a previous rule.
4519
4520 - "set-header" does the same as "add-header" except that the header name
4521 is first removed if it existed. This is useful when passing security
4522 information to the server, where the header must not be manipulated by
4523 external users.
4524
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004525 - "del-header" removes all HTTP header fields whose name is specified in
4526 <name>.
4527
Sasha Pachev218f0642014-06-16 12:05:59 -06004528 - "replace-header" matches the regular expression in all occurrences of
4529 header field <name> according to <match-regex>, and replaces them with
4530 the <replace-fmt> argument. Format characters are allowed in replace-fmt
4531 and work like in <fmt> arguments in "add-header". The match is only
4532 case-sensitive. It is important to understand that this action only
4533 considers whole header lines, regardless of the number of values they
4534 may contain. This usage is suited to headers naturally containing commas
4535 in their value, such as Set-Cookie, Expires and so on.
4536
4537 Example:
4538
4539 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4540
4541 applied to:
4542
4543 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4544
4545 outputs:
4546
4547 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4548
4549 assuming the backend IP is 192.168.1.20.
4550
4551 - "replace-value" works like "replace-header" except that it matches the
4552 regex against every comma-delimited value of the header field <name>
4553 instead of the entire header. This is suited for all headers which are
4554 allowed to carry more than one value. An example could be the Accept
4555 header.
4556
4557 Example:
4558
4559 http-response replace-value Cache-control ^public$ private
4560
4561 applied to:
4562
4563 Cache-Control: max-age=3600, public
4564
4565 outputs:
4566
4567 Cache-Control: max-age=3600, private
4568
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004569 - "set-status" replaces the response status code with <status> which must
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004570 be an integer between 100 and 999. Optionally, a custom reason text can be
4571 provided defined by <str>, or the default reason for the specified code
4572 will be used as a fallback.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004573
4574 Example:
4575
4576 # return "431 Request Header Fields Too Large"
4577 http-response set-status 431
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004578 # return "503 Slow Down", custom reason
4579 http-response set-status 503 reason "Slow Down".
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004580
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004581 - "set-nice" sets the "nice" factor of the current request being processed.
4582 It only has effect against the other requests being processed at the same
4583 time. The default value is 0, unless altered by the "nice" setting on the
4584 "bind" line. The accepted range is -1024..1024. The higher the value, the
4585 nicest the request will be. Lower values will make the request more
4586 important than other ones. This can be useful to improve the speed of
4587 some requests, or lower the priority of non-important requests. Using
4588 this setting without prior experimentation can cause some major slowdown.
4589
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004590 - "set-log-level" is used to change the log level of the current request
4591 when a certain condition is met. Valid levels are the 8 syslog levels
4592 (see the "log" keyword) plus the special level "silent" which disables
4593 logging for this request. This rule is not final so the last matching
4594 rule wins. This rule can be useful to disable health checks coming from
4595 another equipment.
4596
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004597 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
4598 the client to the value passed in <tos> on platforms which support this.
4599 This value represents the whole 8 bits of the IP TOS field, and can be
4600 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
4601 that only the 6 higher bits are used in DSCP or TOS, and the two lower
Davor Ocelice9ed2812017-12-25 17:49:28 +01004602 bits are always 0. This can be used to adjust some routing behavior on
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004603 border routers based on some information from the request. See RFC 2474,
4604 2597, 3260 and 4594 for more information.
4605
Willy Tarreau51347ed2013-06-11 19:34:13 +02004606 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
4607 client to the value passed in <mark> on platforms which support it. This
4608 value is an unsigned 32 bit value which can be matched by netfilter and
4609 by the routing table. It can be expressed both in decimal or hexadecimal
4610 format (prefixed by "0x"). This can be useful to force certain packets to
4611 take a different route (for example a cheaper network path for bulk
4612 downloads). This works on Linux kernels 2.6.32 and above and requires
4613 admin privileges.
4614
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004615 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
4616 from a file (even a dummy empty file). The file name of the ACL to be
4617 updated is passed between parentheses. It takes one argument: <key fmt>,
4618 which follows log-format rules, to collect content of the new entry. It
4619 performs a lookup in the ACL before insertion, to avoid duplicated (or
4620 more) values. This lookup is done by a linear search and can be expensive
4621 with large lists! It is the equivalent of the "add acl" command from the
4622 stats socket, but can be triggered by an HTTP response.
4623
4624 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
4625 from a file (even a dummy empty file). The file name of the ACL to be
4626 updated is passed between parentheses. It takes one argument: <key fmt>,
4627 which follows log-format rules, to collect content of the entry to delete.
4628 It is the equivalent of the "del acl" command from the stats socket, but
4629 can be triggered by an HTTP response.
4630
4631 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
4632 from a file (even a dummy empty file). The file name of the MAP to be
4633 updated is passed between parentheses. It takes one argument: <key fmt>,
4634 which follows log-format rules, to collect content of the entry to delete.
4635 It takes one argument: "file name" It is the equivalent of the "del map"
4636 command from the stats socket, but can be triggered by an HTTP response.
4637
4638 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
4639 from a file (even a dummy empty file). The file name of the MAP to be
4640 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
4641 which follows log-format rules, used to collect MAP key, and <value fmt>,
4642 which follows log-format rules, used to collect content for the new entry.
4643 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4644 more) values. This lookup is done by a linear search and can be expensive
4645 with large lists! It is the equivalent of the "set map" command from the
4646 stats socket, but can be triggered by an HTTP response.
4647
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004648 - capture <sample> id <id> :
4649 captures sample expression <sample> from the response buffer, and converts
4650 it to a string. The resulting string is stored into the next request
4651 "capture" slot, so it will possibly appear next to some captured HTTP
4652 headers. It will then automatically appear in the logs, and it will be
4653 possible to extract it using sample fetch rules to feed it into headers or
4654 anything. Please check section 7.3 (Fetching samples) and "capture
4655 response header" for more information.
4656
4657 The keyword "id" is the id of the capture slot which is used for storing
4658 the string. The capture slot must be defined in an associated frontend.
4659 This is useful to run captures in backends. The slot id can be declared by
4660 a previous directive "http-response capture" or with the "declare capture"
4661 keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01004662 If the slot <id> doesn't exist, then HAProxy fails parsing the
4663 configuration to prevent unexpected behavior at run time.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004664
William Lallemand86d0df02017-11-24 21:36:45 +01004665 - cache-store <name> :
4666 See section 10.2 about cache setup.
4667
Willy Tarreau51d861a2015-05-22 17:30:48 +02004668 - "redirect" : this performs an HTTP redirection based on a redirect rule.
4669 This supports a format string similarly to "http-request redirect" rules,
4670 with the exception that only the "location" type of redirect is possible
4671 on the response. See the "redirect" keyword for the rule's syntax. When
4672 a redirect rule is applied during a response, connections to the server
4673 are closed so that no data can be forwarded from the server to the client.
4674
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004675 - set-var(<var-name>) expr:
4676 Is used to set the contents of a variable. The variable is declared
4677 inline.
4678
Daniel Schneller0b547052016-03-21 20:46:57 +01004679 <var-name> The name of the variable starts with an indication about
4680 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004681 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004682 "sess" : the variable is shared with the whole session
4683 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004684 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004685 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004686 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004687 "res" : the variable is shared only during response
4688 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004689 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01004690 The name may only contain characters 'a-z', 'A-Z', '0-9',
4691 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004692
4693 <expr> Is a standard HAProxy expression formed by a sample-fetch
4694 followed by some converters.
4695
4696 Example:
4697
4698 http-response set-var(sess.last_redir) res.hdr(location)
4699
Christopher Faulet85d79c92016-11-09 16:54:56 +01004700 - unset-var(<var-name>) :
4701 Is used to unset a variable. See above for details about <var-name>.
4702
4703 Example:
4704
4705 http-response unset-var(sess.last_redir)
4706
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004707 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4708 enables tracking of sticky counters from current response. Please refer to
4709 "http-request track-sc" for a complete description. The only difference
4710 from "http-request track-sc" is the <key> sample expression can only make
Davor Ocelice9ed2812017-12-25 17:49:28 +01004711 use of samples in response (e.g. res.*, status etc.) and samples below
4712 Layer 6 (e.g. SSL-related samples, see section 7.3.4). If the sample is
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004713 not supported, haproxy will fail and warn while parsing the config.
4714
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004715 - sc-set-gpt0(<sc-id>) <int> :
4716 This action sets the GPT0 tag according to the sticky counter designated
4717 by <sc-id> and the value of <int>. The expected result is a boolean. If
4718 an error occurs, this action silently fails and the actions evaluation
4719 continues.
4720
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004721 - sc-inc-gpc0(<sc-id>):
4722 This action increments the GPC0 counter according with the sticky counter
4723 designated by <sc-id>. If an error occurs, this action silently fails and
4724 the actions evaluation continues.
4725
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004726 - sc-inc-gpc1(<sc-id>):
4727 Same as "sc-inc-gpc0" action above but for GPC1 counter.
4728
Willy Tarreau2d392c22015-08-24 01:43:45 +02004729 - "silent-drop" : this stops the evaluation of the rules and makes the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004730 client-facing connection suddenly disappear using a system-dependent way
Willy Tarreau2d392c22015-08-24 01:43:45 +02004731 that tries to prevent the client from being notified. The effect it then
4732 that the client still sees an established connection while there's none
4733 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4734 except that it doesn't use any local resource at all on the machine
4735 running HAProxy. It can resist much higher loads than "tarpit", and slow
Davor Ocelice9ed2812017-12-25 17:49:28 +01004736 down stronger attackers. It is important to understand the impact of using
4737 this mechanism. All stateful equipment placed between the client and
Willy Tarreau2d392c22015-08-24 01:43:45 +02004738 HAProxy (firewalls, proxies, load balancers) will also keep the
4739 established connection for a long time and may suffer from this action.
4740 On modern Linux systems running with enough privileges, the TCP_REPAIR
4741 socket option is used to block the emission of a TCP reset. On other
4742 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4743 pass the first router, though it's still delivered to local networks. Do
4744 not use it unless you fully understand how it works.
4745
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004746 - send-spoe-group <engine-name> <group-name> :
4747 This action is used to trigger sending of a group of SPOE messages. To do
4748 so, the SPOE engine used to send messages must be defined, as well as the
4749 SPOE group to send. Of course, the SPOE engine must refer to an existing
4750 SPOE filter. If not engine name is provided on the SPOE filter line, the
4751 SPOE agent name must be used.
4752
4753 <engine-name> The SPOE engine name.
4754
4755 <group-name> The SPOE group name as specified in the engine
4756 configuration.
4757
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004758 There is no limit to the number of http-response statements per instance.
4759
Godbach09250262013-07-02 01:19:15 +08004760 It is important to know that http-response rules are processed very early in
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004761 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
Davor Ocelice9ed2812017-12-25 17:49:28 +01004762 headers added by "add-header"/"set-header" are visible by almost all further
4763 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004764
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004765 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4766 in newer versions (>= 1.5). But if you need to use regular expression to
4767 delete headers, you can still use "rspdel". Also please use
4768 "http-response deny" instead of "rspdeny".
4769
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004770 Example:
4771 acl key_acl res.hdr(X-Acl-Key) -m found
4772
4773 acl myhost hdr(Host) -f myhost.lst
4774
4775 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4776 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4777
4778 Example:
4779 acl value res.hdr(X-Value) -m found
4780
4781 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4782
4783 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4784 http-response del-map(map.lst) %[src] if ! value
4785
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004786 See also : "http-request", section 3.4 about userlists and section 7 about
4787 ACL usage.
4788
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004789
Willy Tarreau30631952015-08-06 15:05:24 +02004790http-reuse { never | safe | aggressive | always }
4791 Declare how idle HTTP connections may be shared between requests
4792
4793 May be used in sections: defaults | frontend | listen | backend
4794 yes | no | yes | yes
4795
4796 By default, a connection established between haproxy and the backend server
4797 belongs to the session that initiated it. The downside is that between the
4798 response and the next request, the connection remains idle and is not used.
4799 In many cases for performance reasons it is desirable to make it possible to
4800 reuse these idle connections to serve other requests from different sessions.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004801 This directive allows to tune this behavior.
Willy Tarreau30631952015-08-06 15:05:24 +02004802
4803 The argument indicates the desired connection reuse strategy :
4804
4805 - "never" : idle connections are never shared between sessions. This is
4806 the default choice. It may be enforced to cancel a different
4807 strategy inherited from a defaults section or for
4808 troubleshooting. For example, if an old bogus application
4809 considers that multiple requests over the same connection come
4810 from the same client and it is not possible to fix the
4811 application, it may be desirable to disable connection sharing
4812 in a single backend. An example of such an application could
4813 be an old haproxy using cookie insertion in tunnel mode and
4814 not checking any request past the first one.
4815
4816 - "safe" : this is the recommended strategy. The first request of a
4817 session is always sent over its own connection, and only
4818 subsequent requests may be dispatched over other existing
4819 connections. This ensures that in case the server closes the
4820 connection when the request is being sent, the browser can
4821 decide to silently retry it. Since it is exactly equivalent to
4822 regular keep-alive, there should be no side effects.
4823
4824 - "aggressive" : this mode may be useful in webservices environments where
4825 all servers are not necessarily known and where it would be
4826 appreciable to deliver most first requests over existing
4827 connections. In this case, first requests are only delivered
4828 over existing connections that have been reused at least once,
4829 proving that the server correctly supports connection reuse.
4830 It should only be used when it's sure that the client can
4831 retry a failed request once in a while and where the benefit
4832 of aggressive connection reuse significantly outweights the
4833 downsides of rare connection failures.
4834
4835 - "always" : this mode is only recommended when the path to the server is
4836 known for never breaking existing connections quickly after
4837 releasing them. It allows the first request of a session to be
4838 sent to an existing connection. This can provide a significant
4839 performance increase over the "safe" strategy when the backend
4840 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004841 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004842 sharing. It is recommended that the "http-keep-alive" timeout
4843 remains low in this mode so that no dead connections remain
4844 usable. In most cases, this will lead to the same performance
4845 gains as "aggressive" but with more risks. It should only be
4846 used when it improves the situation over "aggressive".
4847
4848 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004849 connection properties and compatibility. Specifically :
4850 - connections made with "usesrc" followed by a client-dependent value
4851 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004852
4853 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004854 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004855
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02004856 - connections with certain bogus authentication schemes (relying on the
4857 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004858
4859 No connection pool is involved, once a session dies, the last idle connection
4860 it was attached to is deleted at the same time. This ensures that connections
4861 may not last after all sessions are closed.
4862
4863 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4864 because almost no new connection will be established while idle connections
4865 remain available. This is particularly true with the "always" strategy.
4866
4867 See also : "option http-keep-alive", "server maxconn"
4868
4869
Mark Lamourinec2247f02012-01-04 13:02:01 -05004870http-send-name-header [<header>]
4871 Add the server name to a request. Use the header string given by <header>
4872
4873 May be used in sections: defaults | frontend | listen | backend
4874 yes | no | yes | yes
4875
4876 Arguments :
4877
4878 <header> The header string to use to send the server name
4879
4880 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01004881 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05004882 is added with the header string proved.
4883
4884 See also : "server"
4885
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004886id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02004887 Set a persistent ID to a proxy.
4888 May be used in sections : defaults | frontend | listen | backend
4889 no | yes | yes | yes
4890 Arguments : none
4891
4892 Set a persistent ID for the proxy. This ID must be unique and positive.
4893 An unused ID will automatically be assigned if unset. The first assigned
4894 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004895
4896
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004897ignore-persist { if | unless } <condition>
4898 Declare a condition to ignore persistence
4899 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004900 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004901
4902 By default, when cookie persistence is enabled, every requests containing
4903 the cookie are unconditionally persistent (assuming the target server is up
4904 and running).
4905
4906 The "ignore-persist" statement allows one to declare various ACL-based
4907 conditions which, when met, will cause a request to ignore persistence.
4908 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004909 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004910 persistence for a specific User-Agent (for example, some web crawler bots).
4911
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004912 The persistence is ignored when an "if" condition is met, or unless an
4913 "unless" condition is met.
4914
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004915 Example:
4916 acl url_static path_beg /static /images /img /css
4917 acl url_static path_end .gif .png .jpg .css .js
4918 ignore-persist if url_static
4919
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004920 See also : "force-persist", "cookie", and section 7 about ACL usage.
4921
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004922load-server-state-from-file { global | local | none }
4923 Allow seamless reload of HAProxy
4924 May be used in sections: defaults | frontend | listen | backend
4925 yes | no | yes | yes
4926
4927 This directive points HAProxy to a file where server state from previous
4928 running process has been saved. That way, when starting up, before handling
4929 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01004930 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004931 to tell haproxy which file to use. For now, only 2 arguments to either prevent
4932 loading state or load states from a file containing all backends and servers.
4933 The state file can be generated by running the command "show servers state"
4934 over the stats socket and redirect output.
4935
Davor Ocelice9ed2812017-12-25 17:49:28 +01004936 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004937 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02004938 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004939
4940 Arguments:
4941 global load the content of the file pointed by the global directive
4942 named "server-state-file".
4943
4944 local load the content of the file pointed by the directive
4945 "server-state-file-name" if set. If not set, then the backend
4946 name is used as a file name.
4947
4948 none don't load any stat for this backend
4949
4950 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01004951 - server's IP address is preserved across reloads by default, but the
4952 order can be changed thanks to the server's "init-addr" setting. This
4953 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01004954 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01004955 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004956
4957 - server's weight is applied from previous running process unless it has
4958 has changed between previous and new configuration files.
4959
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004960 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004961
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004962 global
4963 stats socket /tmp/socket
4964 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004965
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004966 defaults
4967 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004968
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004969 backend bk
4970 server s1 127.0.0.1:22 check weight 11
4971 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004972
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004973
4974 Then one can run :
4975
4976 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
4977
4978 Content of the file /tmp/server_state would be like this:
4979
4980 1
4981 # <field names skipped for the doc example>
4982 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
4983 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
4984
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004985 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004986
4987 global
4988 stats socket /tmp/socket
4989 server-state-base /etc/haproxy/states
4990
4991 defaults
4992 load-server-state-from-file local
4993
4994 backend bk
4995 server s1 127.0.0.1:22 check weight 11
4996 server s2 127.0.0.1:22 check weight 12
4997
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004998
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004999 Then one can run :
5000
5001 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5002
5003 Content of the file /etc/haproxy/states/bk would be like this:
5004
5005 1
5006 # <field names skipped for the doc example>
5007 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5008 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5009
5010 See also: "server-state-file", "server-state-file-name", and
5011 "show servers state"
5012
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005013
Willy Tarreau2769aa02007-12-27 18:26:09 +01005014log global
Willy Tarreau18324f52014-06-27 18:10:07 +02005015log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005016no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005017 Enable per-instance logging of events and traffic.
5018 May be used in sections : defaults | frontend | listen | backend
5019 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005020
5021 Prefix :
5022 no should be used when the logger list must be flushed. For example,
5023 if you don't want to inherit from the default logger list. This
5024 prefix does not allow arguments.
5025
Willy Tarreau2769aa02007-12-27 18:26:09 +01005026 Arguments :
5027 global should be used when the instance's logging parameters are the
5028 same as the global ones. This is the most common usage. "global"
5029 replaces <address>, <facility> and <level> with those of the log
5030 entries found in the "global" section. Only one "log global"
5031 statement may be used per instance, and this form takes no other
5032 parameter.
5033
5034 <address> indicates where to send the logs. It takes the same format as
5035 for the "global" section's logs, and can be one of :
5036
5037 - An IPv4 address optionally followed by a colon (':') and a UDP
5038 port. If no port is specified, 514 is used by default (the
5039 standard syslog port).
5040
David du Colombier24bb5f52011-03-17 10:40:23 +01005041 - An IPv6 address followed by a colon (':') and optionally a UDP
5042 port. If no port is specified, 514 is used by default (the
5043 standard syslog port).
5044
Willy Tarreau2769aa02007-12-27 18:26:09 +01005045 - A filesystem path to a UNIX domain socket, keeping in mind
5046 considerations for chroot (be sure the path is accessible
5047 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005048 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005049
William Lallemandb2f07452015-05-12 14:27:13 +02005050 You may want to reference some environment variables in the
5051 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005052
Willy Tarreau18324f52014-06-27 18:10:07 +02005053 <length> is an optional maximum line length. Log lines larger than this
5054 value will be truncated before being sent. The reason is that
5055 syslog servers act differently on log line length. All servers
5056 support the default value of 1024, but some servers simply drop
5057 larger lines while others do log them. If a server supports long
5058 lines, it may make sense to set this value here in order to avoid
5059 truncating long lines. Similarly, if a server drops long lines,
5060 it is preferable to truncate them before sending them. Accepted
5061 values are 80 to 65535 inclusive. The default value of 1024 is
5062 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005063 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005064
Willy Tarreau2769aa02007-12-27 18:26:09 +01005065 <facility> must be one of the 24 standard syslog facilities :
5066
5067 kern user mail daemon auth syslog lpr news
5068 uucp cron auth2 ftp ntp audit alert cron2
5069 local0 local1 local2 local3 local4 local5 local6 local7
5070
5071 <level> is optional and can be specified to filter outgoing messages. By
5072 default, all messages are sent. If a level is specified, only
5073 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005074 will be sent. An optional minimum level can be specified. If it
5075 is set, logs emitted with a more severe level than this one will
5076 be capped to this level. This is used to avoid sending "emerg"
5077 messages on all terminals on some default syslog configurations.
5078 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005079
5080 emerg alert crit err warning notice info debug
5081
William Lallemand0f99e342011-10-12 17:50:54 +02005082 It is important to keep in mind that it is the frontend which decides what to
5083 log from a connection, and that in case of content switching, the log entries
5084 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005085
5086 However, backend log declaration define how and where servers status changes
5087 will be logged. Level "notice" will be used to indicate a server going up,
5088 "warning" will be used for termination signals and definitive service
5089 termination, and "alert" will be used for when a server goes down.
5090
5091 Note : According to RFC3164, messages are truncated to 1024 bytes before
5092 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005093
5094 Example :
5095 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005096 log 127.0.0.1:514 local0 notice # only send important events
5097 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005098 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005099
Willy Tarreau2769aa02007-12-27 18:26:09 +01005100
William Lallemand48940402012-01-30 16:47:22 +01005101log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005102 Specifies the log format string to use for traffic logs
5103 May be used in sections: defaults | frontend | listen | backend
5104 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005105
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005106 This directive specifies the log format string that will be used for all logs
5107 resulting from traffic passing through the frontend using this line. If the
5108 directive is used in a defaults section, all subsequent frontends will use
5109 the same log format. Please see section 8.2.4 which covers the log format
5110 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005111
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005112 "log-format" directive overrides previous "option tcplog", "log-format" and
5113 "option httplog" directives.
5114
Dragan Dosen7ad31542015-09-28 17:16:47 +02005115log-format-sd <string>
5116 Specifies the RFC5424 structured-data log format string
5117 May be used in sections: defaults | frontend | listen | backend
5118 yes | yes | yes | no
5119
5120 This directive specifies the RFC5424 structured-data log format string that
5121 will be used for all logs resulting from traffic passing through the frontend
5122 using this line. If the directive is used in a defaults section, all
5123 subsequent frontends will use the same log format. Please see section 8.2.4
5124 which covers the log format string in depth.
5125
5126 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5127 about the RFC5424 structured-data part.
5128
5129 Note : This log format string will be used only for loggers that have set
5130 log format to "rfc5424".
5131
5132 Example :
5133 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5134
5135
Willy Tarreau094af4e2015-01-07 15:03:42 +01005136log-tag <string>
5137 Specifies the log tag to use for all outgoing logs
5138 May be used in sections: defaults | frontend | listen | backend
5139 yes | yes | yes | yes
5140
5141 Sets the tag field in the syslog header to this string. It defaults to the
5142 log-tag set in the global section, otherwise the program name as launched
5143 from the command line, which usually is "haproxy". Sometimes it can be useful
5144 to differentiate between multiple processes running on the same host, or to
5145 differentiate customer instances running in the same process. In the backend,
5146 logs about servers up/down will use this tag. As a hint, it can be convenient
5147 to set a log-tag related to a hosted customer in a defaults section then put
5148 all the frontends and backends for that customer, then start another customer
5149 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005150
Willy Tarreauc35362a2014-04-25 13:58:37 +02005151max-keep-alive-queue <value>
5152 Set the maximum server queue size for maintaining keep-alive connections
5153 May be used in sections: defaults | frontend | listen | backend
5154 yes | no | yes | yes
5155
5156 HTTP keep-alive tries to reuse the same server connection whenever possible,
5157 but sometimes it can be counter-productive, for example if a server has a lot
5158 of connections while other ones are idle. This is especially true for static
5159 servers.
5160
5161 The purpose of this setting is to set a threshold on the number of queued
5162 connections at which haproxy stops trying to reuse the same server and prefers
5163 to find another one. The default value, -1, means there is no limit. A value
5164 of zero means that keep-alive requests will never be queued. For very close
5165 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005166 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005167 use a value of 10 or less). For remote servers suffering from a high latency,
5168 higher values might be needed to cover for the latency and/or the cost of
5169 picking a different server.
5170
5171 Note that this has no impact on responses which are maintained to the same
5172 server consecutively to a 401 response. They will still go to the same server
5173 even if they have to be queued.
5174
5175 See also : "option http-server-close", "option prefer-last-server", server
5176 "maxconn" and cookie persistence.
5177
5178
Willy Tarreau2769aa02007-12-27 18:26:09 +01005179maxconn <conns>
5180 Fix the maximum number of concurrent connections on a frontend
5181 May be used in sections : defaults | frontend | listen | backend
5182 yes | yes | yes | no
5183 Arguments :
5184 <conns> is the maximum number of concurrent connections the frontend will
5185 accept to serve. Excess connections will be queued by the system
5186 in the socket's listen queue and will be served once a connection
5187 closes.
5188
5189 If the system supports it, it can be useful on big sites to raise this limit
5190 very high so that haproxy manages connection queues, instead of leaving the
5191 clients with unanswered connection attempts. This value should not exceed the
5192 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005193 of tune.bufsize (16kB by default) each, as well as some other data resulting
5194 in about 33 kB of RAM being consumed per established connection. That means
5195 that a medium system equipped with 1GB of RAM can withstand around
5196 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005197
5198 Also, when <conns> is set to large values, it is possible that the servers
5199 are not sized to accept such loads, and for this reason it is generally wise
5200 to assign them some reasonable connection limits.
5201
Vincent Bernat6341be52012-06-27 17:18:30 +02005202 By default, this value is set to 2000.
5203
Willy Tarreau2769aa02007-12-27 18:26:09 +01005204 See also : "server", global section's "maxconn", "fullconn"
5205
5206
5207mode { tcp|http|health }
5208 Set the running mode or protocol of the instance
5209 May be used in sections : defaults | frontend | listen | backend
5210 yes | yes | yes | yes
5211 Arguments :
5212 tcp The instance will work in pure TCP mode. A full-duplex connection
5213 will be established between clients and servers, and no layer 7
5214 examination will be performed. This is the default mode. It
5215 should be used for SSL, SSH, SMTP, ...
5216
5217 http The instance will work in HTTP mode. The client request will be
5218 analyzed in depth before connecting to any server. Any request
5219 which is not RFC-compliant will be rejected. Layer 7 filtering,
5220 processing and switching will be possible. This is the mode which
5221 brings HAProxy most of its value.
5222
5223 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005224 to incoming connections and close the connection. Alternatively,
5225 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5226 instead. Nothing will be logged in either case. This mode is used
5227 to reply to external components health checks. This mode is
5228 deprecated and should not be used anymore as it is possible to do
5229 the same and even better by combining TCP or HTTP modes with the
5230 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005231
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005232 When doing content switching, it is mandatory that the frontend and the
5233 backend are in the same mode (generally HTTP), otherwise the configuration
5234 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005235
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005236 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005237 defaults http_instances
5238 mode http
5239
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005240 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005241
Willy Tarreau0ba27502007-12-24 16:55:16 +01005242
Cyril Bontéf0c60612010-02-06 14:44:47 +01005243monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005244 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005245 May be used in sections : defaults | frontend | listen | backend
5246 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005247 Arguments :
5248 if <cond> the monitor request will fail if the condition is satisfied,
5249 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005250 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005251 are met, for instance a low number of servers both in a
5252 backend and its backup.
5253
5254 unless <cond> the monitor request will succeed only if the condition is
5255 satisfied, and will fail otherwise. Such a condition may be
5256 based on a test on the presence of a minimum number of active
5257 servers in a list of backends.
5258
5259 This statement adds a condition which can force the response to a monitor
5260 request to report a failure. By default, when an external component queries
5261 the URI dedicated to monitoring, a 200 response is returned. When one of the
5262 conditions above is met, haproxy will return 503 instead of 200. This is
5263 very useful to report a site failure to an external component which may base
5264 routing advertisements between multiple sites on the availability reported by
5265 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005266 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5267 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005268
5269 Example:
5270 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005271 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005272 acl site_dead nbsrv(dynamic) lt 2
5273 acl site_dead nbsrv(static) lt 2
5274 monitor-uri /site_alive
5275 monitor fail if site_dead
5276
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005277 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005278
5279
5280monitor-net <source>
5281 Declare a source network which is limited to monitor requests
5282 May be used in sections : defaults | frontend | listen | backend
5283 yes | yes | yes | no
5284 Arguments :
5285 <source> is the source IPv4 address or network which will only be able to
5286 get monitor responses to any request. It can be either an IPv4
5287 address, a host name, or an address followed by a slash ('/')
5288 followed by a mask.
5289
5290 In TCP mode, any connection coming from a source matching <source> will cause
5291 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005292 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005293 forwarding the connection to a remote server.
5294
5295 In HTTP mode, a connection coming from a source matching <source> will be
5296 accepted, the following response will be sent without waiting for a request,
5297 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5298 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005299 running without forwarding the request to a backend server. Note that this
5300 response is sent in raw format, without any transformation. This is important
5301 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005302
Willy Tarreau82569f92012-09-27 23:48:56 +02005303 Monitor requests are processed very early, just after tcp-request connection
5304 ACLs which are the only ones able to block them. These connections are short
5305 lived and never wait for any data from the client. They cannot be logged, and
5306 it is the intended purpose. They are only used to report HAProxy's health to
5307 an upper component, nothing more. Please note that "monitor fail" rules do
5308 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005309
Willy Tarreau95cd2832010-03-04 23:36:33 +01005310 Last, please note that only one "monitor-net" statement can be specified in
5311 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005312
Willy Tarreau2769aa02007-12-27 18:26:09 +01005313 Example :
5314 # addresses .252 and .253 are just probing us.
5315 frontend www
5316 monitor-net 192.168.0.252/31
5317
5318 See also : "monitor fail", "monitor-uri"
5319
5320
5321monitor-uri <uri>
5322 Intercept a URI used by external components' monitor requests
5323 May be used in sections : defaults | frontend | listen | backend
5324 yes | yes | yes | no
5325 Arguments :
5326 <uri> is the exact URI which we want to intercept to return HAProxy's
5327 health status instead of forwarding the request.
5328
5329 When an HTTP request referencing <uri> will be received on a frontend,
5330 HAProxy will not forward it nor log it, but instead will return either
5331 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5332 conditions defined with "monitor fail". This is normally enough for any
5333 front-end HTTP probe to detect that the service is UP and running without
5334 forwarding the request to a backend server. Note that the HTTP method, the
5335 version and all headers are ignored, but the request must at least be valid
5336 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5337
Willy Tarreau721d8e02017-12-01 18:25:08 +01005338 Monitor requests are processed very early, just after the request is parsed
5339 and even before any "http-request" or "block" rulesets. The only rulesets
5340 applied before are the tcp-request ones. They cannot be logged either, and it
5341 is the intended purpose. They are only used to report HAProxy's health to an
5342 upper component, nothing more. However, it is possible to add any number of
5343 conditions using "monitor fail" and ACLs so that the result can be adjusted
5344 to whatever check can be imagined (most often the number of available servers
5345 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005346
5347 Example :
5348 # Use /haproxy_test to report haproxy's status
5349 frontend www
5350 mode http
5351 monitor-uri /haproxy_test
5352
5353 See also : "monitor fail", "monitor-net"
5354
Willy Tarreau0ba27502007-12-24 16:55:16 +01005355
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005356option abortonclose
5357no option abortonclose
5358 Enable or disable early dropping of aborted requests pending in queues.
5359 May be used in sections : defaults | frontend | listen | backend
5360 yes | no | yes | yes
5361 Arguments : none
5362
5363 In presence of very high loads, the servers will take some time to respond.
5364 The per-instance connection queue will inflate, and the response time will
5365 increase respective to the size of the queue times the average per-session
5366 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005367 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005368 the queue, and slowing down other users, and the servers as well, because the
5369 request will eventually be served, then aborted at the first error
5370 encountered while delivering the response.
5371
5372 As there is no way to distinguish between a full STOP and a simple output
5373 close on the client side, HTTP agents should be conservative and consider
5374 that the client might only have closed its output channel while waiting for
5375 the response. However, this introduces risks of congestion when lots of users
5376 do the same, and is completely useless nowadays because probably no client at
5377 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005378 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005379 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005380 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005381 of being the single component to break rare but valid traffic is extremely
5382 low, which adds to the temptation to be able to abort a session early while
5383 still not served and not pollute the servers.
5384
Davor Ocelice9ed2812017-12-25 17:49:28 +01005385 In HAProxy, the user can choose the desired behavior using the option
5386 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005387 compliant and aborted requests will be served. But when the option is
5388 specified, a session with an incoming channel closed will be aborted while
5389 it is still possible, either pending in the queue for a connection slot, or
5390 during the connection establishment if the server has not yet acknowledged
5391 the connection request. This considerably reduces the queue size and the load
5392 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005393 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005394
5395 If this option has been enabled in a "defaults" section, it can be disabled
5396 in a specific instance by prepending the "no" keyword before it.
5397
5398 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5399
5400
Willy Tarreau4076a152009-04-02 15:18:36 +02005401option accept-invalid-http-request
5402no option accept-invalid-http-request
5403 Enable or disable relaxing of HTTP request parsing
5404 May be used in sections : defaults | frontend | listen | backend
5405 yes | yes | yes | no
5406 Arguments : none
5407
Willy Tarreau91852eb2015-05-01 13:26:00 +02005408 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005409 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005410 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005411 forbidden characters are essentially used to build attacks exploiting server
5412 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5413 server will emit invalid header names for whatever reason (configuration,
5414 implementation) and the issue will not be immediately fixed. In such a case,
5415 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005416 even if that does not make sense, by specifying this option. Similarly, the
5417 list of characters allowed to appear in a URI is well defined by RFC3986, and
5418 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5419 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005420 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005421 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005422 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5423 to pass through (no version specified) and multiple digits for both the major
5424 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005425
5426 This option should never be enabled by default as it hides application bugs
5427 and open security breaches. It should only be deployed after a problem has
5428 been confirmed.
5429
5430 When this option is enabled, erroneous header names will still be accepted in
5431 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005432 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5433 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005434 also helps confirming that the issue has been solved.
5435
5436 If this option has been enabled in a "defaults" section, it can be disabled
5437 in a specific instance by prepending the "no" keyword before it.
5438
5439 See also : "option accept-invalid-http-response" and "show errors" on the
5440 stats socket.
5441
5442
5443option accept-invalid-http-response
5444no option accept-invalid-http-response
5445 Enable or disable relaxing of HTTP response parsing
5446 May be used in sections : defaults | frontend | listen | backend
5447 yes | no | yes | yes
5448 Arguments : none
5449
Willy Tarreau91852eb2015-05-01 13:26:00 +02005450 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005451 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005452 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005453 forbidden characters are essentially used to build attacks exploiting server
5454 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5455 server will emit invalid header names for whatever reason (configuration,
5456 implementation) and the issue will not be immediately fixed. In such a case,
5457 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005458 even if that does not make sense, by specifying this option. This option also
5459 relaxes the test on the HTTP version format, it allows multiple digits for
5460 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005461
5462 This option should never be enabled by default as it hides application bugs
5463 and open security breaches. It should only be deployed after a problem has
5464 been confirmed.
5465
5466 When this option is enabled, erroneous header names will still be accepted in
5467 responses, but the complete response will be captured in order to permit
5468 later analysis using the "show errors" request on the UNIX stats socket.
5469 Doing this also helps confirming that the issue has been solved.
5470
5471 If this option has been enabled in a "defaults" section, it can be disabled
5472 in a specific instance by prepending the "no" keyword before it.
5473
5474 See also : "option accept-invalid-http-request" and "show errors" on the
5475 stats socket.
5476
5477
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005478option allbackups
5479no option allbackups
5480 Use either all backup servers at a time or only the first one
5481 May be used in sections : defaults | frontend | listen | backend
5482 yes | no | yes | yes
5483 Arguments : none
5484
5485 By default, the first operational backup server gets all traffic when normal
5486 servers are all down. Sometimes, it may be preferred to use multiple backups
5487 at once, because one will not be enough. When "option allbackups" is enabled,
5488 the load balancing will be performed among all backup servers when all normal
5489 ones are unavailable. The same load balancing algorithm will be used and the
5490 servers' weights will be respected. Thus, there will not be any priority
5491 order between the backup servers anymore.
5492
5493 This option is mostly used with static server farms dedicated to return a
5494 "sorry" page when an application is completely offline.
5495
5496 If this option has been enabled in a "defaults" section, it can be disabled
5497 in a specific instance by prepending the "no" keyword before it.
5498
5499
5500option checkcache
5501no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005502 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005503 May be used in sections : defaults | frontend | listen | backend
5504 yes | no | yes | yes
5505 Arguments : none
5506
5507 Some high-level frameworks set application cookies everywhere and do not
5508 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005509 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005510 high risk of session crossing or stealing between users traversing the same
5511 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005512 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005513
5514 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005515 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005516 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005517 response to check if there's a risk of caching a cookie on a client-side
5518 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005519 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005520 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005521 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5522 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005523 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005524 - all those that result from a request using a method other than GET, HEAD,
5525 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005526 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005527 - those with a 'Pragma: no-cache' header
5528 - those with a 'Cache-control: private' header
5529 - those with a 'Cache-control: no-store' header
5530 - those with a 'Cache-control: max-age=0' header
5531 - those with a 'Cache-control: s-maxage=0' header
5532 - those with a 'Cache-control: no-cache' header
5533 - those with a 'Cache-control: no-cache="set-cookie"' header
5534 - those with a 'Cache-control: no-cache="set-cookie,' header
5535 (allowing other fields after set-cookie)
5536
5537 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005538 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005539 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005540 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005541 that admins are informed that there's something to be fixed.
5542
5543 Due to the high impact on the application, the application should be tested
5544 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005545 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005546 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005547
5548 If this option has been enabled in a "defaults" section, it can be disabled
5549 in a specific instance by prepending the "no" keyword before it.
5550
5551
5552option clitcpka
5553no option clitcpka
5554 Enable or disable the sending of TCP keepalive packets on the client side
5555 May be used in sections : defaults | frontend | listen | backend
5556 yes | yes | yes | no
5557 Arguments : none
5558
5559 When there is a firewall or any session-aware component between a client and
5560 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005561 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005562 components decides to expire a session which has remained idle for too long.
5563
5564 Enabling socket-level TCP keep-alives makes the system regularly send packets
5565 to the other end of the connection, leaving it active. The delay between
5566 keep-alive probes is controlled by the system only and depends both on the
5567 operating system and its tuning parameters.
5568
5569 It is important to understand that keep-alive packets are neither emitted nor
5570 received at the application level. It is only the network stacks which sees
5571 them. For this reason, even if one side of the proxy already uses keep-alives
5572 to maintain its connection alive, those keep-alive packets will not be
5573 forwarded to the other side of the proxy.
5574
5575 Please note that this has nothing to do with HTTP keep-alive.
5576
5577 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5578 client side of a connection, which should help when session expirations are
5579 noticed between HAProxy and a client.
5580
5581 If this option has been enabled in a "defaults" section, it can be disabled
5582 in a specific instance by prepending the "no" keyword before it.
5583
5584 See also : "option srvtcpka", "option tcpka"
5585
5586
Willy Tarreau0ba27502007-12-24 16:55:16 +01005587option contstats
5588 Enable continuous traffic statistics updates
5589 May be used in sections : defaults | frontend | listen | backend
5590 yes | yes | yes | no
5591 Arguments : none
5592
5593 By default, counters used for statistics calculation are incremented
5594 only when a session finishes. It works quite well when serving small
5595 objects, but with big ones (for example large images or archives) or
5596 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005597 a hedgehog. With this option enabled counters get incremented frequently
5598 along the session, typically every 5 seconds, which is often enough to
5599 produce clean graphs. Recounting touches a hotpath directly so it is not
5600 not enabled by default, as it can cause a lot of wakeups for very large
5601 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005602
5603
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005604option dontlog-normal
5605no option dontlog-normal
5606 Enable or disable logging of normal, successful connections
5607 May be used in sections : defaults | frontend | listen | backend
5608 yes | yes | yes | no
5609 Arguments : none
5610
5611 There are large sites dealing with several thousand connections per second
5612 and for which logging is a major pain. Some of them are even forced to turn
5613 logs off and cannot debug production issues. Setting this option ensures that
5614 normal connections, those which experience no error, no timeout, no retry nor
5615 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5616 mode, the response status code is checked and return codes 5xx will still be
5617 logged.
5618
5619 It is strongly discouraged to use this option as most of the time, the key to
5620 complex issues is in the normal logs which will not be logged here. If you
5621 need to separate logs, see the "log-separate-errors" option instead.
5622
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005623 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005624 logging.
5625
5626
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005627option dontlognull
5628no option dontlognull
5629 Enable or disable logging of null connections
5630 May be used in sections : defaults | frontend | listen | backend
5631 yes | yes | yes | no
5632 Arguments : none
5633
5634 In certain environments, there are components which will regularly connect to
5635 various systems to ensure that they are still alive. It can be the case from
5636 another load balancer as well as from monitoring systems. By default, even a
5637 simple port probe or scan will produce a log. If those connections pollute
5638 the logs too much, it is possible to enable option "dontlognull" to indicate
5639 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005640 which typically corresponds to those probes. Note that errors will still be
5641 returned to the client and accounted for in the stats. If this is not what is
5642 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005643
5644 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005645 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005646 would not be logged.
5647
5648 If this option has been enabled in a "defaults" section, it can be disabled
5649 in a specific instance by prepending the "no" keyword before it.
5650
Willy Tarreau0f228a02015-05-01 15:37:53 +02005651 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5652 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005653
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005654
Christopher Faulet315b39c2018-09-21 16:26:19 +02005655option forceclose (deprecated)
5656no option forceclose (deprecated)
5657 This is an alias for "option httpclose". Thus this option is deprecated.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005658
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005659 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005660
5661
Willy Tarreau87cf5142011-08-19 22:57:24 +02005662option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005663 Enable insertion of the X-Forwarded-For header to requests sent to servers
5664 May be used in sections : defaults | frontend | listen | backend
5665 yes | yes | yes | yes
5666 Arguments :
5667 <network> is an optional argument used to disable this option for sources
5668 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005669 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005670 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005671
5672 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5673 their client address. This is sometimes annoying when the client's IP address
5674 is expected in server logs. To solve this problem, the well-known HTTP header
5675 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5676 This header contains a value representing the client's IP address. Since this
5677 header is always appended at the end of the existing header list, the server
5678 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005679 the server's manual to find how to enable use of this standard header. Note
5680 that only the last occurrence of the header must be used, since it is really
5681 possible that the client has already brought one.
5682
Willy Tarreaud72758d2010-01-12 10:42:19 +01005683 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005684 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005685 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005686 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005687 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005688 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005689
5690 Sometimes, a same HAProxy instance may be shared between a direct client
5691 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5692 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5693 header for a known source address or network by adding the "except" keyword
5694 followed by the network address. In this case, any source IP matching the
5695 network will not cause an addition of this header. Most common uses are with
5696 private networks or 127.0.0.1.
5697
Willy Tarreau87cf5142011-08-19 22:57:24 +02005698 Alternatively, the keyword "if-none" states that the header will only be
5699 added if it is not present. This should only be used in perfectly trusted
5700 environment, as this might cause a security issue if headers reaching haproxy
5701 are under the control of the end-user.
5702
Willy Tarreauc27debf2008-01-06 08:57:02 +01005703 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005704 least one of them uses it, the header will be added. Note that the backend's
5705 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005706 both are defined. In the case of the "if-none" argument, if at least one of
5707 the frontend or the backend does not specify it, it wants the addition to be
5708 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005709
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005710 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005711 # Public HTTP address also used by stunnel on the same machine
5712 frontend www
5713 mode http
5714 option forwardfor except 127.0.0.1 # stunnel already adds the header
5715
Ross Westaf72a1d2008-08-03 10:51:45 +02005716 # Those servers want the IP Address in X-Client
5717 backend www
5718 mode http
5719 option forwardfor header X-Client
5720
Willy Tarreau87cf5142011-08-19 22:57:24 +02005721 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005722 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005723
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005724
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005725option http-buffer-request
5726no option http-buffer-request
5727 Enable or disable waiting for whole HTTP request body before proceeding
5728 May be used in sections : defaults | frontend | listen | backend
5729 yes | yes | yes | yes
5730 Arguments : none
5731
5732 It is sometimes desirable to wait for the body of an HTTP request before
5733 taking a decision. This is what is being done by "balance url_param" for
5734 example. The first use case is to buffer requests from slow clients before
5735 connecting to the server. Another use case consists in taking the routing
5736 decision based on the request body's contents. This option placed in a
5737 frontend or backend forces the HTTP processing to wait until either the whole
5738 body is received, or the request buffer is full, or the first chunk is
5739 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005740 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005741 the frontend and the backend, so this should definitely not be used by
5742 default.
5743
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005744 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005745
5746
Willy Tarreau0f228a02015-05-01 15:37:53 +02005747option http-ignore-probes
5748no option http-ignore-probes
5749 Enable or disable logging of null connections and request timeouts
5750 May be used in sections : defaults | frontend | listen | backend
5751 yes | yes | yes | no
5752 Arguments : none
5753
5754 Recently some browsers started to implement a "pre-connect" feature
5755 consisting in speculatively connecting to some recently visited web sites
5756 just in case the user would like to visit them. This results in many
5757 connections being established to web sites, which end up in 408 Request
5758 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5759 decides to close them first. These ones pollute the log and feed the error
5760 counters. There was already "option dontlognull" but it's insufficient in
5761 this case. Instead, this option does the following things :
5762 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005763 was received over a connection before it was closed;
5764 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005765 - prevent any error counter from being incremented
5766
5767 That way the empty connection is silently ignored. Note that it is better
5768 not to use this unless it is clear that it is needed, because it will hide
5769 real problems. The most common reason for not receiving a request and seeing
5770 a 408 is due to an MTU inconsistency between the client and an intermediary
5771 element such as a VPN, which blocks too large packets. These issues are
5772 generally seen with POST requests as well as GET with large cookies. The logs
5773 are often the only way to detect them.
5774
5775 If this option has been enabled in a "defaults" section, it can be disabled
5776 in a specific instance by prepending the "no" keyword before it.
5777
5778 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5779
5780
Willy Tarreau16bfb022010-01-16 19:48:41 +01005781option http-keep-alive
5782no option http-keep-alive
5783 Enable or disable HTTP keep-alive from client to server
5784 May be used in sections : defaults | frontend | listen | backend
5785 yes | yes | yes | yes
5786 Arguments : none
5787
Willy Tarreau70dffda2014-01-30 03:07:23 +01005788 By default HAProxy operates in keep-alive mode with regards to persistent
5789 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02005790 leaves the connection idle on both sides between the end of a response and
5791 the start of a new request. This mode may be changed by several options such
5792 as "option http-server-close", "option httpclose" or "option http-tunnel".
5793 This option allows to set back the keep-alive mode, which can be useful when
5794 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01005795
5796 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5797 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005798 network) and the fastest session reuse on the server side at the expense
5799 of maintaining idle connections to the servers. In general, it is possible
5800 with this option to achieve approximately twice the request rate that the
5801 "http-server-close" option achieves on small objects. There are mainly two
5802 situations where this option may be useful :
5803
5804 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005805 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005806
5807 - when the cost of establishing the connection to the server is significant
5808 compared to the cost of retrieving the associated object from the server.
5809
5810 This last case can happen when the server is a fast static server of cache.
5811 In this case, the server will need to be properly tuned to support high enough
5812 connection counts because connections will last until the client sends another
5813 request.
5814
5815 If the client request has to go to another backend or another server due to
5816 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01005817 immediately be closed and a new one re-opened. Option "prefer-last-server" is
5818 available to try optimize server selection so that if the server currently
5819 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01005820
Willy Tarreau16bfb022010-01-16 19:48:41 +01005821 At the moment, logs will not indicate whether requests came from the same
5822 session or not. The accept date reported in the logs corresponds to the end
5823 of the previous request, and the request time corresponds to the time spent
5824 waiting for a new request. The keep-alive request time is still bound to the
5825 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5826 not set.
5827
Cyril Bonté653dcd62014-02-20 00:13:15 +01005828 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02005829 http-server-close" or "option http-tunnel". When backend and frontend options
5830 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005831
Christopher Faulet315b39c2018-09-21 16:26:19 +02005832 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01005833 "option prefer-last-server", "option http-pretend-keepalive",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005834 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005835
5836
Willy Tarreau96e31212011-05-30 18:10:30 +02005837option http-no-delay
5838no option http-no-delay
5839 Instruct the system to favor low interactive delays over performance in HTTP
5840 May be used in sections : defaults | frontend | listen | backend
5841 yes | yes | yes | yes
5842 Arguments : none
5843
5844 In HTTP, each payload is unidirectional and has no notion of interactivity.
5845 Any agent is expected to queue data somewhat for a reasonably low delay.
5846 There are some very rare server-to-server applications that abuse the HTTP
5847 protocol and expect the payload phase to be highly interactive, with many
5848 interleaved data chunks in both directions within a single request. This is
5849 absolutely not supported by the HTTP specification and will not work across
5850 most proxies or servers. When such applications attempt to do this through
5851 haproxy, it works but they will experience high delays due to the network
5852 optimizations which favor performance by instructing the system to wait for
5853 enough data to be available in order to only send full packets. Typical
5854 delays are around 200 ms per round trip. Note that this only happens with
5855 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
5856 affected.
5857
5858 When "option http-no-delay" is present in either the frontend or the backend
5859 used by a connection, all such optimizations will be disabled in order to
5860 make the exchanges as fast as possible. Of course this offers no guarantee on
5861 the functionality, as it may break at any other place. But if it works via
5862 HAProxy, it will work as fast as possible. This option should never be used
5863 by default, and should never be used at all unless such a buggy application
5864 is discovered. The impact of using this option is an increase of bandwidth
5865 usage and CPU usage, which may significantly lower performance in high
5866 latency environments.
5867
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005868 See also : "option http-buffer-request"
5869
Willy Tarreau96e31212011-05-30 18:10:30 +02005870
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005871option http-pretend-keepalive
5872no option http-pretend-keepalive
5873 Define whether haproxy will announce keepalive to the server or not
5874 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02005875 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005876 Arguments : none
5877
Christopher Faulet315b39c2018-09-21 16:26:19 +02005878 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005879 adds a "Connection: close" header to the request forwarded to the server.
5880 Unfortunately, when some servers see this header, they automatically refrain
5881 from using the chunked encoding for responses of unknown length, while this
5882 is totally unrelated. The immediate effect is that this prevents haproxy from
5883 maintaining the client connection alive. A second effect is that a client or
5884 a cache could receive an incomplete response without being aware of it, and
5885 consider the response complete.
5886
5887 By setting "option http-pretend-keepalive", haproxy will make the server
5888 believe it will keep the connection alive. The server will then not fall back
5889 to the abnormal undesired above. When haproxy gets the whole response, it
5890 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02005891 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005892 connection is correctly closed on the server side.
5893
5894 It is recommended not to enable this option by default, because most servers
5895 will more efficiently close the connection themselves after the last packet,
5896 and release its buffers slightly earlier. Also, the added packet on the
5897 network could slightly reduce the overall peak performance. However it is
5898 worth noting that when this option is enabled, haproxy will have slightly
5899 less work to do. So if haproxy is the bottleneck on the whole architecture,
5900 enabling this option might save a few CPU cycles.
5901
Christopher Faulet98db9762018-09-21 10:25:19 +02005902 This option may be set in backend and listen sections. Using it in a frontend
5903 section will be ignored and a warning will be reported during startup. It is
5904 a backend related option, so there is no real reason to set it on a
5905 frontend. This option may be combined with "option httpclose", which will
5906 cause keepalive to be announced to the server and close to be announced to
5907 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005908
5909 If this option has been enabled in a "defaults" section, it can be disabled
5910 in a specific instance by prepending the "no" keyword before it.
5911
Christopher Faulet315b39c2018-09-21 16:26:19 +02005912 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01005913 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005914
Willy Tarreauc27debf2008-01-06 08:57:02 +01005915
Willy Tarreaub608feb2010-01-02 22:47:18 +01005916option http-server-close
5917no option http-server-close
5918 Enable or disable HTTP connection closing on the server side
5919 May be used in sections : defaults | frontend | listen | backend
5920 yes | yes | yes | yes
5921 Arguments : none
5922
Willy Tarreau70dffda2014-01-30 03:07:23 +01005923 By default HAProxy operates in keep-alive mode with regards to persistent
5924 connections: for each connection it processes each request and response, and
5925 leaves the connection idle on both sides between the end of a response and
5926 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02005927 as "option http-server-close", "option httpclose" or "option http-tunnel".
5928 Setting "option http-server-close" enables HTTP connection-close mode on the
5929 server side while keeping the ability to support HTTP keep-alive and
5930 pipelining on the client side. This provides the lowest latency on the client
5931 side (slow network) and the fastest session reuse on the server side to save
5932 server resources, similarly to "option httpclose". It also permits
5933 non-keepalive capable servers to be served in keep-alive mode to the clients
5934 if they conform to the requirements of RFC7230. Please note that some servers
5935 do not always conform to those requirements when they see "Connection: close"
5936 in the request. The effect will be that keep-alive will never be used. A
5937 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005938
5939 At the moment, logs will not indicate whether requests came from the same
5940 session or not. The accept date reported in the logs corresponds to the end
5941 of the previous request, and the request time corresponds to the time spent
5942 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01005943 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5944 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005945
5946 This option may be set both in a frontend and in a backend. It is enabled if
5947 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02005948 It disables and replaces any previous "option httpclose", "option http-tunnel"
5949 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
5950 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005951
5952 If this option has been enabled in a "defaults" section, it can be disabled
5953 in a specific instance by prepending the "no" keyword before it.
5954
Christopher Faulet315b39c2018-09-21 16:26:19 +02005955 See also : "option httpclose", "option http-pretend-keepalive",
5956 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005957
5958
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005959option http-tunnel
5960no option http-tunnel
5961 Disable or enable HTTP connection processing after first transaction
5962 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02005963 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005964 Arguments : none
5965
Willy Tarreau70dffda2014-01-30 03:07:23 +01005966 By default HAProxy operates in keep-alive mode with regards to persistent
5967 connections: for each connection it processes each request and response, and
5968 leaves the connection idle on both sides between the end of a response and
5969 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02005970 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01005971
5972 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005973 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01005974 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
5975 is normally not needed anymore unless in very specific cases such as when
5976 using an in-house protocol that looks like HTTP but is not compatible, or
5977 just to log one request per client in order to reduce log size. Note that
5978 everything which works at the HTTP level, including header parsing/addition,
5979 cookie processing or content switching will only work for the first request
5980 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005981
Christopher Faulet4212a302018-09-21 10:42:19 +02005982 This option may be set on frontend and listen sections. Using it on a backend
5983 section will be ignored and a warning will be reported during the startup. It
5984 is a frontend related option, so there is no real reason to set it on a
5985 backend.
5986
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005987 If this option has been enabled in a "defaults" section, it can be disabled
5988 in a specific instance by prepending the "no" keyword before it.
5989
Christopher Faulet315b39c2018-09-21 16:26:19 +02005990 See also : "option httpclose", "option http-server-close",
5991 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005992
5993
Willy Tarreau88d349d2010-01-25 12:15:43 +01005994option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01005995no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01005996 Make use of non-standard Proxy-Connection header instead of Connection
5997 May be used in sections : defaults | frontend | listen | backend
5998 yes | yes | yes | no
5999 Arguments : none
6000
Lukas Tribus23953682017-04-28 13:24:30 +00006001 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006002 Connection header to indicate their wish of persistent or non-persistent
6003 connections, both browsers and proxies ignore this header for proxied
6004 connections and make use of the undocumented, non-standard Proxy-Connection
6005 header instead. The issue begins when trying to put a load balancer between
6006 browsers and such proxies, because there will be a difference between what
6007 haproxy understands and what the client and the proxy agree on.
6008
6009 By setting this option in a frontend, haproxy can automatically switch to use
6010 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006011 defined here as one where the URI begins with neither a '/' nor a '*'. This
6012 is incompatible with the HTTP tunnel mode. Note that this option can only be
6013 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006014
Willy Tarreau844a7e72010-01-31 21:46:18 +01006015 Also, when this option is set, a request which requires authentication will
6016 automatically switch to use proxy authentication headers if it is itself a
6017 proxied request. That makes it possible to check or enforce authentication in
6018 front of an existing proxy.
6019
Willy Tarreau88d349d2010-01-25 12:15:43 +01006020 This option should normally never be used, except in front of a proxy.
6021
Christopher Faulet315b39c2018-09-21 16:26:19 +02006022 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006023
6024
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006025option http-use-htx
6026no option http-use-htx
6027 Switch to the new HTX internal representation for HTTP protocol elements
6028 May be used in sections : defaults | frontend | listen | backend
6029 yes | yes | yes | yes
6030 Arguments : none
6031
6032 By default, the HTTP protocol is processed as-is. Inserting, deleting, or
6033 modifying a header field requires to rewrite the affected part in the buffer
6034 and to move the buffer's tail accordingly. Since this principle has deep
6035 roots in haproxy, the HTTP/2 protocol is converted to HTTP/1.1 before being
6036 processed this way. It also results in the inability to establish HTTP/2
6037 connections to servers because of the loss of HTTP/2 semantics in the HTTP/1
6038 representation.
6039
6040 HTX is the name of a totally new native internal representation for the HTTP
6041 protocol, that is agnostic to the version and aims at preserving semantics
6042 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6043 the protocol elements so that no more memory moves are necessary and that
6044 most elements are directly accessed. This mechanism is still limited to the
6045 most basic operations (no compression, filters, Lua, applets, cache, etc).
6046 But it supports using either HTTP/1 or HTTP/2 on any side regardless of the
6047 other side's version.
6048
6049 This option indicates that HTX needs to be used. It will cause errors to be
6050 emitted if incompatible features are used, but will allow H2 to be selected
6051 as a server protocol. It is recommended to use this option on new reasonably
6052 simple configurations, but since the feature still has incomplete functional
6053 coverage, it is not enabled by default.
6054
6055 See also : "mode http"
6056
6057
Willy Tarreaud63335a2010-02-26 12:56:52 +01006058option httpchk
6059option httpchk <uri>
6060option httpchk <method> <uri>
6061option httpchk <method> <uri> <version>
6062 Enable HTTP protocol to check on the servers health
6063 May be used in sections : defaults | frontend | listen | backend
6064 yes | no | yes | yes
6065 Arguments :
6066 <method> is the optional HTTP method used with the requests. When not set,
6067 the "OPTIONS" method is used, as it generally requires low server
6068 processing and is easy to filter out from the logs. Any method
6069 may be used, though it is not recommended to invent non-standard
6070 ones.
6071
6072 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6073 which is accessible by default on almost any server, but may be
6074 changed to any other URI. Query strings are permitted.
6075
6076 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6077 but some servers might behave incorrectly in HTTP 1.0, so turning
6078 it to HTTP/1.1 may sometimes help. Note that the Host field is
6079 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6080 after "\r\n" following the version string.
6081
6082 By default, server health checks only consist in trying to establish a TCP
6083 connection. When "option httpchk" is specified, a complete HTTP request is
6084 sent once the TCP connection is established, and responses 2xx and 3xx are
6085 considered valid, while all other ones indicate a server failure, including
6086 the lack of any response.
6087
6088 The port and interval are specified in the server configuration.
6089
6090 This option does not necessarily require an HTTP backend, it also works with
6091 plain TCP backends. This is particularly useful to check simple scripts bound
6092 to some dedicated ports using the inetd daemon.
6093
6094 Examples :
6095 # Relay HTTPS traffic to Apache instance and check service availability
6096 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6097 backend https_relay
6098 mode tcp
6099 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6100 server apache1 192.168.1.1:443 check port 80
6101
Simon Hormanafc47ee2013-11-25 10:46:35 +09006102 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6103 "option pgsql-check", "http-check" and the "check", "port" and
6104 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006105
6106
Willy Tarreauc27debf2008-01-06 08:57:02 +01006107option httpclose
6108no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006109 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006110 May be used in sections : defaults | frontend | listen | backend
6111 yes | yes | yes | yes
6112 Arguments : none
6113
Willy Tarreau70dffda2014-01-30 03:07:23 +01006114 By default HAProxy operates in keep-alive mode with regards to persistent
6115 connections: for each connection it processes each request and response, and
6116 leaves the connection idle on both sides between the end of a response and
6117 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006118 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006119
Christopher Faulet315b39c2018-09-21 16:26:19 +02006120 If "option httpclose" is set, HAProxy will close connections with the server
6121 and the client as soon as the request and the response are received. It will
6122 alos check if a "Connection: close" header is already set in each direction,
6123 and will add one if missing. Any "Connection" header different from "close"
6124 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006125
Christopher Faulet315b39c2018-09-21 16:26:19 +02006126 This option may also be combined with "option http-pretend-keepalive", which
6127 will disable sending of the "Connection: close" header, but will still cause
6128 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006129
6130 This option may be set both in a frontend and in a backend. It is enabled if
6131 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006132 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006133 "option http-keep-alive" or "option http-tunnel". Please check section 4
6134 ("Proxies") to see how this option combines with others when frontend and
6135 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006136
6137 If this option has been enabled in a "defaults" section, it can be disabled
6138 in a specific instance by prepending the "no" keyword before it.
6139
Christopher Faulet315b39c2018-09-21 16:26:19 +02006140 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006141
6142
Emeric Brun3a058f32009-06-30 18:26:00 +02006143option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006144 Enable logging of HTTP request, session state and timers
6145 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006146 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006147 Arguments :
6148 clf if the "clf" argument is added, then the output format will be
6149 the CLF format instead of HAProxy's default HTTP format. You can
6150 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006151 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006152 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006153
6154 By default, the log output format is very poor, as it only contains the
6155 source and destination addresses, and the instance name. By specifying
6156 "option httplog", each log line turns into a much richer format including,
6157 but not limited to, the HTTP request, the connection timers, the session
6158 status, the connections numbers, the captured headers and cookies, the
6159 frontend, backend and server name, and of course the source address and
6160 ports.
6161
PiBa-NLbd556bf2014-12-11 21:31:54 +01006162 Specifying only "option httplog" will automatically clear the 'clf' mode
6163 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006164
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006165 "option httplog" overrides any previous "log-format" directive.
6166
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006167 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006168
Willy Tarreau55165fe2009-05-10 12:02:55 +02006169
6170option http_proxy
6171no option http_proxy
6172 Enable or disable plain HTTP proxy mode
6173 May be used in sections : defaults | frontend | listen | backend
6174 yes | yes | yes | yes
6175 Arguments : none
6176
6177 It sometimes happens that people need a pure HTTP proxy which understands
6178 basic proxy requests without caching nor any fancy feature. In this case,
6179 it may be worth setting up an HAProxy instance with the "option http_proxy"
6180 set. In this mode, no server is declared, and the connection is forwarded to
6181 the IP address and port found in the URL after the "http://" scheme.
6182
6183 No host address resolution is performed, so this only works when pure IP
6184 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006185 it will probably be used only by experts who know they need exactly it. This
6186 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006187
6188 If this option has been enabled in a "defaults" section, it can be disabled
6189 in a specific instance by prepending the "no" keyword before it.
6190
6191 Example :
6192 # this backend understands HTTP proxy requests and forwards them directly.
6193 backend direct_forward
6194 option httpclose
6195 option http_proxy
6196
6197 See also : "option httpclose"
6198
Willy Tarreau211ad242009-10-03 21:45:07 +02006199
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006200option independent-streams
6201no option independent-streams
6202 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006203 May be used in sections : defaults | frontend | listen | backend
6204 yes | yes | yes | yes
6205 Arguments : none
6206
6207 By default, when data is sent over a socket, both the write timeout and the
6208 read timeout for that socket are refreshed, because we consider that there is
6209 activity on that socket, and we have no other means of guessing if we should
6210 receive data or not.
6211
Davor Ocelice9ed2812017-12-25 17:49:28 +01006212 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006213 exists a situation where it is desirable to disable it, and only refresh the
6214 read timeout if there are incoming data. This happens on sessions with large
6215 timeouts and low amounts of exchanged data such as telnet session. If the
6216 server suddenly disappears, the output data accumulates in the system's
6217 socket buffers, both timeouts are correctly refreshed, and there is no way
6218 to know the server does not receive them, so we don't timeout. However, when
6219 the underlying protocol always echoes sent data, it would be enough by itself
6220 to detect the issue using the read timeout. Note that this problem does not
6221 happen with more verbose protocols because data won't accumulate long in the
6222 socket buffers.
6223
6224 When this option is set on the frontend, it will disable read timeout updates
6225 on data sent to the client. There probably is little use of this case. When
6226 the option is set on the backend, it will disable read timeout updates on
6227 data sent to the server. Doing so will typically break large HTTP posts from
6228 slow lines, so use it with caution.
6229
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006230 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006231 with a spelling mistake. This spelling is still supported but
6232 deprecated.
6233
Willy Tarreauce887fd2012-05-12 12:50:00 +02006234 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006235
6236
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006237option ldap-check
6238 Use LDAPv3 health checks for server testing
6239 May be used in sections : defaults | frontend | listen | backend
6240 yes | no | yes | yes
6241 Arguments : none
6242
6243 It is possible to test that the server correctly talks LDAPv3 instead of just
6244 testing that it accepts the TCP connection. When this option is set, an
6245 LDAPv3 anonymous simple bind message is sent to the server, and the response
6246 is analyzed to find an LDAPv3 bind response message.
6247
6248 The server is considered valid only when the LDAP response contains success
6249 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6250
6251 Logging of bind requests is server dependent see your documentation how to
6252 configure it.
6253
6254 Example :
6255 option ldap-check
6256
6257 See also : "option httpchk"
6258
6259
Simon Horman98637e52014-06-20 12:30:16 +09006260option external-check
6261 Use external processes for server health checks
6262 May be used in sections : defaults | frontend | listen | backend
6263 yes | no | yes | yes
6264
6265 It is possible to test the health of a server using an external command.
6266 This is achieved by running the executable set using "external-check
6267 command".
6268
6269 Requires the "external-check" global to be set.
6270
6271 See also : "external-check", "external-check command", "external-check path"
6272
6273
Willy Tarreau211ad242009-10-03 21:45:07 +02006274option log-health-checks
6275no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006276 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006277 May be used in sections : defaults | frontend | listen | backend
6278 yes | no | yes | yes
6279 Arguments : none
6280
Willy Tarreaubef1b322014-05-13 21:01:39 +02006281 By default, failed health check are logged if server is UP and successful
6282 health checks are logged if server is DOWN, so the amount of additional
6283 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006284
Willy Tarreaubef1b322014-05-13 21:01:39 +02006285 When this option is enabled, any change of the health check status or to
6286 the server's health will be logged, so that it becomes possible to know
6287 that a server was failing occasional checks before crashing, or exactly when
6288 it failed to respond a valid HTTP status, then when the port started to
6289 reject connections, then when the server stopped responding at all.
6290
Davor Ocelice9ed2812017-12-25 17:49:28 +01006291 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006292 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006293
Willy Tarreaubef1b322014-05-13 21:01:39 +02006294 See also: "option httpchk", "option ldap-check", "option mysql-check",
6295 "option pgsql-check", "option redis-check", "option smtpchk",
6296 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006297
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006298
6299option log-separate-errors
6300no option log-separate-errors
6301 Change log level for non-completely successful connections
6302 May be used in sections : defaults | frontend | listen | backend
6303 yes | yes | yes | no
6304 Arguments : none
6305
6306 Sometimes looking for errors in logs is not easy. This option makes haproxy
6307 raise the level of logs containing potentially interesting information such
6308 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6309 level changes from "info" to "err". This makes it possible to log them
6310 separately to a different file with most syslog daemons. Be careful not to
6311 remove them from the original file, otherwise you would lose ordering which
6312 provides very important information.
6313
6314 Using this option, large sites dealing with several thousand connections per
6315 second may log normal traffic to a rotating buffer and only archive smaller
6316 error logs.
6317
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006318 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006319 logging.
6320
Willy Tarreauc27debf2008-01-06 08:57:02 +01006321
6322option logasap
6323no option logasap
6324 Enable or disable early logging of HTTP requests
6325 May be used in sections : defaults | frontend | listen | backend
6326 yes | yes | yes | no
6327 Arguments : none
6328
6329 By default, HTTP requests are logged upon termination so that the total
6330 transfer time and the number of bytes appear in the logs. When large objects
6331 are being transferred, it may take a while before the request appears in the
6332 logs. Using "option logasap", the request gets logged as soon as the server
6333 sends the complete headers. The only missing information in the logs will be
6334 the total number of bytes which will indicate everything except the amount
6335 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006336 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006337 "Content-Length" response header so that the logs at least indicate how many
6338 bytes are expected to be transferred.
6339
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006340 Examples :
6341 listen http_proxy 0.0.0.0:80
6342 mode http
6343 option httplog
6344 option logasap
6345 log 192.168.2.200 local3
6346
6347 >>> Feb 6 12:14:14 localhost \
6348 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6349 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6350 "GET /image.iso HTTP/1.0"
6351
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006352 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006353 logging.
6354
6355
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006356option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006357 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006358 May be used in sections : defaults | frontend | listen | backend
6359 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006360 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006361 <username> This is the username which will be used when connecting to MySQL
6362 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006363 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006364
6365 If you specify a username, the check consists of sending two MySQL packet,
6366 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006367 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006368 Error packet. It is a basic but useful test which does not produce error nor
6369 aborted connect on the server. However, it requires adding an authorization
6370 in the MySQL table, like this :
6371
6372 USE mysql;
6373 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6374 FLUSH PRIVILEGES;
6375
6376 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006377 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006378 Error packet, we don't send anything in this mode. It was reported that it
6379 can generate lockout if check is too frequent and/or if there is not enough
6380 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6381 value as if a connection is established successfully within fewer than MySQL
6382 "max_connect_errors" attempts after a previous connection was interrupted,
6383 the error count for the host is cleared to zero. If HAProxy's server get
6384 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6385
6386 Remember that this does not check database presence nor database consistency.
6387 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006388
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006389 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006390
6391 Most often, an incoming MySQL server needs to see the client's IP address for
6392 various purposes, including IP privilege matching and connection logging.
6393 When possible, it is often wise to masquerade the client's IP address when
6394 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006395 which requires the transparent proxy feature to be compiled in, and the MySQL
6396 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006397
6398 See also: "option httpchk"
6399
6400
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006401option nolinger
6402no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006403 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006404 May be used in sections: defaults | frontend | listen | backend
6405 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006406 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006407
Davor Ocelice9ed2812017-12-25 17:49:28 +01006408 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006409 physically disconnected), the session timeouts triggers and the session is
6410 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6411 using some resources and possibly limiting the ability to establish newer
6412 connections.
6413
6414 When this happens, it is possible to activate "option nolinger" which forces
6415 the system to immediately remove any socket's pending data on close. Thus,
6416 the session is instantly purged from the system's tables. This usually has
6417 side effects such as increased number of TCP resets due to old retransmits
6418 getting immediately rejected. Some firewalls may sometimes complain about
6419 this too.
6420
6421 For this reason, it is not recommended to use this option when not absolutely
6422 needed. You know that you need it when you have thousands of FIN_WAIT1
6423 sessions on your system (TIME_WAIT ones do not count).
6424
6425 This option may be used both on frontends and backends, depending on the side
6426 where it is required. Use it on the frontend for clients, and on the backend
6427 for servers.
6428
6429 If this option has been enabled in a "defaults" section, it can be disabled
6430 in a specific instance by prepending the "no" keyword before it.
6431
6432
Willy Tarreau55165fe2009-05-10 12:02:55 +02006433option originalto [ except <network> ] [ header <name> ]
6434 Enable insertion of the X-Original-To header to requests sent to servers
6435 May be used in sections : defaults | frontend | listen | backend
6436 yes | yes | yes | yes
6437 Arguments :
6438 <network> is an optional argument used to disable this option for sources
6439 matching <network>
6440 <name> an optional argument to specify a different "X-Original-To"
6441 header name.
6442
6443 Since HAProxy can work in transparent mode, every request from a client can
6444 be redirected to the proxy and HAProxy itself can proxy every request to a
6445 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6446 be lost. This is annoying when you want access rules based on destination ip
6447 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6448 added by HAProxy to all requests sent to the server. This header contains a
6449 value representing the original destination IP address. Since this must be
6450 configured to always use the last occurrence of this header only. Note that
6451 only the last occurrence of the header must be used, since it is really
6452 possible that the client has already brought one.
6453
6454 The keyword "header" may be used to supply a different header name to replace
6455 the default "X-Original-To". This can be useful where you might already
6456 have a "X-Original-To" header from a different application, and you need
6457 preserve it. Also if your backend server doesn't use the "X-Original-To"
6458 header and requires different one.
6459
6460 Sometimes, a same HAProxy instance may be shared between a direct client
6461 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6462 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6463 header for a known source address or network by adding the "except" keyword
6464 followed by the network address. In this case, any source IP matching the
6465 network will not cause an addition of this header. Most common uses are with
6466 private networks or 127.0.0.1.
6467
6468 This option may be specified either in the frontend or in the backend. If at
6469 least one of them uses it, the header will be added. Note that the backend's
6470 setting of the header subargument takes precedence over the frontend's if
6471 both are defined.
6472
Willy Tarreau55165fe2009-05-10 12:02:55 +02006473 Examples :
6474 # Original Destination address
6475 frontend www
6476 mode http
6477 option originalto except 127.0.0.1
6478
6479 # Those servers want the IP Address in X-Client-Dst
6480 backend www
6481 mode http
6482 option originalto header X-Client-Dst
6483
Christopher Faulet315b39c2018-09-21 16:26:19 +02006484 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006485
6486
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006487option persist
6488no option persist
6489 Enable or disable forced persistence on down servers
6490 May be used in sections: defaults | frontend | listen | backend
6491 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006492 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006493
6494 When an HTTP request reaches a backend with a cookie which references a dead
6495 server, by default it is redispatched to another server. It is possible to
6496 force the request to be sent to the dead server first using "option persist"
6497 if absolutely needed. A common use case is when servers are under extreme
6498 load and spend their time flapping. In this case, the users would still be
6499 directed to the server they opened the session on, in the hope they would be
6500 correctly served. It is recommended to use "option redispatch" in conjunction
6501 with this option so that in the event it would not be possible to connect to
6502 the server at all (server definitely dead), the client would finally be
6503 redirected to another valid server.
6504
6505 If this option has been enabled in a "defaults" section, it can be disabled
6506 in a specific instance by prepending the "no" keyword before it.
6507
Willy Tarreau4de91492010-01-22 19:10:05 +01006508 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006509
6510
Willy Tarreau0c122822013-12-15 18:49:01 +01006511option pgsql-check [ user <username> ]
6512 Use PostgreSQL health checks for server testing
6513 May be used in sections : defaults | frontend | listen | backend
6514 yes | no | yes | yes
6515 Arguments :
6516 <username> This is the username which will be used when connecting to
6517 PostgreSQL server.
6518
6519 The check sends a PostgreSQL StartupMessage and waits for either
6520 Authentication request or ErrorResponse message. It is a basic but useful
6521 test which does not produce error nor aborted connect on the server.
6522 This check is identical with the "mysql-check".
6523
6524 See also: "option httpchk"
6525
6526
Willy Tarreau9420b122013-12-15 18:58:25 +01006527option prefer-last-server
6528no option prefer-last-server
6529 Allow multiple load balanced requests to remain on the same server
6530 May be used in sections: defaults | frontend | listen | backend
6531 yes | no | yes | yes
6532 Arguments : none
6533
6534 When the load balancing algorithm in use is not deterministic, and a previous
6535 request was sent to a server to which haproxy still holds a connection, it is
6536 sometimes desirable that subsequent requests on a same session go to the same
6537 server as much as possible. Note that this is different from persistence, as
6538 we only indicate a preference which haproxy tries to apply without any form
6539 of warranty. The real use is for keep-alive connections sent to servers. When
6540 this option is used, haproxy will try to reuse the same connection that is
6541 attached to the server instead of rebalancing to another server, causing a
6542 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006543 not make much sense to use this in combination with hashing algorithms. Note,
6544 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006545 to a proxy which sends a 407 (authentication required), when the load
6546 balancing algorithm is not deterministic. This is mandatory for use with the
6547 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006548 troubleshooting some faulty applications. Option prefer-last-server might be
6549 desirable in these environments as well, to avoid redistributing the traffic
6550 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006551
6552 If this option has been enabled in a "defaults" section, it can be disabled
6553 in a specific instance by prepending the "no" keyword before it.
6554
6555 See also: "option http-keep-alive"
6556
6557
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006558option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006559option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006560no option redispatch
6561 Enable or disable session redistribution in case of connection failure
6562 May be used in sections: defaults | frontend | listen | backend
6563 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006564 Arguments :
6565 <interval> The optional integer value that controls how often redispatches
6566 occur when retrying connections. Positive value P indicates a
6567 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006568 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006569 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006570 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006571 positive value of 1 would indicate a redispatch on every retry,
6572 and a positive value of 3 would indicate a redispatch on every
6573 third retry. You can disable redispatches with a value of 0.
6574
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006575
6576 In HTTP mode, if a server designated by a cookie is down, clients may
6577 definitely stick to it because they cannot flush the cookie, so they will not
6578 be able to access the service anymore.
6579
6580 Specifying "option redispatch" will allow the proxy to break their
6581 persistence and redistribute them to a working server.
6582
Joseph Lynch726ab712015-05-11 23:25:34 -07006583 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006584 connection failures. Of course, it requires having "retries" set to a nonzero
6585 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006586
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006587 This form is the preferred form, which replaces both the "redispatch" and
6588 "redisp" keywords.
6589
6590 If this option has been enabled in a "defaults" section, it can be disabled
6591 in a specific instance by prepending the "no" keyword before it.
6592
Willy Tarreau4de91492010-01-22 19:10:05 +01006593 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006594
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006595
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006596option redis-check
6597 Use redis health checks for server testing
6598 May be used in sections : defaults | frontend | listen | backend
6599 yes | no | yes | yes
6600 Arguments : none
6601
6602 It is possible to test that the server correctly talks REDIS protocol instead
6603 of just testing that it accepts the TCP connection. When this option is set,
6604 a PING redis command is sent to the server, and the response is analyzed to
6605 find the "+PONG" response message.
6606
6607 Example :
6608 option redis-check
6609
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006610 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006611
6612
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006613option smtpchk
6614option smtpchk <hello> <domain>
6615 Use SMTP health checks for server testing
6616 May be used in sections : defaults | frontend | listen | backend
6617 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006618 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006619 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006620 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006621 values will be turned into the default command ("HELO").
6622
6623 <domain> is the domain name to present to the server. It may only be
6624 specified (and is mandatory) if the hello command has been
6625 specified. By default, "localhost" is used.
6626
6627 When "option smtpchk" is set, the health checks will consist in TCP
6628 connections followed by an SMTP command. By default, this command is
6629 "HELO localhost". The server's return code is analyzed and only return codes
6630 starting with a "2" will be considered as valid. All other responses,
6631 including a lack of response will constitute an error and will indicate a
6632 dead server.
6633
6634 This test is meant to be used with SMTP servers or relays. Depending on the
6635 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006636 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006637 25 is often easier than adjusting the configuration.
6638
6639 Most often, an incoming SMTP server needs to see the client's IP address for
6640 various purposes, including spam filtering, anti-spoofing and logging. When
6641 possible, it is often wise to masquerade the client's IP address when
6642 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006643 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006644
6645 Example :
6646 option smtpchk HELO mydomain.org
6647
6648 See also : "option httpchk", "source"
6649
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006650
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006651option socket-stats
6652no option socket-stats
6653
6654 Enable or disable collecting & providing separate statistics for each socket.
6655 May be used in sections : defaults | frontend | listen | backend
6656 yes | yes | yes | no
6657
6658 Arguments : none
6659
6660
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006661option splice-auto
6662no option splice-auto
6663 Enable or disable automatic kernel acceleration on sockets in both directions
6664 May be used in sections : defaults | frontend | listen | backend
6665 yes | yes | yes | yes
6666 Arguments : none
6667
6668 When this option is enabled either on a frontend or on a backend, haproxy
6669 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006670 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006671 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006672 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006673 are not much aggressive in order to limit excessive use of splicing. This
6674 option requires splicing to be enabled at compile time, and may be globally
6675 disabled with the global option "nosplice". Since splice uses pipes, using it
6676 requires that there are enough spare pipes.
6677
6678 Important note: kernel-based TCP splicing is a Linux-specific feature which
6679 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6680 transfer data between sockets without copying these data to user-space, thus
6681 providing noticeable performance gains and CPU cycles savings. Since many
6682 early implementations are buggy, corrupt data and/or are inefficient, this
6683 feature is not enabled by default, and it should be used with extreme care.
6684 While it is not possible to detect the correctness of an implementation,
6685 2.6.29 is the first version offering a properly working implementation. In
6686 case of doubt, splicing may be globally disabled using the global "nosplice"
6687 keyword.
6688
6689 Example :
6690 option splice-auto
6691
6692 If this option has been enabled in a "defaults" section, it can be disabled
6693 in a specific instance by prepending the "no" keyword before it.
6694
6695 See also : "option splice-request", "option splice-response", and global
6696 options "nosplice" and "maxpipes"
6697
6698
6699option splice-request
6700no option splice-request
6701 Enable or disable automatic kernel acceleration on sockets for requests
6702 May be used in sections : defaults | frontend | listen | backend
6703 yes | yes | yes | yes
6704 Arguments : none
6705
6706 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006707 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006708 the client to the server. It might still use the recv/send scheme if there
6709 are no spare pipes left. This option requires splicing to be enabled at
6710 compile time, and may be globally disabled with the global option "nosplice".
6711 Since splice uses pipes, using it requires that there are enough spare pipes.
6712
6713 Important note: see "option splice-auto" for usage limitations.
6714
6715 Example :
6716 option splice-request
6717
6718 If this option has been enabled in a "defaults" section, it can be disabled
6719 in a specific instance by prepending the "no" keyword before it.
6720
6721 See also : "option splice-auto", "option splice-response", and global options
6722 "nosplice" and "maxpipes"
6723
6724
6725option splice-response
6726no option splice-response
6727 Enable or disable automatic kernel acceleration on sockets for responses
6728 May be used in sections : defaults | frontend | listen | backend
6729 yes | yes | yes | yes
6730 Arguments : none
6731
6732 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006733 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006734 the server to the client. It might still use the recv/send scheme if there
6735 are no spare pipes left. This option requires splicing to be enabled at
6736 compile time, and may be globally disabled with the global option "nosplice".
6737 Since splice uses pipes, using it requires that there are enough spare pipes.
6738
6739 Important note: see "option splice-auto" for usage limitations.
6740
6741 Example :
6742 option splice-response
6743
6744 If this option has been enabled in a "defaults" section, it can be disabled
6745 in a specific instance by prepending the "no" keyword before it.
6746
6747 See also : "option splice-auto", "option splice-request", and global options
6748 "nosplice" and "maxpipes"
6749
6750
Christopher Fauletba7bc162016-11-07 21:07:38 +01006751option spop-check
6752 Use SPOP health checks for server testing
6753 May be used in sections : defaults | frontend | listen | backend
6754 no | no | no | yes
6755 Arguments : none
6756
6757 It is possible to test that the server correctly talks SPOP protocol instead
6758 of just testing that it accepts the TCP connection. When this option is set,
6759 a HELLO handshake is performed between HAProxy and the server, and the
6760 response is analyzed to check no error is reported.
6761
6762 Example :
6763 option spop-check
6764
6765 See also : "option httpchk"
6766
6767
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006768option srvtcpka
6769no option srvtcpka
6770 Enable or disable the sending of TCP keepalive packets on the server side
6771 May be used in sections : defaults | frontend | listen | backend
6772 yes | no | yes | yes
6773 Arguments : none
6774
6775 When there is a firewall or any session-aware component between a client and
6776 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006777 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006778 components decides to expire a session which has remained idle for too long.
6779
6780 Enabling socket-level TCP keep-alives makes the system regularly send packets
6781 to the other end of the connection, leaving it active. The delay between
6782 keep-alive probes is controlled by the system only and depends both on the
6783 operating system and its tuning parameters.
6784
6785 It is important to understand that keep-alive packets are neither emitted nor
6786 received at the application level. It is only the network stacks which sees
6787 them. For this reason, even if one side of the proxy already uses keep-alives
6788 to maintain its connection alive, those keep-alive packets will not be
6789 forwarded to the other side of the proxy.
6790
6791 Please note that this has nothing to do with HTTP keep-alive.
6792
6793 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6794 server side of a connection, which should help when session expirations are
6795 noticed between HAProxy and a server.
6796
6797 If this option has been enabled in a "defaults" section, it can be disabled
6798 in a specific instance by prepending the "no" keyword before it.
6799
6800 See also : "option clitcpka", "option tcpka"
6801
6802
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006803option ssl-hello-chk
6804 Use SSLv3 client hello health checks for server testing
6805 May be used in sections : defaults | frontend | listen | backend
6806 yes | no | yes | yes
6807 Arguments : none
6808
6809 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6810 possible to test that the server correctly talks SSL instead of just testing
6811 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6812 SSLv3 client hello messages are sent once the connection is established to
6813 the server, and the response is analyzed to find an SSL server hello message.
6814 The server is considered valid only when the response contains this server
6815 hello message.
6816
6817 All servers tested till there correctly reply to SSLv3 client hello messages,
6818 and most servers tested do not even log the requests containing only hello
6819 messages, which is appreciable.
6820
Willy Tarreau763a95b2012-10-04 23:15:39 +02006821 Note that this check works even when SSL support was not built into haproxy
6822 because it forges the SSL message. When SSL support is available, it is best
6823 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006824
Willy Tarreau763a95b2012-10-04 23:15:39 +02006825 See also: "option httpchk", "check-ssl"
6826
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006827
Willy Tarreaued179852013-12-16 01:07:00 +01006828option tcp-check
6829 Perform health checks using tcp-check send/expect sequences
6830 May be used in sections: defaults | frontend | listen | backend
6831 yes | no | yes | yes
6832
6833 This health check method is intended to be combined with "tcp-check" command
6834 lists in order to support send/expect types of health check sequences.
6835
6836 TCP checks currently support 4 modes of operations :
6837 - no "tcp-check" directive : the health check only consists in a connection
6838 attempt, which remains the default mode.
6839
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006840 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01006841 used to send a string along with a connection opening. With some
6842 protocols, it helps sending a "QUIT" message for example that prevents
6843 the server from logging a connection error for each health check. The
6844 check result will still be based on the ability to open the connection
6845 only.
6846
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006847 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01006848 The connection is opened and haproxy waits for the server to present some
6849 contents which must validate some rules. The check result will be based
6850 on the matching between the contents and the rules. This is suited for
6851 POP, IMAP, SMTP, FTP, SSH, TELNET.
6852
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006853 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01006854 used to test a hello-type protocol. HAProxy sends a message, the server
6855 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006856 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01006857 suited for protocols which require a binding or a request/response model.
6858 LDAP, MySQL, Redis and SSL are example of such protocols, though they
6859 already all have their dedicated checks with a deeper understanding of
6860 the respective protocols.
6861 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01006862 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01006863
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006864 A fifth mode can be used to insert comments in different steps of the
6865 script.
6866
6867 For each tcp-check rule you create, you can add a "comment" directive,
6868 followed by a string. This string will be reported in the log and stderr
6869 in debug mode. It is useful to make user-friendly error reporting.
6870 The "comment" is of course optional.
6871
6872
Willy Tarreaued179852013-12-16 01:07:00 +01006873 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006874 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01006875 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006876 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006877
Davor Ocelice9ed2812017-12-25 17:49:28 +01006878 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01006879 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006880 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006881
6882 # look for the redis master server after ensuring it speaks well
6883 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01006884 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01006885 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006886 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006887 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02006888 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006889 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01006890 tcp-check send info\ replication\r\n
6891 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006892 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006893 tcp-check send QUIT\r\n
6894 tcp-check expect string +OK
6895
Davor Ocelice9ed2812017-12-25 17:49:28 +01006896 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01006897 (send many headers before analyzing)
6898 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006899 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01006900 tcp-check send HEAD\ /\ HTTP/1.1\r\n
6901 tcp-check send Host:\ www.mydomain.com\r\n
6902 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
6903 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006904 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01006905
6906
6907 See also : "tcp-check expect", "tcp-check send"
6908
6909
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006910option tcp-smart-accept
6911no option tcp-smart-accept
6912 Enable or disable the saving of one ACK packet during the accept sequence
6913 May be used in sections : defaults | frontend | listen | backend
6914 yes | yes | yes | no
6915 Arguments : none
6916
6917 When an HTTP connection request comes in, the system acknowledges it on
6918 behalf of HAProxy, then the client immediately sends its request, and the
6919 system acknowledges it too while it is notifying HAProxy about the new
6920 connection. HAProxy then reads the request and responds. This means that we
6921 have one TCP ACK sent by the system for nothing, because the request could
6922 very well be acknowledged by HAProxy when it sends its response.
6923
6924 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
6925 sending this useless ACK on platforms which support it (currently at least
6926 Linux). It must not cause any problem, because the system will send it anyway
6927 after 40 ms if the response takes more time than expected to come.
6928
6929 During complex network debugging sessions, it may be desirable to disable
6930 this optimization because delayed ACKs can make troubleshooting more complex
6931 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006932 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006933
6934 It is also possible to force it for non-HTTP proxies by simply specifying
6935 "option tcp-smart-accept". For instance, it can make sense with some services
6936 such as SMTP where the server speaks first.
6937
6938 It is recommended to avoid forcing this option in a defaults section. In case
6939 of doubt, consider setting it back to automatic values by prepending the
6940 "default" keyword before it, or disabling it using the "no" keyword.
6941
Willy Tarreaud88edf22009-06-14 15:48:17 +02006942 See also : "option tcp-smart-connect"
6943
6944
6945option tcp-smart-connect
6946no option tcp-smart-connect
6947 Enable or disable the saving of one ACK packet during the connect sequence
6948 May be used in sections : defaults | frontend | listen | backend
6949 yes | no | yes | yes
6950 Arguments : none
6951
6952 On certain systems (at least Linux), HAProxy can ask the kernel not to
6953 immediately send an empty ACK upon a connection request, but to directly
6954 send the buffer request instead. This saves one packet on the network and
6955 thus boosts performance. It can also be useful for some servers, because they
6956 immediately get the request along with the incoming connection.
6957
6958 This feature is enabled when "option tcp-smart-connect" is set in a backend.
6959 It is not enabled by default because it makes network troubleshooting more
6960 complex.
6961
6962 It only makes sense to enable it with protocols where the client speaks first
6963 such as HTTP. In other situations, if there is no data to send in place of
6964 the ACK, a normal ACK is sent.
6965
6966 If this option has been enabled in a "defaults" section, it can be disabled
6967 in a specific instance by prepending the "no" keyword before it.
6968
6969 See also : "option tcp-smart-accept"
6970
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006971
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006972option tcpka
6973 Enable or disable the sending of TCP keepalive packets on both sides
6974 May be used in sections : defaults | frontend | listen | backend
6975 yes | yes | yes | yes
6976 Arguments : none
6977
6978 When there is a firewall or any session-aware component between a client and
6979 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006980 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006981 components decides to expire a session which has remained idle for too long.
6982
6983 Enabling socket-level TCP keep-alives makes the system regularly send packets
6984 to the other end of the connection, leaving it active. The delay between
6985 keep-alive probes is controlled by the system only and depends both on the
6986 operating system and its tuning parameters.
6987
6988 It is important to understand that keep-alive packets are neither emitted nor
6989 received at the application level. It is only the network stacks which sees
6990 them. For this reason, even if one side of the proxy already uses keep-alives
6991 to maintain its connection alive, those keep-alive packets will not be
6992 forwarded to the other side of the proxy.
6993
6994 Please note that this has nothing to do with HTTP keep-alive.
6995
6996 Using option "tcpka" enables the emission of TCP keep-alive probes on both
6997 the client and server sides of a connection. Note that this is meaningful
6998 only in "defaults" or "listen" sections. If this option is used in a
6999 frontend, only the client side will get keep-alives, and if this option is
7000 used in a backend, only the server side will get keep-alives. For this
7001 reason, it is strongly recommended to explicitly use "option clitcpka" and
7002 "option srvtcpka" when the configuration is split between frontends and
7003 backends.
7004
7005 See also : "option clitcpka", "option srvtcpka"
7006
Willy Tarreau844e3c52008-01-11 16:28:18 +01007007
7008option tcplog
7009 Enable advanced logging of TCP connections with session state and timers
7010 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007011 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007012 Arguments : none
7013
7014 By default, the log output format is very poor, as it only contains the
7015 source and destination addresses, and the instance name. By specifying
7016 "option tcplog", each log line turns into a much richer format including, but
7017 not limited to, the connection timers, the session status, the connections
7018 numbers, the frontend, backend and server name, and of course the source
7019 address and ports. This option is useful for pure TCP proxies in order to
7020 find which of the client or server disconnects or times out. For normal HTTP
7021 proxies, it's better to use "option httplog" which is even more complete.
7022
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007023 "option tcplog" overrides any previous "log-format" directive.
7024
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007025 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007026
7027
Willy Tarreau844e3c52008-01-11 16:28:18 +01007028option transparent
7029no option transparent
7030 Enable client-side transparent proxying
7031 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007032 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007033 Arguments : none
7034
7035 This option was introduced in order to provide layer 7 persistence to layer 3
7036 load balancers. The idea is to use the OS's ability to redirect an incoming
7037 connection for a remote address to a local process (here HAProxy), and let
7038 this process know what address was initially requested. When this option is
7039 used, sessions without cookies will be forwarded to the original destination
7040 IP address of the incoming request (which should match that of another
7041 equipment), while requests with cookies will still be forwarded to the
7042 appropriate server.
7043
7044 Note that contrary to a common belief, this option does NOT make HAProxy
7045 present the client's IP to the server when establishing the connection.
7046
Willy Tarreaua1146052011-03-01 09:51:54 +01007047 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007048 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007049
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007050
Simon Horman98637e52014-06-20 12:30:16 +09007051external-check command <command>
7052 Executable to run when performing an external-check
7053 May be used in sections : defaults | frontend | listen | backend
7054 yes | no | yes | yes
7055
7056 Arguments :
7057 <command> is the external command to run
7058
Simon Horman98637e52014-06-20 12:30:16 +09007059 The arguments passed to the to the command are:
7060
Cyril Bonté777be862014-12-02 21:21:35 +01007061 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007062
Cyril Bonté777be862014-12-02 21:21:35 +01007063 The <proxy_address> and <proxy_port> are derived from the first listener
7064 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7065 listener the proxy_address will be the path of the socket and the
7066 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7067 possible to determine a listener, and both <proxy_address> and <proxy_port>
7068 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007069
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007070 Some values are also provided through environment variables.
7071
7072 Environment variables :
7073 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7074 applicable, for example in a "backend" section).
7075
7076 HAPROXY_PROXY_ID The backend id.
7077
7078 HAPROXY_PROXY_NAME The backend name.
7079
7080 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7081 applicable, for example in a "backend" section or
7082 for a UNIX socket).
7083
7084 HAPROXY_SERVER_ADDR The server address.
7085
7086 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7087
7088 HAPROXY_SERVER_ID The server id.
7089
7090 HAPROXY_SERVER_MAXCONN The server max connections.
7091
7092 HAPROXY_SERVER_NAME The server name.
7093
7094 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7095 socket).
7096
7097 PATH The PATH environment variable used when executing
7098 the command may be set using "external-check path".
7099
Simon Horman98637e52014-06-20 12:30:16 +09007100 If the command executed and exits with a zero status then the check is
7101 considered to have passed, otherwise the check is considered to have
7102 failed.
7103
7104 Example :
7105 external-check command /bin/true
7106
7107 See also : "external-check", "option external-check", "external-check path"
7108
7109
7110external-check path <path>
7111 The value of the PATH environment variable used when running an external-check
7112 May be used in sections : defaults | frontend | listen | backend
7113 yes | no | yes | yes
7114
7115 Arguments :
7116 <path> is the path used when executing external command to run
7117
7118 The default path is "".
7119
7120 Example :
7121 external-check path "/usr/bin:/bin"
7122
7123 See also : "external-check", "option external-check",
7124 "external-check command"
7125
7126
Emeric Brun647caf12009-06-30 17:57:00 +02007127persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007128persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007129 Enable RDP cookie-based persistence
7130 May be used in sections : defaults | frontend | listen | backend
7131 yes | no | yes | yes
7132 Arguments :
7133 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007134 default cookie name "msts" will be used. There currently is no
7135 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007136
7137 This statement enables persistence based on an RDP cookie. The RDP cookie
7138 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007139 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007140 and if an RDP cookie is found, it is decoded. If it matches a known server
7141 which is still UP (or if "option persist" is set), then the connection is
7142 forwarded to this server.
7143
7144 Note that this only makes sense in a TCP backend, but for this to work, the
7145 frontend must have waited long enough to ensure that an RDP cookie is present
7146 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007147 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007148 a single "listen" section.
7149
Willy Tarreau61e28f22010-05-16 22:31:05 +02007150 Also, it is important to understand that the terminal server will emit this
7151 RDP cookie only if it is configured for "token redirection mode", which means
7152 that the "IP address redirection" option is disabled.
7153
Emeric Brun647caf12009-06-30 17:57:00 +02007154 Example :
7155 listen tse-farm
7156 bind :3389
7157 # wait up to 5s for an RDP cookie in the request
7158 tcp-request inspect-delay 5s
7159 tcp-request content accept if RDP_COOKIE
7160 # apply RDP cookie persistence
7161 persist rdp-cookie
7162 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007163 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007164 balance rdp-cookie
7165 server srv1 1.1.1.1:3389
7166 server srv2 1.1.1.2:3389
7167
Simon Hormanab814e02011-06-24 14:50:20 +09007168 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7169 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007170
7171
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007172rate-limit sessions <rate>
7173 Set a limit on the number of new sessions accepted per second on a frontend
7174 May be used in sections : defaults | frontend | listen | backend
7175 yes | yes | yes | no
7176 Arguments :
7177 <rate> The <rate> parameter is an integer designating the maximum number
7178 of new sessions per second to accept on the frontend.
7179
7180 When the frontend reaches the specified number of new sessions per second, it
7181 stops accepting new connections until the rate drops below the limit again.
7182 During this time, the pending sessions will be kept in the socket's backlog
7183 (in system buffers) and haproxy will not even be aware that sessions are
7184 pending. When applying very low limit on a highly loaded service, it may make
7185 sense to increase the socket's backlog using the "backlog" keyword.
7186
7187 This feature is particularly efficient at blocking connection-based attacks
7188 or service abuse on fragile servers. Since the session rate is measured every
7189 millisecond, it is extremely accurate. Also, the limit applies immediately,
7190 no delay is needed at all to detect the threshold.
7191
7192 Example : limit the connection rate on SMTP to 10 per second max
7193 listen smtp
7194 mode tcp
7195 bind :25
7196 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007197 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007198
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007199 Note : when the maximum rate is reached, the frontend's status is not changed
7200 but its sockets appear as "WAITING" in the statistics if the
7201 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007202
7203 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7204
7205
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007206redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7207redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7208redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007209 Return an HTTP redirection if/unless a condition is matched
7210 May be used in sections : defaults | frontend | listen | backend
7211 no | yes | yes | yes
7212
7213 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007214 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007215
Willy Tarreau0140f252008-11-19 21:07:09 +01007216 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007217 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007218 the HTTP "Location" header. When used in an "http-request" rule,
7219 <loc> value follows the log-format rules and can include some
7220 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007221
7222 <pfx> With "redirect prefix", the "Location" header is built from the
7223 concatenation of <pfx> and the complete URI path, including the
7224 query string, unless the "drop-query" option is specified (see
7225 below). As a special case, if <pfx> equals exactly "/", then
7226 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007227 redirect to the same URL (for instance, to insert a cookie). When
7228 used in an "http-request" rule, <pfx> value follows the log-format
7229 rules and can include some dynamic values (see Custom Log Format
7230 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007231
7232 <sch> With "redirect scheme", then the "Location" header is built by
7233 concatenating <sch> with "://" then the first occurrence of the
7234 "Host" header, and then the URI path, including the query string
7235 unless the "drop-query" option is specified (see below). If no
7236 path is found or if the path is "*", then "/" is used instead. If
7237 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007238 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007239 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007240 HTTPS. When used in an "http-request" rule, <sch> value follows
7241 the log-format rules and can include some dynamic values (see
7242 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007243
7244 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007245 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7246 with 302 used by default if no code is specified. 301 means
7247 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007248 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007249 cache the redirection. 303 is equivalent to 302 except that the
7250 browser will fetch the location with a GET method. 307 is just
7251 like 302 but makes it clear that the same method must be reused.
7252 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007253
7254 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007255 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007256
7257 - "drop-query"
7258 When this keyword is used in a prefix-based redirection, then the
7259 location will be set without any possible query-string, which is useful
7260 for directing users to a non-secure page for instance. It has no effect
7261 with a location-type redirect.
7262
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007263 - "append-slash"
7264 This keyword may be used in conjunction with "drop-query" to redirect
7265 users who use a URL not ending with a '/' to the same one with the '/'.
7266 It can be useful to ensure that search engines will only see one URL.
7267 For this, a return code 301 is preferred.
7268
Willy Tarreau0140f252008-11-19 21:07:09 +01007269 - "set-cookie NAME[=value]"
7270 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7271 to the response. This is sometimes used to indicate that a user has
7272 been seen, for instance to protect against some types of DoS. No other
7273 cookie option is added, so the cookie will be a session cookie. Note
7274 that for a browser, a sole cookie name without an equal sign is
7275 different from a cookie with an equal sign.
7276
7277 - "clear-cookie NAME[=]"
7278 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7279 with the "Max-Age" attribute set to zero. This will tell the browser to
7280 delete this cookie. It is useful for instance on logout pages. It is
7281 important to note that clearing the cookie "NAME" will not remove a
7282 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7283 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007284
7285 Example: move the login URL only to HTTPS.
7286 acl clear dst_port 80
7287 acl secure dst_port 8080
7288 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007289 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007290 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007291 acl cookie_set hdr_sub(cookie) SEEN=1
7292
7293 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007294 redirect prefix https://mysite.com if login_page !secure
7295 redirect prefix http://mysite.com drop-query if login_page !uid_given
7296 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007297 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007298
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007299 Example: send redirects for request for articles without a '/'.
7300 acl missing_slash path_reg ^/article/[^/]*$
7301 redirect code 301 prefix / drop-query append-slash if missing_slash
7302
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007303 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007304 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007305
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007306 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007307 http-request redirect code 301 location \
7308 http://www.%[hdr(host)]%[capture.req.uri] \
7309 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007310
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007311 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007312
7313
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007314redisp (deprecated)
7315redispatch (deprecated)
7316 Enable or disable session redistribution in case of connection failure
7317 May be used in sections: defaults | frontend | listen | backend
7318 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007319 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007320
7321 In HTTP mode, if a server designated by a cookie is down, clients may
7322 definitely stick to it because they cannot flush the cookie, so they will not
7323 be able to access the service anymore.
7324
7325 Specifying "redispatch" will allow the proxy to break their persistence and
7326 redistribute them to a working server.
7327
7328 It also allows to retry last connection to another server in case of multiple
7329 connection failures. Of course, it requires having "retries" set to a nonzero
7330 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007331
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007332 This form is deprecated, do not use it in any new configuration, use the new
7333 "option redispatch" instead.
7334
7335 See also : "option redispatch"
7336
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007337
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007338reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007339 Add a header at the end of the HTTP request
7340 May be used in sections : defaults | frontend | listen | backend
7341 no | yes | yes | yes
7342 Arguments :
7343 <string> is the complete line to be added. Any space or known delimiter
7344 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007345 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007346
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007347 <cond> is an optional matching condition built from ACLs. It makes it
7348 possible to ignore this rule when other conditions are not met.
7349
Willy Tarreau303c0352008-01-17 19:01:39 +01007350 A new line consisting in <string> followed by a line feed will be added after
7351 the last header of an HTTP request.
7352
7353 Header transformations only apply to traffic which passes through HAProxy,
7354 and not to traffic generated by HAProxy, such as health-checks or error
7355 responses.
7356
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007357 Example : add "X-Proto: SSL" to requests coming via port 81
7358 acl is-ssl dst_port 81
7359 reqadd X-Proto:\ SSL if is-ssl
7360
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007361 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7362 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007363
7364
Willy Tarreau5321c422010-01-28 20:35:13 +01007365reqallow <search> [{if | unless} <cond>]
7366reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007367 Definitely allow an HTTP request if a line matches a regular expression
7368 May be used in sections : defaults | frontend | listen | backend
7369 no | yes | yes | yes
7370 Arguments :
7371 <search> is the regular expression applied to HTTP headers and to the
7372 request line. This is an extended regular expression. Parenthesis
7373 grouping is supported and no preliminary backslash is required.
7374 Any space or known delimiter must be escaped using a backslash
7375 ('\'). The pattern applies to a full line at a time. The
7376 "reqallow" keyword strictly matches case while "reqiallow"
7377 ignores case.
7378
Willy Tarreau5321c422010-01-28 20:35:13 +01007379 <cond> is an optional matching condition built from ACLs. It makes it
7380 possible to ignore this rule when other conditions are not met.
7381
Willy Tarreau303c0352008-01-17 19:01:39 +01007382 A request containing any line which matches extended regular expression
7383 <search> will mark the request as allowed, even if any later test would
7384 result in a deny. The test applies both to the request line and to request
7385 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007386 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007387
7388 It is easier, faster and more powerful to use ACLs to write access policies.
7389 Reqdeny, reqallow and reqpass should be avoided in new designs.
7390
7391 Example :
7392 # allow www.* but refuse *.local
7393 reqiallow ^Host:\ www\.
7394 reqideny ^Host:\ .*\.local
7395
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007396 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7397 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007398
7399
Willy Tarreau5321c422010-01-28 20:35:13 +01007400reqdel <search> [{if | unless} <cond>]
7401reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007402 Delete all headers matching a regular expression in an HTTP request
7403 May be used in sections : defaults | frontend | listen | backend
7404 no | yes | yes | yes
7405 Arguments :
7406 <search> is the regular expression applied to HTTP headers and to the
7407 request line. This is an extended regular expression. Parenthesis
7408 grouping is supported and no preliminary backslash is required.
7409 Any space or known delimiter must be escaped using a backslash
7410 ('\'). The pattern applies to a full line at a time. The "reqdel"
7411 keyword strictly matches case while "reqidel" ignores case.
7412
Willy Tarreau5321c422010-01-28 20:35:13 +01007413 <cond> is an optional matching condition built from ACLs. It makes it
7414 possible to ignore this rule when other conditions are not met.
7415
Willy Tarreau303c0352008-01-17 19:01:39 +01007416 Any header line matching extended regular expression <search> in the request
7417 will be completely deleted. Most common use of this is to remove unwanted
7418 and/or dangerous headers or cookies from a request before passing it to the
7419 next servers.
7420
7421 Header transformations only apply to traffic which passes through HAProxy,
7422 and not to traffic generated by HAProxy, such as health-checks or error
7423 responses. Keep in mind that header names are not case-sensitive.
7424
7425 Example :
7426 # remove X-Forwarded-For header and SERVER cookie
7427 reqidel ^X-Forwarded-For:.*
7428 reqidel ^Cookie:.*SERVER=
7429
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007430 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7431 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007432
7433
Willy Tarreau5321c422010-01-28 20:35:13 +01007434reqdeny <search> [{if | unless} <cond>]
7435reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007436 Deny an HTTP request if a line matches a regular expression
7437 May be used in sections : defaults | frontend | listen | backend
7438 no | yes | yes | yes
7439 Arguments :
7440 <search> is the regular expression applied to HTTP headers and to the
7441 request line. This is an extended regular expression. Parenthesis
7442 grouping is supported and no preliminary backslash is required.
7443 Any space or known delimiter must be escaped using a backslash
7444 ('\'). The pattern applies to a full line at a time. The
7445 "reqdeny" keyword strictly matches case while "reqideny" ignores
7446 case.
7447
Willy Tarreau5321c422010-01-28 20:35:13 +01007448 <cond> is an optional matching condition built from ACLs. It makes it
7449 possible to ignore this rule when other conditions are not met.
7450
Willy Tarreau303c0352008-01-17 19:01:39 +01007451 A request containing any line which matches extended regular expression
7452 <search> will mark the request as denied, even if any later test would
7453 result in an allow. The test applies both to the request line and to request
7454 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007455 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007456
Willy Tarreauced27012008-01-17 20:35:34 +01007457 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007458 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007459 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007460
Willy Tarreau303c0352008-01-17 19:01:39 +01007461 It is easier, faster and more powerful to use ACLs to write access policies.
7462 Reqdeny, reqallow and reqpass should be avoided in new designs.
7463
7464 Example :
7465 # refuse *.local, then allow www.*
7466 reqideny ^Host:\ .*\.local
7467 reqiallow ^Host:\ www\.
7468
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007469 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7470 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007471
7472
Willy Tarreau5321c422010-01-28 20:35:13 +01007473reqpass <search> [{if | unless} <cond>]
7474reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007475 Ignore any HTTP request line matching a regular expression in next rules
7476 May be used in sections : defaults | frontend | listen | backend
7477 no | yes | yes | yes
7478 Arguments :
7479 <search> is the regular expression applied to HTTP headers and to the
7480 request line. This is an extended regular expression. Parenthesis
7481 grouping is supported and no preliminary backslash is required.
7482 Any space or known delimiter must be escaped using a backslash
7483 ('\'). The pattern applies to a full line at a time. The
7484 "reqpass" keyword strictly matches case while "reqipass" ignores
7485 case.
7486
Willy Tarreau5321c422010-01-28 20:35:13 +01007487 <cond> is an optional matching condition built from ACLs. It makes it
7488 possible to ignore this rule when other conditions are not met.
7489
Willy Tarreau303c0352008-01-17 19:01:39 +01007490 A request containing any line which matches extended regular expression
7491 <search> will skip next rules, without assigning any deny or allow verdict.
7492 The test applies both to the request line and to request headers. Keep in
7493 mind that URLs in request line are case-sensitive while header names are not.
7494
7495 It is easier, faster and more powerful to use ACLs to write access policies.
7496 Reqdeny, reqallow and reqpass should be avoided in new designs.
7497
7498 Example :
7499 # refuse *.local, then allow www.*, but ignore "www.private.local"
7500 reqipass ^Host:\ www.private\.local
7501 reqideny ^Host:\ .*\.local
7502 reqiallow ^Host:\ www\.
7503
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007504 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7505 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007506
7507
Willy Tarreau5321c422010-01-28 20:35:13 +01007508reqrep <search> <string> [{if | unless} <cond>]
7509reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007510 Replace a regular expression with a string in an HTTP request line
7511 May be used in sections : defaults | frontend | listen | backend
7512 no | yes | yes | yes
7513 Arguments :
7514 <search> is the regular expression applied to HTTP headers and to the
7515 request line. This is an extended regular expression. Parenthesis
7516 grouping is supported and no preliminary backslash is required.
7517 Any space or known delimiter must be escaped using a backslash
7518 ('\'). The pattern applies to a full line at a time. The "reqrep"
7519 keyword strictly matches case while "reqirep" ignores case.
7520
7521 <string> is the complete line to be added. Any space or known delimiter
7522 must be escaped using a backslash ('\'). References to matched
7523 pattern groups are possible using the common \N form, with N
7524 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007525 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007526
Willy Tarreau5321c422010-01-28 20:35:13 +01007527 <cond> is an optional matching condition built from ACLs. It makes it
7528 possible to ignore this rule when other conditions are not met.
7529
Willy Tarreau303c0352008-01-17 19:01:39 +01007530 Any line matching extended regular expression <search> in the request (both
7531 the request line and header lines) will be completely replaced with <string>.
7532 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7533
7534 Header transformations only apply to traffic which passes through HAProxy,
7535 and not to traffic generated by HAProxy, such as health-checks or error
7536 responses. Note that for increased readability, it is suggested to add enough
7537 spaces between the request and the response. Keep in mind that URLs in
7538 request line are case-sensitive while header names are not.
7539
7540 Example :
7541 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007542 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007543 # replace "www.mydomain.com" with "www" in the host name.
7544 reqirep ^Host:\ www.mydomain.com Host:\ www
7545
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007546 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7547 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007548
7549
Willy Tarreau5321c422010-01-28 20:35:13 +01007550reqtarpit <search> [{if | unless} <cond>]
7551reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007552 Tarpit an HTTP request containing a line matching a regular expression
7553 May be used in sections : defaults | frontend | listen | backend
7554 no | yes | yes | yes
7555 Arguments :
7556 <search> is the regular expression applied to HTTP headers and to the
7557 request line. This is an extended regular expression. Parenthesis
7558 grouping is supported and no preliminary backslash is required.
7559 Any space or known delimiter must be escaped using a backslash
7560 ('\'). The pattern applies to a full line at a time. The
7561 "reqtarpit" keyword strictly matches case while "reqitarpit"
7562 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 A request containing any line which matches extended regular expression
7568 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007569 be kept open for a pre-defined time, then will return an HTTP error 500 so
7570 that the attacker does not suspect it has been tarpitted. The status 500 will
7571 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007572 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7573 not set.
7574
7575 The goal of the tarpit is to slow down robots attacking servers with
7576 identifiable requests. Many robots limit their outgoing number of connections
7577 and stay connected waiting for a reply which can take several minutes to
7578 come. Depending on the environment and attack, it may be particularly
7579 efficient at reducing the load on the network and firewalls.
7580
Willy Tarreau5321c422010-01-28 20:35:13 +01007581 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007582 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007583 # block all others.
7584 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7585 reqitarpit ^User-Agent:
7586
Willy Tarreau5321c422010-01-28 20:35:13 +01007587 # block bad guys
7588 acl badguys src 10.1.0.3 172.16.13.20/28
7589 reqitarpit . if badguys
7590
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007591 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7592 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007593
7594
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007595retries <value>
7596 Set the number of retries to perform on a server after a connection failure
7597 May be used in sections: defaults | frontend | listen | backend
7598 yes | no | yes | yes
7599 Arguments :
7600 <value> is the number of times a connection attempt should be retried on
7601 a server when a connection either is refused or times out. The
7602 default value is 3.
7603
7604 It is important to understand that this value applies to the number of
7605 connection attempts, not full requests. When a connection has effectively
7606 been established to a server, there will be no more retry.
7607
7608 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007609 a turn-around timer of min("timeout connect", one second) is applied before
7610 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007611
7612 When "option redispatch" is set, the last retry may be performed on another
7613 server even if a cookie references a different server.
7614
7615 See also : "option redispatch"
7616
7617
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007618rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007619 Add a header at the end of the HTTP response
7620 May be used in sections : defaults | frontend | listen | backend
7621 no | yes | yes | yes
7622 Arguments :
7623 <string> is the complete line to be added. Any space or known delimiter
7624 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007625 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007626
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007627 <cond> is an optional matching condition built from ACLs. It makes it
7628 possible to ignore this rule when other conditions are not met.
7629
Willy Tarreau303c0352008-01-17 19:01:39 +01007630 A new line consisting in <string> followed by a line feed will be added after
7631 the last header of an HTTP response.
7632
7633 Header transformations only apply to traffic which passes through HAProxy,
7634 and not to traffic generated by HAProxy, such as health-checks or error
7635 responses.
7636
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007637 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7638 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007639
7640
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007641rspdel <search> [{if | unless} <cond>]
7642rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007643 Delete all headers matching a regular expression in an HTTP response
7644 May be used in sections : defaults | frontend | listen | backend
7645 no | yes | yes | yes
7646 Arguments :
7647 <search> is the regular expression applied to HTTP headers and to the
7648 response line. This is an extended regular expression, so
7649 parenthesis grouping is supported and no preliminary backslash
7650 is required. Any space or known delimiter must be escaped using
7651 a backslash ('\'). The pattern applies to a full line at a time.
7652 The "rspdel" keyword strictly matches case while "rspidel"
7653 ignores case.
7654
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007655 <cond> is an optional matching condition built from ACLs. It makes it
7656 possible to ignore this rule when other conditions are not met.
7657
Willy Tarreau303c0352008-01-17 19:01:39 +01007658 Any header line matching extended regular expression <search> in the response
7659 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007660 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007661 client.
7662
7663 Header transformations only apply to traffic which passes through HAProxy,
7664 and not to traffic generated by HAProxy, such as health-checks or error
7665 responses. Keep in mind that header names are not case-sensitive.
7666
7667 Example :
7668 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007669 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007670
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007671 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7672 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007673
7674
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007675rspdeny <search> [{if | unless} <cond>]
7676rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007677 Block an HTTP response if a line matches a regular expression
7678 May be used in sections : defaults | frontend | listen | backend
7679 no | yes | yes | yes
7680 Arguments :
7681 <search> is the regular expression applied to HTTP headers and to the
7682 response line. This is an extended regular expression, so
7683 parenthesis grouping is supported and no preliminary backslash
7684 is required. Any space or known delimiter must be escaped using
7685 a backslash ('\'). The pattern applies to a full line at a time.
7686 The "rspdeny" keyword strictly matches case while "rspideny"
7687 ignores case.
7688
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007689 <cond> is an optional matching condition built from ACLs. It makes it
7690 possible to ignore this rule when other conditions are not met.
7691
Willy Tarreau303c0352008-01-17 19:01:39 +01007692 A response containing any line which matches extended regular expression
7693 <search> will mark the request as denied. The test applies both to the
7694 response line and to response headers. Keep in mind that header names are not
7695 case-sensitive.
7696
7697 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007698 block the response before it reaches the client. If a response is denied, it
7699 will be replaced with an HTTP 502 error so that the client never retrieves
7700 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007701
7702 It is easier, faster and more powerful to use ACLs to write access policies.
7703 Rspdeny should be avoided in new designs.
7704
7705 Example :
7706 # Ensure that no content type matching ms-word will leak
7707 rspideny ^Content-type:\.*/ms-word
7708
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007709 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7710 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007711
7712
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007713rsprep <search> <string> [{if | unless} <cond>]
7714rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007715 Replace a regular expression with a string in an HTTP response line
7716 May be used in sections : defaults | frontend | listen | backend
7717 no | yes | yes | yes
7718 Arguments :
7719 <search> is the regular expression applied to HTTP headers and to the
7720 response line. This is an extended regular expression, so
7721 parenthesis grouping is supported and no preliminary backslash
7722 is required. Any space or known delimiter must be escaped using
7723 a backslash ('\'). The pattern applies to a full line at a time.
7724 The "rsprep" keyword strictly matches case while "rspirep"
7725 ignores case.
7726
7727 <string> is the complete line to be added. Any space or known delimiter
7728 must be escaped using a backslash ('\'). References to matched
7729 pattern groups are possible using the common \N form, with N
7730 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007731 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007732
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007733 <cond> is an optional matching condition built from ACLs. It makes it
7734 possible to ignore this rule when other conditions are not met.
7735
Willy Tarreau303c0352008-01-17 19:01:39 +01007736 Any line matching extended regular expression <search> in the response (both
7737 the response line and header lines) will be completely replaced with
7738 <string>. Most common use of this is to rewrite Location headers.
7739
7740 Header transformations only apply to traffic which passes through HAProxy,
7741 and not to traffic generated by HAProxy, such as health-checks or error
7742 responses. Note that for increased readability, it is suggested to add enough
7743 spaces between the request and the response. Keep in mind that header names
7744 are not case-sensitive.
7745
7746 Example :
7747 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7748 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7749
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007750 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7751 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007752
7753
David du Colombier486df472011-03-17 10:40:26 +01007754server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007755 Declare a server in a backend
7756 May be used in sections : defaults | frontend | listen | backend
7757 no | no | yes | yes
7758 Arguments :
7759 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007760 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007761 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007762
David du Colombier486df472011-03-17 10:40:26 +01007763 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7764 resolvable hostname is supported, but this name will be resolved
7765 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7766 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007767 address as the one from the client connection. This is useful in
7768 transparent proxy architectures where the client's connection is
7769 intercepted and haproxy must forward to the original destination
7770 address. This is more or less what the "transparent" keyword does
7771 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007772 to report statistics. Optionally, an address family prefix may be
7773 used before the address to force the family regardless of the
7774 address format, which can be useful to specify a path to a unix
7775 socket with no slash ('/'). Currently supported prefixes are :
7776 - 'ipv4@' -> address is always IPv4
7777 - 'ipv6@' -> address is always IPv6
7778 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007779 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007780 - 'sockpair@' -> address is the FD of a connected unix
7781 socket or of a socketpair. During a connection, the
7782 backend creates a pair of connected sockets, and passes
7783 one of them over the FD. The bind part will use the
7784 received socket as the client FD. Should be used
7785 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007786 You may want to reference some environment variables in the
7787 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007788 variables. The "init-addr" setting can be used to modify the way
7789 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007790
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007791 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007792 be sent to this port. If unset, the same port the client
7793 connected to will be used. The port may also be prefixed by a "+"
7794 or a "-". In this case, the server's port will be determined by
7795 adding this value to the client's port.
7796
7797 <param*> is a list of parameters for this server. The "server" keywords
7798 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007799 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007800
7801 Examples :
7802 server first 10.1.1.1:1080 cookie first check inter 1000
7803 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007804 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007805 server backup "${SRV_BACKUP}:1080" backup
7806 server www1_dc1 "${LAN_DC1}.101:80"
7807 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007808
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007809 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7810 sun_path length is used for the address length. Some other programs
7811 such as socat use the string length only by default. Pass the option
7812 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7813 make it compatible with HAProxy's.
7814
Mark Lamourinec2247f02012-01-04 13:02:01 -05007815 See also: "default-server", "http-send-name-header" and section 5 about
7816 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007817
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007818server-state-file-name [<file>]
7819 Set the server state file to read, load and apply to servers available in
7820 this backend. It only applies when the directive "load-server-state-from-file"
7821 is set to "local". When <file> is not provided or if this directive is not
7822 set, then backend name is used. If <file> starts with a slash '/', then it is
7823 considered as an absolute path. Otherwise, <file> is concatenated to the
7824 global directive "server-state-file-base".
7825
7826 Example: the minimal configuration below would make HAProxy look for the
7827 state server file '/etc/haproxy/states/bk':
7828
7829 global
7830 server-state-file-base /etc/haproxy/states
7831
7832 backend bk
7833 load-server-state-from-file
7834
7835 See also: "server-state-file-base", "load-server-state-from-file", and
7836 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007837
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007838server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7839 Set a template to initialize servers with shared parameters.
7840 The names of these servers are built from <prefix> and <num | range> parameters.
7841 May be used in sections : defaults | frontend | listen | backend
7842 no | no | yes | yes
7843
7844 Arguments:
7845 <prefix> A prefix for the server names to be built.
7846
7847 <num | range>
7848 If <num> is provided, this template initializes <num> servers
7849 with 1 up to <num> as server name suffixes. A range of numbers
7850 <num_low>-<num_high> may also be used to use <num_low> up to
7851 <num_high> as server name suffixes.
7852
7853 <fqdn> A FQDN for all the servers this template initializes.
7854
7855 <port> Same meaning as "server" <port> argument (see "server" keyword).
7856
7857 <params*>
7858 Remaining server parameters among all those supported by "server"
7859 keyword.
7860
7861 Examples:
7862 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7863 # google.com as FQDN, and health-check enabled.
7864 server-template srv 1-3 google.com:80 check
7865
7866 # or
7867 server-template srv 3 google.com:80 check
7868
7869 # would be equivalent to:
7870 server srv1 google.com:80 check
7871 server srv2 google.com:80 check
7872 server srv3 google.com:80 check
7873
7874
7875
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007876source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007877source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007878source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007879 Set the source address for outgoing connections
7880 May be used in sections : defaults | frontend | listen | backend
7881 yes | no | yes | yes
7882 Arguments :
7883 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7884 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007885
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007886 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007887 the most appropriate address to reach its destination. Optionally
7888 an address family prefix may be used before the address to force
7889 the family regardless of the address format, which can be useful
7890 to specify a path to a unix socket with no slash ('/'). Currently
7891 supported prefixes are :
7892 - 'ipv4@' -> address is always IPv4
7893 - 'ipv6@' -> address is always IPv6
7894 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007895 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007896 You may want to reference some environment variables in the
7897 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007898
7899 <port> is an optional port. It is normally not needed but may be useful
7900 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007901 the system will select a free port. Note that port ranges are not
7902 supported in the backend. If you want to force port ranges, you
7903 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007904
7905 <addr2> is the IP address to present to the server when connections are
7906 forwarded in full transparent proxy mode. This is currently only
7907 supported on some patched Linux kernels. When this address is
7908 specified, clients connecting to the server will be presented
7909 with this address, while health checks will still use the address
7910 <addr>.
7911
7912 <port2> is the optional port to present to the server when connections
7913 are forwarded in full transparent proxy mode (see <addr2> above).
7914 The default value of zero means the system will select a free
7915 port.
7916
Willy Tarreaubce70882009-09-07 11:51:47 +02007917 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7918 This is the name of a comma-separated header list which can
7919 contain multiple IP addresses. By default, the last occurrence is
7920 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01007921 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02007922 by previous proxy, typically Stunnel. In order to use another
7923 occurrence from the last one, please see the <occ> parameter
7924 below. When the header (or occurrence) is not found, no binding
7925 is performed so that the proxy's default IP address is used. Also
7926 keep in mind that the header name is case insensitive, as for any
7927 HTTP header.
7928
7929 <occ> is the occurrence number of a value to be used in a multi-value
7930 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007931 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02007932 address. Positive values indicate a position from the first
7933 occurrence, 1 being the first one. Negative values indicate
7934 positions relative to the last one, -1 being the last one. This
7935 is helpful for situations where an X-Forwarded-For header is set
7936 at the entry point of an infrastructure and must be used several
7937 proxy layers away. When this value is not specified, -1 is
7938 assumed. Passing a zero here disables the feature.
7939
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007940 <name> is an optional interface name to which to bind to for outgoing
7941 traffic. On systems supporting this features (currently, only
7942 Linux), this allows one to bind all traffic to the server to
7943 this interface even if it is not the one the system would select
7944 based on routing tables. This should be used with extreme care.
7945 Note that using this option requires root privileges.
7946
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007947 The "source" keyword is useful in complex environments where a specific
7948 address only is allowed to connect to the servers. It may be needed when a
7949 private address must be used through a public gateway for instance, and it is
7950 known that the system cannot determine the adequate source address by itself.
7951
7952 An extension which is available on certain patched Linux kernels may be used
7953 through the "usesrc" optional keyword. It makes it possible to connect to the
7954 servers with an IP address which does not belong to the system itself. This
7955 is called "full transparent proxy mode". For this to work, the destination
7956 servers have to route their traffic back to this address through the machine
7957 running HAProxy, and IP forwarding must generally be enabled on this machine.
7958
7959 In this "full transparent proxy" mode, it is possible to force a specific IP
7960 address to be presented to the servers. This is not much used in fact. A more
7961 common use is to tell HAProxy to present the client's IP address. For this,
7962 there are two methods :
7963
7964 - present the client's IP and port addresses. This is the most transparent
7965 mode, but it can cause problems when IP connection tracking is enabled on
7966 the machine, because a same connection may be seen twice with different
7967 states. However, this solution presents the huge advantage of not
7968 limiting the system to the 64k outgoing address+port couples, because all
7969 of the client ranges may be used.
7970
7971 - present only the client's IP address and select a spare port. This
7972 solution is still quite elegant but slightly less transparent (downstream
7973 firewalls logs will not match upstream's). It also presents the downside
7974 of limiting the number of concurrent connections to the usual 64k ports.
7975 However, since the upstream and downstream ports are different, local IP
7976 connection tracking on the machine will not be upset by the reuse of the
7977 same session.
7978
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007979 This option sets the default source for all servers in the backend. It may
7980 also be specified in a "defaults" section. Finer source address specification
7981 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007982 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007983
Baptiste Assmann91bd3372015-07-17 21:59:42 +02007984 In order to work, "usesrc" requires root privileges.
7985
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007986 Examples :
7987 backend private
7988 # Connect to the servers using our 192.168.1.200 source address
7989 source 192.168.1.200
7990
7991 backend transparent_ssl1
7992 # Connect to the SSL farm from the client's source address
7993 source 192.168.1.200 usesrc clientip
7994
7995 backend transparent_ssl2
7996 # Connect to the SSL farm from the client's source address and port
7997 # not recommended if IP conntrack is present on the local machine.
7998 source 192.168.1.200 usesrc client
7999
8000 backend transparent_ssl3
8001 # Connect to the SSL farm from the client's source address. It
8002 # is more conntrack-friendly.
8003 source 192.168.1.200 usesrc clientip
8004
8005 backend transparent_smtp
8006 # Connect to the SMTP farm from the client's source address/port
8007 # with Tproxy version 4.
8008 source 0.0.0.0 usesrc clientip
8009
Willy Tarreaubce70882009-09-07 11:51:47 +02008010 backend transparent_http
8011 # Connect to the servers using the client's IP as seen by previous
8012 # proxy.
8013 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8014
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008015 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008016 the Linux kernel on www.balabit.com, the "bind" keyword.
8017
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008018
Willy Tarreau844e3c52008-01-11 16:28:18 +01008019srvtimeout <timeout> (deprecated)
8020 Set the maximum inactivity time on the server side.
8021 May be used in sections : defaults | frontend | listen | backend
8022 yes | no | yes | yes
8023 Arguments :
8024 <timeout> is the timeout value specified in milliseconds by default, but
8025 can be in any other unit if the number is suffixed by the unit,
8026 as explained at the top of this document.
8027
8028 The inactivity timeout applies when the server is expected to acknowledge or
8029 send data. In HTTP mode, this timeout is particularly important to consider
8030 during the first phase of the server's response, when it has to send the
8031 headers, as it directly represents the server's processing time for the
8032 request. To find out what value to put there, it's often good to start with
8033 what would be considered as unacceptable response times, then check the logs
8034 to observe the response time distribution, and adjust the value accordingly.
8035
8036 The value is specified in milliseconds by default, but can be in any other
8037 unit if the number is suffixed by the unit, as specified at the top of this
8038 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8039 recommended that the client timeout remains equal to the server timeout in
8040 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008041 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008042 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008043 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008044
8045 This parameter is specific to backends, but can be specified once for all in
8046 "defaults" sections. This is in fact one of the easiest solutions not to
8047 forget about it. An unspecified timeout results in an infinite timeout, which
8048 is not recommended. Such a usage is accepted and works but reports a warning
8049 during startup because it may results in accumulation of expired sessions in
8050 the system if the system's timeouts are not configured either.
8051
8052 This parameter is provided for compatibility but is currently deprecated.
8053 Please use "timeout server" instead.
8054
Willy Tarreauce887fd2012-05-12 12:50:00 +02008055 See also : "timeout server", "timeout tunnel", "timeout client" and
8056 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008057
8058
Cyril Bonté66c327d2010-10-12 00:14:37 +02008059stats admin { if | unless } <cond>
8060 Enable statistics admin level if/unless a condition is matched
8061 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008062 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008063
8064 This statement enables the statistics admin level if/unless a condition is
8065 matched.
8066
8067 The admin level allows to enable/disable servers from the web interface. By
8068 default, statistics page is read-only for security reasons.
8069
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008070 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8071 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008072 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008073
Cyril Bonté23b39d92011-02-10 22:54:44 +01008074 Currently, the POST request is limited to the buffer size minus the reserved
8075 buffer space, which means that if the list of servers is too long, the
8076 request won't be processed. It is recommended to alter few servers at a
8077 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008078
8079 Example :
8080 # statistics admin level only for localhost
8081 backend stats_localhost
8082 stats enable
8083 stats admin if LOCALHOST
8084
8085 Example :
8086 # statistics admin level always enabled because of the authentication
8087 backend stats_auth
8088 stats enable
8089 stats auth admin:AdMiN123
8090 stats admin if TRUE
8091
8092 Example :
8093 # statistics admin level depends on the authenticated user
8094 userlist stats-auth
8095 group admin users admin
8096 user admin insecure-password AdMiN123
8097 group readonly users haproxy
8098 user haproxy insecure-password haproxy
8099
8100 backend stats_auth
8101 stats enable
8102 acl AUTH http_auth(stats-auth)
8103 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8104 stats http-request auth unless AUTH
8105 stats admin if AUTH_ADMIN
8106
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008107 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8108 "bind-process", section 3.4 about userlists and section 7 about
8109 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008110
8111
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008112stats auth <user>:<passwd>
8113 Enable statistics with authentication and grant access to an account
8114 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008115 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008116 Arguments :
8117 <user> is a user name to grant access to
8118
8119 <passwd> is the cleartext password associated to this user
8120
8121 This statement enables statistics with default settings, and restricts access
8122 to declared users only. It may be repeated as many times as necessary to
8123 allow as many users as desired. When a user tries to access the statistics
8124 without a valid account, a "401 Forbidden" response will be returned so that
8125 the browser asks the user to provide a valid user and password. The real
8126 which will be returned to the browser is configurable using "stats realm".
8127
8128 Since the authentication method is HTTP Basic Authentication, the passwords
8129 circulate in cleartext on the network. Thus, it was decided that the
8130 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008131 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008132
8133 It is also possible to reduce the scope of the proxies which appear in the
8134 report using "stats scope".
8135
8136 Though this statement alone is enough to enable statistics reporting, it is
8137 recommended to set all other settings in order to avoid relying on default
8138 unobvious parameters.
8139
8140 Example :
8141 # public access (limited to this backend only)
8142 backend public_www
8143 server srv1 192.168.0.1:80
8144 stats enable
8145 stats hide-version
8146 stats scope .
8147 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008148 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008149 stats auth admin1:AdMiN123
8150 stats auth admin2:AdMiN321
8151
8152 # internal monitoring access (unlimited)
8153 backend private_monitoring
8154 stats enable
8155 stats uri /admin?stats
8156 stats refresh 5s
8157
8158 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8159
8160
8161stats enable
8162 Enable statistics reporting with default settings
8163 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008164 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008165 Arguments : none
8166
8167 This statement enables statistics reporting with default settings defined
8168 at build time. Unless stated otherwise, these settings are used :
8169 - stats uri : /haproxy?stats
8170 - stats realm : "HAProxy Statistics"
8171 - stats auth : no authentication
8172 - stats scope : no restriction
8173
8174 Though this statement alone is enough to enable statistics reporting, it is
8175 recommended to set all other settings in order to avoid relying on default
8176 unobvious parameters.
8177
8178 Example :
8179 # public access (limited to this backend only)
8180 backend public_www
8181 server srv1 192.168.0.1:80
8182 stats enable
8183 stats hide-version
8184 stats scope .
8185 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008186 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008187 stats auth admin1:AdMiN123
8188 stats auth admin2:AdMiN321
8189
8190 # internal monitoring access (unlimited)
8191 backend private_monitoring
8192 stats enable
8193 stats uri /admin?stats
8194 stats refresh 5s
8195
8196 See also : "stats auth", "stats realm", "stats uri"
8197
8198
Willy Tarreaud63335a2010-02-26 12:56:52 +01008199stats hide-version
8200 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008201 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008202 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008203 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008204
Willy Tarreaud63335a2010-02-26 12:56:52 +01008205 By default, the stats page reports some useful status information along with
8206 the statistics. Among them is HAProxy's version. However, it is generally
8207 considered dangerous to report precise version to anyone, as it can help them
8208 target known weaknesses with specific attacks. The "stats hide-version"
8209 statement removes the version from the statistics report. This is recommended
8210 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008211
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008212 Though this statement alone is enough to enable statistics reporting, it is
8213 recommended to set all other settings in order to avoid relying on default
8214 unobvious parameters.
8215
Willy Tarreaud63335a2010-02-26 12:56:52 +01008216 Example :
8217 # public access (limited to this backend only)
8218 backend public_www
8219 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008220 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008221 stats hide-version
8222 stats scope .
8223 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008224 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008225 stats auth admin1:AdMiN123
8226 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008227
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008228 # internal monitoring access (unlimited)
8229 backend private_monitoring
8230 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008231 stats uri /admin?stats
8232 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008233
Willy Tarreaud63335a2010-02-26 12:56:52 +01008234 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008235
Willy Tarreau983e01e2010-01-11 18:42:06 +01008236
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008237stats http-request { allow | deny | auth [realm <realm>] }
8238 [ { if | unless } <condition> ]
8239 Access control for statistics
8240
8241 May be used in sections: defaults | frontend | listen | backend
8242 no | no | yes | yes
8243
8244 As "http-request", these set of options allow to fine control access to
8245 statistics. Each option may be followed by if/unless and acl.
8246 First option with matched condition (or option without condition) is final.
8247 For "deny" a 403 error will be returned, for "allow" normal processing is
8248 performed, for "auth" a 401/407 error code is returned so the client
8249 should be asked to enter a username and password.
8250
8251 There is no fixed limit to the number of http-request statements per
8252 instance.
8253
8254 See also : "http-request", section 3.4 about userlists and section 7
8255 about ACL usage.
8256
8257
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008258stats realm <realm>
8259 Enable statistics and set authentication realm
8260 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008261 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008262 Arguments :
8263 <realm> is the name of the HTTP Basic Authentication realm reported to
8264 the browser. The browser uses it to display it in the pop-up
8265 inviting the user to enter a valid username and password.
8266
8267 The realm is read as a single word, so any spaces in it should be escaped
8268 using a backslash ('\').
8269
8270 This statement is useful only in conjunction with "stats auth" since it is
8271 only related to authentication.
8272
8273 Though this statement alone is enough to enable statistics reporting, it is
8274 recommended to set all other settings in order to avoid relying on default
8275 unobvious parameters.
8276
8277 Example :
8278 # public access (limited to this backend only)
8279 backend public_www
8280 server srv1 192.168.0.1:80
8281 stats enable
8282 stats hide-version
8283 stats scope .
8284 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008285 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008286 stats auth admin1:AdMiN123
8287 stats auth admin2:AdMiN321
8288
8289 # internal monitoring access (unlimited)
8290 backend private_monitoring
8291 stats enable
8292 stats uri /admin?stats
8293 stats refresh 5s
8294
8295 See also : "stats auth", "stats enable", "stats uri"
8296
8297
8298stats refresh <delay>
8299 Enable statistics with automatic refresh
8300 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008301 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008302 Arguments :
8303 <delay> is the suggested refresh delay, specified in seconds, which will
8304 be returned to the browser consulting the report page. While the
8305 browser is free to apply any delay, it will generally respect it
8306 and refresh the page this every seconds. The refresh interval may
8307 be specified in any other non-default time unit, by suffixing the
8308 unit after the value, as explained at the top of this document.
8309
8310 This statement is useful on monitoring displays with a permanent page
8311 reporting the load balancer's activity. When set, the HTML report page will
8312 include a link "refresh"/"stop refresh" so that the user can select whether
8313 he wants automatic refresh of the page or not.
8314
8315 Though this statement alone is enough to enable statistics reporting, it is
8316 recommended to set all other settings in order to avoid relying on default
8317 unobvious parameters.
8318
8319 Example :
8320 # public access (limited to this backend only)
8321 backend public_www
8322 server srv1 192.168.0.1:80
8323 stats enable
8324 stats hide-version
8325 stats scope .
8326 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008327 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008328 stats auth admin1:AdMiN123
8329 stats auth admin2:AdMiN321
8330
8331 # internal monitoring access (unlimited)
8332 backend private_monitoring
8333 stats enable
8334 stats uri /admin?stats
8335 stats refresh 5s
8336
8337 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8338
8339
8340stats scope { <name> | "." }
8341 Enable statistics and limit access scope
8342 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008343 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008344 Arguments :
8345 <name> is the name of a listen, frontend or backend section to be
8346 reported. The special name "." (a single dot) designates the
8347 section in which the statement appears.
8348
8349 When this statement is specified, only the sections enumerated with this
8350 statement will appear in the report. All other ones will be hidden. This
8351 statement may appear as many times as needed if multiple sections need to be
8352 reported. Please note that the name checking is performed as simple string
8353 comparisons, and that it is never checked that a give section name really
8354 exists.
8355
8356 Though this statement alone is enough to enable statistics reporting, it is
8357 recommended to set all other settings in order to avoid relying on default
8358 unobvious parameters.
8359
8360 Example :
8361 # public access (limited to this backend only)
8362 backend public_www
8363 server srv1 192.168.0.1:80
8364 stats enable
8365 stats hide-version
8366 stats scope .
8367 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008368 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008369 stats auth admin1:AdMiN123
8370 stats auth admin2:AdMiN321
8371
8372 # internal monitoring access (unlimited)
8373 backend private_monitoring
8374 stats enable
8375 stats uri /admin?stats
8376 stats refresh 5s
8377
8378 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8379
Willy Tarreaud63335a2010-02-26 12:56:52 +01008380
Willy Tarreauc9705a12010-07-27 20:05:50 +02008381stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008382 Enable reporting of a description on the statistics page.
8383 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008384 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008385
Willy Tarreauc9705a12010-07-27 20:05:50 +02008386 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008387 description from global section is automatically used instead.
8388
8389 This statement is useful for users that offer shared services to their
8390 customers, where node or description should be different for each customer.
8391
8392 Though this statement alone is enough to enable statistics reporting, it is
8393 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008394 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008395
8396 Example :
8397 # internal monitoring access (unlimited)
8398 backend private_monitoring
8399 stats enable
8400 stats show-desc Master node for Europe, Asia, Africa
8401 stats uri /admin?stats
8402 stats refresh 5s
8403
8404 See also: "show-node", "stats enable", "stats uri" and "description" in
8405 global section.
8406
8407
8408stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008409 Enable reporting additional information on the statistics page
8410 May be used in sections : defaults | frontend | listen | backend
8411 yes | yes | yes | yes
8412 Arguments : none
8413
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008414 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008415 - cap: capabilities (proxy)
8416 - mode: one of tcp, http or health (proxy)
8417 - id: SNMP ID (proxy, socket, server)
8418 - IP (socket, server)
8419 - cookie (backend, server)
8420
8421 Though this statement alone is enough to enable statistics reporting, it is
8422 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008423 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008424
8425 See also: "stats enable", "stats uri".
8426
8427
8428stats show-node [ <name> ]
8429 Enable reporting of a host name on the statistics page.
8430 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008431 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008432 Arguments:
8433 <name> is an optional name to be reported. If unspecified, the
8434 node name from global section is automatically used instead.
8435
8436 This statement is useful for users that offer shared services to their
8437 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008438 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008439
8440 Though this statement alone is enough to enable statistics reporting, it is
8441 recommended to set all other settings in order to avoid relying on default
8442 unobvious parameters.
8443
8444 Example:
8445 # internal monitoring access (unlimited)
8446 backend private_monitoring
8447 stats enable
8448 stats show-node Europe-1
8449 stats uri /admin?stats
8450 stats refresh 5s
8451
8452 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8453 section.
8454
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008455
8456stats uri <prefix>
8457 Enable statistics and define the URI prefix to access them
8458 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008459 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008460 Arguments :
8461 <prefix> is the prefix of any URI which will be redirected to stats. This
8462 prefix may contain a question mark ('?') to indicate part of a
8463 query string.
8464
8465 The statistics URI is intercepted on the relayed traffic, so it appears as a
8466 page within the normal application. It is strongly advised to ensure that the
8467 selected URI will never appear in the application, otherwise it will never be
8468 possible to reach it in the application.
8469
8470 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008471 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008472 It is generally a good idea to include a question mark in the URI so that
8473 intermediate proxies refrain from caching the results. Also, since any string
8474 beginning with the prefix will be accepted as a stats request, the question
8475 mark helps ensuring that no valid URI will begin with the same words.
8476
8477 It is sometimes very convenient to use "/" as the URI prefix, and put that
8478 statement in a "listen" instance of its own. That makes it easy to dedicate
8479 an address or a port to statistics only.
8480
8481 Though this statement alone is enough to enable statistics reporting, it is
8482 recommended to set all other settings in order to avoid relying on default
8483 unobvious parameters.
8484
8485 Example :
8486 # public access (limited to this backend only)
8487 backend public_www
8488 server srv1 192.168.0.1:80
8489 stats enable
8490 stats hide-version
8491 stats scope .
8492 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008493 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008494 stats auth admin1:AdMiN123
8495 stats auth admin2:AdMiN321
8496
8497 # internal monitoring access (unlimited)
8498 backend private_monitoring
8499 stats enable
8500 stats uri /admin?stats
8501 stats refresh 5s
8502
8503 See also : "stats auth", "stats enable", "stats realm"
8504
8505
Willy Tarreaud63335a2010-02-26 12:56:52 +01008506stick match <pattern> [table <table>] [{if | unless} <cond>]
8507 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008508 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008509 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008510
8511 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008512 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008513 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008514 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008515 stickiness table. This rule is mandatory.
8516
8517 <table> is an optional stickiness table name. If unspecified, the same
8518 backend's table is used. A stickiness table is declared using
8519 the "stick-table" statement.
8520
8521 <cond> is an optional matching condition. It makes it possible to match
8522 on a certain criterion only when other conditions are met (or
8523 not met). For instance, it could be used to match on a source IP
8524 address except when a request passes through a known proxy, in
8525 which case we'd match on a header containing that IP address.
8526
8527 Some protocols or applications require complex stickiness rules and cannot
8528 always simply rely on cookies nor hashing. The "stick match" statement
8529 describes a rule to extract the stickiness criterion from an incoming request
8530 or connection. See section 7 for a complete list of possible patterns and
8531 transformation rules.
8532
8533 The table has to be declared using the "stick-table" statement. It must be of
8534 a type compatible with the pattern. By default it is the one which is present
8535 in the same backend. It is possible to share a table with other backends by
8536 referencing it using the "table" keyword. If another table is referenced,
8537 the server's ID inside the backends are used. By default, all server IDs
8538 start at 1 in each backend, so the server ordering is enough. But in case of
8539 doubt, it is highly recommended to force server IDs using their "id" setting.
8540
8541 It is possible to restrict the conditions where a "stick match" statement
8542 will apply, using "if" or "unless" followed by a condition. See section 7 for
8543 ACL based conditions.
8544
8545 There is no limit on the number of "stick match" statements. The first that
8546 applies and matches will cause the request to be directed to the same server
8547 as was used for the request which created the entry. That way, multiple
8548 matches can be used as fallbacks.
8549
8550 The stick rules are checked after the persistence cookies, so they will not
8551 affect stickiness if a cookie has already been used to select a server. That
8552 way, it becomes very easy to insert cookies and match on IP addresses in
8553 order to maintain stickiness between HTTP and HTTPS.
8554
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008555 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8556 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008557 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008558
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008559 Example :
8560 # forward SMTP users to the same server they just used for POP in the
8561 # last 30 minutes
8562 backend pop
8563 mode tcp
8564 balance roundrobin
8565 stick store-request src
8566 stick-table type ip size 200k expire 30m
8567 server s1 192.168.1.1:110
8568 server s2 192.168.1.1:110
8569
8570 backend smtp
8571 mode tcp
8572 balance roundrobin
8573 stick match src table pop
8574 server s1 192.168.1.1:25
8575 server s2 192.168.1.1:25
8576
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008577 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008578 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008579
8580
8581stick on <pattern> [table <table>] [{if | unless} <condition>]
8582 Define a request pattern to associate a user to a server
8583 May be used in sections : defaults | frontend | listen | backend
8584 no | no | yes | yes
8585
8586 Note : This form is exactly equivalent to "stick match" followed by
8587 "stick store-request", all with the same arguments. Please refer
8588 to both keywords for details. It is only provided as a convenience
8589 for writing more maintainable configurations.
8590
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008591 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8592 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008593 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008594
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008595 Examples :
8596 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008597 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008598
8599 # ...is strictly equivalent to this one :
8600 stick match src table pop if !localhost
8601 stick store-request src table pop if !localhost
8602
8603
8604 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8605 # well as HTTP without cookie. Share the same table between both accesses.
8606 backend http
8607 mode http
8608 balance roundrobin
8609 stick on src table https
8610 cookie SRV insert indirect nocache
8611 server s1 192.168.1.1:80 cookie s1
8612 server s2 192.168.1.1:80 cookie s2
8613
8614 backend https
8615 mode tcp
8616 balance roundrobin
8617 stick-table type ip size 200k expire 30m
8618 stick on src
8619 server s1 192.168.1.1:443
8620 server s2 192.168.1.1:443
8621
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008622 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008623
8624
8625stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8626 Define a request pattern used to create an entry in a stickiness table
8627 May be used in sections : defaults | frontend | listen | backend
8628 no | no | yes | yes
8629
8630 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008631 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008632 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008633 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008634 server is selected.
8635
8636 <table> is an optional stickiness table name. If unspecified, the same
8637 backend's table is used. A stickiness table is declared using
8638 the "stick-table" statement.
8639
8640 <cond> is an optional storage condition. It makes it possible to store
8641 certain criteria only when some conditions are met (or not met).
8642 For instance, it could be used to store the source IP address
8643 except when the request passes through a known proxy, in which
8644 case we'd store a converted form of a header containing that IP
8645 address.
8646
8647 Some protocols or applications require complex stickiness rules and cannot
8648 always simply rely on cookies nor hashing. The "stick store-request" statement
8649 describes a rule to decide what to extract from the request and when to do
8650 it, in order to store it into a stickiness table for further requests to
8651 match it using the "stick match" statement. Obviously the extracted part must
8652 make sense and have a chance to be matched in a further request. Storing a
8653 client's IP address for instance often makes sense. Storing an ID found in a
8654 URL parameter also makes sense. Storing a source port will almost never make
8655 any sense because it will be randomly matched. See section 7 for a complete
8656 list of possible patterns and transformation rules.
8657
8658 The table has to be declared using the "stick-table" statement. It must be of
8659 a type compatible with the pattern. By default it is the one which is present
8660 in the same backend. It is possible to share a table with other backends by
8661 referencing it using the "table" keyword. If another table is referenced,
8662 the server's ID inside the backends are used. By default, all server IDs
8663 start at 1 in each backend, so the server ordering is enough. But in case of
8664 doubt, it is highly recommended to force server IDs using their "id" setting.
8665
8666 It is possible to restrict the conditions where a "stick store-request"
8667 statement will apply, using "if" or "unless" followed by a condition. This
8668 condition will be evaluated while parsing the request, so any criteria can be
8669 used. See section 7 for ACL based conditions.
8670
8671 There is no limit on the number of "stick store-request" statements, but
8672 there is a limit of 8 simultaneous stores per request or response. This
8673 makes it possible to store up to 8 criteria, all extracted from either the
8674 request or the response, regardless of the number of rules. Only the 8 first
8675 ones which match will be kept. Using this, it is possible to feed multiple
8676 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008677 another protocol or access method. Using multiple store-request rules with
8678 the same table is possible and may be used to find the best criterion to rely
8679 on, by arranging the rules by decreasing preference order. Only the first
8680 extracted criterion for a given table will be stored. All subsequent store-
8681 request rules referencing the same table will be skipped and their ACLs will
8682 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008683
8684 The "store-request" rules are evaluated once the server connection has been
8685 established, so that the table will contain the real server that processed
8686 the request.
8687
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008688 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8689 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008690 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008691
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008692 Example :
8693 # forward SMTP users to the same server they just used for POP in the
8694 # last 30 minutes
8695 backend pop
8696 mode tcp
8697 balance roundrobin
8698 stick store-request src
8699 stick-table type ip size 200k expire 30m
8700 server s1 192.168.1.1:110
8701 server s2 192.168.1.1:110
8702
8703 backend smtp
8704 mode tcp
8705 balance roundrobin
8706 stick match src table pop
8707 server s1 192.168.1.1:25
8708 server s2 192.168.1.1:25
8709
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008710 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008711 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008712
8713
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008714stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008715 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8716 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008717 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008718 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008719 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008720
8721 Arguments :
8722 ip a table declared with "type ip" will only store IPv4 addresses.
8723 This form is very compact (about 50 bytes per entry) and allows
8724 very fast entry lookup and stores with almost no overhead. This
8725 is mainly used to store client source IP addresses.
8726
David du Colombier9a6d3c92011-03-17 10:40:24 +01008727 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8728 This form is very compact (about 60 bytes per entry) and allows
8729 very fast entry lookup and stores with almost no overhead. This
8730 is mainly used to store client source IP addresses.
8731
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008732 integer a table declared with "type integer" will store 32bit integers
8733 which can represent a client identifier found in a request for
8734 instance.
8735
8736 string a table declared with "type string" will store substrings of up
8737 to <len> characters. If the string provided by the pattern
8738 extractor is larger than <len>, it will be truncated before
8739 being stored. During matching, at most <len> characters will be
8740 compared between the string in the table and the extracted
8741 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008742 to 32 characters.
8743
8744 binary a table declared with "type binary" will store binary blocks
8745 of <len> bytes. If the block provided by the pattern
8746 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008747 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008748 is shorter than <len>, it will be padded by 0. When not
8749 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008750
8751 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008752 "string" type table (See type "string" above). Or the number
8753 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008754 changing this parameter as memory usage will proportionally
8755 increase.
8756
8757 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008758 value directly impacts memory usage. Count approximately
8759 50 bytes per entry, plus the size of a string if any. The size
8760 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008761
8762 [nopurge] indicates that we refuse to purge older entries when the table
8763 is full. When not specified and the table is full when haproxy
8764 wants to store an entry in it, it will flush a few of the oldest
8765 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008766 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008767 be desirable to refuse new entries instead of purging the older
8768 ones. That may be the case when the amount of data to store is
8769 far above the hardware limits and we prefer not to offer access
8770 to new clients than to reject the ones already connected. When
8771 using this parameter, be sure to properly set the "expire"
8772 parameter (see below).
8773
Emeric Brunf099e792010-09-27 12:05:28 +02008774 <peersect> is the name of the peers section to use for replication. Entries
8775 which associate keys to server IDs are kept synchronized with
8776 the remote peers declared in this section. All entries are also
8777 automatically learned from the local peer (old process) during a
8778 soft restart.
8779
Willy Tarreau1abc6732015-05-01 19:21:02 +02008780 NOTE : each peers section may be referenced only by tables
8781 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008782
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008783 <expire> defines the maximum duration of an entry in the table since it
8784 was last created, refreshed or matched. The expiration delay is
8785 defined using the standard time format, similarly as the various
8786 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008787 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008788 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008789 be removed once full. Be sure not to use the "nopurge" parameter
8790 if not expiration delay is specified.
8791
Willy Tarreau08d5f982010-06-06 13:34:54 +02008792 <data_type> is used to store additional information in the stick-table. This
8793 may be used by ACLs in order to control various criteria related
8794 to the activity of the client matching the stick-table. For each
8795 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008796 that the additional data can fit. Several data types may be
8797 stored with an entry. Multiple data types may be specified after
8798 the "store" keyword, as a comma-separated list. Alternatively,
8799 it is possible to repeat the "store" keyword followed by one or
8800 several data types. Except for the "server_id" type which is
8801 automatically detected and enabled, all data types must be
8802 explicitly declared to be stored. If an ACL references a data
8803 type which is not stored, the ACL will simply not match. Some
8804 data types require an argument which must be passed just after
8805 the type between parenthesis. See below for the supported data
8806 types and their arguments.
8807
8808 The data types that can be stored with an entry are the following :
8809 - server_id : this is an integer which holds the numeric ID of the server a
8810 request was assigned to. It is used by the "stick match", "stick store",
8811 and "stick on" rules. It is automatically enabled when referenced.
8812
8813 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8814 integer which may be used for anything. Most of the time it will be used
8815 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008816 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008817
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008818 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8819 over a period. It is a positive 32-bit integer integer which may be used
8820 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008821 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008822 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008823 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008824
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008825 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8826 integer which may be used for anything. Most of the time it will be used
8827 to put a special tag on some entries, for instance to note that a
8828 specific behavior was detected and must be known for future matches.
8829
8830 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8831 over a period. It is a positive 32-bit integer integer which may be used
8832 for anything. Just like <gpc1>, it counts events, but instead of keeping
8833 a cumulative number, it maintains the rate at which the counter is
8834 incremented. Most of the time it will be used to measure the frequency of
8835 occurrence of certain events (e.g. requests to a specific URL).
8836
Willy Tarreauc9705a12010-07-27 20:05:50 +02008837 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8838 the absolute number of connections received from clients which matched
8839 this entry. It does not mean the connections were accepted, just that
8840 they were received.
8841
8842 - conn_cur : Current Connections. It is a positive 32-bit integer which
8843 stores the concurrent connection counts for the entry. It is incremented
8844 once an incoming connection matches the entry, and decremented once the
8845 connection leaves. That way it is possible to know at any time the exact
8846 number of concurrent connections for an entry.
8847
8848 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8849 integer parameter <period> which indicates in milliseconds the length
8850 of the period over which the average is measured. It reports the average
8851 incoming connection rate over that period, in connections per period. The
8852 result is an integer which can be matched using ACLs.
8853
8854 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8855 the absolute number of sessions received from clients which matched this
8856 entry. A session is a connection that was accepted by the layer 4 rules.
8857
8858 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8859 integer parameter <period> which indicates in milliseconds the length
8860 of the period over which the average is measured. It reports the average
8861 incoming session rate over that period, in sessions per period. The
8862 result is an integer which can be matched using ACLs.
8863
8864 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8865 counts the absolute number of HTTP requests received from clients which
8866 matched this entry. It does not matter whether they are valid requests or
8867 not. Note that this is different from sessions when keep-alive is used on
8868 the client side.
8869
8870 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8871 integer parameter <period> which indicates in milliseconds the length
8872 of the period over which the average is measured. It reports the average
8873 HTTP request rate over that period, in requests per period. The result is
8874 an integer which can be matched using ACLs. It does not matter whether
8875 they are valid requests or not. Note that this is different from sessions
8876 when keep-alive is used on the client side.
8877
8878 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8879 counts the absolute number of HTTP requests errors induced by clients
8880 which matched this entry. Errors are counted on invalid and truncated
8881 requests, as well as on denied or tarpitted requests, and on failed
8882 authentications. If the server responds with 4xx, then the request is
8883 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01008884 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02008885
8886 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8887 integer parameter <period> which indicates in milliseconds the length
8888 of the period over which the average is measured. It reports the average
8889 HTTP request error rate over that period, in requests per period (see
8890 http_err_cnt above for what is accounted as an error). The result is an
8891 integer which can be matched using ACLs.
8892
8893 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008894 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02008895 which matched this entry. Headers are included in the count. This may be
8896 used to limit abuse of upload features on photo or video servers.
8897
8898 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8899 integer parameter <period> which indicates in milliseconds the length
8900 of the period over which the average is measured. It reports the average
8901 incoming bytes rate over that period, in bytes per period. It may be used
8902 to detect users which upload too much and too fast. Warning: with large
8903 uploads, it is possible that the amount of uploaded data will be counted
8904 once upon termination, thus causing spikes in the average transfer speed
8905 instead of having a smooth one. This may partially be smoothed with
8906 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8907 recommended for better fairness.
8908
8909 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008910 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02008911 matched this entry. Headers are included in the count. This may be used
8912 to limit abuse of bots sucking the whole site.
8913
8914 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8915 an integer parameter <period> which indicates in milliseconds the length
8916 of the period over which the average is measured. It reports the average
8917 outgoing bytes rate over that period, in bytes per period. It may be used
8918 to detect users which download too much and too fast. Warning: with large
8919 transfers, it is possible that the amount of transferred data will be
8920 counted once upon termination, thus causing spikes in the average
8921 transfer speed instead of having a smooth one. This may partially be
8922 smoothed with "option contstats" though this is not perfect yet. Use of
8923 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008924
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008925 There is only one stick-table per proxy. At the moment of writing this doc,
8926 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008927 to be required, simply create a dummy backend with a stick-table in it and
8928 reference it.
8929
8930 It is important to understand that stickiness based on learning information
8931 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008932 lost upon restart unless peers are properly configured to transfer such
8933 information upon restart (recommended). In general it can be good as a
8934 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008935
Willy Tarreauc9705a12010-07-27 20:05:50 +02008936 Last, memory requirements may be important when storing many data types.
8937 Indeed, storing all indicators above at once in each entry requires 116 bytes
8938 per entry, or 116 MB for a 1-million entries table. This is definitely not
8939 something that can be ignored.
8940
8941 Example:
8942 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8943 # and store a general purpose counter and the average connection rate
8944 # computed over a sliding window of 30 seconds.
8945 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8946
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008947 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008948 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008949
8950
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008951stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008952 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008953 May be used in sections : defaults | frontend | listen | backend
8954 no | no | yes | yes
8955
8956 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008957 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008958 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008959 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008960 server is selected.
8961
8962 <table> is an optional stickiness table name. If unspecified, the same
8963 backend's table is used. A stickiness table is declared using
8964 the "stick-table" statement.
8965
8966 <cond> is an optional storage condition. It makes it possible to store
8967 certain criteria only when some conditions are met (or not met).
8968 For instance, it could be used to store the SSL session ID only
8969 when the response is a SSL server hello.
8970
8971 Some protocols or applications require complex stickiness rules and cannot
8972 always simply rely on cookies nor hashing. The "stick store-response"
8973 statement describes a rule to decide what to extract from the response and
8974 when to do it, in order to store it into a stickiness table for further
8975 requests to match it using the "stick match" statement. Obviously the
8976 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008977 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008978 See section 7 for a complete list of possible patterns and transformation
8979 rules.
8980
8981 The table has to be declared using the "stick-table" statement. It must be of
8982 a type compatible with the pattern. By default it is the one which is present
8983 in the same backend. It is possible to share a table with other backends by
8984 referencing it using the "table" keyword. If another table is referenced,
8985 the server's ID inside the backends are used. By default, all server IDs
8986 start at 1 in each backend, so the server ordering is enough. But in case of
8987 doubt, it is highly recommended to force server IDs using their "id" setting.
8988
8989 It is possible to restrict the conditions where a "stick store-response"
8990 statement will apply, using "if" or "unless" followed by a condition. This
8991 condition will be evaluated while parsing the response, so any criteria can
8992 be used. See section 7 for ACL based conditions.
8993
8994 There is no limit on the number of "stick store-response" statements, but
8995 there is a limit of 8 simultaneous stores per request or response. This
8996 makes it possible to store up to 8 criteria, all extracted from either the
8997 request or the response, regardless of the number of rules. Only the 8 first
8998 ones which match will be kept. Using this, it is possible to feed multiple
8999 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009000 another protocol or access method. Using multiple store-response rules with
9001 the same table is possible and may be used to find the best criterion to rely
9002 on, by arranging the rules by decreasing preference order. Only the first
9003 extracted criterion for a given table will be stored. All subsequent store-
9004 response rules referencing the same table will be skipped and their ACLs will
9005 not be evaluated. However, even if a store-request rule references a table, a
9006 store-response rule may also use the same table. This means that each table
9007 may learn exactly one element from the request and one element from the
9008 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009009
9010 The table will contain the real server that processed the request.
9011
9012 Example :
9013 # Learn SSL session ID from both request and response and create affinity.
9014 backend https
9015 mode tcp
9016 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009017 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009018 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009019
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009020 acl clienthello req_ssl_hello_type 1
9021 acl serverhello rep_ssl_hello_type 2
9022
9023 # use tcp content accepts to detects ssl client and server hello.
9024 tcp-request inspect-delay 5s
9025 tcp-request content accept if clienthello
9026
9027 # no timeout on response inspect delay by default.
9028 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009029
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009030 # SSL session ID (SSLID) may be present on a client or server hello.
9031 # Its length is coded on 1 byte at offset 43 and its value starts
9032 # at offset 44.
9033
9034 # Match and learn on request if client hello.
9035 stick on payload_lv(43,1) if clienthello
9036
9037 # Learn on response if server hello.
9038 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009039
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009040 server s1 192.168.1.1:443
9041 server s2 192.168.1.1:443
9042
9043 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9044 extraction.
9045
9046
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009047tcp-check connect [params*]
9048 Opens a new connection
9049 May be used in sections: defaults | frontend | listen | backend
9050 no | no | yes | yes
9051
9052 When an application lies on more than a single TCP port or when HAProxy
9053 load-balance many services in a single backend, it makes sense to probe all
9054 the services individually before considering a server as operational.
9055
9056 When there are no TCP port configured on the server line neither server port
9057 directive, then the 'tcp-check connect port <port>' must be the first step
9058 of the sequence.
9059
9060 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9061 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9062 do.
9063
9064 Parameters :
9065 They are optional and can be used to describe how HAProxy should open and
9066 use the TCP connection.
9067
9068 port if not set, check port or server port is used.
9069 It tells HAProxy where to open the connection to.
9070 <port> must be a valid TCP port source integer, from 1 to 65535.
9071
9072 send-proxy send a PROXY protocol string
9073
9074 ssl opens a ciphered connection
9075
9076 Examples:
9077 # check HTTP and HTTPs services on a server.
9078 # first open port 80 thanks to server line port directive, then
9079 # tcp-check opens port 443, ciphered and run a request on it:
9080 option tcp-check
9081 tcp-check connect
9082 tcp-check send GET\ /\ HTTP/1.0\r\n
9083 tcp-check send Host:\ haproxy.1wt.eu\r\n
9084 tcp-check send \r\n
9085 tcp-check expect rstring (2..|3..)
9086 tcp-check connect port 443 ssl
9087 tcp-check send GET\ /\ HTTP/1.0\r\n
9088 tcp-check send Host:\ haproxy.1wt.eu\r\n
9089 tcp-check send \r\n
9090 tcp-check expect rstring (2..|3..)
9091 server www 10.0.0.1 check port 80
9092
9093 # check both POP and IMAP from a single server:
9094 option tcp-check
9095 tcp-check connect port 110
9096 tcp-check expect string +OK\ POP3\ ready
9097 tcp-check connect port 143
9098 tcp-check expect string *\ OK\ IMAP4\ ready
9099 server mail 10.0.0.1 check
9100
9101 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9102
9103
9104tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009105 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009106 May be used in sections: defaults | frontend | listen | backend
9107 no | no | yes | yes
9108
9109 Arguments :
9110 <match> is a keyword indicating how to look for a specific pattern in the
9111 response. The keyword may be one of "string", "rstring" or
9112 binary.
9113 The keyword may be preceded by an exclamation mark ("!") to negate
9114 the match. Spaces are allowed between the exclamation mark and the
9115 keyword. See below for more details on the supported keywords.
9116
9117 <pattern> is the pattern to look for. It may be a string or a regular
9118 expression. If the pattern contains spaces, they must be escaped
9119 with the usual backslash ('\').
9120 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009121 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009122 two digits will represent a byte. The hexadecimal digits may be
9123 used upper or lower case.
9124
9125
9126 The available matches are intentionally similar to their http-check cousins :
9127
9128 string <string> : test the exact string matches in the response buffer.
9129 A health check response will be considered valid if the
9130 response's buffer contains this exact string. If the
9131 "string" keyword is prefixed with "!", then the response
9132 will be considered invalid if the body contains this
9133 string. This can be used to look for a mandatory pattern
9134 in a protocol response, or to detect a failure when a
9135 specific error appears in a protocol banner.
9136
9137 rstring <regex> : test a regular expression on the response buffer.
9138 A health check response will be considered valid if the
9139 response's buffer matches this expression. If the
9140 "rstring" keyword is prefixed with "!", then the response
9141 will be considered invalid if the body matches the
9142 expression.
9143
9144 binary <hexstring> : test the exact string in its hexadecimal form matches
9145 in the response buffer. A health check response will
9146 be considered valid if the response's buffer contains
9147 this exact hexadecimal string.
9148 Purpose is to match data on binary protocols.
9149
9150 It is important to note that the responses will be limited to a certain size
9151 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9152 Thus, too large responses may not contain the mandatory pattern when using
9153 "string", "rstring" or binary. If a large response is absolutely required, it
9154 is possible to change the default max size by setting the global variable.
9155 However, it is worth keeping in mind that parsing very large responses can
9156 waste some CPU cycles, especially when regular expressions are used, and that
9157 it is always better to focus the checks on smaller resources. Also, in its
9158 current state, the check will not find any string nor regex past a null
9159 character in the response. Similarly it is not possible to request matching
9160 the null character.
9161
9162 Examples :
9163 # perform a POP check
9164 option tcp-check
9165 tcp-check expect string +OK\ POP3\ ready
9166
9167 # perform an IMAP check
9168 option tcp-check
9169 tcp-check expect string *\ OK\ IMAP4\ ready
9170
9171 # look for the redis master server
9172 option tcp-check
9173 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009174 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009175 tcp-check send info\ replication\r\n
9176 tcp-check expect string role:master
9177 tcp-check send QUIT\r\n
9178 tcp-check expect string +OK
9179
9180
9181 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9182 "tcp-check send-binary", "http-check expect", tune.chksize
9183
9184
9185tcp-check send <data>
9186 Specify a string to be sent as a question during a generic health check
9187 May be used in sections: defaults | frontend | listen | backend
9188 no | no | yes | yes
9189
9190 <data> : the data to be sent as a question during a generic health check
9191 session. For now, <data> must be a string.
9192
9193 Examples :
9194 # look for the redis master server
9195 option tcp-check
9196 tcp-check send info\ replication\r\n
9197 tcp-check expect string role:master
9198
9199 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9200 "tcp-check send-binary", tune.chksize
9201
9202
Davor Ocelice9ed2812017-12-25 17:49:28 +01009203tcp-check send-binary <hexstring>
9204 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009205 tcp health check
9206 May be used in sections: defaults | frontend | listen | backend
9207 no | no | yes | yes
9208
9209 <data> : the data to be sent as a question during a generic health check
9210 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009211 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009212 response buffer. A health check response will be considered
9213 valid if the response's buffer contains this exact
9214 hexadecimal string.
9215 Purpose is to send binary data to ask on binary protocols.
9216
9217 Examples :
9218 # redis check in binary
9219 option tcp-check
9220 tcp-check send-binary 50494e470d0a # PING\r\n
9221 tcp-check expect binary 2b504F4e47 # +PONG
9222
9223
9224 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9225 "tcp-check send", tune.chksize
9226
9227
Willy Tarreaue9656522010-08-17 15:40:09 +02009228tcp-request connection <action> [{if | unless} <condition>]
9229 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009230 May be used in sections : defaults | frontend | listen | backend
9231 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009232 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009233 <action> defines the action to perform if the condition applies. See
9234 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009235
Willy Tarreaue9656522010-08-17 15:40:09 +02009236 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009237
9238 Immediately after acceptance of a new incoming connection, it is possible to
9239 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009240 or dropped or have its counters tracked. Those conditions cannot make use of
9241 any data contents because the connection has not been read from yet, and the
9242 buffers are not yet allocated. This is used to selectively and very quickly
9243 accept or drop connections from various sources with a very low overhead. If
9244 some contents need to be inspected in order to take the decision, the
9245 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009246
Willy Tarreaue9656522010-08-17 15:40:09 +02009247 The "tcp-request connection" rules are evaluated in their exact declaration
9248 order. If no rule matches or if there is no rule, the default action is to
9249 accept the incoming connection. There is no specific limit to the number of
9250 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009251
Willy Tarreaua9083d02015-05-08 15:27:59 +02009252 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009253 - accept :
9254 accepts the connection if the condition is true (when used with "if")
9255 or false (when used with "unless"). The first such rule executed ends
9256 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009257
Willy Tarreaue9656522010-08-17 15:40:09 +02009258 - reject :
9259 rejects the connection if the condition is true (when used with "if")
9260 or false (when used with "unless"). The first such rule executed ends
9261 the rules evaluation. Rejected connections do not even become a
9262 session, which is why they are accounted separately for in the stats,
9263 as "denied connections". They are not considered for the session
9264 rate-limit and are not logged either. The reason is that these rules
9265 should only be used to filter extremely high connection rates such as
9266 the ones encountered during a massive DDoS attack. Under these extreme
9267 conditions, the simple action of logging each event would make the
9268 system collapse and would considerably lower the filtering capacity. If
9269 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009270 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009271
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009272 - expect-proxy layer4 :
9273 configures the client-facing connection to receive a PROXY protocol
9274 header before any byte is read from the socket. This is equivalent to
9275 having the "accept-proxy" keyword on the "bind" line, except that using
9276 the TCP rule allows the PROXY protocol to be accepted only for certain
9277 IP address ranges using an ACL. This is convenient when multiple layers
9278 of load balancers are passed through by traffic coming from public
9279 hosts.
9280
Bertrand Jacquin90759682016-06-06 15:35:39 +01009281 - expect-netscaler-cip layer4 :
9282 configures the client-facing connection to receive a NetScaler Client
9283 IP insertion protocol header before any byte is read from the socket.
9284 This is equivalent to having the "accept-netscaler-cip" keyword on the
9285 "bind" line, except that using the TCP rule allows the PROXY protocol
9286 to be accepted only for certain IP address ranges using an ACL. This
9287 is convenient when multiple layers of load balancers are passed
9288 through by traffic coming from public hosts.
9289
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009290 - capture <sample> len <length> :
9291 This only applies to "tcp-request content" rules. It captures sample
9292 expression <sample> from the request buffer, and converts it to a
9293 string of at most <len> characters. The resulting string is stored into
9294 the next request "capture" slot, so it will possibly appear next to
9295 some captured HTTP headers. It will then automatically appear in the
9296 logs, and it will be possible to extract it using sample fetch rules to
9297 feed it into headers or anything. The length should be limited given
9298 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009299 session life. Please check section 7.3 (Fetching samples) and "capture
9300 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009301
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009302 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009303 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009304 rules do not stop evaluation and do not change default action. The
9305 number of counters that may be simultaneously tracked by the same
9306 connection is set in MAX_SESS_STKCTR at build time (reported in
9307 haproxy -vv) whichs defaults to 3, so the track-sc number is between 0
9308 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9309 tracking of the counters of the specified table as the first set. The
9310 first "track-sc1" rule executed enables tracking of the counters of the
9311 specified table as the second set. The first "track-sc2" rule executed
9312 enables tracking of the counters of the specified table as the third
9313 set. It is a recommended practice to use the first set of counters for
9314 the per-frontend counters and the second set for the per-backend ones.
9315 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009316
Willy Tarreaue9656522010-08-17 15:40:09 +02009317 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009318 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009319 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009320 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009321 and used to select which table entry to update the counters.
9322 Note that "tcp-request connection" cannot use content-based
9323 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009324
Willy Tarreaue9656522010-08-17 15:40:09 +02009325 <table> is an optional table to be used instead of the default one,
9326 which is the stick-table declared in the current proxy. All
9327 the counters for the matches and updates for the key will
9328 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009329
Willy Tarreaue9656522010-08-17 15:40:09 +02009330 Once a "track-sc*" rule is executed, the key is looked up in the table
9331 and if it is not found, an entry is allocated for it. Then a pointer to
9332 that entry is kept during all the session's life, and this entry's
9333 counters are updated as often as possible, every time the session's
9334 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009335 Counters are only updated for events that happen after the tracking has
9336 been started. For example, connection counters will not be updated when
9337 tracking layer 7 information, since the connection event happens before
9338 layer7 information is extracted.
9339
Willy Tarreaue9656522010-08-17 15:40:09 +02009340 If the entry tracks concurrent connection counters, one connection is
9341 counted for as long as the entry is tracked, and the entry will not
9342 expire during that time. Tracking counters also provides a performance
9343 advantage over just checking the keys, because only one table lookup is
9344 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009345
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009346 - sc-inc-gpc0(<sc-id>):
9347 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9348 counter designated by <sc-id>. If an error occurs, this action silently
9349 fails and the actions evaluation continues.
9350
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009351 - sc-inc-gpc1(<sc-id>):
9352 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9353 counter designated by <sc-id>. If an error occurs, this action silently
9354 fails and the actions evaluation continues.
9355
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009356 - sc-set-gpt0(<sc-id>) <int>:
9357 This action sets the GPT0 tag according to the sticky counter designated
9358 by <sc-id> and the value of <int>. The expected result is a boolean. If
9359 an error occurs, this action silently fails and the actions evaluation
9360 continues.
9361
William Lallemand2e785f22016-05-25 01:48:42 +02009362 - set-src <expr> :
9363 Is used to set the source IP address to the value of specified
9364 expression. Useful if you want to mask source IP for privacy.
9365 If you want to provide an IP from a HTTP header use "http-request
9366 set-src"
9367
9368 <expr> Is a standard HAProxy expression formed by a sample-fetch
9369 followed by some converters.
9370
9371 Example:
9372
9373 tcp-request connection set-src src,ipmask(24)
9374
Willy Tarreau0c630532016-10-21 17:52:58 +02009375 When possible, set-src preserves the original source port as long as the
9376 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009377
William Lallemand44be6402016-05-25 01:51:35 +02009378 - set-src-port <expr> :
9379 Is used to set the source port address to the value of specified
9380 expression.
9381
9382 <expr> Is a standard HAProxy expression formed by a sample-fetch
9383 followed by some converters.
9384
9385 Example:
9386
9387 tcp-request connection set-src-port int(4000)
9388
Willy Tarreau0c630532016-10-21 17:52:58 +02009389 When possible, set-src-port preserves the original source address as long
9390 as the address family supports a port, otherwise it forces the source
9391 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009392
William Lallemand13e9b0c2016-05-25 02:34:07 +02009393 - set-dst <expr> :
9394 Is used to set the destination IP address to the value of specified
9395 expression. Useful if you want to mask IP for privacy in log.
9396 If you want to provide an IP from a HTTP header use "http-request
9397 set-dst". If you want to connect to the new address/port, use
9398 '0.0.0.0:0' as a server address in the backend.
9399
9400 <expr> Is a standard HAProxy expression formed by a sample-fetch
9401 followed by some converters.
9402
9403 Example:
9404
9405 tcp-request connection set-dst dst,ipmask(24)
9406 tcp-request connection set-dst ipv4(10.0.0.1)
9407
Willy Tarreau0c630532016-10-21 17:52:58 +02009408 When possible, set-dst preserves the original destination port as long as
9409 the address family allows it, otherwise the destination port is set to 0.
9410
William Lallemand13e9b0c2016-05-25 02:34:07 +02009411 - set-dst-port <expr> :
9412 Is used to set the destination port address to the value of specified
9413 expression. If you want to connect to the new address/port, use
9414 '0.0.0.0:0' as a server address in the backend.
9415
9416
9417 <expr> Is a standard HAProxy expression formed by a sample-fetch
9418 followed by some converters.
9419
9420 Example:
9421
9422 tcp-request connection set-dst-port int(4000)
9423
Willy Tarreau0c630532016-10-21 17:52:58 +02009424 When possible, set-dst-port preserves the original destination address as
9425 long as the address family supports a port, otherwise it forces the
9426 destination address to IPv4 "0.0.0.0" before rewriting the port.
9427
Willy Tarreau2d392c22015-08-24 01:43:45 +02009428 - "silent-drop" :
9429 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009430 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009431 to prevent the client from being notified. The effect it then that the
9432 client still sees an established connection while there's none on
9433 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9434 except that it doesn't use any local resource at all on the machine
9435 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009436 slow down stronger attackers. It is important to understand the impact
9437 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009438 client and HAProxy (firewalls, proxies, load balancers) will also keep
9439 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009440 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009441 TCP_REPAIR socket option is used to block the emission of a TCP
9442 reset. On other systems, the socket's TTL is reduced to 1 so that the
9443 TCP reset doesn't pass the first router, though it's still delivered to
9444 local networks. Do not use it unless you fully understand how it works.
9445
Willy Tarreaue9656522010-08-17 15:40:09 +02009446 Note that the "if/unless" condition is optional. If no condition is set on
9447 the action, it is simply performed unconditionally. That can be useful for
9448 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009449
Willy Tarreaue9656522010-08-17 15:40:09 +02009450 Example: accept all connections from white-listed hosts, reject too fast
9451 connection without counting them, and track accepted connections.
9452 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009453
Willy Tarreaue9656522010-08-17 15:40:09 +02009454 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009455 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009456 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009457
Willy Tarreaue9656522010-08-17 15:40:09 +02009458 Example: accept all connections from white-listed hosts, count all other
9459 connections and reject too fast ones. This results in abusive ones
9460 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009461
Willy Tarreaue9656522010-08-17 15:40:09 +02009462 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009463 tcp-request connection track-sc0 src
9464 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009465
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009466 Example: enable the PROXY protocol for traffic coming from all known proxies.
9467
9468 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9469
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009470 See section 7 about ACL usage.
9471
Willy Tarreau4f614292016-10-21 17:49:36 +02009472 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009473
9474
Willy Tarreaue9656522010-08-17 15:40:09 +02009475tcp-request content <action> [{if | unless} <condition>]
9476 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009477 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009478 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009479 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009480 <action> defines the action to perform if the condition applies. See
9481 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009482
Willy Tarreaue9656522010-08-17 15:40:09 +02009483 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009484
Davor Ocelice9ed2812017-12-25 17:49:28 +01009485 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009486 called "TCP content inspection". During this stage, ACL-based rules are
9487 evaluated every time the request contents are updated, until either an
9488 "accept" or a "reject" rule matches, or the TCP request inspection delay
9489 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009490
Willy Tarreaue9656522010-08-17 15:40:09 +02009491 The first difference between these rules and "tcp-request connection" rules
9492 is that "tcp-request content" rules can make use of contents to take a
9493 decision. Most often, these decisions will consider a protocol recognition or
9494 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009495 both frontends and backends. In case of HTTP keep-alive with the client, all
9496 tcp-request content rules are evaluated again, so haproxy keeps a record of
9497 what sticky counters were assigned by a "tcp-request connection" versus a
9498 "tcp-request content" rule, and flushes all the content-related ones after
9499 processing an HTTP request, so that they may be evaluated again by the rules
9500 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009501 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009502 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009503
Willy Tarreaue9656522010-08-17 15:40:09 +02009504 Content-based rules are evaluated in their exact declaration order. If no
9505 rule matches or if there is no rule, the default action is to accept the
9506 contents. There is no specific limit to the number of rules which may be
9507 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009508
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009509 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009510 - accept : the request is accepted
9511 - reject : the request is rejected and the connection is closed
9512 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009513 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009514 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009515 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009516 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009517 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009518 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009519 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009520 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009521 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009522
Willy Tarreaue9656522010-08-17 15:40:09 +02009523 They have the same meaning as their counter-parts in "tcp-request connection"
9524 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009525
Willy Tarreauf3338342014-01-28 21:40:28 +01009526 While there is nothing mandatory about it, it is recommended to use the
9527 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9528 content" rules in the frontend, and track-sc2 for "tcp-request content"
9529 rules in the backend, because that makes the configuration more readable
9530 and easier to troubleshoot, but this is just a guideline and all counters
9531 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009532
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009533 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009534 the action, it is simply performed unconditionally. That can be useful for
9535 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009536
Willy Tarreaue9656522010-08-17 15:40:09 +02009537 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009538 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9539 contents of a buffer before extracting the required data. If the buffered
9540 contents do not parse as a valid HTTP message, then the ACL does not match.
9541 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009542 processing, so there is no risk of parsing something differently. In an HTTP
9543 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9544 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009545
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009546 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009547 are present when the rule is processed. The rule processing engine is able to
9548 wait until the inspect delay expires when the data to be tracked is not yet
9549 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009550
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009551 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009552 declared inline. For "tcp-request session" rules, only session-level
9553 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009554
Daniel Schneller0b547052016-03-21 20:46:57 +01009555 <var-name> The name of the variable starts with an indication about
9556 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009557 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009558 "sess" : the variable is shared with the whole session
9559 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009560 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009561 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009562 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009563 "res" : the variable is shared only during response
9564 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009565 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009566 The name may only contain characters 'a-z', 'A-Z', '0-9',
9567 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009568
9569 <expr> Is a standard HAProxy expression formed by a sample-fetch
9570 followed by some converters.
9571
Christopher Faulet85d79c92016-11-09 16:54:56 +01009572 The "unset-var" is used to unset a variable. See above for details about
9573 <var-name>.
9574
Patrick Hemmer268a7072018-05-11 12:52:31 -04009575 The "set-priority-class" is used to set the queue priority class of the
9576 current request. The value must be a sample expression which converts to an
9577 integer in the range -2047..2047. Results outside this range will be
9578 truncated. The priority class determines the order in which queued requests
9579 are processed. Lower values have higher priority.
9580
9581 The "set-priority-offset" is used to set the queue priority timestamp offset
9582 of the current request. The value must be a sample expression which converts
9583 to an integer in the range -524287..524287. Results outside this range will be
9584 truncated. When a request is queued, it is ordered first by the priority
9585 class, then by the current timestamp adjusted by the given offset in
9586 milliseconds. Lower values have higher priority.
9587 Note that the resulting timestamp is is only tracked with enough precision for
9588 524,287ms (8m44s287ms). If the request is queued long enough to where the
9589 adjusted timestamp exceeds this value, it will be misidentified as highest
9590 priority. Thus it is important to set "timeout queue" to a value, where when
9591 combined with the offset, does not exceed this limit.
9592
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009593 The "send-spoe-group" is used to trigger sending of a group of SPOE
9594 messages. To do so, the SPOE engine used to send messages must be defined, as
9595 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9596 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9597 the SPOE agent name must be used.
9598
9599 <engine-name> The SPOE engine name.
9600
9601 <group-name> The SPOE group name as specified in the engine configuration.
9602
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009603 Example:
9604
9605 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009606 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009607
Willy Tarreau62644772008-07-16 18:36:06 +02009608 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009609 # Accept HTTP requests containing a Host header saying "example.com"
9610 # and reject everything else.
9611 acl is_host_com hdr(Host) -i example.com
9612 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009613 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009614 tcp-request content reject
9615
9616 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009617 # reject SMTP connection if client speaks first
9618 tcp-request inspect-delay 30s
9619 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009620 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009621
9622 # Forward HTTPS connection only if client speaks
9623 tcp-request inspect-delay 30s
9624 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009625 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009626 tcp-request content reject
9627
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009628 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009629 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009630 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009631 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009632 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9633 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009634
9635 Example:
9636 # track request counts per "base" (concatenation of Host+URL)
9637 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009638 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009639
Willy Tarreaue9656522010-08-17 15:40:09 +02009640 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009641 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009642
9643 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009644 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009645 # protecting all our sites
9646 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009647 tcp-request connection track-sc0 src
9648 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009649 ...
9650 use_backend http_dynamic if { path_end .php }
9651
9652 backend http_dynamic
9653 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009654 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009655 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009656 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009657 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009658 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009659 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009660
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009661 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009662
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009663 See also : "tcp-request connection", "tcp-request session",
9664 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009665
9666
9667tcp-request inspect-delay <timeout>
9668 Set the maximum allowed time to wait for data during content inspection
9669 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009670 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009671 Arguments :
9672 <timeout> is the timeout value specified in milliseconds by default, but
9673 can be in any other unit if the number is suffixed by the unit,
9674 as explained at the top of this document.
9675
9676 People using haproxy primarily as a TCP relay are often worried about the
9677 risk of passing any type of protocol to a server without any analysis. In
9678 order to be able to analyze the request contents, we must first withhold
9679 the data then analyze them. This statement simply enables withholding of
9680 data for at most the specified amount of time.
9681
Willy Tarreaufb356202010-08-03 14:02:05 +02009682 TCP content inspection applies very early when a connection reaches a
9683 frontend, then very early when the connection is forwarded to a backend. This
9684 means that a connection may experience a first delay in the frontend and a
9685 second delay in the backend if both have tcp-request rules.
9686
Willy Tarreau62644772008-07-16 18:36:06 +02009687 Note that when performing content inspection, haproxy will evaluate the whole
9688 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009689 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009690 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009691 contents are definitive. If no delay is set, haproxy will not wait at all
9692 and will immediately apply a verdict based on the available information.
9693 Obviously this is unlikely to be very useful and might even be racy, so such
9694 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009695
9696 As soon as a rule matches, the request is released and continues as usual. If
9697 the timeout is reached and no rule matches, the default policy will be to let
9698 it pass through unaffected.
9699
9700 For most protocols, it is enough to set it to a few seconds, as most clients
9701 send the full request immediately upon connection. Add 3 or more seconds to
9702 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009703 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009704 before the server (e.g. SMTP), or to wait for a client to talk before passing
9705 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009706 least the inspection delay, otherwise it will expire first. If the client
9707 closes the connection or if the buffer is full, the delay immediately expires
9708 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009709
Willy Tarreau55165fe2009-05-10 12:02:55 +02009710 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009711 "timeout client".
9712
9713
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009714tcp-response content <action> [{if | unless} <condition>]
9715 Perform an action on a session response depending on a layer 4-7 condition
9716 May be used in sections : defaults | frontend | listen | backend
9717 no | no | yes | yes
9718 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009719 <action> defines the action to perform if the condition applies. See
9720 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009721
9722 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9723
Davor Ocelice9ed2812017-12-25 17:49:28 +01009724 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009725 called "TCP content inspection". During this stage, ACL-based rules are
9726 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009727 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9728 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009729
9730 Most often, these decisions will consider a protocol recognition or validity.
9731
9732 Content-based rules are evaluated in their exact declaration order. If no
9733 rule matches or if there is no rule, the default action is to accept the
9734 contents. There is no specific limit to the number of rules which may be
9735 inserted.
9736
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009737 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009738 - accept :
9739 accepts the response if the condition is true (when used with "if")
9740 or false (when used with "unless"). The first such rule executed ends
9741 the rules evaluation.
9742
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009743 - close :
9744 immediately closes the connection with the server if the condition is
9745 true (when used with "if"), or false (when used with "unless"). The
9746 first such rule executed ends the rules evaluation. The main purpose of
9747 this action is to force a connection to be finished between a client
9748 and a server after an exchange when the application protocol expects
9749 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009750 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009751 protocols.
9752
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009753 - reject :
9754 rejects the response if the condition is true (when used with "if")
9755 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009756 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009757
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009758 - set-var(<var-name>) <expr>
9759 Sets a variable.
9760
Christopher Faulet85d79c92016-11-09 16:54:56 +01009761 - unset-var(<var-name>)
9762 Unsets a variable.
9763
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009764 - sc-inc-gpc0(<sc-id>):
9765 This action increments the GPC0 counter according to the sticky
9766 counter designated by <sc-id>. If an error occurs, this action fails
9767 silently and the actions evaluation continues.
9768
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009769 - sc-inc-gpc1(<sc-id>):
9770 This action increments the GPC1 counter according to the sticky
9771 counter designated by <sc-id>. If an error occurs, this action fails
9772 silently and the actions evaluation continues.
9773
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009774 - sc-set-gpt0(<sc-id>) <int> :
9775 This action sets the GPT0 tag according to the sticky counter designated
9776 by <sc-id> and the value of <int>. The expected result is a boolean. If
9777 an error occurs, this action silently fails and the actions evaluation
9778 continues.
9779
Willy Tarreau2d392c22015-08-24 01:43:45 +02009780 - "silent-drop" :
9781 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009782 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009783 to prevent the client from being notified. The effect it then that the
9784 client still sees an established connection while there's none on
9785 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9786 except that it doesn't use any local resource at all on the machine
9787 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009788 slow down stronger attackers. It is important to understand the impact
9789 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009790 client and HAProxy (firewalls, proxies, load balancers) will also keep
9791 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009792 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009793 TCP_REPAIR socket option is used to block the emission of a TCP
9794 reset. On other systems, the socket's TTL is reduced to 1 so that the
9795 TCP reset doesn't pass the first router, though it's still delivered to
9796 local networks. Do not use it unless you fully understand how it works.
9797
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009798 - send-spoe-group <engine-name> <group-name>
9799 Send a group of SPOE messages.
9800
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009801 Note that the "if/unless" condition is optional. If no condition is set on
9802 the action, it is simply performed unconditionally. That can be useful for
9803 for changing the default action to a reject.
9804
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009805 It is perfectly possible to match layer 7 contents with "tcp-response
9806 content" rules, but then it is important to ensure that a full response has
9807 been buffered, otherwise no contents will match. In order to achieve this,
9808 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009809 period.
9810
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009811 The "set-var" is used to set the content of a variable. The variable is
9812 declared inline.
9813
Daniel Schneller0b547052016-03-21 20:46:57 +01009814 <var-name> The name of the variable starts with an indication about
9815 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009816 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009817 "sess" : the variable is shared with the whole session
9818 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009819 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009820 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009821 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009822 "res" : the variable is shared only during response
9823 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009824 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009825 The name may only contain characters 'a-z', 'A-Z', '0-9',
9826 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009827
9828 <expr> Is a standard HAProxy expression formed by a sample-fetch
9829 followed by some converters.
9830
9831 Example:
9832
9833 tcp-request content set-var(sess.my_var) src
9834
Christopher Faulet85d79c92016-11-09 16:54:56 +01009835 The "unset-var" is used to unset a variable. See above for details about
9836 <var-name>.
9837
9838 Example:
9839
9840 tcp-request content unset-var(sess.my_var)
9841
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009842 The "send-spoe-group" is used to trigger sending of a group of SPOE
9843 messages. To do so, the SPOE engine used to send messages must be defined, as
9844 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9845 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9846 the SPOE agent name must be used.
9847
9848 <engine-name> The SPOE engine name.
9849
9850 <group-name> The SPOE group name as specified in the engine configuration.
9851
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009852 See section 7 about ACL usage.
9853
9854 See also : "tcp-request content", "tcp-response inspect-delay"
9855
9856
Willy Tarreau4f614292016-10-21 17:49:36 +02009857tcp-request session <action> [{if | unless} <condition>]
9858 Perform an action on a validated session depending on a layer 5 condition
9859 May be used in sections : defaults | frontend | listen | backend
9860 no | yes | yes | no
9861 Arguments :
9862 <action> defines the action to perform if the condition applies. See
9863 below.
9864
9865 <condition> is a standard layer5-only ACL-based condition (see section 7).
9866
Davor Ocelice9ed2812017-12-25 17:49:28 +01009867 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +02009868 it is possible to evaluate some conditions to decide whether this session
9869 must be accepted or dropped or have its counters tracked. Those conditions
9870 cannot make use of any data contents because no buffers are allocated yet and
9871 the processing cannot wait at this stage. The main use case it to copy some
9872 early information into variables (since variables are accessible in the
9873 session), or to keep track of some information collected after the handshake,
9874 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +01009875 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +02009876 extracted information can thus be copied to a variable or tracked using
9877 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9878 with other rulesets. Most operations performed here could also be performed
9879 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9880 for each new request, and that might not always be acceptable. For example a
9881 rule might increment a counter on each evaluation. It would also be possible
9882 that a country is resolved by geolocation from the source IP address,
9883 assigned to a session-wide variable, then the source address rewritten from
9884 an HTTP header for all requests. If some contents need to be inspected in
9885 order to take the decision, the "tcp-request content" statements must be used
9886 instead.
9887
9888 The "tcp-request session" rules are evaluated in their exact declaration
9889 order. If no rule matches or if there is no rule, the default action is to
9890 accept the incoming session. There is no specific limit to the number of
9891 rules which may be inserted.
9892
9893 Several types of actions are supported :
9894 - accept : the request is accepted
9895 - reject : the request is rejected and the connection is closed
9896 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9897 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009898 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009899 - sc-set-gpt0(<sc-id>) <int>
9900 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009901 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009902 - silent-drop
9903
9904 These actions have the same meaning as their respective counter-parts in
9905 "tcp-request connection" and "tcp-request content", so please refer to these
9906 sections for a complete description.
9907
9908 Note that the "if/unless" condition is optional. If no condition is set on
9909 the action, it is simply performed unconditionally. That can be useful for
9910 "track-sc*" actions as well as for changing the default action to a reject.
9911
9912 Example: track the original source address by default, or the one advertised
9913 in the PROXY protocol header for connection coming from the local
9914 proxies. The first connection-level rule enables receipt of the
9915 PROXY protocol for these ones, the second rule tracks whatever
9916 address we decide to keep after optional decoding.
9917
9918 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9919 tcp-request session track-sc0 src
9920
9921 Example: accept all sessions from white-listed hosts, reject too fast
9922 sessions without counting them, and track accepted sessions.
9923 This results in session rate being capped from abusive sources.
9924
9925 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9926 tcp-request session reject if { src_sess_rate gt 10 }
9927 tcp-request session track-sc0 src
9928
9929 Example: accept all sessions from white-listed hosts, count all other
9930 sessions and reject too fast ones. This results in abusive ones
9931 being blocked as long as they don't slow down.
9932
9933 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9934 tcp-request session track-sc0 src
9935 tcp-request session reject if { sc0_sess_rate gt 10 }
9936
9937 See section 7 about ACL usage.
9938
9939 See also : "tcp-request connection", "tcp-request content", "stick-table"
9940
9941
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009942tcp-response inspect-delay <timeout>
9943 Set the maximum allowed time to wait for a response during content inspection
9944 May be used in sections : defaults | frontend | listen | backend
9945 no | no | yes | yes
9946 Arguments :
9947 <timeout> is the timeout value specified in milliseconds by default, but
9948 can be in any other unit if the number is suffixed by the unit,
9949 as explained at the top of this document.
9950
9951 See also : "tcp-response content", "tcp-request inspect-delay".
9952
9953
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009954timeout check <timeout>
9955 Set additional check timeout, but only after a connection has been already
9956 established.
9957
9958 May be used in sections: defaults | frontend | listen | backend
9959 yes | no | yes | yes
9960 Arguments:
9961 <timeout> is the timeout value specified in milliseconds by default, but
9962 can be in any other unit if the number is suffixed by the unit,
9963 as explained at the top of this document.
9964
9965 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
9966 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009967 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009968 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01009969 (Please also note that there is no valid reason to have such long connect
9970 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
9971 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009972
9973 If "timeout check" is not set haproxy uses "inter" for complete check
9974 timeout (connect + read) exactly like all <1.3.15 version.
9975
9976 In most cases check request is much simpler and faster to handle than normal
9977 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01009978 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009979
9980 This parameter is specific to backends, but can be specified once for all in
9981 "defaults" sections. This is in fact one of the easiest solutions not to
9982 forget about it.
9983
Willy Tarreau41a340d2008-01-22 12:25:31 +01009984 See also: "timeout connect", "timeout queue", "timeout server",
9985 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009986
9987
Willy Tarreau0ba27502007-12-24 16:55:16 +01009988timeout client <timeout>
9989timeout clitimeout <timeout> (deprecated)
9990 Set the maximum inactivity time on the client side.
9991 May be used in sections : defaults | frontend | listen | backend
9992 yes | yes | yes | no
9993 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009994 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009995 can be in any other unit if the number is suffixed by the unit,
9996 as explained at the top of this document.
9997
9998 The inactivity timeout applies when the client is expected to acknowledge or
9999 send data. In HTTP mode, this timeout is particularly important to consider
10000 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010001 response while it is reading data sent by the server. That said, for the
10002 first phase, it is preferable to set the "timeout http-request" to better
10003 protect HAProxy from Slowloris like attacks. The value is specified in
10004 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010005 suffixed by the unit, as specified at the top of this document. In TCP mode
10006 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10007 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010008 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010009 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010010 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10011 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010012 which overrides "timeout client" and "timeout server" for tunnels, as well as
10013 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010014
10015 This parameter is specific to frontends, but can be specified once for all in
10016 "defaults" sections. This is in fact one of the easiest solutions not to
10017 forget about it. An unspecified timeout results in an infinite timeout, which
10018 is not recommended. Such a usage is accepted and works but reports a warning
10019 during startup because it may results in accumulation of expired sessions in
10020 the system if the system's timeouts are not configured either.
10021
Willy Tarreau95c4e142017-11-26 12:18:55 +010010022 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010023
Willy Tarreau0ba27502007-12-24 16:55:16 +010010024 This parameter replaces the old, deprecated "clitimeout". It is recommended
10025 to use it to write new configurations. The form "timeout clitimeout" is
10026 provided only by backwards compatibility but its use is strongly discouraged.
10027
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010028 See also : "clitimeout", "timeout server", "timeout tunnel",
10029 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010030
10031
Willy Tarreau05cdd962014-05-10 14:30:07 +020010032timeout client-fin <timeout>
10033 Set the inactivity timeout on the client side for half-closed connections.
10034 May be used in sections : defaults | frontend | listen | backend
10035 yes | yes | yes | no
10036 Arguments :
10037 <timeout> is the timeout value specified in milliseconds by default, but
10038 can be in any other unit if the number is suffixed by the unit,
10039 as explained at the top of this document.
10040
10041 The inactivity timeout applies when the client is expected to acknowledge or
10042 send data while one direction is already shut down. This timeout is different
10043 from "timeout client" in that it only applies to connections which are closed
10044 in one direction. This is particularly useful to avoid keeping connections in
10045 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10046 problem is particularly common long connections such as RDP or WebSocket.
10047 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010048 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10049 frame was sent, often indicating an expectation that the connection quickly
10050 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010051
10052 This parameter is specific to frontends, but can be specified once for all in
10053 "defaults" sections. By default it is not set, so half-closed connections
10054 will use the other timeouts (timeout.client or timeout.tunnel).
10055
10056 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10057
10058
Willy Tarreau0ba27502007-12-24 16:55:16 +010010059timeout connect <timeout>
10060timeout contimeout <timeout> (deprecated)
10061 Set the maximum time to wait for a connection attempt to a server to succeed.
10062 May be used in sections : defaults | frontend | listen | backend
10063 yes | no | yes | yes
10064 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010065 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010066 can be in any other unit if the number is suffixed by the unit,
10067 as explained at the top of this document.
10068
10069 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010070 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010071 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010072 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010073 connect timeout also presets both queue and tarpit timeouts to the same value
10074 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010075
10076 This parameter is specific to backends, but can be specified once for all in
10077 "defaults" sections. This is in fact one of the easiest solutions not to
10078 forget about it. An unspecified timeout results in an infinite timeout, which
10079 is not recommended. Such a usage is accepted and works but reports a warning
10080 during startup because it may results in accumulation of failed sessions in
10081 the system if the system's timeouts are not configured either.
10082
10083 This parameter replaces the old, deprecated "contimeout". It is recommended
10084 to use it to write new configurations. The form "timeout contimeout" is
10085 provided only by backwards compatibility but its use is strongly discouraged.
10086
Willy Tarreau41a340d2008-01-22 12:25:31 +010010087 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10088 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010089
10090
Willy Tarreaub16a5742010-01-10 14:46:16 +010010091timeout http-keep-alive <timeout>
10092 Set the maximum allowed time to wait for a new HTTP request to appear
10093 May be used in sections : defaults | frontend | listen | backend
10094 yes | yes | yes | yes
10095 Arguments :
10096 <timeout> is the timeout value specified in milliseconds by default, but
10097 can be in any other unit if the number is suffixed by the unit,
10098 as explained at the top of this document.
10099
10100 By default, the time to wait for a new request in case of keep-alive is set
10101 by "timeout http-request". However this is not always convenient because some
10102 people want very short keep-alive timeouts in order to release connections
10103 faster, and others prefer to have larger ones but still have short timeouts
10104 once the request has started to present itself.
10105
10106 The "http-keep-alive" timeout covers these needs. It will define how long to
10107 wait for a new HTTP request to start coming after a response was sent. Once
10108 the first byte of request has been seen, the "http-request" timeout is used
10109 to wait for the complete request to come. Note that empty lines prior to a
10110 new request do not refresh the timeout and are not counted as a new request.
10111
10112 There is also another difference between the two timeouts : when a connection
10113 expires during timeout http-keep-alive, no error is returned, the connection
10114 just closes. If the connection expires in "http-request" while waiting for a
10115 connection to complete, a HTTP 408 error is returned.
10116
10117 In general it is optimal to set this value to a few tens to hundreds of
10118 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010119 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010120 1 millisecond) it will probably only accept pipelined requests but not the
10121 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010122 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010123
10124 If this parameter is not set, the "http-request" timeout applies, and if both
10125 are not set, "timeout client" still applies at the lower level. It should be
10126 set in the frontend to take effect, unless the frontend is in TCP mode, in
10127 which case the HTTP backend's timeout will be used.
10128
Willy Tarreau95c4e142017-11-26 12:18:55 +010010129 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10130 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010131 (where we only have one connection per client and a connection setup).
10132
Willy Tarreaub16a5742010-01-10 14:46:16 +010010133 See also : "timeout http-request", "timeout client".
10134
10135
Willy Tarreau036fae02008-01-06 13:24:40 +010010136timeout http-request <timeout>
10137 Set the maximum allowed time to wait for a complete HTTP request
10138 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010139 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010140 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010141 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010142 can be in any other unit if the number is suffixed by the unit,
10143 as explained at the top of this document.
10144
10145 In order to offer DoS protection, it may be required to lower the maximum
10146 accepted time to receive a complete HTTP request without affecting the client
10147 timeout. This helps protecting against established connections on which
10148 nothing is sent. The client timeout cannot offer a good protection against
10149 this abuse because it is an inactivity timeout, which means that if the
10150 attacker sends one character every now and then, the timeout will not
10151 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010152 types, the request will be aborted if it does not complete in time. When the
10153 timeout expires, an HTTP 408 response is sent to the client to inform it
10154 about the problem, and the connection is closed. The logs will report
10155 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010156 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010157 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10158 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010159
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010160 By default, this timeout only applies to the header part of the request,
10161 and not to any data. As soon as the empty line is received, this timeout is
10162 not used anymore. When combined with "option http-buffer-request", this
10163 timeout also applies to the body of the request..
10164 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010165 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010166
10167 Generally it is enough to set it to a few seconds, as most clients send the
10168 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010169 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010170 generally work on local networks as long as there are no packet losses. This
10171 will prevent people from sending bare HTTP requests using telnet.
10172
10173 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010174 chunk of the incoming request. It should be set in the frontend to take
10175 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10176 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010177
Willy Tarreau0f228a02015-05-01 15:37:53 +020010178 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010179 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010180
Willy Tarreau844e3c52008-01-11 16:28:18 +010010181
10182timeout queue <timeout>
10183 Set the maximum time to wait in the queue for a connection slot to be free
10184 May be used in sections : defaults | frontend | listen | backend
10185 yes | no | yes | yes
10186 Arguments :
10187 <timeout> is the timeout value specified in milliseconds by default, but
10188 can be in any other unit if the number is suffixed by the unit,
10189 as explained at the top of this document.
10190
10191 When a server's maxconn is reached, connections are left pending in a queue
10192 which may be server-specific or global to the backend. In order not to wait
10193 indefinitely, a timeout is applied to requests pending in the queue. If the
10194 timeout is reached, it is considered that the request will almost never be
10195 served, so it is dropped and a 503 error is returned to the client.
10196
10197 The "timeout queue" statement allows to fix the maximum time for a request to
10198 be left pending in a queue. If unspecified, the same value as the backend's
10199 connection timeout ("timeout connect") is used, for backwards compatibility
10200 with older versions with no "timeout queue" parameter.
10201
10202 See also : "timeout connect", "contimeout".
10203
10204
10205timeout server <timeout>
10206timeout srvtimeout <timeout> (deprecated)
10207 Set the maximum inactivity time on the server side.
10208 May be used in sections : defaults | frontend | listen | backend
10209 yes | no | yes | yes
10210 Arguments :
10211 <timeout> is the timeout value specified in milliseconds by default, but
10212 can be in any other unit if the number is suffixed by the unit,
10213 as explained at the top of this document.
10214
10215 The inactivity timeout applies when the server is expected to acknowledge or
10216 send data. In HTTP mode, this timeout is particularly important to consider
10217 during the first phase of the server's response, when it has to send the
10218 headers, as it directly represents the server's processing time for the
10219 request. To find out what value to put there, it's often good to start with
10220 what would be considered as unacceptable response times, then check the logs
10221 to observe the response time distribution, and adjust the value accordingly.
10222
10223 The value is specified in milliseconds by default, but can be in any other
10224 unit if the number is suffixed by the unit, as specified at the top of this
10225 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10226 recommended that the client timeout remains equal to the server timeout in
10227 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010228 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010229 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010230 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10231 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010232 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10233 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010234
10235 This parameter is specific to backends, but can be specified once for all in
10236 "defaults" sections. This is in fact one of the easiest solutions not to
10237 forget about it. An unspecified timeout results in an infinite timeout, which
10238 is not recommended. Such a usage is accepted and works but reports a warning
10239 during startup because it may results in accumulation of expired sessions in
10240 the system if the system's timeouts are not configured either.
10241
10242 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10243 to use it to write new configurations. The form "timeout srvtimeout" is
10244 provided only by backwards compatibility but its use is strongly discouraged.
10245
Willy Tarreauce887fd2012-05-12 12:50:00 +020010246 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010247
Willy Tarreau05cdd962014-05-10 14:30:07 +020010248
10249timeout server-fin <timeout>
10250 Set the inactivity timeout on the server side for half-closed connections.
10251 May be used in sections : defaults | frontend | listen | backend
10252 yes | no | yes | yes
10253 Arguments :
10254 <timeout> is the timeout value specified in milliseconds by default, but
10255 can be in any other unit if the number is suffixed by the unit,
10256 as explained at the top of this document.
10257
10258 The inactivity timeout applies when the server is expected to acknowledge or
10259 send data while one direction is already shut down. This timeout is different
10260 from "timeout server" in that it only applies to connections which are closed
10261 in one direction. This is particularly useful to avoid keeping connections in
10262 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10263 This problem is particularly common long connections such as RDP or WebSocket.
10264 Note that this timeout can override "timeout tunnel" when a connection shuts
10265 down in one direction. This setting was provided for completeness, but in most
10266 situations, it should not be needed.
10267
10268 This parameter is specific to backends, but can be specified once for all in
10269 "defaults" sections. By default it is not set, so half-closed connections
10270 will use the other timeouts (timeout.server or timeout.tunnel).
10271
10272 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10273
Willy Tarreau844e3c52008-01-11 16:28:18 +010010274
10275timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010276 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010277 May be used in sections : defaults | frontend | listen | backend
10278 yes | yes | yes | yes
10279 Arguments :
10280 <timeout> is the tarpit duration specified in milliseconds by default, but
10281 can be in any other unit if the number is suffixed by the unit,
10282 as explained at the top of this document.
10283
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010284 When a connection is tarpitted using "http-request tarpit" or
10285 "reqtarpit", it is maintained open with no activity for a certain
10286 amount of time, then closed. "timeout tarpit" defines how long it will
10287 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010288
10289 The value is specified in milliseconds by default, but can be in any other
10290 unit if the number is suffixed by the unit, as specified at the top of this
10291 document. If unspecified, the same value as the backend's connection timeout
10292 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010293 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010294
10295 See also : "timeout connect", "contimeout".
10296
10297
Willy Tarreauce887fd2012-05-12 12:50:00 +020010298timeout tunnel <timeout>
10299 Set the maximum inactivity time on the client and server side for tunnels.
10300 May be used in sections : defaults | frontend | listen | backend
10301 yes | no | yes | yes
10302 Arguments :
10303 <timeout> is the timeout value specified in milliseconds by default, but
10304 can be in any other unit if the number is suffixed by the unit,
10305 as explained at the top of this document.
10306
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010307 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010308 between a client and a server, and the connection remains inactive in both
10309 directions. This timeout supersedes both the client and server timeouts once
10310 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010311 analyzer remains attached to either connection (e.g. tcp content rules are
10312 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010313 when switching to the WebSocket protocol, or forwarding a CONNECT request
10314 to a proxy), or after the first response when no keepalive/close option is
10315 specified.
10316
Willy Tarreau05cdd962014-05-10 14:30:07 +020010317 Since this timeout is usually used in conjunction with long-lived connections,
10318 it usually is a good idea to also set "timeout client-fin" to handle the
10319 situation where a client suddenly disappears from the net and does not
10320 acknowledge a close, or sends a shutdown and does not acknowledge pending
10321 data anymore. This can happen in lossy networks where firewalls are present,
10322 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10323 state.
10324
Willy Tarreauce887fd2012-05-12 12:50:00 +020010325 The value is specified in milliseconds by default, but can be in any other
10326 unit if the number is suffixed by the unit, as specified at the top of this
10327 document. Whatever the expected normal idle time, it is a good practice to
10328 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010329 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010330
10331 This parameter is specific to backends, but can be specified once for all in
10332 "defaults" sections. This is in fact one of the easiest solutions not to
10333 forget about it.
10334
10335 Example :
10336 defaults http
10337 option http-server-close
10338 timeout connect 5s
10339 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010340 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010341 timeout server 30s
10342 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10343
Willy Tarreau05cdd962014-05-10 14:30:07 +020010344 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010345
10346
Willy Tarreau844e3c52008-01-11 16:28:18 +010010347transparent (deprecated)
10348 Enable client-side transparent proxying
10349 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010350 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010351 Arguments : none
10352
10353 This keyword was introduced in order to provide layer 7 persistence to layer
10354 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10355 connection for a remote address to a local process (here HAProxy), and let
10356 this process know what address was initially requested. When this option is
10357 used, sessions without cookies will be forwarded to the original destination
10358 IP address of the incoming request (which should match that of another
10359 equipment), while requests with cookies will still be forwarded to the
10360 appropriate server.
10361
10362 The "transparent" keyword is deprecated, use "option transparent" instead.
10363
10364 Note that contrary to a common belief, this option does NOT make HAProxy
10365 present the client's IP to the server when establishing the connection.
10366
Willy Tarreau844e3c52008-01-11 16:28:18 +010010367 See also: "option transparent"
10368
William Lallemanda73203e2012-03-12 12:48:57 +010010369unique-id-format <string>
10370 Generate a unique ID for each request.
10371 May be used in sections : defaults | frontend | listen | backend
10372 yes | yes | yes | no
10373 Arguments :
10374 <string> is a log-format string.
10375
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010376 This keyword creates a ID for each request using the custom log format. A
10377 unique ID is useful to trace a request passing through many components of
10378 a complex infrastructure. The newly created ID may also be logged using the
10379 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010380
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010381 The format should be composed from elements that are guaranteed to be
10382 unique when combined together. For instance, if multiple haproxy instances
10383 are involved, it might be important to include the node name. It is often
10384 needed to log the incoming connection's source and destination addresses
10385 and ports. Note that since multiple requests may be performed over the same
10386 connection, including a request counter may help differentiate them.
10387 Similarly, a timestamp may protect against a rollover of the counter.
10388 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010389
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010390 It is recommended to use hexadecimal notation for many fields since it
10391 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010392
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010393 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010394
Julien Vehentf21be322014-03-07 08:27:34 -050010395 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010396
10397 will generate:
10398
10399 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10400
10401 See also: "unique-id-header"
10402
10403unique-id-header <name>
10404 Add a unique ID header in the HTTP request.
10405 May be used in sections : defaults | frontend | listen | backend
10406 yes | yes | yes | no
10407 Arguments :
10408 <name> is the name of the header.
10409
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010410 Add a unique-id header in the HTTP request sent to the server, using the
10411 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010412
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010413 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010414
Julien Vehentf21be322014-03-07 08:27:34 -050010415 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010416 unique-id-header X-Unique-ID
10417
10418 will generate:
10419
10420 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10421
10422 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010423
Willy Tarreauf51658d2014-04-23 01:21:56 +020010424use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010425 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010426 May be used in sections : defaults | frontend | listen | backend
10427 no | yes | yes | no
10428 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010429 <backend> is the name of a valid backend or "listen" section, or a
10430 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010431
Willy Tarreauf51658d2014-04-23 01:21:56 +020010432 <condition> is a condition composed of ACLs, as described in section 7. If
10433 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010434
10435 When doing content-switching, connections arrive on a frontend and are then
10436 dispatched to various backends depending on a number of conditions. The
10437 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010438 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010439 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010440 source address validation) or combined with a "tcp-request" rule to wait for
10441 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010442
10443 There may be as many "use_backend" rules as desired. All of these rules are
10444 evaluated in their declaration order, and the first one which matches will
10445 assign the backend.
10446
10447 In the first form, the backend will be used if the condition is met. In the
10448 second form, the backend will be used if the condition is not met. If no
10449 condition is valid, the backend defined with "default_backend" will be used.
10450 If no default backend is defined, either the servers in the same section are
10451 used (in case of a "listen" section) or, in case of a frontend, no server is
10452 used and a 503 service unavailable response is returned.
10453
Willy Tarreau51aecc72009-07-12 09:47:04 +020010454 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010455 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010456 and backend processing will immediately follow, or the backend will wait for
10457 a complete HTTP request to get in. This feature is useful when a frontend
10458 must decode several protocols on a unique port, one of them being HTTP.
10459
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010460 When <backend> is a simple name, it is resolved at configuration time, and an
10461 error is reported if the specified backend does not exist. If <backend> is
10462 a log-format string instead, no check may be done at configuration time, so
10463 the backend name is resolved dynamically at run time. If the resulting
10464 backend name does not correspond to any valid backend, no other rule is
10465 evaluated, and the default_backend directive is applied instead. Note that
10466 when using dynamic backend names, it is highly recommended to use a prefix
10467 that no other backend uses in order to ensure that an unauthorized backend
10468 cannot be forced from the request.
10469
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010470 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010471 used to detect the association between frontends and backends to compute the
10472 backend's "fullconn" setting. This cannot be done for dynamic names.
10473
10474 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10475 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010476
Willy Tarreau036fae02008-01-06 13:24:40 +010010477
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010478use-server <server> if <condition>
10479use-server <server> unless <condition>
10480 Only use a specific server if/unless an ACL-based condition is matched.
10481 May be used in sections : defaults | frontend | listen | backend
10482 no | no | yes | yes
10483 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010484 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010485
10486 <condition> is a condition composed of ACLs, as described in section 7.
10487
10488 By default, connections which arrive to a backend are load-balanced across
10489 the available servers according to the configured algorithm, unless a
10490 persistence mechanism such as a cookie is used and found in the request.
10491
10492 Sometimes it is desirable to forward a particular request to a specific
10493 server without having to declare a dedicated backend for this server. This
10494 can be achieved using the "use-server" rules. These rules are evaluated after
10495 the "redirect" rules and before evaluating cookies, and they have precedence
10496 on them. There may be as many "use-server" rules as desired. All of these
10497 rules are evaluated in their declaration order, and the first one which
10498 matches will assign the server.
10499
10500 If a rule designates a server which is down, and "option persist" is not used
10501 and no force-persist rule was validated, it is ignored and evaluation goes on
10502 with the next rules until one matches.
10503
10504 In the first form, the server will be used if the condition is met. In the
10505 second form, the server will be used if the condition is not met. If no
10506 condition is valid, the processing continues and the server will be assigned
10507 according to other persistence mechanisms.
10508
10509 Note that even if a rule is matched, cookie processing is still performed but
10510 does not assign the server. This allows prefixed cookies to have their prefix
10511 stripped.
10512
10513 The "use-server" statement works both in HTTP and TCP mode. This makes it
10514 suitable for use with content-based inspection. For instance, a server could
10515 be selected in a farm according to the TLS SNI field. And if these servers
10516 have their weight set to zero, they will not be used for other traffic.
10517
10518 Example :
10519 # intercept incoming TLS requests based on the SNI field
10520 use-server www if { req_ssl_sni -i www.example.com }
10521 server www 192.168.0.1:443 weight 0
10522 use-server mail if { req_ssl_sni -i mail.example.com }
10523 server mail 192.168.0.1:587 weight 0
10524 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010525 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010526 # all the rest is forwarded to this server
10527 server default 192.168.0.2:443 check
10528
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010529 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010530
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010531
Davor Ocelice9ed2812017-12-25 17:49:28 +0100105325. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010533--------------------------
10534
10535The "bind", "server" and "default-server" keywords support a number of settings
10536depending on some build options and on the system HAProxy was built on. These
10537settings generally each consist in one word sometimes followed by a value,
10538written on the same line as the "bind" or "server" line. All these options are
10539described in this section.
10540
10541
105425.1. Bind options
10543-----------------
10544
10545The "bind" keyword supports a certain number of settings which are all passed
10546as arguments on the same line. The order in which those arguments appear makes
10547no importance, provided that they appear after the bind address. All of these
10548parameters are optional. Some of them consist in a single words (booleans),
10549while other ones expect a value after them. In this case, the value must be
10550provided immediately after the setting name.
10551
10552The currently supported settings are the following ones.
10553
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010554accept-netscaler-cip <magic number>
10555 Enforces the use of the NetScaler Client IP insertion protocol over any
10556 connection accepted by any of the TCP sockets declared on the same line. The
10557 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10558 the incoming connection to be used everywhere an address is used, with the
10559 only exception of "tcp-request connection" rules which will only see the
10560 real connection address. Logs will reflect the addresses indicated in the
10561 protocol, unless it is violated, in which case the real address will still
10562 be used. This keyword combined with support from external components can be
10563 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010564 mechanism which is not always reliable and not even always usable. See also
10565 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10566 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010567
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010568accept-proxy
10569 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010570 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10571 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010572 3/4 addresses of the incoming connection to be used everywhere an address is
10573 used, with the only exception of "tcp-request connection" rules which will
10574 only see the real connection address. Logs will reflect the addresses
10575 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010576 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010577 components can be used as an efficient and reliable alternative to the
10578 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010579 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10580 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010581
Olivier Houchardc2aae742017-09-22 18:26:28 +020010582allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010583 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchardc2aae742017-09-22 18:26:28 +020010584 due to security considerations.
10585
Willy Tarreauab861d32013-04-02 02:30:41 +020010586alpn <protocols>
10587 This enables the TLS ALPN extension and advertises the specified protocol
10588 list as supported on top of ALPN. The protocol list consists in a comma-
10589 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10590 quotes). This requires that the SSL library is build with support for TLS
10591 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010592 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10593 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10594 now obsolete NPN extension. At the time of writing this, most browsers still
10595 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10596 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10597 are expected to be supported, both versions can be advertised, in order of
10598 preference, like below :
10599
10600 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010601
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010602backlog <backlog>
10603 Sets the socket's backlog to this value. If unspecified, the frontend's
10604 backlog is used instead, which generally defaults to the maxconn value.
10605
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010606curves <curves>
10607 This setting is only available when support for OpenSSL was built in. It sets
10608 the string describing the list of elliptic curves algorithms ("curve suite")
10609 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10610 string is a colon-delimited list of curve name.
10611 Example: "X25519:P-256" (without quote)
10612 When "curves" is set, "ecdhe" parameter is ignored.
10613
Emeric Brun7fb34422012-09-28 15:26:15 +020010614ecdhe <named curve>
10615 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010616 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10617 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010618
Emeric Brunfd33a262012-10-11 16:28:27 +020010619ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010620 This setting is only available when support for OpenSSL was built in. It
10621 designates a PEM file from which to load CA certificates used to verify
10622 client's certificate.
10623
Emeric Brunb6dc9342012-09-28 17:55:37 +020010624ca-ignore-err [all|<errorID>,...]
10625 This setting is only available when support for OpenSSL was built in.
10626 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10627 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10628 error is ignored.
10629
Christopher Faulet31af49d2015-06-09 17:29:50 +020010630ca-sign-file <cafile>
10631 This setting is only available when support for OpenSSL was built in. It
10632 designates a PEM file containing both the CA certificate and the CA private
10633 key used to create and sign server's certificates. This is a mandatory
10634 setting when the dynamic generation of certificates is enabled. See
10635 'generate-certificates' for details.
10636
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010637ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010638 This setting is only available when support for OpenSSL was built in. It is
10639 the CA private key passphrase. This setting is optional and used only when
10640 the dynamic generation of certificates is enabled. See
10641 'generate-certificates' for details.
10642
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010643ciphers <ciphers>
10644 This setting is only available when support for OpenSSL was built in. It sets
10645 the string describing the list of cipher algorithms ("cipher suite") that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010646 negotiated during the SSL/TLS handshake except for TLSv1.3. The format of the
10647 string is defined in "man 1 ciphers" from OpenSSL man pages, and can be for
10648 instance a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without
10649 quotes). Depending on the compatibility and security requirements, the list
10650 of suitable ciphers depends on a variety of variables. For background
10651 information and recommendations see e.g.
10652 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10653 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10654 cipher configuration, please check the "ciphersuites" keyword.
10655
10656ciphersuites <ciphersuites>
10657 This setting is only available when support for OpenSSL was built in and
10658 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10659 the list of cipher algorithms ("cipher suite") that are negotiated during the
10660 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
10661 OpenSSL man pages under the "ciphersuites" section, and can be for instance a
10662 string such as
10663 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
10664 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
10665 the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010666
Emeric Brunfd33a262012-10-11 16:28:27 +020010667crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010668 This setting is only available when support for OpenSSL was built in. It
10669 designates a PEM file from which to load certificate revocation list used
10670 to verify client's certificate.
10671
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010672crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010673 This setting is only available when support for OpenSSL was built in. It
10674 designates a PEM file containing both the required certificates and any
10675 associated private keys. This file can be built by concatenating multiple
10676 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10677 requires an intermediate certificate, this can also be concatenated into this
10678 file.
10679
10680 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10681 are loaded.
10682
10683 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010684 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010685 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10686 specified multiple times in order to load certificates from multiple files or
10687 directories. The certificates will be presented to clients who provide a
10688 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010689 subjects. Wildcards are supported, where a wildcard character '*' is used
10690 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010691 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010692
10693 If no SNI is provided by the client or if the SSL library does not support
10694 TLS extensions, or if the client provides an SNI hostname which does not
10695 match any certificate, then the first loaded certificate will be presented.
10696 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010697 recommended to load the default one first as a file or to ensure that it will
10698 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010699
Emeric Brune032bfa2012-09-28 13:01:45 +020010700 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010701
Davor Ocelice9ed2812017-12-25 17:49:28 +010010702 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010703 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010704 choose a web server that the CA believes requires an intermediate CA (for
10705 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010706 others, e.g. nginx, result in a wrong bundle that will not work for some
10707 clients).
10708
Emeric Brun4147b2e2014-06-16 18:36:30 +020010709 For each PEM file, haproxy checks for the presence of file at the same path
10710 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10711 Status Request extension (also known as "OCSP stapling") is automatically
10712 enabled. The content of this file is optional. If not empty, it must contain
10713 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10714 must comply with the following rules: it has to indicate a good status,
10715 it has to be a single response for the certificate of the PEM file, and it
10716 has to be valid at the moment of addition. If these rules are not respected
10717 the OCSP Response is ignored and a warning is emitted. In order to identify
10718 which certificate an OCSP Response applies to, the issuer's certificate is
10719 necessary. If the issuer's certificate is not found in the PEM file, it will
10720 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10721 if it exists otherwise it will fail with an error.
10722
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010723 For each PEM file, haproxy also checks for the presence of file at the same
10724 path suffixed by ".sctl". If such file is found, support for Certificate
10725 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10726 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10727 to check basic syntax, but no signatures are verified.
10728
yanbzhu6c25e9e2016-01-05 12:52:02 -050010729 There are cases where it is desirable to support multiple key types, e.g. RSA
10730 and ECDSA in the cipher suites offered to the clients. This allows clients
10731 that support EC certificates to be able to use EC ciphers, while
10732 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010733
10734 In order to provide this functionality, multiple PEM files, each with a
10735 different key type, are required. To associate these PEM files into a
10736 "cert bundle" that is recognized by haproxy, they must be named in the
10737 following way: All PEM files that are to be bundled must have the same base
10738 name, with a suffix indicating the key type. Currently, three suffixes are
10739 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10740 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10741 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10742 suffix matters. To load this bundle into haproxy, specify the base name only:
10743
10744 Example : bind :8443 ssl crt example.pem
10745
yanbzhu6c25e9e2016-01-05 12:52:02 -050010746 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010747 a cert bundle.
10748
Davor Ocelice9ed2812017-12-25 17:49:28 +010010749 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010750 support multiple key types. PEM files are combined based on Common Name
10751 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10752 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10753 entries in the certificates in that bundle, haproxy will not be able to
10754 provide multi-cert support.
10755
10756 Assuming bundle in the example above contained the following:
10757
10758 Filename | CN | SAN
10759 -------------------+-----------------+-------------------
10760 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010761 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010762 example.pem.ecdsa | www.example.com | ecdsa.example.com
10763 -------------------+-----------------+-------------------
10764
10765 Users connecting with an SNI of "www.example.com" will be able
10766 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10767 "rsa.example.com" will only be able to use RSA cipher suites, and users
10768 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010769 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10770 no need to bundle certificates. ECDSA certificate will be preferred if client
10771 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010772
10773 If a directory name is given as the <cert> argument, haproxy will
10774 automatically search and load bundled files in that directory.
10775
10776 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10777 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10778 time, sctl is not supported in multi-certificate bundling.
10779
Emeric Brunb6dc9342012-09-28 17:55:37 +020010780crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010781 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010782 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010783 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010784 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010785
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010786crt-list <file>
10787 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010788 designates a list of PEM file with an optional ssl configuration and a SNI
10789 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010790
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010791 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10792
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010793 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10794 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010795 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010796 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010797
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010798 Wildcards are supported in the SNI filter. Negative filter are also supported,
10799 only useful in combination with a wildcard filter to exclude a particular SNI.
10800 The certificates will be presented to clients who provide a valid TLS Server
10801 Name Indication field matching one of the SNI filters. If no SNI filter is
10802 specified, the CN and alt subjects are used. This directive may be specified
10803 multiple times. See the "crt" option for more information. The default
10804 certificate is still needed to meet OpenSSL expectations. If it is not used,
10805 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010806
yanbzhu6c25e9e2016-01-05 12:52:02 -050010807 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010808 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010809 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10810 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10811 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010812
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010813 crt-list file example:
10814 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010815 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010816 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010817 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010818
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010819defer-accept
10820 Is an optional keyword which is supported only on certain Linux kernels. It
10821 states that a connection will only be accepted once some data arrive on it,
10822 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010823 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010824 performance by ensuring that most of the request is already available when
10825 the connection is accepted. On the other hand, it will not be able to detect
10826 connections which don't talk. It is important to note that this option is
10827 broken in all kernels up to 2.6.31, as the connection is never accepted until
10828 the client talks. This can cause issues with front firewalls which would see
10829 an established connection while the proxy will only see it in SYN_RECV. This
10830 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10831
William Lallemandf6975e92017-05-26 17:42:10 +020010832expose-fd listeners
10833 This option is only usable with the stats socket. It gives your stats socket
10834 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010835 During a reload with the master-worker mode, the process is automatically
10836 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010837 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010838
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010839force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010840 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010841 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010842 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010843 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010844
10845force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010846 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010847 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010848 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010849
10850force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010851 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010852 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010853 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010854
10855force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010856 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010857 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010858 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010859
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010860force-tlsv13
10861 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10862 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010863 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010864
Christopher Faulet31af49d2015-06-09 17:29:50 +020010865generate-certificates
10866 This setting is only available when support for OpenSSL was built in. It
10867 enables the dynamic SSL certificates generation. A CA certificate and its
10868 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10869 a transparent forward proxy, SSL requests generate errors because of a common
10870 name mismatch on the certificate presented to the client. With this option
10871 enabled, HAProxy will try to forge a certificate using the SNI hostname
10872 indicated by the client. This is done only if no certificate matches the SNI
10873 hostname (see 'crt-list'). If an error occurs, the default certificate is
10874 used, else the 'strict-sni' option is set.
10875 It can also be used when HAProxy is configured as a reverse proxy to ease the
10876 deployment of an architecture with many backends.
10877
10878 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10879 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010010880 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020010881 certificate is used many times.
10882
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010883gid <gid>
10884 Sets the group of the UNIX sockets to the designated system gid. It can also
10885 be set by default in the global section's "unix-bind" statement. Note that
10886 some platforms simply ignore this. This setting is equivalent to the "group"
10887 setting except that the group ID is used instead of its name. This setting is
10888 ignored by non UNIX sockets.
10889
10890group <group>
10891 Sets the group of the UNIX sockets to the designated system group. It can
10892 also be set by default in the global section's "unix-bind" statement. Note
10893 that some platforms simply ignore this. This setting is equivalent to the
10894 "gid" setting except that the group name is used instead of its gid. This
10895 setting is ignored by non UNIX sockets.
10896
10897id <id>
10898 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10899 sometimes it is more convenient to fix them to ease monitoring. This value
10900 must be strictly positive and unique within the listener/frontend. This
10901 option can only be used when defining only a single socket.
10902
10903interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010904 Restricts the socket to a specific interface. When specified, only packets
10905 received from that particular interface are processed by the socket. This is
10906 currently only supported on Linux. The interface must be a primary system
10907 interface, not an aliased interface. It is also possible to bind multiple
10908 frontends to the same address if they are bound to different interfaces. Note
10909 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010010910 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
10911 uses the same interface as inbound traffic, and its associated routing table,
10912 even if there are explicit routes through different interfaces configured.
10913 This can prove useful to address asymmetric routing issues when the same
10914 client IP addresses need to be able to reach frontends hosted on different
10915 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010916
Willy Tarreauabb175f2012-09-24 12:43:26 +020010917level <level>
10918 This setting is used with the stats sockets only to restrict the nature of
10919 the commands that can be issued on the socket. It is ignored by other
10920 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010010921 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020010922 read, and no change is allowed. It would make sense on systems where it
10923 is not easy to restrict access to the socket.
10924 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010925 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020010926 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010927 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020010928 all counters).
10929
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010930severity-output <format>
10931 This setting is used with the stats sockets only to configure severity
10932 level output prepended to informational feedback messages. Severity
10933 level of messages can range between 0 and 7, conforming to syslog
10934 rfc5424. Valid and successful socket commands requesting data
10935 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10936 prepended. It is ignored by other sockets. <format> can be one of :
10937 - "none" (default) no severity level is prepended to feedback messages.
10938 - "number" severity level is prepended as a number.
10939 - "string" severity level is prepended as a string following the
10940 rfc5424 convention.
10941
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010942maxconn <maxconn>
10943 Limits the sockets to this number of concurrent connections. Extraneous
10944 connections will remain in the system's backlog until a connection is
10945 released. If unspecified, the limit will be the same as the frontend's
10946 maxconn. Note that in case of port ranges or multiple addresses, the same
10947 value will be applied to each socket. This setting enables different
10948 limitations on expensive sockets, for instance SSL entries which may easily
10949 eat all memory.
10950
10951mode <mode>
10952 Sets the octal mode used to define access permissions on the UNIX socket. It
10953 can also be set by default in the global section's "unix-bind" statement.
10954 Note that some platforms simply ignore this. This setting is ignored by non
10955 UNIX sockets.
10956
10957mss <maxseg>
10958 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10959 connections. This can be used to force a lower MSS for certain specific
10960 ports, for instance for connections passing through a VPN. Note that this
10961 relies on a kernel feature which is theoretically supported under Linux but
10962 was buggy in all versions prior to 2.6.28. It may or may not work on other
10963 operating systems. It may also not change the advertised value but change the
10964 effective size of outgoing segments. The commonly advertised value for TCPv4
10965 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
10966 positive, it will be used as the advertised MSS. If it is negative, it will
10967 indicate by how much to reduce the incoming connection's advertised MSS for
10968 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
10969
10970name <name>
10971 Sets an optional name for these sockets, which will be reported on the stats
10972 page.
10973
Willy Tarreaud72f0f32015-10-13 14:50:22 +020010974namespace <name>
10975 On Linux, it is possible to specify which network namespace a socket will
10976 belong to. This directive makes it possible to explicitly bind a listener to
10977 a namespace different from the default one. Please refer to your operating
10978 system's documentation to find more details about network namespaces.
10979
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010980nice <nice>
10981 Sets the 'niceness' of connections initiated from the socket. Value must be
10982 in the range -1024..1024 inclusive, and defaults to zero. Positive values
10983 means that such connections are more friendly to others and easily offer
10984 their place in the scheduler. On the opposite, negative values mean that
10985 connections want to run with a higher priority than others. The difference
10986 only happens under high loads when the system is close to saturation.
10987 Negative values are appropriate for low-latency or administration services,
10988 and high values are generally recommended for CPU intensive tasks such as SSL
10989 processing or bulk transfers which are less sensible to latency. For example,
10990 it may make sense to use a positive value for an SMTP socket and a negative
10991 one for an RDP socket.
10992
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010993no-ca-names
10994 This setting is only available when support for OpenSSL was built in. It
10995 prevents from send CA names in server hello message when ca-file is used.
10996
Emeric Brun9b3009b2012-10-05 11:55:06 +020010997no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010998 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010999 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011000 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011001 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011002 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11003 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011004
Emeric Brun90ad8722012-10-02 14:00:59 +020011005no-tls-tickets
11006 This setting is only available when support for OpenSSL was built in. It
11007 disables the stateless session resumption (RFC 5077 TLS Ticket
11008 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011009 session resumption is more expensive in CPU usage. This option is also
11010 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011011
Emeric Brun9b3009b2012-10-05 11:55:06 +020011012no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011013 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011014 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011015 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011016 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011017 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11018 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011019
Emeric Brun9b3009b2012-10-05 11:55:06 +020011020no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011021 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011022 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011023 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011024 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011025 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11026 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011027
Emeric Brun9b3009b2012-10-05 11:55:06 +020011028no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011029 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011030 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011031 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011032 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011033 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11034 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011035
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011036no-tlsv13
11037 This setting is only available when support for OpenSSL was built in. It
11038 disables support for TLSv1.3 on any sockets instantiated from the listener
11039 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11040 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011041 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11042 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011043
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011044npn <protocols>
11045 This enables the NPN TLS extension and advertises the specified protocol list
11046 as supported on top of NPN. The protocol list consists in a comma-delimited
11047 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11048 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011049 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011050 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11051 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11052 version of OpenSSL, NPN might still be used as most clients still support it
11053 at the time of writing this. It is possible to enable both NPN and ALPN
11054 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011055
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011056prefer-client-ciphers
11057 Use the client's preference when selecting the cipher suite, by default
11058 the server's preference is enforced. This option is also available on
11059 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011060 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11061 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11062 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011063
Christopher Fauletc644fa92017-11-23 22:44:11 +010011064process <process-set>[/<thread-set>]
11065 This restricts the list of processes and/or threads on which this listener is
11066 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011067 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011068 between the two is applied. If in the end the listener is not allowed to run
11069 on any remaining process, a warning is emitted, and the listener will either
11070 run on the first process of the listener if a single process was specified,
11071 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011072 set is specified, it limits the threads allowed to process incoming
Christopher Fauletc644fa92017-11-23 22:44:11 +010011073 connections for this listener, for the corresponding process set. For the
11074 unlikely case where several ranges are needed, this directive may be
11075 repeated. <process-set> and <thread-set> must use the format
11076
11077 all | odd | even | number[-[number]]
11078
11079 Ranges can be partially defined. The higher bound can be omitted. In such
11080 case, it is replaced by the corresponding maximum value. The main purpose of
11081 this directive is to be used with the stats sockets and have one different
11082 socket per process. The second purpose is to have multiple bind lines sharing
11083 the same IP:port but not the same process in a listener, so that the system
11084 can distribute the incoming connections into multiple queues and allow a
11085 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11086 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011087
Christopher Fauleta717b992018-04-10 14:43:00 +020011088proto <name>
11089 Forces the multiplexer's protocol to use for the incoming connections. It
11090 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11091 be usable on the frontend side. The list of available protocols is reported
11092 in haproxy -vv.
11093 Idea behind this optipon is to bypass the selection of the best multiplexer's
11094 protocol for all connections instantiated from this listening socket. For
11095 instance, it is possible to force the http/2 on clear TCP by specifing "proto
11096 h2" on the bind line.
11097
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011098ssl
11099 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011100 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011101 certificate is necessary (see "crt" above). All contents in the buffers will
11102 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011103 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11104 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011105
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011106ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11107 This option enforces use of <version> or lower on SSL connections instantiated
11108 from this listener. This option is also available on global statement
11109 "ssl-default-bind-options". See also "ssl-min-ver".
11110
11111ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11112 This option enforces use of <version> or upper on SSL connections instantiated
11113 from this listener. This option is also available on global statement
11114 "ssl-default-bind-options". See also "ssl-max-ver".
11115
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011116strict-sni
11117 This setting is only available when support for OpenSSL was built in. The
11118 SSL/TLS negotiation is allow only if the client provided an SNI which match
11119 a certificate. The default certificate is not used.
11120 See the "crt" option for more information.
11121
Willy Tarreau2af207a2015-02-04 00:45:58 +010011122tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011123 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011124 listening socket. This option is available on Linux since version 2.6.37. It
11125 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011126 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011127 useful on long-lived connections experiencing long idle periods such as
11128 remote terminals or database connection pools, where the client and server
11129 timeouts must remain high to allow a long period of idle, but where it is
11130 important to detect that the client has disappeared in order to release all
11131 resources associated with its connection (and the server's session). The
11132 argument is a delay expressed in milliseconds by default. This only works
11133 for regular TCP connections, and is ignored for other protocols.
11134
Willy Tarreau1c862c52012-10-05 16:21:00 +020011135tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011136 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011137 enables TCP Fast Open on the listening socket, which means that clients which
11138 support this feature will be able to send a request and receive a response
11139 during the 3-way handshake starting from second connection, thus saving one
11140 round-trip after the first connection. This only makes sense with protocols
11141 that use high connection rates and where each round trip matters. This can
11142 possibly cause issues with many firewalls which do not accept data on SYN
11143 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011144 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11145 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11146 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011147
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011148tls-ticket-keys <keyfile>
11149 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
11150 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
11151 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
11152 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
11153 used for decryption and the penultimate one for encryption. This enables easy
11154 key rotation by just appending new key to the file and reloading the process.
11155 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
11156 is compromised. It is also a good idea to keep the keys off any permanent
11157 storage such as hard drives (hint: use tmpfs and don't swap those files).
11158 Lifetime hint can be changed using tune.ssl.timeout.
11159
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011160transparent
11161 Is an optional keyword which is supported only on certain Linux kernels. It
11162 indicates that the addresses will be bound even if they do not belong to the
11163 local machine, and that packets targeting any of these addresses will be
11164 intercepted just as if the addresses were locally configured. This normally
11165 requires that IP forwarding is enabled. Caution! do not use this with the
11166 default address '*', as it would redirect any traffic for the specified port.
11167 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11168 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11169 kernel version. Some distribution kernels include backports of the feature,
11170 so check for support with your vendor.
11171
Willy Tarreau77e3af92012-11-24 15:07:23 +010011172v4v6
11173 Is an optional keyword which is supported only on most recent systems
11174 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11175 and IPv6 when it uses the default address. Doing so is sometimes necessary
11176 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011177 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011178
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011179v6only
11180 Is an optional keyword which is supported only on most recent systems
11181 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11182 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011183 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11184 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011185
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011186uid <uid>
11187 Sets the owner of the UNIX sockets to the designated system uid. It can also
11188 be set by default in the global section's "unix-bind" statement. Note that
11189 some platforms simply ignore this. This setting is equivalent to the "user"
11190 setting except that the user numeric ID is used instead of its name. This
11191 setting is ignored by non UNIX sockets.
11192
11193user <user>
11194 Sets the owner of the UNIX sockets to the designated system user. It can also
11195 be set by default in the global section's "unix-bind" statement. Note that
11196 some platforms simply ignore this. This setting is equivalent to the "uid"
11197 setting except that the user name is used instead of its uid. This setting is
11198 ignored by non UNIX sockets.
11199
Emeric Brun1a073b42012-09-28 17:07:34 +020011200verify [none|optional|required]
11201 This setting is only available when support for OpenSSL was built in. If set
11202 to 'none', client certificate is not requested. This is the default. In other
11203 cases, a client certificate is requested. If the client does not provide a
11204 certificate after the request and if 'verify' is set to 'required', then the
11205 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011206 certificate provided by the client is always verified using CAs from
11207 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11208 is aborted, regardless of the 'verify' option, unless the error code exactly
11209 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011210
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200112115.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011212------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011213
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011214The "server" and "default-server" keywords support a certain number of settings
11215which are all passed as arguments on the server line. The order in which those
11216arguments appear does not count, and they are all optional. Some of those
11217settings are single words (booleans) while others expect one or several values
11218after them. In this case, the values must immediately follow the setting name.
11219Except default-server, all those settings must be specified after the server's
11220address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011221
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011222 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011223 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011224
Frédéric Lécailled2376272017-03-21 18:52:12 +010011225Note that all these settings are supported both by "server" and "default-server"
11226keywords, except "id" which is only supported by "server".
11227
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011228The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011229
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011230addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011231 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011232 to send health-checks or to probe the agent-check. On some servers, it may be
11233 desirable to dedicate an IP address to specific component able to perform
11234 complex tests which are more suitable to health-checks than the application.
11235 This parameter is ignored if the "check" parameter is not set. See also the
11236 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011237
Simon Hormand60d6912013-11-25 10:46:36 +090011238agent-check
11239 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011240 health check. An agent health check is performed by making a TCP connection
11241 to the port set by the "agent-port" parameter and reading an ASCII string.
11242 The string is made of a series of words delimited by spaces, tabs or commas
11243 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011244
Willy Tarreau81f5d942013-12-09 20:51:51 +010011245 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011246 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011247 weight of a server as configured when haproxy starts. Note that a zero
11248 weight is reported on the stats page as "DRAIN" since it has the same
11249 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011250
Davor Ocelice9ed2812017-12-25 17:49:28 +010011251 - The string "maxconn:" followed by an integer (no space between). Values
11252 in this format will set the maxconn of a server. The maximum number of
11253 connections advertised needs to be multiplied by the number of load
11254 balancers and different backends that use this health check to get the
11255 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011256
Willy Tarreau81f5d942013-12-09 20:51:51 +010011257 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011258 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011259
Willy Tarreau81f5d942013-12-09 20:51:51 +010011260 - The word "drain". This will turn the server's administrative state to the
11261 DRAIN mode, thus it will not accept any new connections other than those
11262 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011263
Willy Tarreau81f5d942013-12-09 20:51:51 +010011264 - The word "maint". This will turn the server's administrative state to the
11265 MAINT mode, thus it will not accept any new connections at all, and health
11266 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011267
Willy Tarreau81f5d942013-12-09 20:51:51 +010011268 - The words "down", "failed", or "stopped", optionally followed by a
11269 description string after a sharp ('#'). All of these mark the server's
11270 operating state as DOWN, but since the word itself is reported on the stats
11271 page, the difference allows an administrator to know if the situation was
11272 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011273 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011274 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011275
Willy Tarreau81f5d942013-12-09 20:51:51 +010011276 - The word "up" sets back the server's operating state as UP if health checks
11277 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011278
Willy Tarreau81f5d942013-12-09 20:51:51 +010011279 Parameters which are not advertised by the agent are not changed. For
11280 example, an agent might be designed to monitor CPU usage and only report a
11281 relative weight and never interact with the operating status. Similarly, an
11282 agent could be designed as an end-user interface with 3 radio buttons
11283 allowing an administrator to change only the administrative state. However,
11284 it is important to consider that only the agent may revert its own actions,
11285 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11286 agent must implement the other equivalent actions to bring the service into
11287 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011288
Simon Horman2f1f9552013-11-25 10:46:37 +090011289 Failure to connect to the agent is not considered an error as connectivity
11290 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011291 parameter. Warning though, it is not a good idea to stop an agent after it
11292 reports "down", since only an agent reporting "up" will be able to turn the
11293 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011294 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011295
Willy Tarreau81f5d942013-12-09 20:51:51 +010011296 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011297 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011298
James Brown55f9ff12015-10-21 18:19:05 -070011299agent-send <string>
11300 If this option is specified, haproxy will send the given string (verbatim)
11301 to the agent server upon connection. You could, for example, encode
11302 the backend name into this string, which would enable your agent to send
11303 different responses based on the backend. Make sure to include a '\n' if
11304 you want to terminate your request with a newline.
11305
Simon Hormand60d6912013-11-25 10:46:36 +090011306agent-inter <delay>
11307 The "agent-inter" parameter sets the interval between two agent checks
11308 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11309
11310 Just as with every other time-based parameter, it may be entered in any
11311 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11312 parameter also serves as a timeout for agent checks "timeout check" is
11313 not set. In order to reduce "resonance" effects when multiple servers are
11314 hosted on the same hardware, the agent and health checks of all servers
11315 are started with a small time offset between them. It is also possible to
11316 add some random noise in the agent and health checks interval using the
11317 global "spread-checks" keyword. This makes sense for instance when a lot
11318 of backends use the same servers.
11319
11320 See also the "agent-check" and "agent-port" parameters.
11321
Misiek768d8602017-01-09 09:52:43 +010011322agent-addr <addr>
11323 The "agent-addr" parameter sets address for agent check.
11324
11325 You can offload agent-check to another target, so you can make single place
11326 managing status and weights of servers defined in haproxy in case you can't
11327 make self-aware and self-managing services. You can specify both IP or
11328 hostname, it will be resolved.
11329
Simon Hormand60d6912013-11-25 10:46:36 +090011330agent-port <port>
11331 The "agent-port" parameter sets the TCP port used for agent checks.
11332
11333 See also the "agent-check" and "agent-inter" parameters.
11334
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011335backup
11336 When "backup" is present on a server line, the server is only used in load
11337 balancing when all other non-backup servers are unavailable. Requests coming
11338 with a persistence cookie referencing the server will always be served
11339 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011340 the "allbackups" option is set in the backend. See also the "no-backup" and
11341 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011342
Emeric Brunef42d922012-10-11 16:11:36 +020011343ca-file <cafile>
11344 This setting is only available when support for OpenSSL was built in. It
11345 designates a PEM file from which to load CA certificates used to verify
11346 server's certificate.
11347
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011348check
11349 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011350 always considered available. If "check" is set, the server is available when
11351 accepting periodic TCP connections, to ensure that it is really able to serve
11352 requests. The default address and port to send the tests to are those of the
11353 server, and the default source is the same as the one defined in the
11354 backend. It is possible to change the address using the "addr" parameter, the
11355 port using the "port" parameter, the source address using the "source"
11356 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011357 parameters. The request method is define in the backend using the "httpchk",
11358 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011359 refer to those options and parameters for more information. See also
11360 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011361
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011362check-send-proxy
11363 This option forces emission of a PROXY protocol line with outgoing health
11364 checks, regardless of whether the server uses send-proxy or not for the
11365 normal traffic. By default, the PROXY protocol is enabled for health checks
11366 if it is already enabled for normal traffic and if no "port" nor "addr"
11367 directive is present. However, if such a directive is present, the
11368 "check-send-proxy" option needs to be used to force the use of the
11369 protocol. See also the "send-proxy" option for more information.
11370
Olivier Houchard9130a962017-10-17 17:33:43 +020011371check-sni
11372 This option allows you to specify the SNI to be used when doing health checks
11373 over SSL.
11374
Willy Tarreau763a95b2012-10-04 23:15:39 +020011375check-ssl
11376 This option forces encryption of all health checks over SSL, regardless of
11377 whether the server uses SSL or not for the normal traffic. This is generally
11378 used when an explicit "port" or "addr" directive is specified and SSL health
11379 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011380 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011381 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11382 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011383 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011384 See the "ssl" option for more information and "no-check-ssl" to disable
11385 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011386
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011387ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011388 This setting is only available when support for OpenSSL was built in. This
11389 option sets the string describing the list of cipher algorithms that is
11390 negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011391 string is defined in "man 1 ciphers". When SSL is used to communicate with
11392 servers on the local network, it is common to see a weaker set of algorithms
11393 than what is used over the internet. Doing so reduces CPU usage on both the
11394 server and haproxy while still keeping it compatible with deployed software.
11395 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
11396 is needed and just connectivity, using DES can be appropriate.
11397
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011398ciphersuites <ciphersuites>
11399 This setting is only available when support for OpenSSL was built in and
11400 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11401 describing the list of cipher algorithms that is negotiated during the TLS
11402 1.3 handshake with the server. The format of the string is defined in
11403 "man 1 ciphers" under the "ciphersuites" section.
11404
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011405cookie <value>
11406 The "cookie" parameter sets the cookie value assigned to the server to
11407 <value>. This value will be checked in incoming requests, and the first
11408 operational server possessing the same value will be selected. In return, in
11409 cookie insertion or rewrite modes, this value will be assigned to the cookie
11410 sent to the client. There is nothing wrong in having several servers sharing
11411 the same cookie value, and it is in fact somewhat common between normal and
11412 backup servers. See also the "cookie" keyword in backend section.
11413
Emeric Brunef42d922012-10-11 16:11:36 +020011414crl-file <crlfile>
11415 This setting is only available when support for OpenSSL was built in. It
11416 designates a PEM file from which to load certificate revocation list used
11417 to verify server's certificate.
11418
Emeric Bruna7aa3092012-10-26 12:58:00 +020011419crt <cert>
11420 This setting is only available when support for OpenSSL was built in.
11421 It designates a PEM file from which to load both a certificate and the
11422 associated private key. This file can be built by concatenating both PEM
11423 files into one. This certificate will be sent if the server send a client
11424 certificate request.
11425
Willy Tarreau96839092010-03-29 10:02:24 +020011426disabled
11427 The "disabled" keyword starts the server in the "disabled" state. That means
11428 that it is marked down in maintenance mode, and no connection other than the
11429 ones allowed by persist mode will reach it. It is very well suited to setup
11430 new servers, because normal traffic will never reach them, while it is still
11431 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011432 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011433
Frédéric Lécailled2376272017-03-21 18:52:12 +010011434enabled
11435 This option may be used as 'server' setting to reset any 'disabled'
11436 setting which would have been inherited from 'default-server' directive as
11437 default value.
11438 It may also be used as 'default-server' setting to reset any previous
11439 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011440
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011441error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011442 If health observing is enabled, the "error-limit" parameter specifies the
11443 number of consecutive errors that triggers event selected by the "on-error"
11444 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011445
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011446 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011447
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011448fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011449 The "fall" parameter states that a server will be considered as dead after
11450 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11451 unspecified. See also the "check", "inter" and "rise" parameters.
11452
Emeric Brun8694b9a2012-10-05 14:39:07 +020011453force-sslv3
11454 This option enforces use of SSLv3 only when SSL is used to communicate with
11455 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011456 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011457 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011458
11459force-tlsv10
11460 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011461 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011462 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011463
11464force-tlsv11
11465 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011466 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011467 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011468
11469force-tlsv12
11470 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011471 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011472 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011473
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011474force-tlsv13
11475 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11476 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011477 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011478
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011479id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011480 Set a persistent ID for the server. This ID must be positive and unique for
11481 the proxy. An unused ID will automatically be assigned if unset. The first
11482 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011483
Willy Tarreau6a031d12016-11-07 19:42:35 +010011484init-addr {last | libc | none | <ip>},[...]*
11485 Indicate in what order the server's address should be resolved upon startup
11486 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011487 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011488 method which succeeds is used. If the end of the list is reached without
11489 finding a working method, an error is thrown. Method "last" suggests to pick
11490 the address which appears in the state file (see "server-state-file"). Method
11491 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11492 depending on the operating system and build options). Method "none"
11493 specifically indicates that the server should start without any valid IP
11494 address in a down state. It can be useful to ignore some DNS issues upon
11495 startup, waiting for the situation to get fixed later. Finally, an IP address
11496 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011497 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011498 server used to catch old sessions and present them with a decent error
11499 message for example. When the "first" load balancing algorithm is used, this
11500 IP address could point to a fake server used to trigger the creation of new
11501 instances on the fly. This option defaults to "last,libc" indicating that the
11502 previous address found in the state file (if any) is used first, otherwise
11503 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011504 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011505
11506 Example:
11507 defaults
11508 # never fail on address resolution
11509 default-server init-addr last,libc,none
11510
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011511inter <delay>
11512fastinter <delay>
11513downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011514 The "inter" parameter sets the interval between two consecutive health checks
11515 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11516 It is also possible to use "fastinter" and "downinter" to optimize delays
11517 between checks depending on the server state :
11518
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011519 Server state | Interval used
11520 ----------------------------------------+----------------------------------
11521 UP 100% (non-transitional) | "inter"
11522 ----------------------------------------+----------------------------------
11523 Transitionally UP (going down "fall"), | "fastinter" if set,
11524 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11525 or yet unchecked. |
11526 ----------------------------------------+----------------------------------
11527 DOWN 100% (non-transitional) | "downinter" if set,
11528 | "inter" otherwise.
11529 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011530
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011531 Just as with every other time-based parameter, they can be entered in any
11532 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11533 serves as a timeout for health checks sent to servers if "timeout check" is
11534 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011535 hosted on the same hardware, the agent and health checks of all servers
11536 are started with a small time offset between them. It is also possible to
11537 add some random noise in the agent and health checks interval using the
11538 global "spread-checks" keyword. This makes sense for instance when a lot
11539 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011540
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011541maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011542 The "maxconn" parameter specifies the maximal number of concurrent
11543 connections that will be sent to this server. If the number of incoming
11544 concurrent requests goes higher than this value, they will be queued, waiting
11545 for a connection to be released. This parameter is very important as it can
11546 save fragile servers from going down under extreme loads. If a "minconn"
11547 parameter is specified, the limit becomes dynamic. The default value is "0"
11548 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11549 the backend's "fullconn" keyword.
11550
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011551maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011552 The "maxqueue" parameter specifies the maximal number of connections which
11553 will wait in the queue for this server. If this limit is reached, next
11554 requests will be redispatched to other servers instead of indefinitely
11555 waiting to be served. This will break persistence but may allow people to
11556 quickly re-log in when the server they try to connect to is dying. The
11557 default value is "0" which means the queue is unlimited. See also the
11558 "maxconn" and "minconn" parameters.
11559
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011560minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011561 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11562 limit following the backend's load. The server will always accept at least
11563 <minconn> connections, never more than <maxconn>, and the limit will be on
11564 the ramp between both values when the backend has less than <fullconn>
11565 concurrent connections. This makes it possible to limit the load on the
11566 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011567 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011568 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011569
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011570namespace <name>
11571 On Linux, it is possible to specify which network namespace a socket will
11572 belong to. This directive makes it possible to explicitly bind a server to
11573 a namespace different from the default one. Please refer to your operating
11574 system's documentation to find more details about network namespaces.
11575
Frédéric Lécailled2376272017-03-21 18:52:12 +010011576no-agent-check
11577 This option may be used as "server" setting to reset any "agent-check"
11578 setting which would have been inherited from "default-server" directive as
11579 default value.
11580 It may also be used as "default-server" setting to reset any previous
11581 "default-server" "agent-check" setting.
11582
11583no-backup
11584 This option may be used as "server" setting to reset any "backup"
11585 setting which would have been inherited from "default-server" directive as
11586 default value.
11587 It may also be used as "default-server" setting to reset any previous
11588 "default-server" "backup" setting.
11589
11590no-check
11591 This option may be used as "server" setting to reset any "check"
11592 setting which would have been inherited from "default-server" directive as
11593 default value.
11594 It may also be used as "default-server" setting to reset any previous
11595 "default-server" "check" setting.
11596
11597no-check-ssl
11598 This option may be used as "server" setting to reset any "check-ssl"
11599 setting which would have been inherited from "default-server" directive as
11600 default value.
11601 It may also be used as "default-server" setting to reset any previous
11602 "default-server" "check-ssl" setting.
11603
Frédéric Lécailled2376272017-03-21 18:52:12 +010011604no-send-proxy
11605 This option may be used as "server" setting to reset any "send-proxy"
11606 setting which would have been inherited from "default-server" directive as
11607 default value.
11608 It may also be used as "default-server" setting to reset any previous
11609 "default-server" "send-proxy" setting.
11610
11611no-send-proxy-v2
11612 This option may be used as "server" setting to reset any "send-proxy-v2"
11613 setting which would have been inherited from "default-server" directive as
11614 default value.
11615 It may also be used as "default-server" setting to reset any previous
11616 "default-server" "send-proxy-v2" setting.
11617
11618no-send-proxy-v2-ssl
11619 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11620 setting which would have been inherited from "default-server" directive as
11621 default value.
11622 It may also be used as "default-server" setting to reset any previous
11623 "default-server" "send-proxy-v2-ssl" setting.
11624
11625no-send-proxy-v2-ssl-cn
11626 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11627 setting which would have been inherited from "default-server" directive as
11628 default value.
11629 It may also be used as "default-server" setting to reset any previous
11630 "default-server" "send-proxy-v2-ssl-cn" setting.
11631
11632no-ssl
11633 This option may be used as "server" setting to reset any "ssl"
11634 setting which would have been inherited from "default-server" directive as
11635 default value.
11636 It may also be used as "default-server" setting to reset any previous
11637 "default-server" "ssl" setting.
11638
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011639no-ssl-reuse
11640 This option disables SSL session reuse when SSL is used to communicate with
11641 the server. It will force the server to perform a full handshake for every
11642 new connection. It's probably only useful for benchmarking, troubleshooting,
11643 and for paranoid users.
11644
Emeric Brun9b3009b2012-10-05 11:55:06 +020011645no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011646 This option disables support for SSLv3 when SSL is used to communicate with
11647 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011648 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011649
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011650 Supported in default-server: No
11651
Emeric Brunf9c5c472012-10-11 15:28:34 +020011652no-tls-tickets
11653 This setting is only available when support for OpenSSL was built in. It
11654 disables the stateless session resumption (RFC 5077 TLS Ticket
11655 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011656 session resumption is more expensive in CPU usage for servers. This option
11657 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011658 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011659
Emeric Brun9b3009b2012-10-05 11:55:06 +020011660no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011661 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011662 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11663 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011664 often makes sense to disable it when communicating with local servers. This
11665 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011666 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011667
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011668 Supported in default-server: No
11669
Emeric Brun9b3009b2012-10-05 11:55:06 +020011670no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011671 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011672 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11673 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011674 often makes sense to disable it when communicating with local servers. This
11675 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011676 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011677
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011678 Supported in default-server: No
11679
Emeric Brun9b3009b2012-10-05 11:55:06 +020011680no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011681 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011682 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11683 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011684 often makes sense to disable it when communicating with local servers. This
11685 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011686 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011687
11688 Supported in default-server: No
11689
11690no-tlsv13
11691 This option disables support for TLSv1.3 when SSL is used to communicate with
11692 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11693 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11694 often makes sense to disable it when communicating with local servers. This
11695 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011696 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011697
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011698 Supported in default-server: No
11699
Frédéric Lécailled2376272017-03-21 18:52:12 +010011700no-verifyhost
11701 This option may be used as "server" setting to reset any "verifyhost"
11702 setting which would have been inherited from "default-server" directive as
11703 default value.
11704 It may also be used as "default-server" setting to reset any previous
11705 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011706
Simon Hormanfa461682011-06-25 09:39:49 +090011707non-stick
11708 Never add connections allocated to this sever to a stick-table.
11709 This may be used in conjunction with backup to ensure that
11710 stick-table persistence is disabled for backup servers.
11711
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011712observe <mode>
11713 This option enables health adjusting based on observing communication with
11714 the server. By default this functionality is disabled and enabling it also
11715 requires to enable health checks. There are two supported modes: "layer4" and
11716 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11717 significant. In layer7, which is only allowed for http proxies, responses
11718 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011719 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011720
11721 See also the "check", "on-error" and "error-limit".
11722
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011723on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011724 Select what should happen when enough consecutive errors are detected.
11725 Currently, four modes are available:
11726 - fastinter: force fastinter
11727 - fail-check: simulate a failed check, also forces fastinter (default)
11728 - sudden-death: simulate a pre-fatal failed health check, one more failed
11729 check will mark a server down, forces fastinter
11730 - mark-down: mark the server immediately down and force fastinter
11731
11732 See also the "check", "observe" and "error-limit".
11733
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011734on-marked-down <action>
11735 Modify what occurs when a server is marked down.
11736 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011737 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11738 all connections to the server are immediately terminated when the server
11739 goes down. It might be used if the health check detects more complex cases
11740 than a simple connection status, and long timeouts would cause the service
11741 to remain unresponsive for too long a time. For instance, a health check
11742 might detect that a database is stuck and that there's no chance to reuse
11743 existing connections anymore. Connections killed this way are logged with
11744 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011745
11746 Actions are disabled by default
11747
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011748on-marked-up <action>
11749 Modify what occurs when a server is marked up.
11750 Currently one action is available:
11751 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11752 done only if the server is not in backup state and if it is not disabled
11753 (it must have an effective weight > 0). This can be used sometimes to force
11754 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011755 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11756 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011757 with extreme care. Sessions killed because a server comes up are logged
11758 with an 'U' termination code (for "Up").
11759
11760 Actions are disabled by default
11761
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011762port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011763 Using the "port" parameter, it becomes possible to use a different port to
11764 send health-checks. On some servers, it may be desirable to dedicate a port
11765 to a specific component able to perform complex tests which are more suitable
11766 to health-checks than the application. It is common to run a simple script in
11767 inetd for instance. This parameter is ignored if the "check" parameter is not
11768 set. See also the "addr" parameter.
11769
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011770proto <name>
11771
11772 Forces the multiplexer's protocol to use for the outgoing connections to this
11773 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11774 must also be usable on the backend side. The list of available protocols is
11775 reported in haproxy -vv.
11776 Idea behind this optipon is to bypass the selection of the best multiplexer's
11777 protocol for all connections established to this server.
11778
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011779redir <prefix>
11780 The "redir" parameter enables the redirection mode for all GET and HEAD
11781 requests addressing this server. This means that instead of having HAProxy
11782 forward the request to the server, it will send an "HTTP 302" response with
11783 the "Location" header composed of this prefix immediately followed by the
11784 requested URI beginning at the leading '/' of the path component. That means
11785 that no trailing slash should be used after <prefix>. All invalid requests
11786 will be rejected, and all non-GET or HEAD requests will be normally served by
11787 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011788 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011789 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011790 users to a remote location in case of local disaster. Main use consists in
11791 increasing bandwidth for static servers by having the clients directly
11792 connect to them. Note: never use a relative location here, it would cause a
11793 loop between the client and HAProxy!
11794
11795 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11796
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011797rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011798 The "rise" parameter states that a server will be considered as operational
11799 after <count> consecutive successful health checks. This value defaults to 2
11800 if unspecified. See also the "check", "inter" and "fall" parameters.
11801
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020011802resolve-opts <option>,<option>,...
11803 Comma separated list of options to apply to DNS resolution linked to this
11804 server.
11805
11806 Available options:
11807
11808 * allow-dup-ip
11809 By default, HAProxy prevents IP address duplication in a backend when DNS
11810 resolution at runtime is in operation.
11811 That said, for some cases, it makes sense that two servers (in the same
11812 backend, being resolved by the same FQDN) have the same IP address.
11813 For such case, simply enable this option.
11814 This is the opposite of prevent-dup-ip.
11815
11816 * prevent-dup-ip
11817 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
11818 an IP address already set to a server in the same backend and sharing the
11819 same fqdn.
11820 This is the opposite of allow-dup-ip.
11821
11822 Example:
11823 backend b_myapp
11824 default-server init-addr none resolvers dns
11825 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
11826 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
11827
11828 With the option allow-dup-ip set:
11829 * if the nameserver returns a single IP address, then both servers will use
11830 it
11831 * If the nameserver returns 2 IP addresses, then each server will pick up a
11832 different address
11833
11834 Default value: not set
11835
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011836resolve-prefer <family>
11837 When DNS resolution is enabled for a server and multiple IP addresses from
11838 different families are returned, HAProxy will prefer using an IP address
11839 from the family mentioned in the "resolve-prefer" parameter.
11840 Available families: "ipv4" and "ipv6"
11841
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011842 Default value: ipv6
11843
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011844 Example:
11845
11846 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011847
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011848resolve-net <network>[,<network[,...]]
11849 This options prioritize th choice of an ip address matching a network. This is
11850 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011851 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010011852 different datacenters. The latency between datacenter is not negligible, so
11853 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011854 configured network, another address is selected.
11855
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011856 Example:
11857
11858 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011859
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011860resolvers <id>
11861 Points to an existing "resolvers" section to resolve current server's
11862 hostname.
11863
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011864 Example:
11865
11866 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011867
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011868 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011869
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011870send-proxy
11871 The "send-proxy" parameter enforces use of the PROXY protocol over any
11872 connection established to this server. The PROXY protocol informs the other
11873 end about the layer 3/4 addresses of the incoming connection, so that it can
11874 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011875 upper layer protocol. For connections accepted by an "accept-proxy" or
11876 "accept-netscaler-cip" listener, the advertised address will be used. Only
11877 TCPv4 and TCPv6 address families are supported. Other families such as
11878 Unix sockets, will report an UNKNOWN family. Servers using this option can
11879 fully be chained to another instance of haproxy listening with an
11880 "accept-proxy" setting. This setting must not be used if the server isn't
11881 aware of the protocol. When health checks are sent to the server, the PROXY
11882 protocol is automatically used when this option is set, unless there is an
11883 explicit "port" or "addr" directive, in which case an explicit
11884 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011885 See also the "no-send-proxy" option of this section and "accept-proxy" and
11886 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011887
David Safb76832014-05-08 23:42:08 -040011888send-proxy-v2
11889 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
11890 over any connection established to this server. The PROXY protocol informs
11891 the other end about the layer 3/4 addresses of the incoming connection, so
11892 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020011893 whatever the upper layer protocol. It also send ALPN information if an alpn
11894 have been negotiated. This setting must not be used if the server isn't aware
11895 of this version of the protocol. See also the "no-send-proxy-v2" option of
11896 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011897
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011898proxy-v2-options <option>[,<option>]*
11899 The "proxy-v2-options" parameter add option to send in PROXY protocol version
11900 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010011901 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
11902 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010011903 certificate, "cert-key": key algorithm of the used certificate), "authority":
11904 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010011905 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011906
David Safb76832014-05-08 23:42:08 -040011907send-proxy-v2-ssl
11908 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11909 2 over any connection established to this server. The PROXY protocol informs
11910 the other end about the layer 3/4 addresses of the incoming connection, so
11911 that it can know the client's address or the public address it accessed to,
11912 whatever the upper layer protocol. In addition, the SSL information extension
11913 of the PROXY protocol is added to the PROXY protocol header. This setting
11914 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011915 See also the "no-send-proxy-v2-ssl" option of this section and the
11916 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011917
11918send-proxy-v2-ssl-cn
11919 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11920 2 over any connection established to this server. The PROXY protocol informs
11921 the other end about the layer 3/4 addresses of the incoming connection, so
11922 that it can know the client's address or the public address it accessed to,
11923 whatever the upper layer protocol. In addition, the SSL information extension
11924 of the PROXY protocol, along along with the Common Name from the subject of
11925 the client certificate (if any), is added to the PROXY protocol header. This
11926 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011927 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
11928 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011929
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011930slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011931 The "slowstart" parameter for a server accepts a value in milliseconds which
11932 indicates after how long a server which has just come back up will run at
11933 full speed. Just as with every other time-based parameter, it can be entered
11934 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
11935 linearly from 0 to 100% during this time. The limitation applies to two
11936 parameters :
11937
11938 - maxconn: the number of connections accepted by the server will grow from 1
11939 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
11940
11941 - weight: when the backend uses a dynamic weighted algorithm, the weight
11942 grows linearly from 1 to 100%. In this case, the weight is updated at every
11943 health-check. For this reason, it is important that the "inter" parameter
11944 is smaller than the "slowstart", in order to maximize the number of steps.
11945
11946 The slowstart never applies when haproxy starts, otherwise it would cause
11947 trouble to running servers. It only applies when a server has been previously
11948 seen as failed.
11949
Willy Tarreau732eac42015-07-09 11:40:25 +020011950sni <expression>
11951 The "sni" parameter evaluates the sample fetch expression, converts it to a
11952 string and uses the result as the host name sent in the SNI TLS extension to
11953 the server. A typical use case is to send the SNI received from the client in
11954 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020011955 expression, though alternatives such as req.hdr(host) can also make sense. If
11956 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011957 name will also be matched against the server certificate's names. See the
11958 "verify" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020011959
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011960source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020011961source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011962source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011963 The "source" parameter sets the source address which will be used when
11964 connecting to the server. It follows the exact same parameters and principle
11965 as the backend "source" keyword, except that it only applies to the server
11966 referencing it. Please consult the "source" keyword for details.
11967
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011968 Additionally, the "source" statement on a server line allows one to specify a
11969 source port range by indicating the lower and higher bounds delimited by a
11970 dash ('-'). Some operating systems might require a valid IP address when a
11971 source port range is specified. It is permitted to have the same IP/range for
11972 several servers. Doing so makes it possible to bypass the maximum of 64k
11973 total concurrent connections. The limit will then reach 64k connections per
11974 server.
11975
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000011976 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
11977 specifying the source address without port(s).
11978
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011979ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020011980 This option enables SSL ciphering on outgoing connections to the server. It
11981 is critical to verify server certificates using "verify" when using SSL to
11982 connect to servers, otherwise the communication is prone to trivial man in
11983 the-middle attacks rendering SSL useless. When this option is used, health
11984 checks are automatically sent in SSL too unless there is a "port" or an
11985 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011986 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
11987 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011988
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011989ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11990 This option enforces use of <version> or lower when SSL is used to communicate
11991 with the server. This option is also available on global statement
11992 "ssl-default-server-options". See also "ssl-min-ver".
11993
11994ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11995 This option enforces use of <version> or upper when SSL is used to communicate
11996 with the server. This option is also available on global statement
11997 "ssl-default-server-options". See also "ssl-max-ver".
11998
Frédéric Lécailled2376272017-03-21 18:52:12 +010011999ssl-reuse
12000 This option may be used as "server" setting to reset any "no-ssl-reuse"
12001 setting which would have been inherited from "default-server" directive as
12002 default value.
12003 It may also be used as "default-server" setting to reset any previous
12004 "default-server" "no-ssl-reuse" setting.
12005
12006stick
12007 This option may be used as "server" setting to reset any "non-stick"
12008 setting which would have been inherited from "default-server" directive as
12009 default value.
12010 It may also be used as "default-server" setting to reset any previous
12011 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012012
Willy Tarreau163d4622015-10-13 16:16:41 +020012013tcp-ut <delay>
12014 Sets the TCP User Timeout for all outgoing connections to this server. This
12015 option is available on Linux since version 2.6.37. It allows haproxy to
12016 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012017 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012018 long-lived connections experiencing long idle periods such as remote
12019 terminals or database connection pools, where the client and server timeouts
12020 must remain high to allow a long period of idle, but where it is important to
12021 detect that the server has disappeared in order to release all resources
12022 associated with its connection (and the client's session). One typical use
12023 case is also to force dead server connections to die when health checks are
12024 too slow or during a soft reload since health checks are then disabled. The
12025 argument is a delay expressed in milliseconds by default. This only works for
12026 regular TCP connections, and is ignored for other protocols.
12027
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012028track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012029 This option enables ability to set the current state of the server by tracking
12030 another one. It is possible to track a server which itself tracks another
12031 server, provided that at the end of the chain, a server has health checks
12032 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012033 used, it has to be enabled on both proxies.
12034
Frédéric Lécailled2376272017-03-21 18:52:12 +010012035tls-tickets
12036 This option may be used as "server" setting to reset any "no-tls-tickets"
12037 setting which would have been inherited from "default-server" directive as
12038 default value.
12039 It may also be used as "default-server" setting to reset any previous
12040 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012041
Emeric Brunef42d922012-10-11 16:11:36 +020012042verify [none|required]
12043 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012044 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012045 certificate provided by the server is verified using CAs from 'ca-file' and
12046 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012047 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012048 the name passed using the "sni" directive, or if not provided, the static
12049 host name passed using the "verifyhost" directive. When no name is found, the
12050 certificate's names are ignored. For this reason, without SNI it's important
12051 to use "verifyhost". On verification failure the handshake is aborted. It is
12052 critically important to verify server certificates when using SSL to connect
12053 to servers, otherwise the communication is prone to trivial man-in-the-middle
12054 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12055 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012056
Evan Broderbe554312013-06-27 00:05:25 -070012057verifyhost <hostname>
12058 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012059 only takes effect if 'verify required' is also specified. This directive sets
12060 a default static hostname to check the server's certificate against when no
12061 SNI was used to connect to the server. If SNI is not used, this is the only
12062 way to enable hostname verification. This static hostname, when set, will
12063 also be used for health checks (which cannot provide an SNI value). If none
12064 of the hostnames in the certificate match the specified hostname, the
12065 handshake is aborted. The hostnames in the server-provided certificate may
12066 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012067
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012068weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012069 The "weight" parameter is used to adjust the server's weight relative to
12070 other servers. All servers will receive a load proportional to their weight
12071 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012072 load. The default weight is 1, and the maximal value is 256. A value of 0
12073 means the server will not participate in load-balancing but will still accept
12074 persistent connections. If this parameter is used to distribute the load
12075 according to server's capacity, it is recommended to start with values which
12076 can both grow and shrink, for instance between 10 and 100 to leave enough
12077 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012078
12079
Cyril Bonté46175dd2015-07-02 22:45:32 +0200120805.3. Server IP address resolution using DNS
12081-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012082
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012083HAProxy allows using a host name on the server line to retrieve its IP address
12084using name servers. By default, HAProxy resolves the name when parsing the
12085configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012086This is not sufficient in some cases, such as in Amazon where a server's IP
12087can change after a reboot or an ELB Virtual IP can change based on current
12088workload.
12089This chapter describes how HAProxy can be configured to process server's name
12090resolution at run time.
12091Whether run time server name resolution has been enable or not, HAProxy will
12092carry on doing the first resolution when parsing the configuration.
12093
12094
Cyril Bonté46175dd2015-07-02 22:45:32 +0200120955.3.1. Global overview
12096----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012097
12098As we've seen in introduction, name resolution in HAProxy occurs at two
12099different steps of the process life:
12100
12101 1. when starting up, HAProxy parses the server line definition and matches a
12102 host name. It uses libc functions to get the host name resolved. This
12103 resolution relies on /etc/resolv.conf file.
12104
Christopher Faulet67957bd2017-09-27 11:00:59 +020012105 2. at run time, HAProxy performs periodically name resolutions for servers
12106 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012107
12108A few other events can trigger a name resolution at run time:
12109 - when a server's health check ends up in a connection timeout: this may be
12110 because the server has a new IP address. So we need to trigger a name
12111 resolution to know this new IP.
12112
Christopher Faulet67957bd2017-09-27 11:00:59 +020012113When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012114HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012115SRV label is specified, then the corresponding SRV records will be retrieved
12116from the DNS server, and the provided hostnames will be used. The SRV label
12117will be checked periodically, and if any server are added or removed, haproxy
12118will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012119
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012120A few things important to notice:
12121 - all the name servers are queried in the mean time. HAProxy will process the
12122 first valid response.
12123
12124 - a resolution is considered as invalid (NX, timeout, refused), when all the
12125 servers return an error.
12126
12127
Cyril Bonté46175dd2015-07-02 22:45:32 +0200121285.3.2. The resolvers section
12129----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012130
12131This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012132HAProxy. There can be as many as resolvers section as needed. Each section can
12133contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012134
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012135When multiple name servers are configured in a resolvers section, then HAProxy
12136uses the first valid response. In case of invalid responses, only the last one
12137is treated. Purpose is to give the chance to a slow server to deliver a valid
12138answer after a fast faulty or outdated server.
12139
12140When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012141used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012142
Christopher Faulet67957bd2017-09-27 11:00:59 +020012143 1. HAProxy retries the same DNS query with a new query type. The A queries are
12144 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12145 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012146
Christopher Faulet67957bd2017-09-27 11:00:59 +020012147 2. When the fallback on the query type was done (or not applicable), HAProxy
12148 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012149
Christopher Faulet67957bd2017-09-27 11:00:59 +020012150 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12151 response is received after that, it stops the DNS resolution and reports
12152 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012153
Christopher Faulet67957bd2017-09-27 11:00:59 +020012154For example, with 2 name servers configured in a resolvers section, the
12155following scenarios are possible:
12156
12157 - First response is valid and is applied directly, second response is
12158 ignored
12159
12160 - First response is invalid and second one is valid, then second response is
12161 applied
12162
12163 - First response is a NX domain and second one a truncated response, then
12164 HAProxy retries the query with a new type
12165
12166 - First response is a NX domain and second one is a timeout, then HAProxy
12167 retries the query with a new type
12168
12169 - Query timed out for both name servers, then HAProxy retries it with the
12170 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012171
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012172As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12173a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012174<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012175
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012176
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012177resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012178 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012179
12180A resolvers section accept the following parameters:
12181
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012182accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012183 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012184 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012185 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12186 by RFC 6891)
12187
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012188 Note: the maximum allowed value is 8192.
12189
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012190nameserver <id> <ip>:<port>
12191 DNS server description:
12192 <id> : label of the server, should be unique
12193 <ip> : IP address of the server
12194 <port> : port where the DNS service actually runs
12195
Ben Draut44e609b2018-05-29 15:40:08 -060012196parse-resolv-conf
12197 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12198 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12199 placed in the resolvers section in place of this directive.
12200
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012201hold <status> <period>
12202 Defines <period> during which the last name resolution should be kept based
12203 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012204 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012205 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012206 <period> : interval between two successive name resolution when the last
12207 answer was in <status>. It follows the HAProxy time format.
12208 <period> is in milliseconds by default.
12209
Baptiste Assmann686408b2017-08-18 10:15:42 +020012210 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012211
Christopher Faulet67957bd2017-09-27 11:00:59 +020012212resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012213 Defines the number of resolutions available in the pool for this resolvers.
12214 If not defines, it defaults to 64. If your configuration requires more than
12215 <nb>, then HAProxy will return an error when parsing the configuration.
12216
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012217resolve_retries <nb>
12218 Defines the number <nb> of queries to send to resolve a server name before
12219 giving up.
12220 Default value: 3
12221
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012222 A retry occurs on name server timeout or when the full sequence of DNS query
12223 type failover is over and we need to start up from the default ANY query
12224 type.
12225
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012226timeout <event> <time>
12227 Defines timeouts related to name resolution
12228 <event> : the event on which the <time> timeout period applies to.
12229 events available are:
Christopher Faulet67957bd2017-09-27 11:00:59 +020012230 - resolve : default time to trigger name resolutions when no
12231 other time applied.
12232 Default value: 1s
12233 - retry : time between two DNS queries, when no valid response
12234 have been received.
12235 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012236 <time> : time related to the event. It follows the HAProxy time format.
12237 <time> is expressed in milliseconds.
12238
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012239 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012240
12241 resolvers mydns
12242 nameserver dns1 10.0.0.1:53
12243 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012244 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012245 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012246 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012247 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012248 hold other 30s
12249 hold refused 30s
12250 hold nx 30s
12251 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012252 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012253 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012254
12255
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122566. HTTP header manipulation
12257---------------------------
12258
12259In HTTP mode, it is possible to rewrite, add or delete some of the request and
12260response headers based on regular expressions. It is also possible to block a
12261request or a response if a particular header matches a regular expression,
12262which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012263against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012264
Willy Tarreau70dffda2014-01-30 03:07:23 +010012265If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12266to process all rsp* rules which can allow, deny, rewrite or delete a header,
12267but it will refuse to add a header to any such messages as this is not
12268HTTP-compliant. The reason for still processing headers in such responses is to
12269stop and/or fix any possible information leak which may happen, for instance
12270because another downstream equipment would unconditionally add a header, or if
12271a server name appears there. When such messages are seen, normal processing
12272still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012273
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012274This section covers common usage of the following keywords, described in detail
12275in section 4.2 :
12276
12277 - reqadd <string>
12278 - reqallow <search>
12279 - reqiallow <search>
12280 - reqdel <search>
12281 - reqidel <search>
12282 - reqdeny <search>
12283 - reqideny <search>
12284 - reqpass <search>
12285 - reqipass <search>
12286 - reqrep <search> <replace>
12287 - reqirep <search> <replace>
12288 - reqtarpit <search>
12289 - reqitarpit <search>
12290 - rspadd <string>
12291 - rspdel <search>
12292 - rspidel <search>
12293 - rspdeny <search>
12294 - rspideny <search>
12295 - rsprep <search> <replace>
12296 - rspirep <search> <replace>
12297
12298With all these keywords, the same conventions are used. The <search> parameter
12299is a POSIX extended regular expression (regex) which supports grouping through
12300parenthesis (without the backslash). Spaces and other delimiters must be
12301prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12302Other characters may be prefixed with a backslash to change their meaning :
12303
12304 \t for a tab
12305 \r for a carriage return (CR)
12306 \n for a new line (LF)
12307 \ to mark a space and differentiate it from a delimiter
12308 \# to mark a sharp and differentiate it from a comment
12309 \\ to use a backslash in a regex
12310 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12311 \xXX to write the ASCII hex code XX as in the C language
12312
12313The <replace> parameter contains the string to be used to replace the largest
12314portion of text matching the regex. It can make use of the special characters
12315above, and can reference a substring which is delimited by parenthesis in the
12316regex, by writing a backslash ('\') immediately followed by one digit from 0 to
123179 indicating the group position (0 designating the entire line). This practice
12318is very common to users of the "sed" program.
12319
12320The <string> parameter represents the string which will systematically be added
12321after the last header line. It can also use special character sequences above.
12322
12323Notes related to these keywords :
12324---------------------------------
12325 - these keywords are not always convenient to allow/deny based on header
12326 contents. It is strongly recommended to use ACLs with the "block" keyword
12327 instead, resulting in far more flexible and manageable rules.
12328
12329 - lines are always considered as a whole. It is not possible to reference
12330 a header name only or a value only. This is important because of the way
12331 headers are written (notably the number of spaces after the colon).
12332
12333 - the first line is always considered as a header, which makes it possible to
12334 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12335 it harder to distinguish between headers and request line. The regex prefix
12336 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12337 ^[^ \t:]*: matches any header name followed by a colon.
12338
12339 - for performances reasons, the number of characters added to a request or to
12340 a response is limited at build time to values between 1 and 4 kB. This
12341 should normally be far more than enough for most usages. If it is too short
12342 on occasional usages, it is possible to gain some space by removing some
12343 useless headers before adding new ones.
12344
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012345 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012346 without the 'i' letter except that they ignore case when matching patterns.
12347
12348 - when a request passes through a frontend then a backend, all req* rules
12349 from the frontend will be evaluated, then all req* rules from the backend
12350 will be evaluated. The reverse path is applied to responses.
12351
12352 - req* statements are applied after "block" statements, so that "block" is
12353 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012354 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012355
12356
Willy Tarreau74ca5042013-06-11 23:12:07 +0200123577. Using ACLs and fetching samples
12358----------------------------------
12359
Davor Ocelice9ed2812017-12-25 17:49:28 +010012360HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012361client or server information, from tables, environmental information etc...
12362The action of extracting such data is called fetching a sample. Once retrieved,
12363these samples may be used for various purposes such as a key to a stick-table,
12364but most common usages consist in matching them against predefined constant
12365data called patterns.
12366
12367
123687.1. ACL basics
12369---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012370
12371The use of Access Control Lists (ACL) provides a flexible solution to perform
12372content switching and generally to take decisions based on content extracted
12373from the request, the response or any environmental status. The principle is
12374simple :
12375
Willy Tarreau74ca5042013-06-11 23:12:07 +020012376 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012377 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012378 - apply one or multiple pattern matching methods on this sample
12379 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012380
Willy Tarreau74ca5042013-06-11 23:12:07 +020012381The actions generally consist in blocking a request, selecting a backend, or
12382adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012383
12384In order to define a test, the "acl" keyword is used. The syntax is :
12385
Willy Tarreau74ca5042013-06-11 23:12:07 +020012386 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012387
12388This creates a new ACL <aclname> or completes an existing one with new tests.
12389Those tests apply to the portion of request/response specified in <criterion>
12390and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012391an operator which may be specified before the set of values. Optionally some
12392conversion operators may be applied to the sample, and they will be specified
12393as a comma-delimited list of keywords just after the first keyword. The values
12394are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012395
12396ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12397'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12398which means that "my_acl" and "My_Acl" are two different ACLs.
12399
12400There is no enforced limit to the number of ACLs. The unused ones do not affect
12401performance, they just consume a small amount of memory.
12402
Willy Tarreau74ca5042013-06-11 23:12:07 +020012403The criterion generally is the name of a sample fetch method, or one of its ACL
12404specific declinations. The default test method is implied by the output type of
12405this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012406methods of a same sample fetch method. The sample fetch methods are the only
12407ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012408
12409Sample fetch methods return data which can be of the following types :
12410 - boolean
12411 - integer (signed or unsigned)
12412 - IPv4 or IPv6 address
12413 - string
12414 - data block
12415
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012416Converters transform any of these data into any of these. For example, some
12417converters might convert a string to a lower-case string while other ones
12418would turn a string to an IPv4 address, or apply a netmask to an IP address.
12419The resulting sample is of the type of the last converter applied to the list,
12420which defaults to the type of the sample fetch method.
12421
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012422Each sample or converter returns data of a specific type, specified with its
12423keyword in this documentation. When an ACL is declared using a standard sample
12424fetch method, certain types automatically involved a default matching method
12425which are summarized in the table below :
12426
12427 +---------------------+-----------------+
12428 | Sample or converter | Default |
12429 | output type | matching method |
12430 +---------------------+-----------------+
12431 | boolean | bool |
12432 +---------------------+-----------------+
12433 | integer | int |
12434 +---------------------+-----------------+
12435 | ip | ip |
12436 +---------------------+-----------------+
12437 | string | str |
12438 +---------------------+-----------------+
12439 | binary | none, use "-m" |
12440 +---------------------+-----------------+
12441
12442Note that in order to match a binary samples, it is mandatory to specify a
12443matching method, see below.
12444
Willy Tarreau74ca5042013-06-11 23:12:07 +020012445The ACL engine can match these types against patterns of the following types :
12446 - boolean
12447 - integer or integer range
12448 - IP address / network
12449 - string (exact, substring, suffix, prefix, subdir, domain)
12450 - regular expression
12451 - hex block
12452
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012453The following ACL flags are currently supported :
12454
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012455 -i : ignore case during matching of all subsequent patterns.
12456 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012457 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012458 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012459 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012460 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012461 -- : force end of flags. Useful when a string looks like one of the flags.
12462
Willy Tarreau74ca5042013-06-11 23:12:07 +020012463The "-f" flag is followed by the name of a file from which all lines will be
12464read as individual values. It is even possible to pass multiple "-f" arguments
12465if the patterns are to be loaded from multiple files. Empty lines as well as
12466lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12467will be stripped. If it is absolutely necessary to insert a valid pattern
12468beginning with a sharp, just prefix it with a space so that it is not taken for
12469a comment. Depending on the data type and match method, haproxy may load the
12470lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12471exact string matching. In this case, duplicates will automatically be removed.
12472
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012473The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12474parsed as two column file. The first column contains the patterns used by the
12475ACL, and the second column contain the samples. The sample can be used later by
12476a map. This can be useful in some rare cases where an ACL would just be used to
12477check for the existence of a pattern in a map before a mapping is applied.
12478
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012479The "-u" flag forces the unique id of the ACL. This unique id is used with the
12480socket interface to identify ACL and dynamically change its values. Note that a
12481file is always identified by its name even if an id is set.
12482
Willy Tarreau74ca5042013-06-11 23:12:07 +020012483Also, note that the "-i" flag applies to subsequent entries and not to entries
12484loaded from files preceding it. For instance :
12485
12486 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12487
12488In this example, each line of "exact-ua.lst" will be exactly matched against
12489the "user-agent" header of the request. Then each line of "generic-ua" will be
12490case-insensitively matched. Then the word "test" will be insensitively matched
12491as well.
12492
12493The "-m" flag is used to select a specific pattern matching method on the input
12494sample. All ACL-specific criteria imply a pattern matching method and generally
12495do not need this flag. However, this flag is useful with generic sample fetch
12496methods to describe how they're going to be matched against the patterns. This
12497is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012498obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012499followed by a pattern matching method name, this method is used instead of the
12500default one for the criterion. This makes it possible to match contents in ways
12501that were not initially planned, or with sample fetch methods which return a
12502string. The matching method also affects the way the patterns are parsed.
12503
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012504The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12505By default, if the parser cannot parse ip address it considers that the parsed
12506string is maybe a domain name and try dns resolution. The flag "-n" disable this
12507resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12508server is not reachable, the haproxy configuration parsing may last many minutes
12509waiting fir the timeout. During this time no error messages are displayed. The
12510flag "-n" disable this behavior. Note also that during the runtime, this
12511function is disabled for the dynamic acl modifications.
12512
Willy Tarreau74ca5042013-06-11 23:12:07 +020012513There are some restrictions however. Not all methods can be used with all
12514sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12515be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012516
12517 - "found" : only check if the requested sample could be found in the stream,
12518 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012519 to pass any pattern to avoid confusion. This matching method is
12520 particularly useful to detect presence of certain contents such
12521 as headers, cookies, etc... even if they are empty and without
12522 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012523
12524 - "bool" : check the value as a boolean. It can only be applied to fetches
12525 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012526 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012527
12528 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012529 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012530
12531 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012532 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012533
Davor Ocelice9ed2812017-12-25 17:49:28 +010012534 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012535 binary sequence. This may be used with binary or string samples.
12536
12537 - "len" : match the sample's length as an integer. This may be used with
12538 binary or string samples.
12539
Willy Tarreau74ca5042013-06-11 23:12:07 +020012540 - "str" : exact match : match the contents against a string. This may be
12541 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012542
Willy Tarreau74ca5042013-06-11 23:12:07 +020012543 - "sub" : substring match : check that the contents contain at least one of
12544 the provided string patterns. This may be used with binary or
12545 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012546
Willy Tarreau74ca5042013-06-11 23:12:07 +020012547 - "reg" : regex match : match the contents against a list of regular
12548 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012549
Willy Tarreau74ca5042013-06-11 23:12:07 +020012550 - "beg" : prefix match : check that the contents begin like the provided
12551 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012552
Willy Tarreau74ca5042013-06-11 23:12:07 +020012553 - "end" : suffix match : check that the contents end like the provided
12554 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012555
Willy Tarreau74ca5042013-06-11 23:12:07 +020012556 - "dir" : subdir match : check that a slash-delimited portion of the
12557 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012558 This may be used with binary or string samples.
12559
Willy Tarreau74ca5042013-06-11 23:12:07 +020012560 - "dom" : domain match : check that a dot-delimited portion of the contents
12561 exactly match one of the provided string patterns. This may be
12562 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012563
12564For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12565request, it is possible to do :
12566
12567 acl jsess_present cook(JSESSIONID) -m found
12568
12569In order to apply a regular expression on the 500 first bytes of data in the
12570buffer, one would use the following acl :
12571
12572 acl script_tag payload(0,500) -m reg -i <script>
12573
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012574On systems where the regex library is much slower when using "-i", it is
12575possible to convert the sample to lowercase before matching, like this :
12576
12577 acl script_tag payload(0,500),lower -m reg <script>
12578
Willy Tarreau74ca5042013-06-11 23:12:07 +020012579All ACL-specific criteria imply a default matching method. Most often, these
12580criteria are composed by concatenating the name of the original sample fetch
12581method and the matching method. For example, "hdr_beg" applies the "beg" match
12582to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12583criteria rely on a sample fetch method, it is always possible instead to use
12584the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012585
Willy Tarreau74ca5042013-06-11 23:12:07 +020012586If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012587the matching method is simply applied to the underlying sample fetch method.
12588For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012589
Willy Tarreau74ca5042013-06-11 23:12:07 +020012590 acl short_form hdr_beg(host) www.
12591 acl alternate1 hdr_beg(host) -m beg www.
12592 acl alternate2 hdr_dom(host) -m beg www.
12593 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012594
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012595
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012596The table below summarizes the compatibility matrix between sample or converter
12597types and the pattern types to fetch against. It indicates for each compatible
12598combination the name of the matching method to be used, surrounded with angle
12599brackets ">" and "<" when the method is the default one and will work by
12600default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012601
Willy Tarreau74ca5042013-06-11 23:12:07 +020012602 +-------------------------------------------------+
12603 | Input sample type |
12604 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012605 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012606 +----------------------+---------+---------+---------+---------+---------+
12607 | none (presence only) | found | found | found | found | found |
12608 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012609 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012610 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012611 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012612 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012613 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012614 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012615 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012616 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012617 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012618 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012619 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012620 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012621 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012622 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012623 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012624 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012625 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012626 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012627 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012628 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012629 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012630 +----------------------+---------+---------+---------+---------+---------+
12631 | hex block | | | | bin | bin |
12632 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012633
12634
Willy Tarreau74ca5042013-06-11 23:12:07 +0200126357.1.1. Matching booleans
12636------------------------
12637
12638In order to match a boolean, no value is needed and all values are ignored.
12639Boolean matching is used by default for all fetch methods of type "boolean".
12640When boolean matching is used, the fetched value is returned as-is, which means
12641that a boolean "true" will always match and a boolean "false" will never match.
12642
12643Boolean matching may also be enforced using "-m bool" on fetch methods which
12644return an integer value. Then, integer value 0 is converted to the boolean
12645"false" and all other values are converted to "true".
12646
Willy Tarreau6a06a402007-07-15 20:15:28 +020012647
Willy Tarreau74ca5042013-06-11 23:12:07 +0200126487.1.2. Matching integers
12649------------------------
12650
12651Integer matching applies by default to integer fetch methods. It can also be
12652enforced on boolean fetches using "-m int". In this case, "false" is converted
12653to the integer 0, and "true" is converted to the integer 1.
12654
12655Integer matching also supports integer ranges and operators. Note that integer
12656matching only applies to positive values. A range is a value expressed with a
12657lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012658
12659For instance, "1024:65535" is a valid range to represent a range of
12660unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12661representation of privileged ports, and ":1023" would also work.
12662
Willy Tarreau62644772008-07-16 18:36:06 +020012663As a special case, some ACL functions support decimal numbers which are in fact
12664two integers separated by a dot. This is used with some version checks for
12665instance. All integer properties apply to those decimal numbers, including
12666ranges and operators.
12667
Willy Tarreau6a06a402007-07-15 20:15:28 +020012668For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012669operators with ranges does not make much sense and is strongly discouraged.
12670Similarly, it does not make much sense to perform order comparisons with a set
12671of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012672
Willy Tarreau0ba27502007-12-24 16:55:16 +010012673Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012674
12675 eq : true if the tested value equals at least one value
12676 ge : true if the tested value is greater than or equal to at least one value
12677 gt : true if the tested value is greater than at least one value
12678 le : true if the tested value is less than or equal to at least one value
12679 lt : true if the tested value is less than at least one value
12680
Willy Tarreau0ba27502007-12-24 16:55:16 +010012681For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012682
12683 acl negative-length hdr_val(content-length) lt 0
12684
Willy Tarreau62644772008-07-16 18:36:06 +020012685This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12686
12687 acl sslv3 req_ssl_ver 3:3.1
12688
Willy Tarreau6a06a402007-07-15 20:15:28 +020012689
Willy Tarreau74ca5042013-06-11 23:12:07 +0200126907.1.3. Matching strings
12691-----------------------
12692
12693String matching applies to string or binary fetch methods, and exists in 6
12694different forms :
12695
12696 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012697 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012698
12699 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012700 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012701
12702 - prefix match (-m beg) : the patterns are compared with the beginning of
12703 the extracted string, and the ACL matches if any of them matches.
12704
12705 - suffix match (-m end) : the patterns are compared with the end of the
12706 extracted string, and the ACL matches if any of them matches.
12707
Baptiste Assmann33db6002016-03-06 23:32:10 +010012708 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012709 string, delimited with slashes ("/"), and the ACL matches if any of them
12710 matches.
12711
12712 - domain match (-m dom) : the patterns are looked up inside the extracted
12713 string, delimited with dots ("."), and the ACL matches if any of them
12714 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012715
12716String matching applies to verbatim strings as they are passed, with the
12717exception of the backslash ("\") which makes it possible to escape some
12718characters such as the space. If the "-i" flag is passed before the first
12719string, then the matching will be performed ignoring the case. In order
12720to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012721before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012722
12723
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127247.1.4. Matching regular expressions (regexes)
12725---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012726
12727Just like with string matching, regex matching applies to verbatim strings as
12728they are passed, with the exception of the backslash ("\") which makes it
12729possible to escape some characters such as the space. If the "-i" flag is
12730passed before the first regex, then the matching will be performed ignoring
12731the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012732the "--" flag before the first string. Same principle applies of course to
12733match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012734
12735
Willy Tarreau74ca5042013-06-11 23:12:07 +0200127367.1.5. Matching arbitrary data blocks
12737-------------------------------------
12738
12739It is possible to match some extracted samples against a binary block which may
12740not safely be represented as a string. For this, the patterns must be passed as
12741a series of hexadecimal digits in an even number, when the match method is set
12742to binary. Each sequence of two digits will represent a byte. The hexadecimal
12743digits may be used upper or lower case.
12744
12745Example :
12746 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12747 acl hello payload(0,6) -m bin 48656c6c6f0a
12748
12749
127507.1.6. Matching IPv4 and IPv6 addresses
12751---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012752
12753IPv4 addresses values can be specified either as plain addresses or with a
12754netmask appended, in which case the IPv4 address matches whenever it is
12755within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012756host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012757difficult to read and debug configurations. If hostnames are used, you should
12758at least ensure that they are present in /etc/hosts so that the configuration
12759does not depend on any random DNS match at the moment the configuration is
12760parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012761
Daniel Schnellereba56342016-04-13 00:26:52 +020012762The dotted IPv4 address notation is supported in both regular as well as the
12763abbreviated form with all-0-octets omitted:
12764
12765 +------------------+------------------+------------------+
12766 | Example 1 | Example 2 | Example 3 |
12767 +------------------+------------------+------------------+
12768 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12769 | 192.168.1 | 10.12 | 127.1 |
12770 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12771 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12772 +------------------+------------------+------------------+
12773
12774Notice that this is different from RFC 4632 CIDR address notation in which
12775192.168.42/24 would be equivalent to 192.168.42.0/24.
12776
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012777IPv6 may be entered in their usual form, with or without a netmask appended.
12778Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12779trouble with randomly resolved IP addresses, host names are never allowed in
12780IPv6 patterns.
12781
12782HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12783following situations :
12784 - tested address is IPv4, pattern address is IPv4, the match applies
12785 in IPv4 using the supplied mask if any.
12786 - tested address is IPv6, pattern address is IPv6, the match applies
12787 in IPv6 using the supplied mask if any.
12788 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12789 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12790 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12791 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12792 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12793 applied in IPv6 using the supplied IPv6 mask.
12794
Willy Tarreau74ca5042013-06-11 23:12:07 +020012795
127967.2. Using ACLs to form conditions
12797----------------------------------
12798
12799Some actions are only performed upon a valid condition. A condition is a
12800combination of ACLs with operators. 3 operators are supported :
12801
12802 - AND (implicit)
12803 - OR (explicit with the "or" keyword or the "||" operator)
12804 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012805
Willy Tarreau74ca5042013-06-11 23:12:07 +020012806A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012807
Willy Tarreau74ca5042013-06-11 23:12:07 +020012808 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012809
Willy Tarreau74ca5042013-06-11 23:12:07 +020012810Such conditions are generally used after an "if" or "unless" statement,
12811indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012812
Willy Tarreau74ca5042013-06-11 23:12:07 +020012813For instance, to block HTTP requests to the "*" URL with methods other than
12814"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12815requests with a content-length greater than 0, and finally every request which
12816is not either GET/HEAD/POST/OPTIONS !
12817
12818 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012819 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12820 http-request deny if METH_GET HTTP_CONTENT
12821 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012822
12823To select a different backend for requests to static contents on the "www" site
12824and to every request on the "img", "video", "download" and "ftp" hosts :
12825
12826 acl url_static path_beg /static /images /img /css
12827 acl url_static path_end .gif .png .jpg .css .js
12828 acl host_www hdr_beg(host) -i www
12829 acl host_static hdr_beg(host) -i img. video. download. ftp.
12830
Davor Ocelice9ed2812017-12-25 17:49:28 +010012831 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020012832 # of host "www". Use backend "www" for the rest.
12833 use_backend static if host_static or host_www url_static
12834 use_backend www if host_www
12835
12836It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
12837expressions that are built on the fly without needing to be declared. They must
12838be enclosed between braces, with a space before and after each brace (because
12839the braces must be seen as independent words). Example :
12840
12841 The following rule :
12842
12843 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012844 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020012845
12846 Can also be written that way :
12847
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012848 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020012849
12850It is generally not recommended to use this construct because it's a lot easier
12851to leave errors in the configuration when written that way. However, for very
12852simple rules matching only one source IP address for instance, it can make more
12853sense to use them than to declare ACLs with random names. Another example of
12854good use is the following :
12855
12856 With named ACLs :
12857
12858 acl site_dead nbsrv(dynamic) lt 2
12859 acl site_dead nbsrv(static) lt 2
12860 monitor fail if site_dead
12861
12862 With anonymous ACLs :
12863
12864 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
12865
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012866See section 4.2 for detailed help on the "http-request deny" and "use_backend"
12867keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012868
12869
128707.3. Fetching samples
12871---------------------
12872
12873Historically, sample fetch methods were only used to retrieve data to match
12874against patterns using ACLs. With the arrival of stick-tables, a new class of
12875sample fetch methods was created, most often sharing the same syntax as their
12876ACL counterpart. These sample fetch methods are also known as "fetches". As
12877of now, ACLs and fetches have converged. All ACL fetch methods have been made
12878available as fetch methods, and ACLs may use any sample fetch method as well.
12879
12880This section details all available sample fetch methods and their output type.
12881Some sample fetch methods have deprecated aliases that are used to maintain
12882compatibility with existing configurations. They are then explicitly marked as
12883deprecated and should not be used in new setups.
12884
12885The ACL derivatives are also indicated when available, with their respective
12886matching methods. These ones all have a well defined default pattern matching
12887method, so it is never necessary (though allowed) to pass the "-m" option to
12888indicate how the sample will be matched using ACLs.
12889
12890As indicated in the sample type versus matching compatibility matrix above,
12891when using a generic sample fetch method in an ACL, the "-m" option is
12892mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
12893the same keyword exists as an ACL keyword and as a standard fetch method, the
12894ACL engine will automatically pick the ACL-only one by default.
12895
12896Some of these keywords support one or multiple mandatory arguments, and one or
12897multiple optional arguments. These arguments are strongly typed and are checked
12898when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012899incorrect argument (e.g. an unresolved backend name). Fetch function arguments
12900are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020012901is optional, it will be indicated below between square brackets ('[ ]'). When
12902all arguments are optional, the parenthesis may be omitted.
12903
12904Thus, the syntax of a standard sample fetch method is one of the following :
12905 - name
12906 - name(arg1)
12907 - name(arg1,arg2)
12908
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012909
129107.3.1. Converters
12911-----------------
12912
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012913Sample fetch methods may be combined with transformations to be applied on top
12914of the fetched sample (also called "converters"). These combinations form what
12915is called "sample expressions" and the result is a "sample". Initially this
12916was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012917has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012918unique-id-format, add-header, ...).
12919
12920These transformations are enumerated as a series of specific keywords after the
12921sample fetch method. These keywords may equally be appended immediately after
12922the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010012923support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012924
Willy Tarreau97707872015-01-27 15:12:13 +010012925A certain category of converters are bitwise and arithmetic operators which
12926support performing basic operations on integers. Some bitwise operations are
12927supported (and, or, xor, cpl) and some arithmetic operations are supported
12928(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
12929bool) which make it possible to report a match without having to write an ACL.
12930
Willy Tarreau74ca5042013-06-11 23:12:07 +020012931The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010012932
Ben Shillitof25e8e52016-12-02 14:25:37 +00001293351d.single(<prop>[,<prop>*])
12934 Returns values for the properties requested as a string, where values are
12935 separated by the delimiter specified with "51degrees-property-separator".
12936 The device is identified using the User-Agent header passed to the
12937 converter. The function can be passed up to five property names, and if a
12938 property name can't be found, the value "NoData" is returned.
12939
12940 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010012941 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
12942 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000012943 # User-Agent passed to the converter.
12944 frontend http-in
12945 bind *:8081
12946 default_backend servers
12947 http-request set-header X-51D-DeviceTypeMobileTablet \
12948 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
12949
Willy Tarreau97707872015-01-27 15:12:13 +010012950add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012951 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012952 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010012953 name. The name of the variable starts with an indication about its scope. The
12954 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012955 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012956 "sess" : the variable is shared with the whole session
12957 "txn" : the variable is shared with the transaction (request and response)
12958 "req" : the variable is shared only during request processing
12959 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012960 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012961 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012962
12963and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012964 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012965 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010012966 numeric value or a variable name. The name of the variable starts with an
12967 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012968 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012969 "sess" : the variable is shared with the whole session
12970 "txn" : the variable is shared with the transaction (request and response)
12971 "req" : the variable is shared only during request processing
12972 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012973 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012974 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012975
Holger Just1bfc24b2017-05-06 00:56:53 +020012976b64dec
12977 Converts (decodes) a base64 encoded input string to its binary
12978 representation. It performs the inverse operation of base64().
12979
Emeric Brun53d1a982014-04-30 18:21:37 +020012980base64
12981 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010012982 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020012983 an SSL ID can be copied in a header).
12984
Willy Tarreau97707872015-01-27 15:12:13 +010012985bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012986 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010012987 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010012988 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010012989 presence of a flag).
12990
Emeric Brun54c4ac82014-11-03 15:32:43 +010012991bytes(<offset>[,<length>])
12992 Extracts some bytes from an input binary sample. The result is a binary
12993 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010012994 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010012995
Willy Tarreau280f42b2018-02-19 15:34:12 +010012996concat([<start>],[<var>],[<end>])
12997 Concatenates up to 3 fields after the current sample which is then turned to
12998 a string. The first one, <start>, is a constant string, that will be appended
12999 immediately after the existing sample. It may be omitted if not used. The
13000 second one, <var>, is a variable name. The variable will be looked up, its
13001 contents converted to a string, and it will be appended immediately after the
13002 <first> part. If the variable is not found, nothing is appended. It may be
13003 omitted as well. The third field, <end> is a constant string that will be
13004 appended after the variable. It may also be omitted. Together, these elements
13005 allow to concatenate variables with delimiters to an existing set of
13006 variables. This can be used to build new variables made of a succession of
13007 other variables, such as colon-delimited varlues. Note that due to the config
13008 parser, it is not possible to use a comma nor a closing parenthesis as
13009 delimitors.
13010
13011 Example:
13012 tcp-request session set-var(sess.src) src
13013 tcp-request session set-var(sess.dn) ssl_c_s_dn
13014 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13015 http-request set-header x-hap-sig %[var(txn.sig)]
13016
Willy Tarreau97707872015-01-27 15:12:13 +010013017cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013018 Takes the input value of type signed integer, applies a ones-complement
13019 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013020
Willy Tarreau80599772015-01-20 19:35:24 +010013021crc32([<avalanche>])
13022 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13023 hash function. Optionally, it is possible to apply a full avalanche hash
13024 function to the output if the optional <avalanche> argument equals 1. This
13025 converter uses the same functions as used by the various hash-based load
13026 balancing algorithms, so it will provide exactly the same results. It is
13027 provided for compatibility with other software which want a CRC32 to be
13028 computed on some input keys, so it follows the most common implementation as
13029 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13030 but may provide a better or at least less predictable distribution. It must
13031 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013032 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13033
13034crc32c([<avalanche>])
13035 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13036 hash function. Optionally, it is possible to apply a full avalanche hash
13037 function to the output if the optional <avalanche> argument equals 1. This
13038 converter uses the same functions as described in RFC4960, Appendix B [8].
13039 It is provided for compatibility with other software which want a CRC32C to be
13040 computed on some input keys. It is slower than the other algorithms and it must
13041 not be used for security purposes as a 32-bit hash is trivial to break. See
13042 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013043
David Carlier29b3ca32015-09-25 14:09:21 +010013044da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013045 Asks the DeviceAtlas converter to identify the User Agent string passed on
13046 input, and to emit a string made of the concatenation of the properties
13047 enumerated in argument, delimited by the separator defined by the global
13048 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013049 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013050 configuration language.
13051
13052 Example:
13053 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013054 bind *:8881
13055 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013056 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 +020013057
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013058debug
13059 This converter is used as debug tool. It dumps on screen the content and the
13060 type of the input sample. The sample is returned as is on its output. This
13061 converter only exists when haproxy was built with debugging enabled.
13062
Willy Tarreau97707872015-01-27 15:12:13 +010013063div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013064 Divides the input value of type signed integer by <value>, and returns the
13065 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013066 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013067 variable name. The name of the variable starts with an indication about its
13068 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013069 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013070 "sess" : the variable is shared with the whole session
13071 "txn" : the variable is shared with the transaction (request and response)
13072 "req" : the variable is shared only during request processing
13073 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013074 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013075 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013076
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013077djb2([<avalanche>])
13078 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13079 hash function. Optionally, it is possible to apply a full avalanche hash
13080 function to the output if the optional <avalanche> argument equals 1. This
13081 converter uses the same functions as used by the various hash-based load
13082 balancing algorithms, so it will provide exactly the same results. It is
13083 mostly intended for debugging, but can be used as a stick-table entry to
13084 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013085 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13086 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013087
Willy Tarreau97707872015-01-27 15:12:13 +010013088even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013089 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013090 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13091
Marcin Deranek9631a282018-04-16 14:30:46 +020013092field(<index>,<delimiters>[,<count>])
13093 Extracts the substring at the given index counting from the beginning
13094 (positive index) or from the end (negative index) considering given delimiters
13095 from an input string. Indexes start at 1 or -1 and delimiters are a string
13096 formatted list of chars. Optionally you can specify <count> of fields to
13097 extract (default: 1). Value of 0 indicates extraction of all remaining
13098 fields.
13099
13100 Example :
13101 str(f1_f2_f3__f5),field(5,_) # f5
13102 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13103 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13104 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13105 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013106
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013107hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013108 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013109 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013110 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 +020013111 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013112
Dragan Dosen3f957b22017-10-24 09:27:34 +020013113hex2i
13114 Converts a hex string containing two hex digits per input byte to an
13115 integer. If the input value can not be converted, then zero is returned.
13116
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013117http_date([<offset>])
13118 Converts an integer supposed to contain a date since epoch to a string
13119 representing this date in a format suitable for use in HTTP header fields. If
13120 an offset value is specified, then it is a number of seconds that is added to
13121 the date before the conversion is operated. This is particularly useful to
13122 emit Date header fields, Expires values in responses when combined with a
13123 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013124
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013125in_table(<table>)
13126 Uses the string representation of the input sample to perform a look up in
13127 the specified table. If the key is not found in the table, a boolean false
13128 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013129 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013130 or not a source IP address or an Authorization header was already seen).
13131
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013132ipmask(<mask4>, [<mask6>])
13133 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013134 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013135 table entries and as such use the same server. The mask4 can be passed in
13136 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13137 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13138 If no mask6 is given IPv6 addresses will fail to convert for backwards
13139 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013140
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013141json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013142 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013143 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013144 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013145 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13146 of errors:
13147 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13148 bytes, ...)
13149 - invalid range (the decoded value is within a UTF-8 prohibited range),
13150 - code overlong (the value is encoded with more bytes than necessary).
13151
13152 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13153 character is greater than 0xffff because the JSON string escape specification
13154 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13155 in 4 variants designated by a combination of two suffix letters : "p" for
13156 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13157 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013158 - "ascii" : never fails;
13159 - "utf8" : fails on any detected errors;
13160 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013161 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013162 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013163 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13164 characters corresponding to the other errors.
13165
13166 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013167 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013168
13169 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013170 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013171 capture request header user-agent len 150
13172 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013173
13174 Input request from client 127.0.0.1:
13175 GET / HTTP/1.0
13176 User-Agent: Very "Ugly" UA 1/2
13177
13178 Output log:
13179 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13180
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013181language(<value>[,<default>])
13182 Returns the value with the highest q-factor from a list as extracted from the
13183 "accept-language" header using "req.fhdr". Values with no q-factor have a
13184 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13185 belong to the list of semi-colon delimited <values> will be considered. The
13186 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13187 given list and a default value is provided, it is returned. Note that language
13188 names may have a variant after a dash ('-'). If this variant is present in the
13189 list, it will be matched, but if it is not, only the base language is checked.
13190 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013191 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013192 ordering of the values in the request counts, as the first value among
13193 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013194
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013195 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013196
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013197 # this configuration switches to the backend matching a
13198 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013199
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013200 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13201 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13202 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13203 use_backend spanish if es
13204 use_backend french if fr
13205 use_backend english if en
13206 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013207
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013208length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013209 Get the length of the string. This can only be placed after a string
13210 sample fetch function or after a transformation keyword returning a string
13211 type. The result is of type integer.
13212
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013213lower
13214 Convert a string sample to lower case. This can only be placed after a string
13215 sample fetch function or after a transformation keyword returning a string
13216 type. The result is of type string.
13217
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013218ltime(<format>[,<offset>])
13219 Converts an integer supposed to contain a date since epoch to a string
13220 representing this date in local time using a format defined by the <format>
13221 string using strftime(3). The purpose is to allow any date format to be used
13222 in logs. An optional <offset> in seconds may be applied to the input date
13223 (positive or negative). See the strftime() man page for the format supported
13224 by your operating system. See also the utime converter.
13225
13226 Example :
13227
13228 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013229 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013230 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13231
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013232map(<map_file>[,<default_value>])
13233map_<match_type>(<map_file>[,<default_value>])
13234map_<match_type>_<output_type>(<map_file>[,<default_value>])
13235 Search the input value from <map_file> using the <match_type> matching method,
13236 and return the associated value converted to the type <output_type>. If the
13237 input value cannot be found in the <map_file>, the converter returns the
13238 <default_value>. If the <default_value> is not set, the converter fails and
13239 acts as if no input value could be fetched. If the <match_type> is not set, it
13240 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13241 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13242 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013243
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013244 It is important to avoid overlapping between the keys : IP addresses and
13245 strings are stored in trees, so the first of the finest match will be used.
13246 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013247
Tim Düsterhus4896c442016-11-29 02:15:19 +010013248 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013249 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013250
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013251 input type | match method | output type str | output type int | output type ip
13252 -----------+--------------+-----------------+-----------------+---------------
13253 str | str | map_str | map_str_int | map_str_ip
13254 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013255 str | beg | map_beg | map_beg_int | map_end_ip
13256 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013257 str | sub | map_sub | map_sub_int | map_sub_ip
13258 -----------+--------------+-----------------+-----------------+---------------
13259 str | dir | map_dir | map_dir_int | map_dir_ip
13260 -----------+--------------+-----------------+-----------------+---------------
13261 str | dom | map_dom | map_dom_int | map_dom_ip
13262 -----------+--------------+-----------------+-----------------+---------------
13263 str | end | map_end | map_end_int | map_end_ip
13264 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013265 str | reg | map_reg | map_reg_int | map_reg_ip
13266 -----------+--------------+-----------------+-----------------+---------------
13267 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013268 -----------+--------------+-----------------+-----------------+---------------
13269 int | int | map_int | map_int_int | map_int_ip
13270 -----------+--------------+-----------------+-----------------+---------------
13271 ip | ip | map_ip | map_ip_int | map_ip_ip
13272 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013273
Thierry Fournier8feaa662016-02-10 22:55:20 +010013274 The special map called "map_regm" expect matching zone in the regular
13275 expression and modify the output replacing back reference (like "\1") by
13276 the corresponding match text.
13277
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013278 The file contains one key + value per line. Lines which start with '#' are
13279 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13280 is then the first "word" (series of non-space/tabs characters), and the value
13281 is what follows this series of space/tab till the end of the line excluding
13282 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013283
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013284 Example :
13285
13286 # this is a comment and is ignored
13287 2.22.246.0/23 United Kingdom \n
13288 <-><-----------><--><------------><---->
13289 | | | | `- trailing spaces ignored
13290 | | | `---------- value
13291 | | `-------------------- middle spaces ignored
13292 | `---------------------------- key
13293 `------------------------------------ leading spaces ignored
13294
Willy Tarreau97707872015-01-27 15:12:13 +010013295mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013296 Divides the input value of type signed integer by <value>, and returns the
13297 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013298 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013299 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013300 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013301 "sess" : the variable is shared with the whole session
13302 "txn" : the variable is shared with the transaction (request and response)
13303 "req" : the variable is shared only during request processing
13304 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013305 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013306 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013307
13308mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013309 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013310 the product as an signed integer. In case of overflow, the largest possible
13311 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013312 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013313 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013314 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013315 "sess" : the variable is shared with the whole session
13316 "txn" : the variable is shared with the transaction (request and response)
13317 "req" : the variable is shared only during request processing
13318 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013319 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013320 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013321
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013322nbsrv
13323 Takes an input value of type string, interprets it as a backend name and
13324 returns the number of usable servers in that backend. Can be used in places
13325 where we want to look up a backend from a dynamic name, like a result of a
13326 map lookup.
13327
Willy Tarreau97707872015-01-27 15:12:13 +010013328neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013329 Takes the input value of type signed integer, computes the opposite value,
13330 and returns the remainder as an signed integer. 0 is identity. This operator
13331 is provided for reversed subtracts : in order to subtract the input from a
13332 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013333
13334not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013335 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013336 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013337 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013338 absence of a flag).
13339
13340odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013341 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013342 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13343
13344or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013345 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013346 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013347 numeric value or a variable name. The name of the variable starts with an
13348 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013349 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013350 "sess" : the variable is shared with the whole session
13351 "txn" : the variable is shared with the transaction (request and response)
13352 "req" : the variable is shared only during request processing
13353 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013354 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013355 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013356
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013357regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013358 Applies a regex-based substitution to the input string. It does the same
13359 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13360 default it will replace in the input string the first occurrence of the
13361 largest part matching the regular expression <regex> with the substitution
13362 string <subst>. It is possible to replace all occurrences instead by adding
13363 the flag "g" in the third argument <flags>. It is also possible to make the
13364 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13365 string, it is made up from the concatenation of all desired flags. Thus if
13366 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13367 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013368 configuration parser, some characters such as closing parenthesis, closing
13369 square brackets or comma are not possible to use in the arguments. The first
13370 use of this converter is to replace certain characters or sequence of
13371 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013372
13373 Example :
13374
13375 # de-duplicate "/" in header "x-path".
13376 # input: x-path: /////a///b/c/xzxyz/
13377 # output: x-path: /a/b/c/xzxyz/
13378 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13379
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013380capture-req(<id>)
13381 Capture the string entry in the request slot <id> and returns the entry as
13382 is. If the slot doesn't exist, the capture fails silently.
13383
13384 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013385 "http-response capture", "capture.req.hdr" and
13386 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013387
13388capture-res(<id>)
13389 Capture the string entry in the response slot <id> and returns the entry as
13390 is. If the slot doesn't exist, the capture fails silently.
13391
13392 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013393 "http-response capture", "capture.req.hdr" and
13394 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013395
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013396sdbm([<avalanche>])
13397 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13398 hash function. Optionally, it is possible to apply a full avalanche hash
13399 function to the output if the optional <avalanche> argument equals 1. This
13400 converter uses the same functions as used by the various hash-based load
13401 balancing algorithms, so it will provide exactly the same results. It is
13402 mostly intended for debugging, but can be used as a stick-table entry to
13403 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013404 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13405 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013406
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013407set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013408 Sets a variable with the input content and returns the content on the output
13409 as-is. The variable keeps the value and the associated input type. The name of
13410 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013411 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013412 "sess" : the variable is shared with the whole session
13413 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013414 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013415 "req" : the variable is shared only during request processing,
13416 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013417 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013418 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013419
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013420sha1
13421 Converts a binary input sample to a SHA1 digest. The result is a binary
13422 sample with length of 20 bytes.
13423
Tim Duesterhusca097c12018-04-27 21:18:45 +020013424strcmp(<var>)
13425 Compares the contents of <var> with the input value of type string. Returns
13426 the result as a signed integer compatible with strcmp(3): 0 if both strings
13427 are identical. A value less than 0 if the left string is lexicographically
13428 smaller than the right string or if the left string is shorter. A value greater
13429 than 0 otherwise (right string greater than left string or the right string is
13430 shorter).
13431
13432 Example :
13433
13434 http-request set-var(txn.host) hdr(host)
13435 # Check whether the client is attempting domain fronting.
13436 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13437
13438
Willy Tarreau97707872015-01-27 15:12:13 +010013439sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013440 Subtracts <value> from the input value of type signed integer, and returns
13441 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013442 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013443 or a variable name. The name of the variable starts with an indication about
13444 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013445 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013446 "sess" : the variable is shared with the whole session
13447 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013448 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013449 "req" : the variable is shared only during request processing,
13450 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013451 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013452 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013453
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013454table_bytes_in_rate(<table>)
13455 Uses the string representation of the input sample to perform a look up in
13456 the specified table. If the key is not found in the table, integer value zero
13457 is returned. Otherwise the converter returns the average client-to-server
13458 bytes rate associated with the input sample in the designated table, measured
13459 in amount of bytes over the period configured in the table. See also the
13460 sc_bytes_in_rate sample fetch keyword.
13461
13462
13463table_bytes_out_rate(<table>)
13464 Uses the string representation of the input sample to perform a look up in
13465 the specified table. If the key is not found in the table, integer value zero
13466 is returned. Otherwise the converter returns the average server-to-client
13467 bytes rate associated with the input sample in the designated table, measured
13468 in amount of bytes over the period configured in the table. See also the
13469 sc_bytes_out_rate sample fetch keyword.
13470
13471table_conn_cnt(<table>)
13472 Uses the string representation of the input sample to perform a look up in
13473 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013474 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013475 connections associated with the input sample in the designated table. See
13476 also the sc_conn_cnt sample fetch keyword.
13477
13478table_conn_cur(<table>)
13479 Uses the string representation of the input sample to perform a look up in
13480 the specified table. If the key is not found in the table, integer value zero
13481 is returned. Otherwise the converter returns the current amount of concurrent
13482 tracked connections associated with the input sample in the designated table.
13483 See also the sc_conn_cur sample fetch keyword.
13484
13485table_conn_rate(<table>)
13486 Uses the string representation of the input sample to perform a look up in
13487 the specified table. If the key is not found in the table, integer value zero
13488 is returned. Otherwise the converter returns the average incoming connection
13489 rate associated with the input sample in the designated table. See also the
13490 sc_conn_rate sample fetch keyword.
13491
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013492table_gpt0(<table>)
13493 Uses the string representation of the input sample to perform a look up in
13494 the specified table. If the key is not found in the table, boolean value zero
13495 is returned. Otherwise the converter returns the current value of the first
13496 general purpose tag associated with the input sample in the designated table.
13497 See also the sc_get_gpt0 sample fetch keyword.
13498
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013499table_gpc0(<table>)
13500 Uses the string representation of the input sample to perform a look up in
13501 the specified table. If the key is not found in the table, integer value zero
13502 is returned. Otherwise the converter returns the current value of the first
13503 general purpose counter associated with the input sample in the designated
13504 table. See also the sc_get_gpc0 sample fetch keyword.
13505
13506table_gpc0_rate(<table>)
13507 Uses the string representation of the input sample to perform a look up in
13508 the specified table. If the key is not found in the table, integer value zero
13509 is returned. Otherwise the converter returns the frequency which the gpc0
13510 counter was incremented over the configured period in the table, associated
13511 with the input sample in the designated table. See also the sc_get_gpc0_rate
13512 sample fetch keyword.
13513
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013514table_gpc1(<table>)
13515 Uses the string representation of the input sample to perform a look up in
13516 the specified table. If the key is not found in the table, integer value zero
13517 is returned. Otherwise the converter returns the current value of the second
13518 general purpose counter associated with the input sample in the designated
13519 table. See also the sc_get_gpc1 sample fetch keyword.
13520
13521table_gpc1_rate(<table>)
13522 Uses the string representation of the input sample to perform a look up in
13523 the specified table. If the key is not found in the table, integer value zero
13524 is returned. Otherwise the converter returns the frequency which the gpc1
13525 counter was incremented over the configured period in the table, associated
13526 with the input sample in the designated table. See also the sc_get_gpc1_rate
13527 sample fetch keyword.
13528
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013529table_http_err_cnt(<table>)
13530 Uses the string representation of the input sample to perform a look up in
13531 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013532 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013533 errors associated with the input sample in the designated table. See also the
13534 sc_http_err_cnt sample fetch keyword.
13535
13536table_http_err_rate(<table>)
13537 Uses the string representation of the input sample to perform a look up in
13538 the specified table. If the key is not found in the table, integer value zero
13539 is returned. Otherwise the average rate of HTTP errors associated with the
13540 input sample in the designated table, measured in amount of errors over the
13541 period configured in the table. See also the sc_http_err_rate sample fetch
13542 keyword.
13543
13544table_http_req_cnt(<table>)
13545 Uses the string representation of the input sample to perform a look up in
13546 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013547 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013548 requests associated with the input sample in the designated table. See also
13549 the sc_http_req_cnt sample fetch keyword.
13550
13551table_http_req_rate(<table>)
13552 Uses the string representation of the input sample to perform a look up in
13553 the specified table. If the key is not found in the table, integer value zero
13554 is returned. Otherwise the average rate of HTTP requests associated with the
13555 input sample in the designated table, measured in amount of requests over the
13556 period configured in the table. See also the sc_http_req_rate sample fetch
13557 keyword.
13558
13559table_kbytes_in(<table>)
13560 Uses the string representation of the input sample to perform a look up in
13561 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013562 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013563 to-server data associated with the input sample in the designated table,
13564 measured in kilobytes. The test is currently performed on 32-bit integers,
13565 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13566 keyword.
13567
13568table_kbytes_out(<table>)
13569 Uses the string representation of the input sample to perform a look up in
13570 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013571 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013572 to-client data associated with the input sample in the designated table,
13573 measured in kilobytes. The test is currently performed on 32-bit integers,
13574 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13575 keyword.
13576
13577table_server_id(<table>)
13578 Uses the string representation of the input sample to perform a look up in
13579 the specified table. If the key is not found in the table, integer value zero
13580 is returned. Otherwise the converter returns the server ID associated with
13581 the input sample in the designated table. A server ID is associated to a
13582 sample by a "stick" rule when a connection to a server succeeds. A server ID
13583 zero means that no server is associated with this key.
13584
13585table_sess_cnt(<table>)
13586 Uses the string representation of the input sample to perform a look up in
13587 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013588 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013589 sessions associated with the input sample in the designated table. Note that
13590 a session here refers to an incoming connection being accepted by the
13591 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13592 keyword.
13593
13594table_sess_rate(<table>)
13595 Uses the string representation of the input sample to perform a look up in
13596 the specified table. If the key is not found in the table, integer value zero
13597 is returned. Otherwise the converter returns the average incoming session
13598 rate associated with the input sample in the designated table. Note that a
13599 session here refers to an incoming connection being accepted by the
13600 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13601 keyword.
13602
13603table_trackers(<table>)
13604 Uses the string representation of the input sample to perform a look up in
13605 the specified table. If the key is not found in the table, integer value zero
13606 is returned. Otherwise the converter returns the current amount of concurrent
13607 connections tracking the same key as the input sample in the designated
13608 table. It differs from table_conn_cur in that it does not rely on any stored
13609 information but on the table's reference count (the "use" value which is
13610 returned by "show table" on the CLI). This may sometimes be more suited for
13611 layer7 tracking. It can be used to tell a server how many concurrent
13612 connections there are from a given address for example. See also the
13613 sc_trackers sample fetch keyword.
13614
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013615upper
13616 Convert a string sample to upper case. This can only be placed after a string
13617 sample fetch function or after a transformation keyword returning a string
13618 type. The result is of type string.
13619
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013620url_dec
13621 Takes an url-encoded string provided as input and returns the decoded
13622 version as output. The input and the output are of type string.
13623
Christopher Faulet85d79c92016-11-09 16:54:56 +010013624unset-var(<var name>)
13625 Unsets a variable if the input content is defined. The name of the variable
13626 starts with an indication about its scope. The scopes allowed are:
13627 "proc" : the variable is shared with the whole process
13628 "sess" : the variable is shared with the whole session
13629 "txn" : the variable is shared with the transaction (request and
13630 response),
13631 "req" : the variable is shared only during request processing,
13632 "res" : the variable is shared only during response processing.
13633 This prefix is followed by a name. The separator is a '.'. The name may only
13634 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13635
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013636utime(<format>[,<offset>])
13637 Converts an integer supposed to contain a date since epoch to a string
13638 representing this date in UTC time using a format defined by the <format>
13639 string using strftime(3). The purpose is to allow any date format to be used
13640 in logs. An optional <offset> in seconds may be applied to the input date
13641 (positive or negative). See the strftime() man page for the format supported
13642 by your operating system. See also the ltime converter.
13643
13644 Example :
13645
13646 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013647 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013648 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13649
Marcin Deranek9631a282018-04-16 14:30:46 +020013650word(<index>,<delimiters>[,<count>])
13651 Extracts the nth word counting from the beginning (positive index) or from
13652 the end (negative index) considering given delimiters from an input string.
13653 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13654 Optionally you can specify <count> of words to extract (default: 1).
13655 Value of 0 indicates extraction of all remaining words.
13656
13657 Example :
13658 str(f1_f2_f3__f5),word(4,_) # f5
13659 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13660 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13661 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13662 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013663
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013664wt6([<avalanche>])
13665 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13666 hash function. Optionally, it is possible to apply a full avalanche hash
13667 function to the output if the optional <avalanche> argument equals 1. This
13668 converter uses the same functions as used by the various hash-based load
13669 balancing algorithms, so it will provide exactly the same results. It is
13670 mostly intended for debugging, but can be used as a stick-table entry to
13671 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013672 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13673 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013674
Willy Tarreau97707872015-01-27 15:12:13 +010013675xor(<value>)
13676 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013677 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013678 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013679 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013680 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013681 "sess" : the variable is shared with the whole session
13682 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013683 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013684 "req" : the variable is shared only during request processing,
13685 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013686 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013687 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013688
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013689xxh32([<seed>])
13690 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13691 variant of the XXHash hash function. This hash supports a seed which defaults
13692 to zero but a different value maybe passed as the <seed> argument. This hash
13693 is known to be very good and very fast so it can be used to hash URLs and/or
13694 URL parameters for use as stick-table keys to collect statistics with a low
13695 collision rate, though care must be taken as the algorithm is not considered
13696 as cryptographically secure.
13697
13698xxh64([<seed>])
13699 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13700 variant of the XXHash hash function. This hash supports a seed which defaults
13701 to zero but a different value maybe passed as the <seed> argument. This hash
13702 is known to be very good and very fast so it can be used to hash URLs and/or
13703 URL parameters for use as stick-table keys to collect statistics with a low
13704 collision rate, though care must be taken as the algorithm is not considered
13705 as cryptographically secure.
13706
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013707
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200137087.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013709--------------------------------------------
13710
13711A first set of sample fetch methods applies to internal information which does
13712not even relate to any client information. These ones are sometimes used with
13713"monitor-fail" directives to report an internal status to external watchers.
13714The sample fetch methods described in this section are usable anywhere.
13715
13716always_false : boolean
13717 Always returns the boolean "false" value. It may be used with ACLs as a
13718 temporary replacement for another one when adjusting configurations.
13719
13720always_true : boolean
13721 Always returns the boolean "true" value. It may be used with ACLs as a
13722 temporary replacement for another one when adjusting configurations.
13723
13724avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013725 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013726 divided by the number of active servers. The current backend is used if no
13727 backend is specified. This is very similar to "queue" except that the size of
13728 the farm is considered, in order to give a more accurate measurement of the
13729 time it may take for a new connection to be processed. The main usage is with
13730 ACL to return a sorry page to new users when it becomes certain they will get
13731 a degraded service, or to pass to the backend servers in a header so that
13732 they decide to work in degraded mode or to disable some functions to speed up
13733 the processing a bit. Note that in the event there would not be any active
13734 server anymore, twice the number of queued connections would be considered as
13735 the measured value. This is a fair estimate, as we expect one server to get
13736 back soon anyway, but we still prefer to send new traffic to another backend
13737 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13738 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013739
Willy Tarreau74ca5042013-06-11 23:12:07 +020013740be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013741 Applies to the number of currently established connections on the backend,
13742 possibly including the connection being evaluated. If no backend name is
13743 specified, the current one is used. But it is also possible to check another
13744 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013745 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
13746
13747be_conn_free([<backend>]) : integer
13748 Returns an integer value corresponding to the number of available connections
13749 across available servers in the backend. Queue slots are not included. Backup
13750 servers are also not included, unless all other servers are down. If no
13751 backend name is specified, the current one is used. But it is also possible
13752 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013753 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
13754 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013755
13756 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
13757 (meaning unlimited), then this fetch clearly does not make sense, in which
13758 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013759
Willy Tarreau74ca5042013-06-11 23:12:07 +020013760be_sess_rate([<backend>]) : integer
13761 Returns an integer value corresponding to the sessions creation rate on the
13762 backend, in number of new sessions per second. This is used with ACLs to
13763 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013764 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013765 online dictionary). It can also be useful to add this element to logs using a
13766 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013767
13768 Example :
13769 # Redirect to an error page if the dictionary is requested too often
13770 backend dynamic
13771 mode http
13772 acl being_scanned be_sess_rate gt 100
13773 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013774
Davor Ocelice9ed2812017-12-25 17:49:28 +010013775bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013776 Returns a binary chain. The input is the hexadecimal representation
13777 of the string.
13778
13779bool(<bool>) : bool
13780 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
13781 'false' and '0' are the same. 'true' and '1' are the same.
13782
Willy Tarreau74ca5042013-06-11 23:12:07 +020013783connslots([<backend>]) : integer
13784 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013785 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020013786 connections on all servers and the maximum queue size. This is probably only
13787 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050013788
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013789 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020013790 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013791 usage; see "use_backend" keyword) can be redirected to a different backend.
13792
Willy Tarreau55165fe2009-05-10 12:02:55 +020013793 'connslots' = number of available server connection slots, + number of
13794 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013795
Willy Tarreaua36af912009-10-10 12:02:45 +020013796 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020013797 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020013798 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020013799 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010013800 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020013801 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020013802 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013803
Willy Tarreau55165fe2009-05-10 12:02:55 +020013804 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
13805 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020013806 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020013807 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013808
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013809date([<offset>]) : integer
13810 Returns the current date as the epoch (number of seconds since 01/01/1970).
13811 If an offset value is specified, then it is a number of seconds that is added
13812 to the current date before returning the value. This is particularly useful
13813 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020013814 It is useful combined with the http_date converter.
13815
13816 Example :
13817
13818 # set an expires header to now+1 hour in every response
13819 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013820
Etienne Carrierea792a0a2018-01-17 13:43:24 +010013821date_us : integer
13822 Return the microseconds part of the date (the "second" part is returned by
13823 date sample). This sample is coherent with the date sample as it is comes
13824 from the same timeval structure.
13825
Willy Tarreaud716f9b2017-10-13 11:03:15 +020013826distcc_body(<token>[,<occ>]) : binary
13827 Parses a distcc message and returns the body associated to occurrence #<occ>
13828 of the token <token>. Occurrences start at 1, and when unspecified, any may
13829 match though in practice only the first one is checked for now. This can be
13830 used to extract file names or arguments in files built using distcc through
13831 haproxy. Please refer to distcc's protocol documentation for the complete
13832 list of supported tokens.
13833
13834distcc_param(<token>[,<occ>]) : integer
13835 Parses a distcc message and returns the parameter associated to occurrence
13836 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
13837 any may match though in practice only the first one is checked for now. This
13838 can be used to extract certain information such as the protocol version, the
13839 file size or the argument in files built using distcc through haproxy.
13840 Another use case consists in waiting for the start of the preprocessed file
13841 contents before connecting to the server to avoid keeping idle connections.
13842 Please refer to distcc's protocol documentation for the complete list of
13843 supported tokens.
13844
13845 Example :
13846 # wait up to 20s for the pre-processed file to be uploaded
13847 tcp-request inspect-delay 20s
13848 tcp-request content accept if { distcc_param(DOTI) -m found }
13849 # send large files to the big farm
13850 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
13851
Willy Tarreau595ec542013-06-12 21:34:28 +020013852env(<name>) : string
13853 Returns a string containing the value of environment variable <name>. As a
13854 reminder, environment variables are per-process and are sampled when the
13855 process starts. This can be useful to pass some information to a next hop
13856 server, or with ACLs to take specific action when the process is started a
13857 certain way.
13858
13859 Examples :
13860 # Pass the Via header to next hop with the local hostname in it
13861 http-request add-header Via 1.1\ %[env(HOSTNAME)]
13862
13863 # reject cookie-less requests when the STOP environment variable is set
13864 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
13865
Willy Tarreau74ca5042013-06-11 23:12:07 +020013866fe_conn([<frontend>]) : integer
13867 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010013868 possibly including the connection being evaluated. If no frontend name is
13869 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013870 frontend. It can be used to return a sorry page before hard-blocking, or to
13871 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010013872 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020013873 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
13874 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020013875
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020013876fe_req_rate([<frontend>]) : integer
13877 Returns an integer value corresponding to the number of HTTP requests per
13878 second sent to a frontend. This number can differ from "fe_sess_rate" in
13879 situations where client-side keep-alive is enabled.
13880
Willy Tarreau74ca5042013-06-11 23:12:07 +020013881fe_sess_rate([<frontend>]) : integer
13882 Returns an integer value corresponding to the sessions creation rate on the
13883 frontend, in number of new sessions per second. This is used with ACLs to
13884 limit the incoming session rate to an acceptable range in order to prevent
13885 abuse of service at the earliest moment, for example when combined with other
13886 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
13887 down below the limit. It can also be useful to add this element to logs using
13888 a log-format directive. See also the "rate-limit sessions" directive for use
13889 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013890
13891 Example :
13892 # This frontend limits incoming mails to 10/s with a max of 100
13893 # concurrent connections. We accept any connection below 10/s, and
13894 # force excess clients to wait for 100 ms. Since clients are limited to
13895 # 100 max, there cannot be more than 10 incoming mails per second.
13896 frontend mail
13897 bind :25
13898 mode tcp
13899 maxconn 100
13900 acl too_fast fe_sess_rate ge 10
13901 tcp-request inspect-delay 100ms
13902 tcp-request content accept if ! too_fast
13903 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010013904
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010013905hostname : string
13906 Returns the system hostname.
13907
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013908int(<integer>) : signed integer
13909 Returns a signed integer.
13910
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013911ipv4(<ipv4>) : ipv4
13912 Returns an ipv4.
13913
13914ipv6(<ipv6>) : ipv6
13915 Returns an ipv6.
13916
13917meth(<method>) : method
13918 Returns a method.
13919
Willy Tarreau0f30d262014-11-24 16:02:05 +010013920nbproc : integer
13921 Returns an integer value corresponding to the number of processes that were
13922 started (it equals the global "nbproc" setting). This is useful for logging
13923 and debugging purposes.
13924
Willy Tarreau74ca5042013-06-11 23:12:07 +020013925nbsrv([<backend>]) : integer
13926 Returns an integer value corresponding to the number of usable servers of
13927 either the current backend or the named backend. This is mostly used with
13928 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010013929 switch to an alternate backend when the number of servers is too low to
13930 to handle some load. It is useful to report a failure when combined with
13931 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013932
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040013933prio_class : integer
13934 Returns the priority class of the current session for http mode or connection
13935 for tcp mode. The value will be that set by the last call to "http-request
13936 set-priority-class" or "tcp-request content set-priority-class".
13937
13938prio_offset : integer
13939 Returns the priority offset of the current session for http mode or
13940 connection for tcp mode. The value will be that set by the last call to
13941 "http-request set-priority-offset" or "tcp-request content
13942 set-priority-offset".
13943
Willy Tarreau0f30d262014-11-24 16:02:05 +010013944proc : integer
13945 Returns an integer value corresponding to the position of the process calling
13946 the function, between 1 and global.nbproc. This is useful for logging and
13947 debugging purposes.
13948
Willy Tarreau74ca5042013-06-11 23:12:07 +020013949queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013950 Returns the total number of queued connections of the designated backend,
13951 including all the connections in server queues. If no backend name is
13952 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013953 one. This is useful with ACLs or to pass statistics to backend servers. This
13954 can be used to take actions when queuing goes above a known level, generally
13955 indicating a surge of traffic or a massive slowdown on the servers. One
13956 possible action could be to reject new users but still accept old ones. See
13957 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
13958
Willy Tarreau84310e22014-02-14 11:59:04 +010013959rand([<range>]) : integer
13960 Returns a random integer value within a range of <range> possible values,
13961 starting at zero. If the range is not specified, it defaults to 2^32, which
13962 gives numbers between 0 and 4294967295. It can be useful to pass some values
13963 needed to take some routing decisions for example, or just for debugging
13964 purposes. This random must not be used for security purposes.
13965
Willy Tarreau74ca5042013-06-11 23:12:07 +020013966srv_conn([<backend>/]<server>) : integer
13967 Returns an integer value corresponding to the number of currently established
13968 connections on the designated server, possibly including the connection being
13969 evaluated. If <backend> is omitted, then the server is looked up in the
13970 current backend. It can be used to use a specific farm when one server is
13971 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013972 connections with it. See also the "fe_conn", "be_conn", "queue", and
13973 "srv_conn_free" fetch methods.
13974
13975srv_conn_free([<backend>/]<server>) : integer
13976 Returns an integer value corresponding to the number of available connections
13977 on the designated server, possibly including the connection being evaluated.
13978 The value does not include queue slots. If <backend> is omitted, then the
13979 server is looked up in the current backend. It can be used to use a specific
13980 farm when one server is full, or to inform the server about our view of the
13981 number of active connections with it. See also the "be_conn_free" and
13982 "srv_conn" fetch methods.
13983
13984 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
13985 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013986
13987srv_is_up([<backend>/]<server>) : boolean
13988 Returns true when the designated server is UP, and false when it is either
13989 DOWN or in maintenance mode. If <backend> is omitted, then the server is
13990 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010013991 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020013992 availability). Another possible use which is more of a hack consists in
13993 using dummy servers as boolean variables that can be enabled or disabled from
13994 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
13995
Willy Tarreauff2b7af2017-10-13 11:46:26 +020013996srv_queue([<backend>/]<server>) : integer
13997 Returns an integer value corresponding to the number of connections currently
13998 pending in the designated server's queue. If <backend> is omitted, then the
13999 server is looked up in the current backend. It can sometimes be used together
14000 with the "use-server" directive to force to use a known faster server when it
14001 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14002 fetch methods.
14003
Willy Tarreau74ca5042013-06-11 23:12:07 +020014004srv_sess_rate([<backend>/]<server>) : integer
14005 Returns an integer corresponding to the sessions creation rate on the
14006 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014007 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014008 used with ACLs but can make sense with logs too. This is used to switch to an
14009 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014010 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014011 overloading servers).
14012
14013 Example :
14014 # Redirect to a separate back
14015 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14016 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14017 use_backend be2 if srv1_full or srv2_full
14018
Willy Tarreau0f30d262014-11-24 16:02:05 +010014019stopping : boolean
14020 Returns TRUE if the process calling the function is currently stopping. This
14021 can be useful for logging, or for relaxing certain checks or helping close
14022 certain connections upon graceful shutdown.
14023
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014024str(<string>) : string
14025 Returns a string.
14026
Willy Tarreau74ca5042013-06-11 23:12:07 +020014027table_avl([<table>]) : integer
14028 Returns the total number of available entries in the current proxy's
14029 stick-table or in the designated stick-table. See also table_cnt.
14030
14031table_cnt([<table>]) : integer
14032 Returns the total number of entries currently in use in the current proxy's
14033 stick-table or in the designated stick-table. See also src_conn_cnt and
14034 table_avl for other entry counting methods.
14035
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014036thread : integer
14037 Returns an integer value corresponding to the position of the thread calling
14038 the function, between 0 and (global.nbthread-1). This is useful for logging
14039 and debugging purposes.
14040
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014041var(<var-name>) : undefined
14042 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014043 sample fetch fails. The name of the variable starts with an indication
14044 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014045 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014046 "sess" : the variable is shared with the whole session
14047 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014048 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014049 "req" : the variable is shared only during request processing,
14050 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014051 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014052 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014053
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200140547.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014055----------------------------------
14056
14057The layer 4 usually describes just the transport layer which in haproxy is
14058closest to the connection, where no content is yet made available. The fetch
14059methods described here are usable as low as the "tcp-request connection" rule
14060sets unless they require some future information. Those generally include
14061TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014062the incoming connection. For retrieving a value from a sticky counters, the
14063counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014064"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14065used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14066can be specified as the first integer argument when using the "sc_" prefix.
14067Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14068table may be specified with the "sc*" form, in which case the currently
14069tracked key will be looked up into this alternate table instead of the table
14070currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014071
14072be_id : integer
14073 Returns an integer containing the current backend's id. It can be used in
14074 frontends with responses to check which backend processed the request.
14075
Marcin Deranekd2471c22016-12-12 14:08:05 +010014076be_name : string
14077 Returns a string containing the current backend's name. It can be used in
14078 frontends with responses to check which backend processed the request.
14079
Willy Tarreau74ca5042013-06-11 23:12:07 +020014080dst : ip
14081 This is the destination IPv4 address of the connection on the client side,
14082 which is the address the client connected to. It can be useful when running
14083 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14084 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
14085 RFC 4291.
14086
14087dst_conn : integer
14088 Returns an integer value corresponding to the number of currently established
14089 connections on the same socket including the one being evaluated. It is
14090 normally used with ACLs but can as well be used to pass the information to
14091 servers in an HTTP header or in logs. It can be used to either return a sorry
14092 page before hard-blocking, or to use a specific backend to drain new requests
14093 when the socket is considered saturated. This offers the ability to assign
14094 different limits to different listening ports or addresses. See also the
14095 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014096
Willy Tarreau16e01562016-08-09 16:46:18 +020014097dst_is_local : boolean
14098 Returns true if the destination address of the incoming connection is local
14099 to the system, or false if the address doesn't exist on the system, meaning
14100 that it was intercepted in transparent mode. It can be useful to apply
14101 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014102 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014103 be delivered only on this address, or SSH access could be locally redirected.
14104 Please note that the check involves a few system calls, so it's better to do
14105 it only once per connection.
14106
Willy Tarreau74ca5042013-06-11 23:12:07 +020014107dst_port : integer
14108 Returns an integer value corresponding to the destination TCP port of the
14109 connection on the client side, which is the port the client connected to.
14110 This might be used when running in transparent mode, when assigning dynamic
14111 ports to some clients for a whole application session, to stick all users to
14112 a same server, or to pass the destination port information to a server using
14113 an HTTP header.
14114
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014115fc_http_major : integer
14116 Reports the front connection's HTTP major version encoding, which may be 1
14117 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14118 encoding and not on the version present in the request header.
14119
Emeric Brun4f603012017-01-05 15:11:44 +010014120fc_rcvd_proxy : boolean
14121 Returns true if the client initiated the connection with a PROXY protocol
14122 header.
14123
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014124fc_rtt(<unit>) : integer
14125 Returns the Round Trip Time (RTT) measured by the kernel for the client
14126 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14127 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14128 connection is not established, if the connection is not TCP or if the
14129 operating system does not support TCP_INFO, for example Linux kernels before
14130 2.4, the sample fetch fails.
14131
14132fc_rttvar(<unit>) : integer
14133 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14134 client connection. <unit> is facultative, by default the unit is milliseconds.
14135 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14136 server connection is not established, if the connection is not TCP or if the
14137 operating system does not support TCP_INFO, for example Linux kernels before
14138 2.4, the sample fetch fails.
14139
Joe Williams30fcd392016-08-10 07:06:44 -070014140fc_unacked(<unit>) : integer
14141 Returns the unacked counter measured by the kernel for the client connection.
14142 If the server connection is not established, if the connection is not TCP or
14143 if the operating system does not support TCP_INFO, for example Linux kernels
14144 before 2.4, the sample fetch fails.
14145
14146fc_sacked(<unit>) : integer
14147 Returns the sacked counter measured by the kernel for the client connection.
14148 If the server connection is not established, if the connection is not TCP or
14149 if the operating system does not support TCP_INFO, for example Linux kernels
14150 before 2.4, the sample fetch fails.
14151
14152fc_retrans(<unit>) : integer
14153 Returns the retransmits counter measured by the kernel for the client
14154 connection. If the server connection is not established, if the connection is
14155 not TCP or if the operating system does not support TCP_INFO, for example
14156 Linux kernels before 2.4, the sample fetch fails.
14157
14158fc_fackets(<unit>) : integer
14159 Returns the fack counter measured by the kernel for the client
14160 connection. If the server connection is not established, if the connection is
14161 not TCP or if the operating system does not support TCP_INFO, for example
14162 Linux kernels before 2.4, the sample fetch fails.
14163
14164fc_lost(<unit>) : integer
14165 Returns the lost counter measured by the kernel for the client
14166 connection. If the server connection is not established, if the connection is
14167 not TCP or if the operating system does not support TCP_INFO, for example
14168 Linux kernels before 2.4, the sample fetch fails.
14169
14170fc_reordering(<unit>) : integer
14171 Returns the reordering counter measured by the kernel for the client
14172 connection. If the server connection is not established, if the connection is
14173 not TCP or if the operating system does not support TCP_INFO, for example
14174 Linux kernels before 2.4, the sample fetch fails.
14175
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014176fe_defbe : string
14177 Returns a string containing the frontend's default backend name. It can be
14178 used in frontends to check which backend will handle requests by default.
14179
Willy Tarreau74ca5042013-06-11 23:12:07 +020014180fe_id : integer
14181 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014182 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014183 coming via a same frontend to the same server.
14184
Marcin Deranekd2471c22016-12-12 14:08:05 +010014185fe_name : string
14186 Returns a string containing the current frontend's name. It can be used in
14187 backends to check from which frontend it was called, or to stick all users
14188 coming via a same frontend to the same server.
14189
Cyril Bonté62ba8702014-04-22 23:52:25 +020014190sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014191sc0_bytes_in_rate([<table>]) : integer
14192sc1_bytes_in_rate([<table>]) : integer
14193sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014194 Returns the average client-to-server bytes rate from the currently tracked
14195 counters, measured in amount of bytes over the period configured in the
14196 table. See also src_bytes_in_rate.
14197
Cyril Bonté62ba8702014-04-22 23:52:25 +020014198sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014199sc0_bytes_out_rate([<table>]) : integer
14200sc1_bytes_out_rate([<table>]) : integer
14201sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014202 Returns the average server-to-client bytes rate from the currently tracked
14203 counters, measured in amount of bytes over the period configured in the
14204 table. See also src_bytes_out_rate.
14205
Cyril Bonté62ba8702014-04-22 23:52:25 +020014206sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014207sc0_clr_gpc0([<table>]) : integer
14208sc1_clr_gpc0([<table>]) : integer
14209sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014210 Clears the first General Purpose Counter associated to the currently tracked
14211 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014212 stored value is zero, so first invocation will always return zero. This is
14213 typically used as a second ACL in an expression in order to mark a connection
14214 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014215
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014216 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014217 # block if 5 consecutive requests continue to come faster than 10 sess
14218 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014219 acl abuse sc0_http_req_rate gt 10
14220 acl kill sc0_inc_gpc0 gt 5
14221 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014222 tcp-request connection accept if !abuse save
14223 tcp-request connection reject if abuse kill
14224
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014225sc_clr_gpc1(<ctr>[,<table>]) : integer
14226sc0_clr_gpc1([<table>]) : integer
14227sc1_clr_gpc1([<table>]) : integer
14228sc2_clr_gpc1([<table>]) : integer
14229 Clears the second General Purpose Counter associated to the currently tracked
14230 counters, and returns its previous value. Before the first invocation, the
14231 stored value is zero, so first invocation will always return zero. This is
14232 typically used as a second ACL in an expression in order to mark a connection
14233 when a first ACL was verified.
14234
Cyril Bonté62ba8702014-04-22 23:52:25 +020014235sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014236sc0_conn_cnt([<table>]) : integer
14237sc1_conn_cnt([<table>]) : integer
14238sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014239 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014240 counters. See also src_conn_cnt.
14241
Cyril Bonté62ba8702014-04-22 23:52:25 +020014242sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014243sc0_conn_cur([<table>]) : integer
14244sc1_conn_cur([<table>]) : integer
14245sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014246 Returns the current amount of concurrent connections tracking the same
14247 tracked counters. This number is automatically incremented when tracking
14248 begins and decremented when tracking stops. See also src_conn_cur.
14249
Cyril Bonté62ba8702014-04-22 23:52:25 +020014250sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014251sc0_conn_rate([<table>]) : integer
14252sc1_conn_rate([<table>]) : integer
14253sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014254 Returns the average connection rate from the currently tracked counters,
14255 measured in amount of connections over the period configured in the table.
14256 See also src_conn_rate.
14257
Cyril Bonté62ba8702014-04-22 23:52:25 +020014258sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014259sc0_get_gpc0([<table>]) : integer
14260sc1_get_gpc0([<table>]) : integer
14261sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014262 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014263 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014264
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014265sc_get_gpc1(<ctr>[,<table>]) : integer
14266sc0_get_gpc1([<table>]) : integer
14267sc1_get_gpc1([<table>]) : integer
14268sc2_get_gpc1([<table>]) : integer
14269 Returns the value of the second General Purpose Counter associated to the
14270 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14271
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014272sc_get_gpt0(<ctr>[,<table>]) : integer
14273sc0_get_gpt0([<table>]) : integer
14274sc1_get_gpt0([<table>]) : integer
14275sc2_get_gpt0([<table>]) : integer
14276 Returns the value of the first General Purpose Tag associated to the
14277 currently tracked counters. See also src_get_gpt0.
14278
Cyril Bonté62ba8702014-04-22 23:52:25 +020014279sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014280sc0_gpc0_rate([<table>]) : integer
14281sc1_gpc0_rate([<table>]) : integer
14282sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014283 Returns the average increment rate of the first General Purpose Counter
14284 associated to the currently tracked counters. It reports the frequency
14285 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014286 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14287 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14288 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014289
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014290sc_gpc1_rate(<ctr>[,<table>]) : integer
14291sc0_gpc1_rate([<table>]) : integer
14292sc1_gpc1_rate([<table>]) : integer
14293sc2_gpc1_rate([<table>]) : integer
14294 Returns the average increment rate of the second General Purpose Counter
14295 associated to the currently tracked counters. It reports the frequency
14296 which the gpc1 counter was incremented over the configured period. See also
14297 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14298 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14299 be returned, as "gpc1" only holds the event count.
14300
Cyril Bonté62ba8702014-04-22 23:52:25 +020014301sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014302sc0_http_err_cnt([<table>]) : integer
14303sc1_http_err_cnt([<table>]) : integer
14304sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014305 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014306 counters. This includes the both request errors and 4xx error responses.
14307 See also src_http_err_cnt.
14308
Cyril Bonté62ba8702014-04-22 23:52:25 +020014309sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014310sc0_http_err_rate([<table>]) : integer
14311sc1_http_err_rate([<table>]) : integer
14312sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014313 Returns the average rate of HTTP errors from the currently tracked counters,
14314 measured in amount of errors over the period configured in the table. This
14315 includes the both request errors and 4xx error responses. See also
14316 src_http_err_rate.
14317
Cyril Bonté62ba8702014-04-22 23:52:25 +020014318sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014319sc0_http_req_cnt([<table>]) : integer
14320sc1_http_req_cnt([<table>]) : integer
14321sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014322 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014323 counters. This includes every started request, valid or not. See also
14324 src_http_req_cnt.
14325
Cyril Bonté62ba8702014-04-22 23:52:25 +020014326sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014327sc0_http_req_rate([<table>]) : integer
14328sc1_http_req_rate([<table>]) : integer
14329sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014330 Returns the average rate of HTTP requests from the currently tracked
14331 counters, measured in amount of requests over the period configured in
14332 the table. This includes every started request, valid or not. See also
14333 src_http_req_rate.
14334
Cyril Bonté62ba8702014-04-22 23:52:25 +020014335sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014336sc0_inc_gpc0([<table>]) : integer
14337sc1_inc_gpc0([<table>]) : integer
14338sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014339 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014340 tracked counters, and returns its new value. Before the first invocation,
14341 the stored value is zero, so first invocation will increase it to 1 and will
14342 return 1. This is typically used as a second ACL in an expression in order
14343 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014344
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014345 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014346 acl abuse sc0_http_req_rate gt 10
14347 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014348 tcp-request connection reject if abuse kill
14349
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014350sc_inc_gpc1(<ctr>[,<table>]) : integer
14351sc0_inc_gpc1([<table>]) : integer
14352sc1_inc_gpc1([<table>]) : integer
14353sc2_inc_gpc1([<table>]) : integer
14354 Increments the second General Purpose Counter associated to the currently
14355 tracked counters, and returns its new value. Before the first invocation,
14356 the stored value is zero, so first invocation will increase it to 1 and will
14357 return 1. This is typically used as a second ACL in an expression in order
14358 to mark a connection when a first ACL was verified.
14359
Cyril Bonté62ba8702014-04-22 23:52:25 +020014360sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014361sc0_kbytes_in([<table>]) : integer
14362sc1_kbytes_in([<table>]) : integer
14363sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014364 Returns the total amount of client-to-server data from the currently tracked
14365 counters, measured in kilobytes. The test is currently performed on 32-bit
14366 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014367
Cyril Bonté62ba8702014-04-22 23:52:25 +020014368sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014369sc0_kbytes_out([<table>]) : integer
14370sc1_kbytes_out([<table>]) : integer
14371sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014372 Returns the total amount of server-to-client data from the currently tracked
14373 counters, measured in kilobytes. The test is currently performed on 32-bit
14374 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014375
Cyril Bonté62ba8702014-04-22 23:52:25 +020014376sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014377sc0_sess_cnt([<table>]) : integer
14378sc1_sess_cnt([<table>]) : integer
14379sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014380 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014381 into sessions, which means that they were accepted by a "tcp-request
14382 connection" rule, from the currently tracked counters. A backend may count
14383 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014384 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014385 with the client. See also src_sess_cnt.
14386
Cyril Bonté62ba8702014-04-22 23:52:25 +020014387sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014388sc0_sess_rate([<table>]) : integer
14389sc1_sess_rate([<table>]) : integer
14390sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014391 Returns the average session rate from the currently tracked counters,
14392 measured in amount of sessions over the period configured in the table. A
14393 session is a connection that got past the early "tcp-request connection"
14394 rules. A backend may count more sessions than connections because each
14395 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014396 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014397
Cyril Bonté62ba8702014-04-22 23:52:25 +020014398sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014399sc0_tracked([<table>]) : boolean
14400sc1_tracked([<table>]) : boolean
14401sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014402 Returns true if the designated session counter is currently being tracked by
14403 the current session. This can be useful when deciding whether or not we want
14404 to set some values in a header passed to the server.
14405
Cyril Bonté62ba8702014-04-22 23:52:25 +020014406sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014407sc0_trackers([<table>]) : integer
14408sc1_trackers([<table>]) : integer
14409sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014410 Returns the current amount of concurrent connections tracking the same
14411 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014412 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014413 that it does not rely on any stored information but on the table's reference
14414 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014415 may sometimes be more suited for layer7 tracking. It can be used to tell a
14416 server how many concurrent connections there are from a given address for
14417 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014418
Willy Tarreau74ca5042013-06-11 23:12:07 +020014419so_id : integer
14420 Returns an integer containing the current listening socket's id. It is useful
14421 in frontends involving many "bind" lines, or to stick all users coming via a
14422 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014423
Willy Tarreau74ca5042013-06-11 23:12:07 +020014424src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014425 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014426 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14427 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14428 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014429 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14430 directive is used, it can be the address of a client behind another
14431 PROXY-protocol compatible component for all rule sets except
14432 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014433
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014434 Example:
14435 # add an HTTP header in requests with the originating address' country
14436 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14437
Willy Tarreau74ca5042013-06-11 23:12:07 +020014438src_bytes_in_rate([<table>]) : integer
14439 Returns the average bytes rate from the incoming connection's source address
14440 in the current proxy's stick-table or in the designated stick-table, measured
14441 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014442 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014443
Willy Tarreau74ca5042013-06-11 23:12:07 +020014444src_bytes_out_rate([<table>]) : integer
14445 Returns the average bytes rate to the incoming connection's source address in
14446 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014447 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014448 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014449
Willy Tarreau74ca5042013-06-11 23:12:07 +020014450src_clr_gpc0([<table>]) : integer
14451 Clears the first General Purpose Counter associated to the incoming
14452 connection's source address in the current proxy's stick-table or in the
14453 designated stick-table, and returns its previous value. If the address is not
14454 found, an entry is created and 0 is returned. This is typically used as a
14455 second ACL in an expression in order to mark a connection when a first ACL
14456 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014457
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014458 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014459 # block if 5 consecutive requests continue to come faster than 10 sess
14460 # per second, and reset the counter as soon as the traffic slows down.
14461 acl abuse src_http_req_rate gt 10
14462 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014463 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014464 tcp-request connection accept if !abuse save
14465 tcp-request connection reject if abuse kill
14466
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014467src_clr_gpc1([<table>]) : integer
14468 Clears the second General Purpose Counter associated to the incoming
14469 connection's source address in the current proxy's stick-table or in the
14470 designated stick-table, and returns its previous value. If the address is not
14471 found, an entry is created and 0 is returned. This is typically used as a
14472 second ACL in an expression in order to mark a connection when a first ACL
14473 was verified.
14474
Willy Tarreau74ca5042013-06-11 23:12:07 +020014475src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014476 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014477 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014478 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014479 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014480
Willy Tarreau74ca5042013-06-11 23:12:07 +020014481src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014482 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014483 current incoming connection's source address in the current proxy's
14484 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014485 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014486
Willy Tarreau74ca5042013-06-11 23:12:07 +020014487src_conn_rate([<table>]) : integer
14488 Returns the average connection rate from the incoming connection's source
14489 address in the current proxy's stick-table or in the designated stick-table,
14490 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014491 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014492
Willy Tarreau74ca5042013-06-11 23:12:07 +020014493src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014494 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014495 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014496 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014497 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014498
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014499src_get_gpc1([<table>]) : integer
14500 Returns the value of the second General Purpose Counter associated to the
14501 incoming connection's source address in the current proxy's stick-table or in
14502 the designated stick-table. If the address is not found, zero is returned.
14503 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14504
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014505src_get_gpt0([<table>]) : integer
14506 Returns the value of the first General Purpose Tag associated to the
14507 incoming connection's source address in the current proxy's stick-table or in
14508 the designated stick-table. If the address is not found, zero is returned.
14509 See also sc/sc0/sc1/sc2_get_gpt0.
14510
Willy Tarreau74ca5042013-06-11 23:12:07 +020014511src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014512 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014513 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014514 stick-table or in the designated stick-table. It reports the frequency
14515 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014516 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14517 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14518 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014519
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014520src_gpc1_rate([<table>]) : integer
14521 Returns the average increment rate of the second General Purpose Counter
14522 associated to the incoming connection's source address in the current proxy's
14523 stick-table or in the designated stick-table. It reports the frequency
14524 which the gpc1 counter was incremented over the configured period. See also
14525 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14526 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14527 be returned, as "gpc1" only holds the event count.
14528
Willy Tarreau74ca5042013-06-11 23:12:07 +020014529src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014530 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014531 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014532 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014533 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014534 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014535
Willy Tarreau74ca5042013-06-11 23:12:07 +020014536src_http_err_rate([<table>]) : integer
14537 Returns the average rate of HTTP errors from the incoming connection's source
14538 address in the current proxy's stick-table or in the designated stick-table,
14539 measured in amount of errors over the period configured in the table. This
14540 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014541 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014542
Willy Tarreau74ca5042013-06-11 23:12:07 +020014543src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014544 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014545 source address in the current proxy's stick-table or in the designated stick-
14546 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014547 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014548
Willy Tarreau74ca5042013-06-11 23:12:07 +020014549src_http_req_rate([<table>]) : integer
14550 Returns the average rate of HTTP requests from the incoming connection's
14551 source address in the current proxy's stick-table or in the designated stick-
14552 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014553 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014554 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014555
Willy Tarreau74ca5042013-06-11 23:12:07 +020014556src_inc_gpc0([<table>]) : integer
14557 Increments the first General Purpose Counter associated to the incoming
14558 connection's source address in the current proxy's stick-table or in the
14559 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014560 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014561 This is typically used as a second ACL in an expression in order to mark a
14562 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014563
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014564 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014565 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014566 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014567 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014568
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014569src_inc_gpc1([<table>]) : integer
14570 Increments the second General Purpose Counter associated to the incoming
14571 connection's source address in the current proxy's stick-table or in the
14572 designated stick-table, and returns its new value. If the address is not
14573 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14574 This is typically used as a second ACL in an expression in order to mark a
14575 connection when a first ACL was verified.
14576
Willy Tarreau16e01562016-08-09 16:46:18 +020014577src_is_local : boolean
14578 Returns true if the source address of the incoming connection is local to the
14579 system, or false if the address doesn't exist on the system, meaning that it
14580 comes from a remote machine. Note that UNIX addresses are considered local.
14581 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014582 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014583 note that the check involves a few system calls, so it's better to do it only
14584 once per connection.
14585
Willy Tarreau74ca5042013-06-11 23:12:07 +020014586src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014587 Returns the total amount of data received from the incoming connection's
14588 source address in the current proxy's stick-table or in the designated
14589 stick-table, measured in kilobytes. If the address is not found, zero is
14590 returned. The test is currently performed on 32-bit integers, which limits
14591 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014592
Willy Tarreau74ca5042013-06-11 23:12:07 +020014593src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014594 Returns the total amount of data sent to the incoming connection's source
14595 address in the current proxy's stick-table or in the designated stick-table,
14596 measured in kilobytes. If the address is not found, zero is returned. The
14597 test is currently performed on 32-bit integers, which limits values to 4
14598 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014599
Willy Tarreau74ca5042013-06-11 23:12:07 +020014600src_port : integer
14601 Returns an integer value corresponding to the TCP source port of the
14602 connection on the client side, which is the port the client connected from.
14603 Usage of this function is very limited as modern protocols do not care much
14604 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014605
Willy Tarreau74ca5042013-06-11 23:12:07 +020014606src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014607 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014608 connection's source IPv4 address in the current proxy's stick-table or in the
14609 designated stick-table, that were transformed into sessions, which means that
14610 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014611 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014612
Willy Tarreau74ca5042013-06-11 23:12:07 +020014613src_sess_rate([<table>]) : integer
14614 Returns the average session rate from the incoming connection's source
14615 address in the current proxy's stick-table or in the designated stick-table,
14616 measured in amount of sessions over the period configured in the table. A
14617 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014618 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014619
Willy Tarreau74ca5042013-06-11 23:12:07 +020014620src_updt_conn_cnt([<table>]) : integer
14621 Creates or updates the entry associated to the incoming connection's source
14622 address in the current proxy's stick-table or in the designated stick-table.
14623 This table must be configured to store the "conn_cnt" data type, otherwise
14624 the match will be ignored. The current count is incremented by one, and the
14625 expiration timer refreshed. The updated count is returned, so this match
14626 can't return zero. This was used to reject service abusers based on their
14627 source address. Note: it is recommended to use the more complete "track-sc*"
14628 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014629
14630 Example :
14631 # This frontend limits incoming SSH connections to 3 per 10 second for
14632 # each source address, and rejects excess connections until a 10 second
14633 # silence is observed. At most 20 addresses are tracked.
14634 listen ssh
14635 bind :22
14636 mode tcp
14637 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014638 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014639 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014640 server local 127.0.0.1:22
14641
Willy Tarreau74ca5042013-06-11 23:12:07 +020014642srv_id : integer
14643 Returns an integer containing the server's id when processing the response.
14644 While it's almost only used with ACLs, it may be used for logging or
14645 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014646
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200146477.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014648----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014649
Willy Tarreau74ca5042013-06-11 23:12:07 +020014650The layer 5 usually describes just the session layer which in haproxy is
14651closest to the session once all the connection handshakes are finished, but
14652when no content is yet made available. The fetch methods described here are
14653usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014654future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014655
Ben Shillitof25e8e52016-12-02 14:25:37 +00001465651d.all(<prop>[,<prop>*]) : string
14657 Returns values for the properties requested as a string, where values are
14658 separated by the delimiter specified with "51degrees-property-separator".
14659 The device is identified using all the important HTTP headers from the
14660 request. The function can be passed up to five property names, and if a
14661 property name can't be found, the value "NoData" is returned.
14662
14663 Example :
14664 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14665 # containing the three properties requested using all relevant headers from
14666 # the request.
14667 frontend http-in
14668 bind *:8081
14669 default_backend servers
14670 http-request set-header X-51D-DeviceTypeMobileTablet \
14671 %[51d.all(DeviceType,IsMobile,IsTablet)]
14672
Emeric Brun645ae792014-04-30 14:21:06 +020014673ssl_bc : boolean
14674 Returns true when the back connection was made via an SSL/TLS transport
14675 layer and is locally deciphered. This means the outgoing connection was made
14676 other a server with the "ssl" option.
14677
14678ssl_bc_alg_keysize : integer
14679 Returns the symmetric cipher key size supported in bits when the outgoing
14680 connection was made over an SSL/TLS transport layer.
14681
14682ssl_bc_cipher : string
14683 Returns the name of the used cipher when the outgoing connection was made
14684 over an SSL/TLS transport layer.
14685
Emeric Brun74f7ffa2018-02-19 16:14:12 +010014686ssl_bc_is_resumed : boolean
14687 Returns true when the back connection was made over an SSL/TLS transport
14688 layer and the newly created SSL session was resumed using a cached
14689 session or a TLS ticket.
14690
Emeric Brun645ae792014-04-30 14:21:06 +020014691ssl_bc_protocol : string
14692 Returns the name of the used protocol when the outgoing connection was made
14693 over an SSL/TLS transport layer.
14694
Emeric Brunb73a9b02014-04-30 18:49:19 +020014695ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020014696 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014697 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14698 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020014699
14700ssl_bc_session_id : binary
14701 Returns the SSL ID of the back connection when the outgoing connection was
14702 made over an SSL/TLS transport layer. It is useful to log if we want to know
14703 if session was reused or not.
14704
Patrick Hemmere0275472018-04-28 19:15:51 -040014705ssl_bc_session_key : binary
14706 Returns the SSL session master key of the back connection when the outgoing
14707 connection was made over an SSL/TLS transport layer. It is useful to decrypt
14708 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
14709 BoringSSL.
14710
Emeric Brun645ae792014-04-30 14:21:06 +020014711ssl_bc_use_keysize : integer
14712 Returns the symmetric cipher key size used in bits when the outgoing
14713 connection was made over an SSL/TLS transport layer.
14714
Willy Tarreau74ca5042013-06-11 23:12:07 +020014715ssl_c_ca_err : integer
14716 When the incoming connection was made over an SSL/TLS transport layer,
14717 returns the ID of the first error detected during verification of the client
14718 certificate at depth > 0, or 0 if no error was encountered during this
14719 verification process. Please refer to your SSL library's documentation to
14720 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020014721
Willy Tarreau74ca5042013-06-11 23:12:07 +020014722ssl_c_ca_err_depth : integer
14723 When the incoming connection was made over an SSL/TLS transport layer,
14724 returns the depth in the CA chain of the first error detected during the
14725 verification of the client certificate. If no error is encountered, 0 is
14726 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014727
Emeric Brun43e79582014-10-29 19:03:26 +010014728ssl_c_der : binary
14729 Returns the DER formatted certificate presented by the client when the
14730 incoming connection was made over an SSL/TLS transport layer. When used for
14731 an ACL, the value(s) to match against can be passed in hexadecimal form.
14732
Willy Tarreau74ca5042013-06-11 23:12:07 +020014733ssl_c_err : integer
14734 When the incoming connection was made over an SSL/TLS transport layer,
14735 returns the ID of the first error detected during verification at depth 0, or
14736 0 if no error was encountered during this verification process. Please refer
14737 to your SSL library's documentation to find the exhaustive list of error
14738 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020014739
Willy Tarreau74ca5042013-06-11 23:12:07 +020014740ssl_c_i_dn([<entry>[,<occ>]]) : string
14741 When the incoming connection was made over an SSL/TLS transport layer,
14742 returns the full distinguished name of the issuer of the certificate
14743 presented by the client when no <entry> is specified, or the value of the
14744 first given entry found from the beginning of the DN. If a positive/negative
14745 occurrence number is specified as the optional second argument, it returns
14746 the value of the nth given entry value from the beginning/end of the DN.
14747 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
14748 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020014749
Willy Tarreau74ca5042013-06-11 23:12:07 +020014750ssl_c_key_alg : string
14751 Returns the name of the algorithm used to generate the key of the certificate
14752 presented by the client when the incoming connection was made over an SSL/TLS
14753 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020014754
Willy Tarreau74ca5042013-06-11 23:12:07 +020014755ssl_c_notafter : string
14756 Returns the end date presented by the client as a formatted string
14757 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14758 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020014759
Willy Tarreau74ca5042013-06-11 23:12:07 +020014760ssl_c_notbefore : string
14761 Returns the start date presented by the client as a formatted string
14762 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14763 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014764
Willy Tarreau74ca5042013-06-11 23:12:07 +020014765ssl_c_s_dn([<entry>[,<occ>]]) : string
14766 When the incoming connection was made over an SSL/TLS transport layer,
14767 returns the full distinguished name of the subject of the certificate
14768 presented by the client when no <entry> is specified, or the value of the
14769 first given entry found from the beginning of the DN. If a positive/negative
14770 occurrence number is specified as the optional second argument, it returns
14771 the value of the nth given entry value from the beginning/end of the DN.
14772 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
14773 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014774
Willy Tarreau74ca5042013-06-11 23:12:07 +020014775ssl_c_serial : binary
14776 Returns the serial of the certificate presented by the client when the
14777 incoming connection was made over an SSL/TLS transport layer. When used for
14778 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014779
Willy Tarreau74ca5042013-06-11 23:12:07 +020014780ssl_c_sha1 : binary
14781 Returns the SHA-1 fingerprint of the certificate presented by the client when
14782 the incoming connection was made over an SSL/TLS transport layer. This can be
14783 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014784 Note that the output is binary, so if you want to pass that signature to the
14785 server, you need to encode it in hex or base64, such as in the example below:
14786
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014787 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014788 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020014789
Willy Tarreau74ca5042013-06-11 23:12:07 +020014790ssl_c_sig_alg : string
14791 Returns the name of the algorithm used to sign the certificate presented by
14792 the client when the incoming connection was made over an SSL/TLS transport
14793 layer.
Emeric Brun87855892012-10-17 17:39:35 +020014794
Willy Tarreau74ca5042013-06-11 23:12:07 +020014795ssl_c_used : boolean
14796 Returns true if current SSL session uses a client certificate even if current
14797 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020014798
Willy Tarreau74ca5042013-06-11 23:12:07 +020014799ssl_c_verify : integer
14800 Returns the verify result error ID when the incoming connection was made over
14801 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
14802 refer to your SSL library's documentation for an exhaustive list of error
14803 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020014804
Willy Tarreau74ca5042013-06-11 23:12:07 +020014805ssl_c_version : integer
14806 Returns the version of the certificate presented by the client when the
14807 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020014808
Emeric Brun43e79582014-10-29 19:03:26 +010014809ssl_f_der : binary
14810 Returns the DER formatted certificate presented by the frontend when the
14811 incoming connection was made over an SSL/TLS transport layer. When used for
14812 an ACL, the value(s) to match against can be passed in hexadecimal form.
14813
Willy Tarreau74ca5042013-06-11 23:12:07 +020014814ssl_f_i_dn([<entry>[,<occ>]]) : string
14815 When the incoming connection was made over an SSL/TLS transport layer,
14816 returns the full distinguished name of the issuer of the certificate
14817 presented by the frontend when no <entry> is specified, or the value of the
14818 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020014819 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020014820 the value of the nth given entry value from the beginning/end of the DN.
14821 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
14822 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020014823
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824ssl_f_key_alg : string
14825 Returns the name of the algorithm used to generate the key of the certificate
14826 presented by the frontend when the incoming connection was made over an
14827 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014828
Willy Tarreau74ca5042013-06-11 23:12:07 +020014829ssl_f_notafter : string
14830 Returns the end date presented by the frontend as a formatted string
14831 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14832 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014833
Willy Tarreau74ca5042013-06-11 23:12:07 +020014834ssl_f_notbefore : string
14835 Returns the start date presented by the frontend as a formatted string
14836 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14837 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020014838
Willy Tarreau74ca5042013-06-11 23:12:07 +020014839ssl_f_s_dn([<entry>[,<occ>]]) : string
14840 When the incoming connection was made over an SSL/TLS transport layer,
14841 returns the full distinguished name of the subject of the certificate
14842 presented by the frontend when no <entry> is specified, or the value of the
14843 first given entry found from the beginning of the DN. If a positive/negative
14844 occurrence number is specified as the optional second argument, it returns
14845 the value of the nth given entry value from the beginning/end of the DN.
14846 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
14847 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020014848
Willy Tarreau74ca5042013-06-11 23:12:07 +020014849ssl_f_serial : binary
14850 Returns the serial of the certificate presented by the frontend when the
14851 incoming connection was made over an SSL/TLS transport layer. When used for
14852 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020014853
Emeric Brun55f4fa82014-04-30 17:11:25 +020014854ssl_f_sha1 : binary
14855 Returns the SHA-1 fingerprint of the certificate presented by the frontend
14856 when the incoming connection was made over an SSL/TLS transport layer. This
14857 can be used to know which certificate was chosen using SNI.
14858
Willy Tarreau74ca5042013-06-11 23:12:07 +020014859ssl_f_sig_alg : string
14860 Returns the name of the algorithm used to sign the certificate presented by
14861 the frontend when the incoming connection was made over an SSL/TLS transport
14862 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014863
Willy Tarreau74ca5042013-06-11 23:12:07 +020014864ssl_f_version : integer
14865 Returns the version of the certificate presented by the frontend when the
14866 incoming connection was made over an SSL/TLS transport layer.
14867
14868ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014869 Returns true when the front connection was made via an SSL/TLS transport
14870 layer and is locally deciphered. This means it has matched a socket declared
14871 with a "bind" line having the "ssl" option.
14872
Willy Tarreau74ca5042013-06-11 23:12:07 +020014873 Example :
14874 # This passes "X-Proto: https" to servers when client connects over SSL
14875 listen http-https
14876 bind :80
14877 bind :443 ssl crt /etc/haproxy.pem
14878 http-request add-header X-Proto https if { ssl_fc }
14879
14880ssl_fc_alg_keysize : integer
14881 Returns the symmetric cipher key size supported in bits when the incoming
14882 connection was made over an SSL/TLS transport layer.
14883
14884ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014885 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014886 incoming connection made via a TLS transport layer and locally deciphered by
14887 haproxy. The result is a string containing the protocol name advertised by
14888 the client. The SSL library must have been built with support for TLS
14889 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
14890 not advertised unless the "alpn" keyword on the "bind" line specifies a
14891 protocol list. Also, nothing forces the client to pick a protocol from this
14892 list, any other one may be requested. The TLS ALPN extension is meant to
14893 replace the TLS NPN extension. See also "ssl_fc_npn".
14894
Willy Tarreau74ca5042013-06-11 23:12:07 +020014895ssl_fc_cipher : string
14896 Returns the name of the used cipher when the incoming connection was made
14897 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020014898
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014899ssl_fc_cipherlist_bin : binary
14900 Returns the binary form of the client hello cipher list. The maximum returned
14901 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014902 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014903
14904ssl_fc_cipherlist_hex : string
14905 Returns the binary form of the client hello cipher list encoded as
14906 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014907 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014908
14909ssl_fc_cipherlist_str : string
14910 Returns the decoded text form of the client hello cipher list. The maximum
14911 number of ciphers returned is according with the value of
14912 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010014913 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020014914 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014915
14916ssl_fc_cipherlist_xxh : integer
14917 Returns a xxh64 of the cipher list. This hash can be return only is the value
14918 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014919 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014920
Willy Tarreau74ca5042013-06-11 23:12:07 +020014921ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014922 Returns true if a client certificate is present in an incoming connection over
14923 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010014924 Note: on SSL session resumption with Session ID or TLS ticket, client
14925 certificate is not present in the current connection but may be retrieved
14926 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
14927 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014928
Olivier Houchardccaa7de2017-10-02 11:51:03 +020014929ssl_fc_has_early : boolean
14930 Returns true if early data were sent, and the handshake didn't happen yet. As
14931 it has security implications, it is useful to be able to refuse those, or
14932 wait until the handshake happened.
14933
Willy Tarreau74ca5042013-06-11 23:12:07 +020014934ssl_fc_has_sni : boolean
14935 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020014936 in an incoming connection was made over an SSL/TLS transport layer. Returns
14937 true when the incoming connection presents a TLS SNI field. This requires
14938 that the SSL library is build with support for TLS extensions enabled (check
14939 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020014940
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020014941ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020014942 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010014943 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
14944 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020014945
Willy Tarreau74ca5042013-06-11 23:12:07 +020014946ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014947 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020014948 made via a TLS transport layer and locally deciphered by haproxy. The result
14949 is a string containing the protocol name advertised by the client. The SSL
14950 library must have been built with support for TLS extensions enabled (check
14951 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
14952 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
14953 forces the client to pick a protocol from this list, any other one may be
14954 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020014955
Willy Tarreau74ca5042013-06-11 23:12:07 +020014956ssl_fc_protocol : string
14957 Returns the name of the used protocol when the incoming connection was made
14958 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020014959
Emeric Brunb73a9b02014-04-30 18:49:19 +020014960ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040014961 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014962 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14963 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040014964
Willy Tarreau74ca5042013-06-11 23:12:07 +020014965ssl_fc_session_id : binary
14966 Returns the SSL ID of the front connection when the incoming connection was
14967 made over an SSL/TLS transport layer. It is useful to stick a given client to
14968 a server. It is important to note that some browsers refresh their session ID
14969 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020014970
Patrick Hemmere0275472018-04-28 19:15:51 -040014971ssl_fc_session_key : binary
14972 Returns the SSL session master key of the front connection when the incoming
14973 connection was made over an SSL/TLS transport layer. It is useful to decrypt
14974 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
14975 BoringSSL.
14976
14977
Willy Tarreau74ca5042013-06-11 23:12:07 +020014978ssl_fc_sni : string
14979 This extracts the Server Name Indication TLS extension (SNI) field from an
14980 incoming connection made via an SSL/TLS transport layer and locally
14981 deciphered by haproxy. The result (when present) typically is a string
14982 matching the HTTPS host name (253 chars or less). The SSL library must have
14983 been built with support for TLS extensions enabled (check haproxy -vv).
14984
14985 This fetch is different from "req_ssl_sni" above in that it applies to the
14986 connection being deciphered by haproxy and not to SSL contents being blindly
14987 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020014988 requires that the SSL library is build with support for TLS extensions
14989 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020014990
Willy Tarreau74ca5042013-06-11 23:12:07 +020014991 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020014992 ssl_fc_sni_end : suffix match
14993 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020014994
Willy Tarreau74ca5042013-06-11 23:12:07 +020014995ssl_fc_use_keysize : integer
14996 Returns the symmetric cipher key size used in bits when the incoming
14997 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014998
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014999
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200150007.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015001------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015002
Willy Tarreau74ca5042013-06-11 23:12:07 +020015003Fetching samples from buffer contents is a bit different from the previous
15004sample fetches above because the sampled data are ephemeral. These data can
15005only be used when they're available and will be lost when they're forwarded.
15006For this reason, samples fetched from buffer contents during a request cannot
15007be used in a response for example. Even while the data are being fetched, they
15008can change. Sometimes it is necessary to set some delays or combine multiple
15009sample fetch methods to ensure that the expected data are complete and usable,
15010for example through TCP request content inspection. Please see the "tcp-request
15011content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015012
Willy Tarreau74ca5042013-06-11 23:12:07 +020015013payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015014 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015015 "stick on", "stick match"), and for "res.payload" when used in the context of
15016 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015017
Willy Tarreau74ca5042013-06-11 23:12:07 +020015018payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15019 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015020 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015021 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015022
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015023req.hdrs : string
15024 Returns the current request headers as string including the last empty line
15025 separating headers from the request body. The last empty line can be used to
15026 detect a truncated header block. This sample fetch is useful for some SPOE
15027 headers analyzers and for advanced logging.
15028
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015029req.hdrs_bin : binary
15030 Returns the current request headers contained in preparsed binary form. This
15031 is useful for offloading some processing with SPOE. Each string is described
15032 by a length followed by the number of bytes indicated in the length. The
15033 length is represented using the variable integer encoding detailed in the
15034 SPOE documentation. The end of the list is marked by a couple of empty header
15035 names and values (length of 0 for both).
15036
15037 *(<str:header-name><str:header-value>)<empty string><empty string>
15038
15039 int: refer to the SPOE documentation for the encoding
15040 str: <int:length><bytes>
15041
Willy Tarreau74ca5042013-06-11 23:12:07 +020015042req.len : integer
15043req_len : integer (deprecated)
15044 Returns an integer value corresponding to the number of bytes present in the
15045 request buffer. This is mostly used in ACL. It is important to understand
15046 that this test does not return false as long as the buffer is changing. This
15047 means that a check with equality to zero will almost always immediately match
15048 at the beginning of the session, while a test for more data will wait for
15049 that data to come in and return false only when haproxy is certain that no
15050 more data will come in. This test was designed to be used with TCP request
15051 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015052
Willy Tarreau74ca5042013-06-11 23:12:07 +020015053req.payload(<offset>,<length>) : binary
15054 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015055 in the request buffer. As a special case, if the <length> argument is zero,
15056 the the whole buffer from <offset> to the end is extracted. This can be used
15057 with ACLs in order to check for the presence of some content in a buffer at
15058 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015059
Willy Tarreau74ca5042013-06-11 23:12:07 +020015060 ACL alternatives :
15061 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015062
Willy Tarreau74ca5042013-06-11 23:12:07 +020015063req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15064 This extracts a binary block whose size is specified at <offset1> for <length>
15065 bytes, and which starts at <offset2> if specified or just after the length in
15066 the request buffer. The <offset2> parameter also supports relative offsets if
15067 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015068
Willy Tarreau74ca5042013-06-11 23:12:07 +020015069 ACL alternatives :
15070 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015071
Willy Tarreau74ca5042013-06-11 23:12:07 +020015072 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015073
Willy Tarreau74ca5042013-06-11 23:12:07 +020015074req.proto_http : boolean
15075req_proto_http : boolean (deprecated)
15076 Returns true when data in the request buffer look like HTTP and correctly
15077 parses as such. It is the same parser as the common HTTP request parser which
15078 is used so there should be no surprises. The test does not match until the
15079 request is complete, failed or timed out. This test may be used to report the
15080 protocol in TCP logs, but the biggest use is to block TCP request analysis
15081 until a complete HTTP request is present in the buffer, for example to track
15082 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015083
Willy Tarreau74ca5042013-06-11 23:12:07 +020015084 Example:
15085 # track request counts per "base" (concatenation of Host+URL)
15086 tcp-request inspect-delay 10s
15087 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015088 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015089
Willy Tarreau74ca5042013-06-11 23:12:07 +020015090req.rdp_cookie([<name>]) : string
15091rdp_cookie([<name>]) : string (deprecated)
15092 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15093 <name>, or any cookie if unspecified. The parser only checks for the first
15094 cookie, as illustrated in the RDP protocol specification. The cookie name is
15095 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15096 contain the user name of the client connecting to the server if properly
15097 configured on the client. The "MSTSHASH" cookie is often used as well for
15098 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015099
Willy Tarreau74ca5042013-06-11 23:12:07 +020015100 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15101 used and thus the distribution of clients to backend servers is not linked to
15102 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15103 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15104 distribution of clients to backend servers than the hash used by "balance
15105 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015106
Willy Tarreau74ca5042013-06-11 23:12:07 +020015107 ACL derivatives :
15108 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015109
Willy Tarreau74ca5042013-06-11 23:12:07 +020015110 Example :
15111 listen tse-farm
15112 bind 0.0.0.0:3389
15113 # wait up to 5s for an RDP cookie in the request
15114 tcp-request inspect-delay 5s
15115 tcp-request content accept if RDP_COOKIE
15116 # apply RDP cookie persistence
15117 persist rdp-cookie
15118 # Persist based on the mstshash cookie
15119 # This is only useful makes sense if
15120 # balance rdp-cookie is not used
15121 stick-table type string size 204800
15122 stick on req.rdp_cookie(mstshash)
15123 server srv1 1.1.1.1:3389
15124 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015125
Willy Tarreau74ca5042013-06-11 23:12:07 +020015126 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15127 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015128
Willy Tarreau74ca5042013-06-11 23:12:07 +020015129req.rdp_cookie_cnt([name]) : integer
15130rdp_cookie_cnt([name]) : integer (deprecated)
15131 Tries to parse the request buffer as RDP protocol, then returns an integer
15132 corresponding to the number of RDP cookies found. If an optional cookie name
15133 is passed, only cookies matching this name are considered. This is mostly
15134 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015135
Willy Tarreau74ca5042013-06-11 23:12:07 +020015136 ACL derivatives :
15137 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015138
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015139req.ssl_ec_ext : boolean
15140 Returns a boolean identifying if client sent the Supported Elliptic Curves
15141 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015142 message. This can be used to present ECC compatible clients with EC
15143 certificate and to use RSA for all others, on the same IP address. Note that
15144 this only applies to raw contents found in the request buffer and not to
15145 contents deciphered via an SSL data layer, so this will not work with "bind"
15146 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015147
Willy Tarreau74ca5042013-06-11 23:12:07 +020015148req.ssl_hello_type : integer
15149req_ssl_hello_type : integer (deprecated)
15150 Returns an integer value containing the type of the SSL hello message found
15151 in the request buffer if the buffer contains data that parse as a complete
15152 SSL (v3 or superior) client hello message. Note that this only applies to raw
15153 contents found in the request buffer and not to contents deciphered via an
15154 SSL data layer, so this will not work with "bind" lines having the "ssl"
15155 option. This is mostly used in ACL to detect presence of an SSL hello message
15156 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015157
Willy Tarreau74ca5042013-06-11 23:12:07 +020015158req.ssl_sni : string
15159req_ssl_sni : string (deprecated)
15160 Returns a string containing the value of the Server Name TLS extension sent
15161 by a client in a TLS stream passing through the request buffer if the buffer
15162 contains data that parse as a complete SSL (v3 or superior) client hello
15163 message. Note that this only applies to raw contents found in the request
15164 buffer and not to contents deciphered via an SSL data layer, so this will not
15165 work with "bind" lines having the "ssl" option. SNI normally contains the
15166 name of the host the client tries to connect to (for recent browsers). SNI is
15167 useful for allowing or denying access to certain hosts when SSL/TLS is used
15168 by the client. This test was designed to be used with TCP request content
15169 inspection. If content switching is needed, it is recommended to first wait
15170 for a complete client hello (type 1), like in the example below. See also
15171 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015172
Willy Tarreau74ca5042013-06-11 23:12:07 +020015173 ACL derivatives :
15174 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015175
Willy Tarreau74ca5042013-06-11 23:12:07 +020015176 Examples :
15177 # Wait for a client hello for at most 5 seconds
15178 tcp-request inspect-delay 5s
15179 tcp-request content accept if { req_ssl_hello_type 1 }
15180 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15181 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015182
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015183req.ssl_st_ext : integer
15184 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15185 Returns 1 if the client sent SessionTicket TLS Extension
15186 Returns 2 if the client also sent non-zero length TLS SessionTicket
15187 Note that this only applies to raw contents found in the request buffer and
15188 not to contents deciphered via an SSL data layer, so this will not work with
15189 "bind" lines having the "ssl" option. This can for example be used to detect
15190 whether the client sent a SessionTicket or not and stick it accordingly, if
15191 no SessionTicket then stick on SessionID or don't stick as there's no server
15192 side state is there when SessionTickets are in use.
15193
Willy Tarreau74ca5042013-06-11 23:12:07 +020015194req.ssl_ver : integer
15195req_ssl_ver : integer (deprecated)
15196 Returns an integer value containing the version of the SSL/TLS protocol of a
15197 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15198 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15199 composed of the major version multiplied by 65536, added to the minor
15200 version. Note that this only applies to raw contents found in the request
15201 buffer and not to contents deciphered via an SSL data layer, so this will not
15202 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015203 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015204 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015205
Willy Tarreau74ca5042013-06-11 23:12:07 +020015206 ACL derivatives :
15207 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015208
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015209res.len : integer
15210 Returns an integer value corresponding to the number of bytes present in the
15211 response buffer. This is mostly used in ACL. It is important to understand
15212 that this test does not return false as long as the buffer is changing. This
15213 means that a check with equality to zero will almost always immediately match
15214 at the beginning of the session, while a test for more data will wait for
15215 that data to come in and return false only when haproxy is certain that no
15216 more data will come in. This test was designed to be used with TCP response
15217 content inspection.
15218
Willy Tarreau74ca5042013-06-11 23:12:07 +020015219res.payload(<offset>,<length>) : binary
15220 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015221 in the response buffer. As a special case, if the <length> argument is zero,
15222 the the whole buffer from <offset> to the end is extracted. This can be used
15223 with ACLs in order to check for the presence of some content in a buffer at
15224 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015225
Willy Tarreau74ca5042013-06-11 23:12:07 +020015226res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15227 This extracts a binary block whose size is specified at <offset1> for <length>
15228 bytes, and which starts at <offset2> if specified or just after the length in
15229 the response buffer. The <offset2> parameter also supports relative offsets
15230 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015231
Willy Tarreau74ca5042013-06-11 23:12:07 +020015232 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015233
Willy Tarreau971f7b62015-09-29 14:06:59 +020015234res.ssl_hello_type : integer
15235rep_ssl_hello_type : integer (deprecated)
15236 Returns an integer value containing the type of the SSL hello message found
15237 in the response buffer if the buffer contains data that parses as a complete
15238 SSL (v3 or superior) hello message. Note that this only applies to raw
15239 contents found in the response buffer and not to contents deciphered via an
15240 SSL data layer, so this will not work with "server" lines having the "ssl"
15241 option. This is mostly used in ACL to detect presence of an SSL hello message
15242 that is supposed to contain an SSL session ID usable for stickiness.
15243
Willy Tarreau74ca5042013-06-11 23:12:07 +020015244wait_end : boolean
15245 This fetch either returns true when the inspection period is over, or does
15246 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015247 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015248 actions, such as a delayed reject for some special addresses. Since it either
15249 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015250 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015251 "WAIT_END" is always usable without prior declaration. This test was designed
15252 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015253
Willy Tarreau74ca5042013-06-11 23:12:07 +020015254 Examples :
15255 # delay every incoming request by 2 seconds
15256 tcp-request inspect-delay 2s
15257 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015258
Willy Tarreau74ca5042013-06-11 23:12:07 +020015259 # don't immediately tell bad guys they are rejected
15260 tcp-request inspect-delay 10s
15261 acl goodguys src 10.0.0.0/24
15262 acl badguys src 10.0.1.0/24
15263 tcp-request content accept if goodguys
15264 tcp-request content reject if badguys WAIT_END
15265 tcp-request content reject
15266
15267
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200152687.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015269--------------------------------------
15270
15271It is possible to fetch samples from HTTP contents, requests and responses.
15272This application layer is also called layer 7. It is only possible to fetch the
15273data in this section when a full HTTP request or response has been parsed from
15274its respective request or response buffer. This is always the case with all
15275HTTP specific rules and for sections running with "mode http". When using TCP
15276content inspection, it may be necessary to support an inspection delay in order
15277to let the request or response come in first. These fetches may require a bit
15278more CPU resources than the layer 4 ones, but not much since the request and
15279response are indexed.
15280
15281base : string
15282 This returns the concatenation of the first Host header and the path part of
15283 the request, which starts at the first slash and ends before the question
15284 mark. It can be useful in virtual hosted environments to detect URL abuses as
15285 well as to improve shared caches efficiency. Using this with a limited size
15286 stick table also allows one to collect statistics about most commonly
15287 requested objects by host/path. With ACLs it can allow simple content
15288 switching rules involving the host and the path at the same time, such as
15289 "www.example.com/favicon.ico". See also "path" and "uri".
15290
15291 ACL derivatives :
15292 base : exact string match
15293 base_beg : prefix match
15294 base_dir : subdir match
15295 base_dom : domain match
15296 base_end : suffix match
15297 base_len : length match
15298 base_reg : regex match
15299 base_sub : substring match
15300
15301base32 : integer
15302 This returns a 32-bit hash of the value returned by the "base" fetch method
15303 above. This is useful to track per-URL activity on high traffic sites without
15304 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015305 memory. The output type is an unsigned integer. The hash function used is
15306 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15307 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015308
15309base32+src : binary
15310 This returns the concatenation of the base32 fetch above and the src fetch
15311 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15312 depending on the source address family. This can be used to track per-IP,
15313 per-URL counters.
15314
William Lallemand65ad6e12014-01-31 15:08:02 +010015315capture.req.hdr(<idx>) : string
15316 This extracts the content of the header captured by the "capture request
15317 header", idx is the position of the capture keyword in the configuration.
15318 The first entry is an index of 0. See also: "capture request header".
15319
15320capture.req.method : string
15321 This extracts the METHOD of an HTTP request. It can be used in both request
15322 and response. Unlike "method", it can be used in both request and response
15323 because it's allocated.
15324
15325capture.req.uri : string
15326 This extracts the request's URI, which starts at the first slash and ends
15327 before the first space in the request (without the host part). Unlike "path"
15328 and "url", it can be used in both request and response because it's
15329 allocated.
15330
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015331capture.req.ver : string
15332 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15333 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15334 logs because it relies on a persistent flag.
15335
William Lallemand65ad6e12014-01-31 15:08:02 +010015336capture.res.hdr(<idx>) : string
15337 This extracts the content of the header captured by the "capture response
15338 header", idx is the position of the capture keyword in the configuration.
15339 The first entry is an index of 0.
15340 See also: "capture response header"
15341
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015342capture.res.ver : string
15343 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15344 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15345 persistent flag.
15346
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015347req.body : binary
15348 This returns the HTTP request's available body as a block of data. It
15349 requires that the request body has been buffered made available using
15350 "option http-buffer-request". In case of chunked-encoded body, currently only
15351 the first chunk is analyzed.
15352
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015353req.body_param([<name>) : string
15354 This fetch assumes that the body of the POST request is url-encoded. The user
15355 can check if the "content-type" contains the value
15356 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15357 parameter <name> in the body, which ends before '&'. The parameter name is
15358 case-sensitive. If no name is given, any parameter will match, and the first
15359 one will be returned. The result is a string corresponding to the value of the
15360 parameter <name> as presented in the request body (no URL decoding is
15361 performed). Note that the ACL version of this fetch iterates over multiple
15362 parameters and will iteratively report all parameters values if no name is
15363 given.
15364
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015365req.body_len : integer
15366 This returns the length of the HTTP request's available body in bytes. It may
15367 be lower than the advertised length if the body is larger than the buffer. It
15368 requires that the request body has been buffered made available using
15369 "option http-buffer-request".
15370
15371req.body_size : integer
15372 This returns the advertised length of the HTTP request's body in bytes. It
15373 will represent the advertised Content-Length header, or the size of the first
15374 chunk in case of chunked encoding. In order to parse the chunks, it requires
15375 that the request body has been buffered made available using
15376 "option http-buffer-request".
15377
Willy Tarreau74ca5042013-06-11 23:12:07 +020015378req.cook([<name>]) : string
15379cook([<name>]) : string (deprecated)
15380 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15381 header line from the request, and returns its value as string. If no name is
15382 specified, the first cookie value is returned. When used with ACLs, all
15383 matching cookies are evaluated. Spaces around the name and the value are
15384 ignored as requested by the Cookie header specification (RFC6265). The cookie
15385 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15386 well return an empty value if it is present. Use the "found" match to detect
15387 presence. Use the res.cook() variant for response cookies sent by the server.
15388
15389 ACL derivatives :
15390 cook([<name>]) : exact string match
15391 cook_beg([<name>]) : prefix match
15392 cook_dir([<name>]) : subdir match
15393 cook_dom([<name>]) : domain match
15394 cook_end([<name>]) : suffix match
15395 cook_len([<name>]) : length match
15396 cook_reg([<name>]) : regex match
15397 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015398
Willy Tarreau74ca5042013-06-11 23:12:07 +020015399req.cook_cnt([<name>]) : integer
15400cook_cnt([<name>]) : integer (deprecated)
15401 Returns an integer value representing the number of occurrences of the cookie
15402 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015403
Willy Tarreau74ca5042013-06-11 23:12:07 +020015404req.cook_val([<name>]) : integer
15405cook_val([<name>]) : integer (deprecated)
15406 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15407 header line from the request, and converts its value to an integer which is
15408 returned. If no name is specified, the first cookie value is returned. When
15409 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015410
Willy Tarreau74ca5042013-06-11 23:12:07 +020015411cookie([<name>]) : string (deprecated)
15412 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15413 header line from the request, or a "Set-Cookie" header from the response, and
15414 returns its value as a string. A typical use is to get multiple clients
15415 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015416 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015417 multi-peer synchronization and state keeping across restarts. If no name is
15418 specified, the first cookie value is returned. This fetch should not be used
15419 anymore and should be replaced by req.cook() or res.cook() instead as it
15420 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015421
Willy Tarreau74ca5042013-06-11 23:12:07 +020015422hdr([<name>[,<occ>]]) : string
15423 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15424 used on responses. Please refer to these respective fetches for more details.
15425 In case of doubt about the fetch direction, please use the explicit ones.
15426 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015427 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015428
Willy Tarreau74ca5042013-06-11 23:12:07 +020015429req.fhdr(<name>[,<occ>]) : string
15430 This extracts the last occurrence of header <name> in an HTTP request. When
15431 used from an ACL, all occurrences are iterated over until a match is found.
15432 Optionally, a specific occurrence might be specified as a position number.
15433 Positive values indicate a position from the first occurrence, with 1 being
15434 the first one. Negative values indicate positions relative to the last one,
15435 with -1 being the last one. It differs from req.hdr() in that any commas
15436 present in the value are returned and are not used as delimiters. This is
15437 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015438
Willy Tarreau74ca5042013-06-11 23:12:07 +020015439req.fhdr_cnt([<name>]) : integer
15440 Returns an integer value representing the number of occurrences of request
15441 header field name <name>, or the total number of header fields if <name> is
15442 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15443 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015444
Willy Tarreau74ca5042013-06-11 23:12:07 +020015445req.hdr([<name>[,<occ>]]) : string
15446 This extracts the last occurrence of header <name> in an HTTP request. When
15447 used from an ACL, all occurrences are iterated over until a match is found.
15448 Optionally, a specific occurrence might be specified as a position number.
15449 Positive values indicate a position from the first occurrence, with 1 being
15450 the first one. Negative values indicate positions relative to the last one,
15451 with -1 being the last one. A typical use is with the X-Forwarded-For header
15452 once converted to IP, associated with an IP stick-table. The function
15453 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015454 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015455 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015456 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015457
Willy Tarreau74ca5042013-06-11 23:12:07 +020015458 ACL derivatives :
15459 hdr([<name>[,<occ>]]) : exact string match
15460 hdr_beg([<name>[,<occ>]]) : prefix match
15461 hdr_dir([<name>[,<occ>]]) : subdir match
15462 hdr_dom([<name>[,<occ>]]) : domain match
15463 hdr_end([<name>[,<occ>]]) : suffix match
15464 hdr_len([<name>[,<occ>]]) : length match
15465 hdr_reg([<name>[,<occ>]]) : regex match
15466 hdr_sub([<name>[,<occ>]]) : substring match
15467
15468req.hdr_cnt([<name>]) : integer
15469hdr_cnt([<header>]) : integer (deprecated)
15470 Returns an integer value representing the number of occurrences of request
15471 header field name <name>, or the total number of header field values if
15472 <name> is not specified. It is important to remember that one header line may
15473 count as several headers if it has several values. The function considers any
15474 comma as a delimiter for distinct values. If full-line headers are desired
15475 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15476 detect presence, absence or abuse of a specific header, as well as to block
15477 request smuggling attacks by rejecting requests which contain more than one
15478 of certain headers. See "req.hdr" for more information on header matching.
15479
15480req.hdr_ip([<name>[,<occ>]]) : ip
15481hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15482 This extracts the last occurrence of header <name> in an HTTP request,
15483 converts it to an IPv4 or IPv6 address and returns this address. When used
15484 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15485 of every header is checked. Optionally, a specific occurrence might be
15486 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015487 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015488 positions relative to the last one, with -1 being the last one. A typical use
15489 is with the X-Forwarded-For and X-Client-IP headers.
15490
15491req.hdr_val([<name>[,<occ>]]) : integer
15492hdr_val([<name>[,<occ>]]) : integer (deprecated)
15493 This extracts the last occurrence of header <name> in an HTTP request, and
15494 converts it to an integer value. When used with ACLs, all occurrences are
15495 checked, and if <name> is omitted, every value of every header is checked.
15496 Optionally, a specific occurrence might be specified as a position number.
15497 Positive values indicate a position from the first occurrence, with 1 being
15498 the first one. Negative values indicate positions relative to the last one,
15499 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15500
15501http_auth(<userlist>) : boolean
15502 Returns a boolean indicating whether the authentication data received from
15503 the client match a username & password stored in the specified userlist. This
15504 fetch function is not really useful outside of ACLs. Currently only http
15505 basic auth is supported.
15506
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015507http_auth_group(<userlist>) : string
15508 Returns a string corresponding to the user name found in the authentication
15509 data received from the client if both the user name and password are valid
15510 according to the specified userlist. The main purpose is to use it in ACLs
15511 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015512 This fetch function is not really useful outside of ACLs. Currently only http
15513 basic auth is supported.
15514
15515 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015516 http_auth_group(<userlist>) : group ...
15517 Returns true when the user extracted from the request and whose password is
15518 valid according to the specified userlist belongs to at least one of the
15519 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015520
15521http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015522 Returns true when the request being processed is the first one of the
15523 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015524 from some requests when a request is not the first one, or to help grouping
15525 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015526
Willy Tarreau74ca5042013-06-11 23:12:07 +020015527method : integer + string
15528 Returns an integer value corresponding to the method in the HTTP request. For
15529 example, "GET" equals 1 (check sources to establish the matching). Value 9
15530 means "other method" and may be converted to a string extracted from the
15531 stream. This should not be used directly as a sample, this is only meant to
15532 be used from ACLs, which transparently convert methods from patterns to these
15533 integer + string values. Some predefined ACL already check for most common
15534 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015535
Willy Tarreau74ca5042013-06-11 23:12:07 +020015536 ACL derivatives :
15537 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015538
Willy Tarreau74ca5042013-06-11 23:12:07 +020015539 Example :
15540 # only accept GET and HEAD requests
15541 acl valid_method method GET HEAD
15542 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015543
Willy Tarreau74ca5042013-06-11 23:12:07 +020015544path : string
15545 This extracts the request's URL path, which starts at the first slash and
15546 ends before the question mark (without the host part). A typical use is with
15547 prefetch-capable caches, and with portals which need to aggregate multiple
15548 information from databases and keep them in caches. Note that with outgoing
15549 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015550 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015551 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015552
Willy Tarreau74ca5042013-06-11 23:12:07 +020015553 ACL derivatives :
15554 path : exact string match
15555 path_beg : prefix match
15556 path_dir : subdir match
15557 path_dom : domain match
15558 path_end : suffix match
15559 path_len : length match
15560 path_reg : regex match
15561 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015562
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015563query : string
15564 This extracts the request's query string, which starts after the first
15565 question mark. If no question mark is present, this fetch returns nothing. If
15566 a question mark is present but nothing follows, it returns an empty string.
15567 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015568 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015569 which stops before the question mark.
15570
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015571req.hdr_names([<delim>]) : string
15572 This builds a string made from the concatenation of all header names as they
15573 appear in the request when the rule is evaluated. The default delimiter is
15574 the comma (',') but it may be overridden as an optional argument <delim>. In
15575 this case, only the first character of <delim> is considered.
15576
Willy Tarreau74ca5042013-06-11 23:12:07 +020015577req.ver : string
15578req_ver : string (deprecated)
15579 Returns the version string from the HTTP request, for example "1.1". This can
15580 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15581 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015582
Willy Tarreau74ca5042013-06-11 23:12:07 +020015583 ACL derivatives :
15584 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015585
Willy Tarreau74ca5042013-06-11 23:12:07 +020015586res.comp : boolean
15587 Returns the boolean "true" value if the response has been compressed by
15588 HAProxy, otherwise returns boolean "false". This may be used to add
15589 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015590
Willy Tarreau74ca5042013-06-11 23:12:07 +020015591res.comp_algo : string
15592 Returns a string containing the name of the algorithm used if the response
15593 was compressed by HAProxy, for example : "deflate". This may be used to add
15594 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015595
Willy Tarreau74ca5042013-06-11 23:12:07 +020015596res.cook([<name>]) : string
15597scook([<name>]) : string (deprecated)
15598 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15599 header line from the response, and returns its value as string. If no name is
15600 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015601
Willy Tarreau74ca5042013-06-11 23:12:07 +020015602 ACL derivatives :
15603 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015604
Willy Tarreau74ca5042013-06-11 23:12:07 +020015605res.cook_cnt([<name>]) : integer
15606scook_cnt([<name>]) : integer (deprecated)
15607 Returns an integer value representing the number of occurrences of the cookie
15608 <name> in the response, or all cookies if <name> is not specified. This is
15609 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015610
Willy Tarreau74ca5042013-06-11 23:12:07 +020015611res.cook_val([<name>]) : integer
15612scook_val([<name>]) : integer (deprecated)
15613 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15614 header line from the response, and converts its value to an integer which is
15615 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015616
Willy Tarreau74ca5042013-06-11 23:12:07 +020015617res.fhdr([<name>[,<occ>]]) : string
15618 This extracts the last occurrence of header <name> in an HTTP response, or of
15619 the last header if no <name> is specified. Optionally, a specific occurrence
15620 might be specified as a position number. Positive values indicate a position
15621 from the first occurrence, with 1 being the first one. Negative values
15622 indicate positions relative to the last one, with -1 being the last one. It
15623 differs from res.hdr() in that any commas present in the value are returned
15624 and are not used as delimiters. If this is not desired, the res.hdr() fetch
15625 should be used instead. This is sometimes useful with headers such as Date or
15626 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015627
Willy Tarreau74ca5042013-06-11 23:12:07 +020015628res.fhdr_cnt([<name>]) : integer
15629 Returns an integer value representing the number of occurrences of response
15630 header field name <name>, or the total number of header fields if <name> is
15631 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
15632 the number of full line headers and does not stop on commas. If this is not
15633 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015634
Willy Tarreau74ca5042013-06-11 23:12:07 +020015635res.hdr([<name>[,<occ>]]) : string
15636shdr([<name>[,<occ>]]) : string (deprecated)
15637 This extracts the last occurrence of header <name> in an HTTP response, or of
15638 the last header if no <name> is specified. Optionally, a specific occurrence
15639 might be specified as a position number. Positive values indicate a position
15640 from the first occurrence, with 1 being the first one. Negative values
15641 indicate positions relative to the last one, with -1 being the last one. This
15642 can be useful to learn some data into a stick-table. The function considers
15643 any comma as a delimiter for distinct values. If this is not desired, the
15644 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015645
Willy Tarreau74ca5042013-06-11 23:12:07 +020015646 ACL derivatives :
15647 shdr([<name>[,<occ>]]) : exact string match
15648 shdr_beg([<name>[,<occ>]]) : prefix match
15649 shdr_dir([<name>[,<occ>]]) : subdir match
15650 shdr_dom([<name>[,<occ>]]) : domain match
15651 shdr_end([<name>[,<occ>]]) : suffix match
15652 shdr_len([<name>[,<occ>]]) : length match
15653 shdr_reg([<name>[,<occ>]]) : regex match
15654 shdr_sub([<name>[,<occ>]]) : substring match
15655
15656res.hdr_cnt([<name>]) : integer
15657shdr_cnt([<name>]) : integer (deprecated)
15658 Returns an integer value representing the number of occurrences of response
15659 header field name <name>, or the total number of header fields if <name> is
15660 not specified. The function considers any comma as a delimiter for distinct
15661 values. If this is not desired, the res.fhdr_cnt() fetch should be used
15662 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015663
Willy Tarreau74ca5042013-06-11 23:12:07 +020015664res.hdr_ip([<name>[,<occ>]]) : ip
15665shdr_ip([<name>[,<occ>]]) : ip (deprecated)
15666 This extracts the last occurrence of header <name> in an HTTP response,
15667 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
15668 specific occurrence might be specified as a position number. Positive values
15669 indicate a position from the first occurrence, with 1 being the first one.
15670 Negative values indicate positions relative to the last one, with -1 being
15671 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015672
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015673res.hdr_names([<delim>]) : string
15674 This builds a string made from the concatenation of all header names as they
15675 appear in the response when the rule is evaluated. The default delimiter is
15676 the comma (',') but it may be overridden as an optional argument <delim>. In
15677 this case, only the first character of <delim> is considered.
15678
Willy Tarreau74ca5042013-06-11 23:12:07 +020015679res.hdr_val([<name>[,<occ>]]) : integer
15680shdr_val([<name>[,<occ>]]) : integer (deprecated)
15681 This extracts the last occurrence of header <name> in an HTTP response, and
15682 converts it to an integer value. Optionally, a specific occurrence might be
15683 specified as a position number. Positive values indicate a position from the
15684 first occurrence, with 1 being the first one. Negative values indicate
15685 positions relative to the last one, with -1 being the last one. This can be
15686 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015687
Willy Tarreau74ca5042013-06-11 23:12:07 +020015688res.ver : string
15689resp_ver : string (deprecated)
15690 Returns the version string from the HTTP response, for example "1.1". This
15691 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020015692
Willy Tarreau74ca5042013-06-11 23:12:07 +020015693 ACL derivatives :
15694 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015695
Willy Tarreau74ca5042013-06-11 23:12:07 +020015696set-cookie([<name>]) : string (deprecated)
15697 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15698 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020015699 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020015700 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015701
Willy Tarreau74ca5042013-06-11 23:12:07 +020015702 This fetch function is deprecated and has been superseded by the "res.cook"
15703 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015704
Willy Tarreau74ca5042013-06-11 23:12:07 +020015705status : integer
15706 Returns an integer containing the HTTP status code in the HTTP response, for
15707 example, 302. It is mostly used within ACLs and integer ranges, for example,
15708 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015709
Thierry Fournier0e00dca2016-04-07 15:47:40 +020015710unique-id : string
15711 Returns the unique-id attached to the request. The directive
15712 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
15713 fails. Note that the unique-id is usually used with HTTP requests, however this
15714 sample fetch can be used with other protocols. Obviously, if it is used with
15715 other protocols than HTTP, the unique-id-format directive must not contain
15716 HTTP parts. See: unique-id-format and unique-id-header
15717
Willy Tarreau74ca5042013-06-11 23:12:07 +020015718url : string
15719 This extracts the request's URL as presented in the request. A typical use is
15720 with prefetch-capable caches, and with portals which need to aggregate
15721 multiple information from databases and keep them in caches. With ACLs, using
15722 "path" is preferred over using "url", because clients may send a full URL as
15723 is normally done with proxies. The only real use is to match "*" which does
15724 not match in "path", and for which there is already a predefined ACL. See
15725 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015726
Willy Tarreau74ca5042013-06-11 23:12:07 +020015727 ACL derivatives :
15728 url : exact string match
15729 url_beg : prefix match
15730 url_dir : subdir match
15731 url_dom : domain match
15732 url_end : suffix match
15733 url_len : length match
15734 url_reg : regex match
15735 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015736
Willy Tarreau74ca5042013-06-11 23:12:07 +020015737url_ip : ip
15738 This extracts the IP address from the request's URL when the host part is
15739 presented as an IP address. Its use is very limited. For instance, a
15740 monitoring system might use this field as an alternative for the source IP in
15741 order to test what path a given source address would follow, or to force an
15742 entry in a table for a given source address. With ACLs it can be used to
15743 restrict access to certain systems through a proxy, for example when combined
15744 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015745
Willy Tarreau74ca5042013-06-11 23:12:07 +020015746url_port : integer
15747 This extracts the port part from the request's URL. Note that if the port is
15748 not specified in the request, port 80 is assumed. With ACLs it can be used to
15749 restrict access to certain systems through a proxy, for example when combined
15750 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015751
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015752urlp([<name>[,<delim>]]) : string
15753url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015754 This extracts the first occurrence of the parameter <name> in the query
15755 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015756 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
15757 any parameter will match, and the first one will be returned. The result is
15758 a string corresponding to the value of the parameter <name> as presented in
15759 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020015760 stickiness based on a client ID, to extract an application cookie passed as a
15761 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015762 this fetch iterates over multiple parameters and will iteratively report all
15763 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015764
Willy Tarreau74ca5042013-06-11 23:12:07 +020015765 ACL derivatives :
15766 urlp(<name>[,<delim>]) : exact string match
15767 urlp_beg(<name>[,<delim>]) : prefix match
15768 urlp_dir(<name>[,<delim>]) : subdir match
15769 urlp_dom(<name>[,<delim>]) : domain match
15770 urlp_end(<name>[,<delim>]) : suffix match
15771 urlp_len(<name>[,<delim>]) : length match
15772 urlp_reg(<name>[,<delim>]) : regex match
15773 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015774
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015775
Willy Tarreau74ca5042013-06-11 23:12:07 +020015776 Example :
15777 # match http://example.com/foo?PHPSESSIONID=some_id
15778 stick on urlp(PHPSESSIONID)
15779 # match http://example.com/foo;JSESSIONID=some_id
15780 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015781
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015782urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020015783 See "urlp" above. This one extracts the URL parameter <name> in the request
15784 and converts it to an integer value. This can be used for session stickiness
15785 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020015786
Dragan Dosen0070cd52016-06-16 12:19:49 +020015787url32 : integer
15788 This returns a 32-bit hash of the value obtained by concatenating the first
15789 Host header and the whole URL including parameters (not only the path part of
15790 the request, as in the "base32" fetch above). This is useful to track per-URL
15791 activity. A shorter hash is stored, saving a lot of memory. The output type
15792 is an unsigned integer.
15793
15794url32+src : binary
15795 This returns the concatenation of the "url32" fetch and the "src" fetch. The
15796 resulting type is of type binary, with a size of 8 or 20 bytes depending on
15797 the source address family. This can be used to track per-IP, per-URL counters.
15798
Willy Tarreau198a7442008-01-17 12:05:32 +010015799
Willy Tarreau74ca5042013-06-11 23:12:07 +0200158007.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015801---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015802
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015803Some predefined ACLs are hard-coded so that they do not have to be declared in
15804every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020015805order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010015806
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015807ACL name Equivalent to Usage
15808---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015809FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020015810HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015811HTTP_1.0 req_ver 1.0 match HTTP version 1.0
15812HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010015813HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
15814HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
15815HTTP_URL_SLASH url_beg / match URL beginning with "/"
15816HTTP_URL_STAR url * match URL equal to "*"
15817LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015818METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015819METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015820METH_GET method GET HEAD match HTTP GET or HEAD method
15821METH_HEAD method HEAD match HTTP HEAD method
15822METH_OPTIONS method OPTIONS match HTTP OPTIONS method
15823METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015824METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015825METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020015826RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015827REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015828TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015829WAIT_END wait_end wait for end of content analysis
15830---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015831
Willy Tarreaub937b7e2010-01-12 15:27:54 +010015832
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158338. Logging
15834----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010015835
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015836One of HAProxy's strong points certainly lies is its precise logs. It probably
15837provides the finest level of information available for such a product, which is
15838very important for troubleshooting complex environments. Standard information
15839provided in logs include client ports, TCP/HTTP state timers, precise session
15840state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010015841to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015842headers.
15843
15844In order to improve administrators reactivity, it offers a great transparency
15845about encountered problems, both internal and external, and it is possible to
15846send logs to different sources at the same time with different level filters :
15847
15848 - global process-level logs (system errors, start/stop, etc..)
15849 - per-instance system and internal errors (lack of resource, bugs, ...)
15850 - per-instance external troubles (servers up/down, max connections)
15851 - per-instance activity (client connections), either at the establishment or
15852 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010015853 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060015854 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015855
15856The ability to distribute different levels of logs to different log servers
15857allow several production teams to interact and to fix their problems as soon
15858as possible. For example, the system team might monitor system-wide errors,
15859while the application team might be monitoring the up/down for their servers in
15860real time, and the security team might analyze the activity logs with one hour
15861delay.
15862
15863
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158648.1. Log levels
15865---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015866
Simon Hormandf791f52011-05-29 15:01:10 +090015867TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015868source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090015869HTTP request, HTTP return code, number of bytes transmitted, conditions
15870in which the session ended, and even exchanged cookies values. For example
15871track a particular user's problems. All messages may be sent to up to two
15872syslog servers. Check the "log" keyword in section 4.2 for more information
15873about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015874
15875
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158768.2. Log formats
15877----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015878
William Lallemand48940402012-01-30 16:47:22 +010015879HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090015880and will be detailed in the following sections. A few of them may vary
15881slightly with the configuration, due to indicators specific to certain
15882options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015883
15884 - the default format, which is very basic and very rarely used. It only
15885 provides very basic information about the incoming connection at the moment
15886 it is accepted : source IP:port, destination IP:port, and frontend-name.
15887 This mode will eventually disappear so it will not be described to great
15888 extents.
15889
15890 - the TCP format, which is more advanced. This format is enabled when "option
15891 tcplog" is set on the frontend. HAProxy will then usually wait for the
15892 connection to terminate before logging. This format provides much richer
15893 information, such as timers, connection counts, queue size, etc... This
15894 format is recommended for pure TCP proxies.
15895
15896 - the HTTP format, which is the most advanced for HTTP proxying. This format
15897 is enabled when "option httplog" is set on the frontend. It provides the
15898 same information as the TCP format with some HTTP-specific fields such as
15899 the request, the status code, and captures of headers and cookies. This
15900 format is recommended for HTTP proxies.
15901
Emeric Brun3a058f32009-06-30 18:26:00 +020015902 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
15903 fields arranged in the same order as the CLF format. In this mode, all
15904 timers, captures, flags, etc... appear one per field after the end of the
15905 common fields, in the same order they appear in the standard HTTP format.
15906
William Lallemand48940402012-01-30 16:47:22 +010015907 - the custom log format, allows you to make your own log line.
15908
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015909Next sections will go deeper into details for each of these formats. Format
15910specification will be performed on a "field" basis. Unless stated otherwise, a
15911field is a portion of text delimited by any number of spaces. Since syslog
15912servers are susceptible of inserting fields at the beginning of a line, it is
15913always assumed that the first field is the one containing the process name and
15914identifier.
15915
15916Note : Since log lines may be quite long, the log examples in sections below
15917 might be broken into multiple lines. The example log lines will be
15918 prefixed with 3 closing angle brackets ('>>>') and each time a log is
15919 broken into multiple lines, each non-final line will end with a
15920 backslash ('\') and the next line will start indented by two characters.
15921
15922
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200159238.2.1. Default log format
15924-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015925
15926This format is used when no specific option is set. The log is emitted as soon
15927as the connection is accepted. One should note that this currently is the only
15928format which logs the request's destination IP and ports.
15929
15930 Example :
15931 listen www
15932 mode http
15933 log global
15934 server srv1 127.0.0.1:8000
15935
15936 >>> Feb 6 12:12:09 localhost \
15937 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
15938 (www/HTTP)
15939
15940 Field Format Extract from the example above
15941 1 process_name '[' pid ']:' haproxy[14385]:
15942 2 'Connect from' Connect from
15943 3 source_ip ':' source_port 10.0.1.2:33312
15944 4 'to' to
15945 5 destination_ip ':' destination_port 10.0.3.31:8012
15946 6 '(' frontend_name '/' mode ')' (www/HTTP)
15947
15948Detailed fields description :
15949 - "source_ip" is the IP address of the client which initiated the connection.
15950 - "source_port" is the TCP port of the client which initiated the connection.
15951 - "destination_ip" is the IP address the client connected to.
15952 - "destination_port" is the TCP port the client connected to.
15953 - "frontend_name" is the name of the frontend (or listener) which received
15954 and processed the connection.
15955 - "mode is the mode the frontend is operating (TCP or HTTP).
15956
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015957In case of a UNIX socket, the source and destination addresses are marked as
15958"unix:" and the ports reflect the internal ID of the socket which accepted the
15959connection (the same ID as reported in the stats).
15960
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015961It is advised not to use this deprecated format for newer installations as it
15962will eventually disappear.
15963
15964
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200159658.2.2. TCP log format
15966---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015967
15968The TCP format is used when "option tcplog" is specified in the frontend, and
15969is the recommended format for pure TCP proxies. It provides a lot of precious
15970information for troubleshooting. Since this format includes timers and byte
15971counts, the log is normally emitted at the end of the session. It can be
15972emitted earlier if "option logasap" is specified, which makes sense in most
15973environments with long sessions such as remote terminals. Sessions which match
15974the "monitor" rules are never logged. It is also possible not to emit logs for
15975sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015976specifying "option dontlognull" in the frontend. Successful connections will
15977not be logged if "option dontlog-normal" is specified in the frontend. A few
15978fields may slightly vary depending on some configuration options, those are
15979marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015980
15981 Example :
15982 frontend fnt
15983 mode tcp
15984 option tcplog
15985 log global
15986 default_backend bck
15987
15988 backend bck
15989 server srv1 127.0.0.1:8000
15990
15991 >>> Feb 6 12:12:56 localhost \
15992 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
15993 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
15994
15995 Field Format Extract from the example above
15996 1 process_name '[' pid ']:' haproxy[14387]:
15997 2 client_ip ':' client_port 10.0.1.2:33313
15998 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
15999 4 frontend_name fnt
16000 5 backend_name '/' server_name bck/srv1
16001 6 Tw '/' Tc '/' Tt* 0/0/5007
16002 7 bytes_read* 212
16003 8 termination_state --
16004 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16005 10 srv_queue '/' backend_queue 0/0
16006
16007Detailed fields description :
16008 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016009 connection to haproxy. If the connection was accepted on a UNIX socket
16010 instead, the IP address would be replaced with the word "unix". Note that
16011 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016012 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016013 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016014 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016015
16016 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016017 If the connection was accepted on a UNIX socket instead, the port would be
16018 replaced with the ID of the accepting socket, which is also reported in the
16019 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016020
16021 - "accept_date" is the exact date when the connection was received by haproxy
16022 (which might be very slightly different from the date observed on the
16023 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016024 the same date which may appear in any upstream firewall's log. When used in
16025 HTTP mode, the accept_date field will be reset to the first moment the
16026 connection is ready to receive a new request (end of previous response for
16027 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016028
16029 - "frontend_name" is the name of the frontend (or listener) which received
16030 and processed the connection.
16031
16032 - "backend_name" is the name of the backend (or listener) which was selected
16033 to manage the connection to the server. This will be the same as the
16034 frontend if no switching rule has been applied, which is common for TCP
16035 applications.
16036
16037 - "server_name" is the name of the last server to which the connection was
16038 sent, which might differ from the first one if there were connection errors
16039 and a redispatch occurred. Note that this server belongs to the backend
16040 which processed the request. If the connection was aborted before reaching
16041 a server, "<NOSRV>" is indicated instead of a server name.
16042
16043 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16044 It can be "-1" if the connection was aborted before reaching the queue.
16045 See "Timers" below for more details.
16046
16047 - "Tc" is the total time in milliseconds spent waiting for the connection to
16048 establish to the final server, including retries. It can be "-1" if the
16049 connection was aborted before a connection could be established. See
16050 "Timers" below for more details.
16051
16052 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016053 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016054 "option logasap" was specified, then the time counting stops at the moment
16055 the log is emitted. In this case, a '+' sign is prepended before the value,
16056 indicating that the final one will be larger. See "Timers" below for more
16057 details.
16058
16059 - "bytes_read" is the total number of bytes transmitted from the server to
16060 the client when the log is emitted. If "option logasap" is specified, the
16061 this value will be prefixed with a '+' sign indicating that the final one
16062 may be larger. Please note that this value is a 64-bit counter, so log
16063 analysis tools must be able to handle it without overflowing.
16064
16065 - "termination_state" is the condition the session was in when the session
16066 ended. This indicates the session state, which side caused the end of
16067 session to happen, and for what reason (timeout, error, ...). The normal
16068 flags should be "--", indicating the session was closed by either end with
16069 no data remaining in buffers. See below "Session state at disconnection"
16070 for more details.
16071
16072 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016073 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016074 limits have been reached. For instance, if actconn is close to 512 when
16075 multiple connection errors occur, chances are high that the system limits
16076 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016077 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016078
16079 - "feconn" is the total number of concurrent connections on the frontend when
16080 the session was logged. It is useful to estimate the amount of resource
16081 required to sustain high loads, and to detect when the frontend's "maxconn"
16082 has been reached. Most often when this value increases by huge jumps, it is
16083 because there is congestion on the backend servers, but sometimes it can be
16084 caused by a denial of service attack.
16085
16086 - "beconn" is the total number of concurrent connections handled by the
16087 backend when the session was logged. It includes the total number of
16088 concurrent connections active on servers as well as the number of
16089 connections pending in queues. It is useful to estimate the amount of
16090 additional servers needed to support high loads for a given application.
16091 Most often when this value increases by huge jumps, it is because there is
16092 congestion on the backend servers, but sometimes it can be caused by a
16093 denial of service attack.
16094
16095 - "srv_conn" is the total number of concurrent connections still active on
16096 the server when the session was logged. It can never exceed the server's
16097 configured "maxconn" parameter. If this value is very often close or equal
16098 to the server's "maxconn", it means that traffic regulation is involved a
16099 lot, meaning that either the server's maxconn value is too low, or that
16100 there aren't enough servers to process the load with an optimal response
16101 time. When only one of the server's "srv_conn" is high, it usually means
16102 that this server has some trouble causing the connections to take longer to
16103 be processed than on other servers.
16104
16105 - "retries" is the number of connection retries experienced by this session
16106 when trying to connect to the server. It must normally be zero, unless a
16107 server is being stopped at the same moment the connection was attempted.
16108 Frequent retries generally indicate either a network problem between
16109 haproxy and the server, or a misconfigured system backlog on the server
16110 preventing new connections from being queued. This field may optionally be
16111 prefixed with a '+' sign, indicating that the session has experienced a
16112 redispatch after the maximal retry count has been reached on the initial
16113 server. In this case, the server name appearing in the log is the one the
16114 connection was redispatched to, and not the first one, though both may
16115 sometimes be the same in case of hashing for instance. So as a general rule
16116 of thumb, when a '+' is present in front of the retry count, this count
16117 should not be attributed to the logged server.
16118
16119 - "srv_queue" is the total number of requests which were processed before
16120 this one in the server queue. It is zero when the request has not gone
16121 through the server queue. It makes it possible to estimate the approximate
16122 server's response time by dividing the time spent in queue by the number of
16123 requests in the queue. It is worth noting that if a session experiences a
16124 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016125 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016126 backend queue unless a redispatch occurs.
16127
16128 - "backend_queue" is the total number of requests which were processed before
16129 this one in the backend's global queue. It is zero when the request has not
16130 gone through the global queue. It makes it possible to estimate the average
16131 queue length, which easily translates into a number of missing servers when
16132 divided by a server's "maxconn" parameter. It is worth noting that if a
16133 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016134 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016135 through both the server queue and the backend queue unless a redispatch
16136 occurs.
16137
16138
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161398.2.3. HTTP log format
16140----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016141
16142The HTTP format is the most complete and the best suited for HTTP proxies. It
16143is enabled by when "option httplog" is specified in the frontend. It provides
16144the same level of information as the TCP format with additional features which
16145are specific to the HTTP protocol. Just like the TCP format, the log is usually
16146emitted at the end of the session, unless "option logasap" is specified, which
16147generally only makes sense for download sites. A session which matches the
16148"monitor" rules will never logged. It is also possible not to log sessions for
16149which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016150frontend. Successful connections will not be logged if "option dontlog-normal"
16151is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016152
16153Most fields are shared with the TCP log, some being different. A few fields may
16154slightly vary depending on some configuration options. Those ones are marked
16155with a star ('*') after the field name below.
16156
16157 Example :
16158 frontend http-in
16159 mode http
16160 option httplog
16161 log global
16162 default_backend bck
16163
16164 backend static
16165 server srv1 127.0.0.1:8000
16166
16167 >>> Feb 6 12:14:14 localhost \
16168 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16169 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 +010016170 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016171
16172 Field Format Extract from the example above
16173 1 process_name '[' pid ']:' haproxy[14389]:
16174 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016175 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016176 4 frontend_name http-in
16177 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016178 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016179 7 status_code 200
16180 8 bytes_read* 2750
16181 9 captured_request_cookie -
16182 10 captured_response_cookie -
16183 11 termination_state ----
16184 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16185 13 srv_queue '/' backend_queue 0/0
16186 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16187 15 '{' captured_response_headers* '}' {}
16188 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016189
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016190Detailed fields description :
16191 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016192 connection to haproxy. If the connection was accepted on a UNIX socket
16193 instead, the IP address would be replaced with the word "unix". Note that
16194 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016195 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016196 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016197 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016198
16199 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016200 If the connection was accepted on a UNIX socket instead, the port would be
16201 replaced with the ID of the accepting socket, which is also reported in the
16202 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016203
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016204 - "request_date" is the exact date when the first byte of the HTTP request
16205 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016206
16207 - "frontend_name" is the name of the frontend (or listener) which received
16208 and processed the connection.
16209
16210 - "backend_name" is the name of the backend (or listener) which was selected
16211 to manage the connection to the server. This will be the same as the
16212 frontend if no switching rule has been applied.
16213
16214 - "server_name" is the name of the last server to which the connection was
16215 sent, which might differ from the first one if there were connection errors
16216 and a redispatch occurred. Note that this server belongs to the backend
16217 which processed the request. If the request was aborted before reaching a
16218 server, "<NOSRV>" is indicated instead of a server name. If the request was
16219 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16220
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016221 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16222 request from the client (not counting body) after the first byte was
16223 received. It can be "-1" if the connection was aborted before a complete
16224 request could be received or the a bad request was received. It should
16225 always be very small because a request generally fits in one single packet.
16226 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016227 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16228 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016229
16230 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16231 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016232 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016233
16234 - "Tc" is the total time in milliseconds spent waiting for the connection to
16235 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016236 request was aborted before a connection could be established. See section
16237 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016238
16239 - "Tr" is the total time in milliseconds spent waiting for the server to send
16240 a full HTTP response, not counting data. It can be "-1" if the request was
16241 aborted before a complete response could be received. It generally matches
16242 the server's processing time for the request, though it may be altered by
16243 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016244 "GET" requests generally indicate an overloaded server. See section 8.4
16245 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016246
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016247 - "Ta" is the time the request remained active in haproxy, which is the total
16248 time in milliseconds elapsed between the first byte of the request was
16249 received and the last byte of response was sent. It covers all possible
16250 processing except the handshake (see Th) and idle time (see Ti). There is
16251 one exception, if "option logasap" was specified, then the time counting
16252 stops at the moment the log is emitted. In this case, a '+' sign is
16253 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016254 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016255
16256 - "status_code" is the HTTP status code returned to the client. This status
16257 is generally set by the server, but it might also be set by haproxy when
16258 the server cannot be reached or when its response is blocked by haproxy.
16259
16260 - "bytes_read" is the total number of bytes transmitted to the client when
16261 the log is emitted. This does include HTTP headers. If "option logasap" is
16262 specified, the this value will be prefixed with a '+' sign indicating that
16263 the final one may be larger. Please note that this value is a 64-bit
16264 counter, so log analysis tools must be able to handle it without
16265 overflowing.
16266
16267 - "captured_request_cookie" is an optional "name=value" entry indicating that
16268 the client had this cookie in the request. The cookie name and its maximum
16269 length are defined by the "capture cookie" statement in the frontend
16270 configuration. The field is a single dash ('-') when the option is not
16271 set. Only one cookie may be captured, it is generally used to track session
16272 ID exchanges between a client and a server to detect session crossing
16273 between clients due to application bugs. For more details, please consult
16274 the section "Capturing HTTP headers and cookies" below.
16275
16276 - "captured_response_cookie" is an optional "name=value" entry indicating
16277 that the server has returned a cookie with its response. The cookie name
16278 and its maximum length are defined by the "capture cookie" statement in the
16279 frontend configuration. The field is a single dash ('-') when the option is
16280 not set. Only one cookie may be captured, it is generally used to track
16281 session ID exchanges between a client and a server to detect session
16282 crossing between clients due to application bugs. For more details, please
16283 consult the section "Capturing HTTP headers and cookies" below.
16284
16285 - "termination_state" is the condition the session was in when the session
16286 ended. This indicates the session state, which side caused the end of
16287 session to happen, for what reason (timeout, error, ...), just like in TCP
16288 logs, and information about persistence operations on cookies in the last
16289 two characters. The normal flags should begin with "--", indicating the
16290 session was closed by either end with no data remaining in buffers. See
16291 below "Session state at disconnection" for more details.
16292
16293 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016294 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016295 limits have been reached. For instance, if actconn is close to 512 or 1024
16296 when multiple connection errors occur, chances are high that the system
16297 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016298 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016299 system.
16300
16301 - "feconn" is the total number of concurrent connections on the frontend when
16302 the session was logged. It is useful to estimate the amount of resource
16303 required to sustain high loads, and to detect when the frontend's "maxconn"
16304 has been reached. Most often when this value increases by huge jumps, it is
16305 because there is congestion on the backend servers, but sometimes it can be
16306 caused by a denial of service attack.
16307
16308 - "beconn" is the total number of concurrent connections handled by the
16309 backend when the session was logged. It includes the total number of
16310 concurrent connections active on servers as well as the number of
16311 connections pending in queues. It is useful to estimate the amount of
16312 additional servers needed to support high loads for a given application.
16313 Most often when this value increases by huge jumps, it is because there is
16314 congestion on the backend servers, but sometimes it can be caused by a
16315 denial of service attack.
16316
16317 - "srv_conn" is the total number of concurrent connections still active on
16318 the server when the session was logged. It can never exceed the server's
16319 configured "maxconn" parameter. If this value is very often close or equal
16320 to the server's "maxconn", it means that traffic regulation is involved a
16321 lot, meaning that either the server's maxconn value is too low, or that
16322 there aren't enough servers to process the load with an optimal response
16323 time. When only one of the server's "srv_conn" is high, it usually means
16324 that this server has some trouble causing the requests to take longer to be
16325 processed than on other servers.
16326
16327 - "retries" is the number of connection retries experienced by this session
16328 when trying to connect to the server. It must normally be zero, unless a
16329 server is being stopped at the same moment the connection was attempted.
16330 Frequent retries generally indicate either a network problem between
16331 haproxy and the server, or a misconfigured system backlog on the server
16332 preventing new connections from being queued. This field may optionally be
16333 prefixed with a '+' sign, indicating that the session has experienced a
16334 redispatch after the maximal retry count has been reached on the initial
16335 server. In this case, the server name appearing in the log is the one the
16336 connection was redispatched to, and not the first one, though both may
16337 sometimes be the same in case of hashing for instance. So as a general rule
16338 of thumb, when a '+' is present in front of the retry count, this count
16339 should not be attributed to the logged server.
16340
16341 - "srv_queue" is the total number of requests which were processed before
16342 this one in the server queue. It is zero when the request has not gone
16343 through the server queue. It makes it possible to estimate the approximate
16344 server's response time by dividing the time spent in queue by the number of
16345 requests in the queue. It is worth noting that if a session experiences a
16346 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016347 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016348 backend queue unless a redispatch occurs.
16349
16350 - "backend_queue" is the total number of requests which were processed before
16351 this one in the backend's global queue. It is zero when the request has not
16352 gone through the global queue. It makes it possible to estimate the average
16353 queue length, which easily translates into a number of missing servers when
16354 divided by a server's "maxconn" parameter. It is worth noting that if a
16355 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016356 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016357 through both the server queue and the backend queue unless a redispatch
16358 occurs.
16359
16360 - "captured_request_headers" is a list of headers captured in the request due
16361 to the presence of the "capture request header" statement in the frontend.
16362 Multiple headers can be captured, they will be delimited by a vertical bar
16363 ('|'). When no capture is enabled, the braces do not appear, causing a
16364 shift of remaining fields. It is important to note that this field may
16365 contain spaces, and that using it requires a smarter log parser than when
16366 it's not used. Please consult the section "Capturing HTTP headers and
16367 cookies" below for more details.
16368
16369 - "captured_response_headers" is a list of headers captured in the response
16370 due to the presence of the "capture response header" statement in the
16371 frontend. Multiple headers can be captured, they will be delimited by a
16372 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16373 causing a shift of remaining fields. It is important to note that this
16374 field may contain spaces, and that using it requires a smarter log parser
16375 than when it's not used. Please consult the section "Capturing HTTP headers
16376 and cookies" below for more details.
16377
16378 - "http_request" is the complete HTTP request line, including the method,
16379 request and HTTP version string. Non-printable characters are encoded (see
16380 below the section "Non-printable characters"). This is always the last
16381 field, and it is always delimited by quotes and is the only one which can
16382 contain quotes. If new fields are added to the log format, they will be
16383 added before this field. This field might be truncated if the request is
16384 huge and does not fit in the standard syslog buffer (1024 characters). This
16385 is the reason why this field must always remain the last one.
16386
16387
Cyril Bontédc4d9032012-04-08 21:57:39 +0200163888.2.4. Custom log format
16389------------------------
William Lallemand48940402012-01-30 16:47:22 +010016390
Willy Tarreau2beef582012-12-20 17:22:52 +010016391The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016392mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016393
Davor Ocelice9ed2812017-12-25 17:49:28 +010016394HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016395Variables can take arguments using braces ('{}'), and multiple arguments are
16396separated by commas within the braces. Flags may be added or removed by
16397prefixing them with a '+' or '-' sign.
16398
16399Special variable "%o" may be used to propagate its flags to all other
16400variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016401("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016402
Willy Tarreauc8368452012-12-21 00:09:23 +010016403If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016404as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016405less common information such as the client's SSL certificate's DN, or to log
16406the key that would be used to store an entry into a stick table.
16407
William Lallemand48940402012-01-30 16:47:22 +010016408Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016409In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016410in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016411
Dragan Dosen835b9212016-02-12 13:23:03 +010016412Note: when using the RFC5424 syslog message format, the characters '"',
16413'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16414https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16415such cases, the use of the flag "E" should be considered.
16416
William Lallemand48940402012-01-30 16:47:22 +010016417Flags are :
16418 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016419 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016420 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16421 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016422
16423 Example:
16424
16425 log-format %T\ %t\ Some\ Text
16426 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16427
Dragan Dosen835b9212016-02-12 13:23:03 +010016428 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16429
William Lallemand48940402012-01-30 16:47:22 +010016430At the moment, the default HTTP format is defined this way :
16431
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016432 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16433 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016434
William Lallemandbddd4fd2012-02-27 11:23:10 +010016435the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016436
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016437 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16438 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16439 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016440
William Lallemandbddd4fd2012-02-27 11:23:10 +010016441and the default TCP format is defined this way :
16442
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016443 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16444 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016445
William Lallemand48940402012-01-30 16:47:22 +010016446Please refer to the table below for currently defined variables :
16447
William Lallemandbddd4fd2012-02-27 11:23:10 +010016448 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016449 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016450 +---+------+-----------------------------------------------+-------------+
16451 | | %o | special variable, apply flags on all next var | |
16452 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016453 | | %B | bytes_read (from server to client) | numeric |
16454 | H | %CC | captured_request_cookie | string |
16455 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016456 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016457 | H | %HM | HTTP method (ex: POST) | string |
16458 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016459 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016460 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16461 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016462 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016463 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016464 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016465 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016466 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016467 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016468 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016469 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16470 | H | %Ti | idle time before the HTTP request | numeric |
16471 | H | %Tq | Th + Ti + TR | numeric |
16472 | H | %TR | time to receive the full request from 1st byte| numeric |
16473 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016474 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016475 | | %Tt | Tt | numeric |
16476 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016477 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016478 | | %ac | actconn | numeric |
16479 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016480 | | %bc | beconn (backend concurrent connections) | numeric |
16481 | | %bi | backend_source_ip (connecting address) | IP |
16482 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016483 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016484 | | %ci | client_ip (accepted address) | IP |
16485 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016486 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016487 | | %fc | feconn (frontend concurrent connections) | numeric |
16488 | | %fi | frontend_ip (accepting address) | IP |
16489 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016490 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016491 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016492 | | %hr | captured_request_headers default style | string |
16493 | | %hrl | captured_request_headers CLF style | string list |
16494 | | %hs | captured_response_headers default style | string |
16495 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016496 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016497 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016498 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016499 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016500 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016501 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016502 | | %sc | srv_conn (server concurrent connections) | numeric |
16503 | | %si | server_IP (target address) | IP |
16504 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016505 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016506 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16507 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016508 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016509 | H | %tr | date_time of HTTP request | date |
16510 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016511 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016512 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016513 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016514 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016515
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016516 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016517
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016518
165198.2.5. Error log format
16520-----------------------
16521
16522When an incoming connection fails due to an SSL handshake or an invalid PROXY
16523protocol header, haproxy will log the event using a shorter, fixed line format.
16524By default, logs are emitted at the LOG_INFO level, unless the option
16525"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016526will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016527logged if the "dontlognull" option is set.
16528
16529The format looks like this :
16530
16531 >>> Dec 3 18:27:14 localhost \
16532 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16533 Connection error during SSL handshake
16534
16535 Field Format Extract from the example above
16536 1 process_name '[' pid ']:' haproxy[6103]:
16537 2 client_ip ':' client_port 127.0.0.1:56059
16538 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16539 4 frontend_name "/" bind_name ":" frt/f1:
16540 5 message Connection error during SSL handshake
16541
16542These fields just provide minimal information to help debugging connection
16543failures.
16544
16545
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165468.3. Advanced logging options
16547-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016548
16549Some advanced logging options are often looked for but are not easy to find out
16550just by looking at the various options. Here is an entry point for the few
16551options which can enable better logging. Please refer to the keywords reference
16552for more information about their usage.
16553
16554
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165558.3.1. Disabling logging of external tests
16556------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016557
16558It is quite common to have some monitoring tools perform health checks on
16559haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16560commercial load-balancer, and sometimes it will simply be a more complete
16561monitoring system such as Nagios. When the tests are very frequent, users often
16562ask how to disable logging for those checks. There are three possibilities :
16563
16564 - if connections come from everywhere and are just TCP probes, it is often
16565 desired to simply disable logging of connections without data exchange, by
16566 setting "option dontlognull" in the frontend. It also disables logging of
16567 port scans, which may or may not be desired.
16568
16569 - if the connection come from a known source network, use "monitor-net" to
16570 declare this network as monitoring only. Any host in this network will then
16571 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016572 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016573 such as other load-balancers.
16574
16575 - if the tests are performed on a known URI, use "monitor-uri" to declare
16576 this URI as dedicated to monitoring. Any host sending this request will
16577 only get the result of a health-check, and the request will not be logged.
16578
16579
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165808.3.2. Logging before waiting for the session to terminate
16581----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016582
16583The problem with logging at end of connection is that you have no clue about
16584what is happening during very long sessions, such as remote terminal sessions
16585or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016586"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016587just before data transfer begins. This means that in case of TCP, it will still
16588log the connection status to the server, and in case of HTTP, it will log just
16589after processing the server headers. In this case, the number of bytes reported
16590is the number of header bytes sent to the client. In order to avoid confusion
16591with normal logs, the total time field and the number of bytes are prefixed
16592with a '+' sign which means that real numbers are certainly larger.
16593
16594
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165958.3.3. Raising log level upon errors
16596------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016597
16598Sometimes it is more convenient to separate normal traffic from errors logs,
16599for instance in order to ease error monitoring from log files. When the option
16600"log-separate-errors" is used, connections which experience errors, timeouts,
16601retries, redispatches or HTTP status codes 5xx will see their syslog level
16602raised from "info" to "err". This will help a syslog daemon store the log in
16603a separate file. It is very important to keep the errors in the normal traffic
16604file too, so that log ordering is not altered. You should also be careful if
16605you already have configured your syslog daemon to store all logs higher than
16606"notice" in an "admin" file, because the "err" level is higher than "notice".
16607
16608
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166098.3.4. Disabling logging of successful connections
16610--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016611
16612Although this may sound strange at first, some large sites have to deal with
16613multiple thousands of logs per second and are experiencing difficulties keeping
16614them intact for a long time or detecting errors within them. If the option
16615"dontlog-normal" is set on the frontend, all normal connections will not be
16616logged. In this regard, a normal connection is defined as one without any
16617error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
16618and a response with a status 5xx is not considered normal and will be logged
16619too. Of course, doing is is really discouraged as it will remove most of the
16620useful information from the logs. Do this only if you have no other
16621alternative.
16622
16623
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200166248.4. Timing events
16625------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016626
16627Timers provide a great help in troubleshooting network problems. All values are
16628reported in milliseconds (ms). These timers should be used in conjunction with
16629the session termination flags. In TCP mode with "option tcplog" set on the
16630frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016631mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
16632addition, three other measures are provided, "Th", "Ti", and "Tq".
16633
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010016634Timings events in HTTP mode:
16635
16636 first request 2nd request
16637 |<-------------------------------->|<-------------- ...
16638 t tr t tr ...
16639 ---|----|----|----|----|----|----|----|----|--
16640 : Th Ti TR Tw Tc Tr Td : Ti ...
16641 :<---- Tq ---->: :
16642 :<-------------- Tt -------------->:
16643 :<--------- Ta --------->:
16644
16645Timings events in TCP mode:
16646
16647 TCP session
16648 |<----------------->|
16649 t t
16650 ---|----|----|----|----|---
16651 | Th Tw Tc Td |
16652 |<------ Tt ------->|
16653
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016654 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016655 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016656 only happen once during the whole connection's lifetime. A large time here
16657 may indicate that the client only pre-established the connection without
16658 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010016659 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020016660 SSL handshake was very expensive to compute. Please note that this time is
16661 reported only before the first request, so it is safe to average it over
16662 all request to calculate the amortized value. The second and subsequent
16663 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016664
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016665 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
16666 counts between the end of the handshakes and the first byte of the HTTP
16667 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020016668 to count after the end of the transmission the previous response. When a
16669 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
16670 after the previous request. Some browsers pre-establish connections to a
16671 server in order to reduce the latency of a future request, and keep them
16672 pending until they need it. This delay will be reported as the idle time. A
16673 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016674
16675 - TR: total time to get the client request (HTTP mode only). It's the time
16676 elapsed between the first bytes received and the moment the proxy received
16677 the empty line marking the end of the HTTP headers. The value "-1"
16678 indicates that the end of headers has never been seen. This happens when
16679 the client closes prematurely or times out. This time is usually very short
16680 since most requests fit in a single packet. A large time may indicate a
16681 request typed by hand during a test.
16682
16683 - Tq: total time to get the client request from the accept date or since the
16684 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010016685 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 +020016686 returns -1 as well. This timer used to be very useful before the arrival of
16687 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
16688 it in favor of TR nowadays, as the idle time adds a lot of noise to the
16689 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016690
16691 - Tw: total time spent in the queues waiting for a connection slot. It
16692 accounts for backend queue as well as the server queues, and depends on the
16693 queue size, and the time needed for the server to complete previous
16694 requests. The value "-1" means that the request was killed before reaching
16695 the queue, which is generally what happens with invalid or denied requests.
16696
16697 - Tc: total time to establish the TCP connection to the server. It's the time
16698 elapsed between the moment the proxy sent the connection request, and the
16699 moment it was acknowledged by the server, or between the TCP SYN packet and
16700 the matching SYN/ACK packet in return. The value "-1" means that the
16701 connection never established.
16702
16703 - Tr: server response time (HTTP mode only). It's the time elapsed between
16704 the moment the TCP connection was established to the server and the moment
16705 the server sent its complete response headers. It purely shows its request
16706 processing time, without the network overhead due to the data transmission.
16707 It is worth noting that when the client has data to send to the server, for
16708 instance during a POST request, the time already runs, and this can distort
16709 apparent response time. For this reason, it's generally wise not to trust
16710 too much this field for POST requests initiated from clients behind an
16711 untrusted network. A value of "-1" here means that the last the response
16712 header (empty line) was never seen, most likely because the server timeout
16713 stroke before the server managed to process the request.
16714
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016715 - Ta: total active time for the HTTP request, between the moment the proxy
16716 received the first byte of the request header and the emission of the last
16717 byte of the response body. The exception is when the "logasap" option is
16718 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
16719 a '+' sign. From this field, we can deduce "Td", the data transmission time,
16720 by subtracting other timers when valid :
16721
16722 Td = Ta - (TR + Tw + Tc + Tr)
16723
16724 Timers with "-1" values have to be excluded from this equation. Note that
16725 "Ta" can never be negative.
16726
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016727 - Tt: total session duration time, between the moment the proxy accepted it
16728 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016729 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
16730 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016731 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016732
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016733 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016734
16735 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016736 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
16737 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016738
16739These timers provide precious indications on trouble causes. Since the TCP
16740protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
16741that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016742due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
16743"Tt" is close to a timeout value specified in the configuration, it often means
16744that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016745
16746Most common cases :
16747
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016748 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
16749 the client and the proxy. This is very rare on local networks but might
16750 happen when clients are on far remote networks and send large requests. It
16751 may happen that values larger than usual appear here without any network
16752 cause. Sometimes, during an attack or just after a resource starvation has
16753 ended, haproxy may accept thousands of connections in a few milliseconds.
16754 The time spent accepting these connections will inevitably slightly delay
16755 processing of other connections, and it can happen that request times in the
16756 order of a few tens of milliseconds are measured after a few thousands of
16757 new connections have been accepted at once. Using one of the keep-alive
16758 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020016759 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016760
16761 - If "Tc" is close to 3000, a packet has probably been lost between the
16762 server and the proxy during the server connection phase. This value should
16763 always be very low, such as 1 ms on local networks and less than a few tens
16764 of ms on remote networks.
16765
Willy Tarreau55165fe2009-05-10 12:02:55 +020016766 - If "Tr" is nearly always lower than 3000 except some rare values which seem
16767 to be the average majored by 3000, there are probably some packets lost
16768 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016769
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016770 - If "Ta" is large even for small byte counts, it generally is because
16771 neither the client nor the server decides to close the connection while
16772 haproxy is running in tunnel mode and both have agreed on a keep-alive
16773 connection mode. In order to solve this issue, it will be needed to specify
16774 one of the HTTP options to manipulate keep-alive or close options on either
16775 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
16776 important when connection regulation is used with the "maxconn" option on
16777 the servers, since no new connection will be sent to the server until
16778 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016779
16780Other noticeable HTTP log cases ('xx' means any value to be ignored) :
16781
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016782 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016783 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016784 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016785
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016786 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016787 or it aborted too early. Check the session termination flags
16788 then "timeout http-request" and "timeout client" settings.
16789
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016790 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016791 servers were out of order, because the request was invalid
16792 or forbidden by ACL rules. Check the session termination
16793 flags.
16794
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016795 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
16796 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016797 Check the session termination flags, then check the
16798 "timeout connect" setting. Note that the tarpit action might
16799 return similar-looking patterns, with "Tw" equal to the time
16800 the client connection was maintained open.
16801
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016802 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016803 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016804 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016805 termination flags, then check the "timeout server" setting.
16806
16807
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168088.5. Session state at disconnection
16809-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016810
16811TCP and HTTP logs provide a session termination indicator in the
16812"termination_state" field, just before the number of active connections. It is
168132-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
16814each of which has a special meaning :
16815
16816 - On the first character, a code reporting the first event which caused the
16817 session to terminate :
16818
16819 C : the TCP session was unexpectedly aborted by the client.
16820
16821 S : the TCP session was unexpectedly aborted by the server, or the
16822 server explicitly refused it.
16823
16824 P : the session was prematurely aborted by the proxy, because of a
16825 connection limit enforcement, because a DENY filter was matched,
16826 because of a security check which detected and blocked a dangerous
16827 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010016828 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020016829
16830 L : the session was locally processed by haproxy and was not passed to
16831 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016832
16833 R : a resource on the proxy has been exhausted (memory, sockets, source
16834 ports, ...). Usually, this appears during the connection phase, and
16835 system logs should contain a copy of the precise error. If this
16836 happens, it must be considered as a very serious anomaly which
16837 should be fixed as soon as possible by any means.
16838
16839 I : an internal error was identified by the proxy during a self-check.
16840 This should NEVER happen, and you are encouraged to report any log
16841 containing this, because this would almost certainly be a bug. It
16842 would be wise to preventively restart the process after such an
16843 event too, in case it would be caused by memory corruption.
16844
Simon Horman752dc4a2011-06-21 14:34:59 +090016845 D : the session was killed by haproxy because the server was detected
16846 as down and was configured to kill all connections when going down.
16847
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070016848 U : the session was killed by haproxy on this backup server because an
16849 active server was detected as up and was configured to kill all
16850 backup connections when going up.
16851
Willy Tarreaua2a64e92011-09-07 23:01:56 +020016852 K : the session was actively killed by an admin operating on haproxy.
16853
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016854 c : the client-side timeout expired while waiting for the client to
16855 send or receive data.
16856
16857 s : the server-side timeout expired while waiting for the server to
16858 send or receive data.
16859
16860 - : normal session completion, both the client and the server closed
16861 with nothing left in the buffers.
16862
16863 - on the second character, the TCP or HTTP session state when it was closed :
16864
Willy Tarreauf7b30a92010-12-06 22:59:17 +010016865 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016866 (HTTP mode only). Nothing was sent to any server.
16867
16868 Q : the proxy was waiting in the QUEUE for a connection slot. This can
16869 only happen when servers have a 'maxconn' parameter set. It can
16870 also happen in the global queue after a redispatch consecutive to
16871 a failed attempt to connect to a dying server. If no redispatch is
16872 reported, then no connection attempt was made to any server.
16873
16874 C : the proxy was waiting for the CONNECTION to establish on the
16875 server. The server might at most have noticed a connection attempt.
16876
16877 H : the proxy was waiting for complete, valid response HEADERS from the
16878 server (HTTP only).
16879
16880 D : the session was in the DATA phase.
16881
16882 L : the proxy was still transmitting LAST data to the client while the
16883 server had already finished. This one is very rare as it can only
16884 happen when the client dies while receiving the last packets.
16885
16886 T : the request was tarpitted. It has been held open with the client
16887 during the whole "timeout tarpit" duration or until the client
16888 closed, both of which will be reported in the "Tw" timer.
16889
16890 - : normal session completion after end of data transfer.
16891
16892 - the third character tells whether the persistence cookie was provided by
16893 the client (only in HTTP mode) :
16894
16895 N : the client provided NO cookie. This is usually the case for new
16896 visitors, so counting the number of occurrences of this flag in the
16897 logs generally indicate a valid trend for the site frequentation.
16898
16899 I : the client provided an INVALID cookie matching no known server.
16900 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020016901 cookies between HTTP/HTTPS sites, persistence conditionally
16902 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016903
16904 D : the client provided a cookie designating a server which was DOWN,
16905 so either "option persist" was used and the client was sent to
16906 this server, or it was not set and the client was redispatched to
16907 another server.
16908
Willy Tarreau996a92c2010-10-13 19:30:47 +020016909 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016910 server.
16911
Willy Tarreau996a92c2010-10-13 19:30:47 +020016912 E : the client provided a valid cookie, but with a last date which was
16913 older than what is allowed by the "maxidle" cookie parameter, so
16914 the cookie is consider EXPIRED and is ignored. The request will be
16915 redispatched just as if there was no cookie.
16916
16917 O : the client provided a valid cookie, but with a first date which was
16918 older than what is allowed by the "maxlife" cookie parameter, so
16919 the cookie is consider too OLD and is ignored. The request will be
16920 redispatched just as if there was no cookie.
16921
Willy Tarreauc89ccb62012-04-05 21:18:22 +020016922 U : a cookie was present but was not used to select the server because
16923 some other server selection mechanism was used instead (typically a
16924 "use-server" rule).
16925
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016926 - : does not apply (no cookie set in configuration).
16927
16928 - the last character reports what operations were performed on the persistence
16929 cookie returned by the server (only in HTTP mode) :
16930
16931 N : NO cookie was provided by the server, and none was inserted either.
16932
16933 I : no cookie was provided by the server, and the proxy INSERTED one.
16934 Note that in "cookie insert" mode, if the server provides a cookie,
16935 it will still be overwritten and reported as "I" here.
16936
Willy Tarreau996a92c2010-10-13 19:30:47 +020016937 U : the proxy UPDATED the last date in the cookie that was presented by
16938 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016939 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020016940 date indicated in the cookie. If any other change happens, such as
16941 a redispatch, then the cookie will be marked as inserted instead.
16942
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016943 P : a cookie was PROVIDED by the server and transmitted as-is.
16944
16945 R : the cookie provided by the server was REWRITTEN by the proxy, which
16946 happens in "cookie rewrite" or "cookie prefix" modes.
16947
16948 D : the cookie provided by the server was DELETED by the proxy.
16949
16950 - : does not apply (no cookie set in configuration).
16951
Willy Tarreau996a92c2010-10-13 19:30:47 +020016952The combination of the two first flags gives a lot of information about what
16953was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016954helpful to detect server saturation, network troubles, local system resource
16955starvation, attacks, etc...
16956
16957The most common termination flags combinations are indicated below. They are
16958alphabetically sorted, with the lowercase set just after the upper case for
16959easier finding and understanding.
16960
16961 Flags Reason
16962
16963 -- Normal termination.
16964
16965 CC The client aborted before the connection could be established to the
16966 server. This can happen when haproxy tries to connect to a recently
16967 dead (or unchecked) server, and the client aborts while haproxy is
16968 waiting for the server to respond or for "timeout connect" to expire.
16969
16970 CD The client unexpectedly aborted during data transfer. This can be
16971 caused by a browser crash, by an intermediate equipment between the
16972 client and haproxy which decided to actively break the connection,
16973 by network routing issues between the client and haproxy, or by a
16974 keep-alive session between the server and the client terminated first
16975 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010016976
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016977 cD The client did not send nor acknowledge any data for as long as the
16978 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020016979 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016980
16981 CH The client aborted while waiting for the server to start responding.
16982 It might be the server taking too long to respond or the client
16983 clicking the 'Stop' button too fast.
16984
16985 cH The "timeout client" stroke while waiting for client data during a
16986 POST request. This is sometimes caused by too large TCP MSS values
16987 for PPPoE networks which cannot transport full-sized packets. It can
16988 also happen when client timeout is smaller than server timeout and
16989 the server takes too long to respond.
16990
16991 CQ The client aborted while its session was queued, waiting for a server
16992 with enough empty slots to accept it. It might be that either all the
16993 servers were saturated or that the assigned server was taking too
16994 long a time to respond.
16995
16996 CR The client aborted before sending a full HTTP request. Most likely
16997 the request was typed by hand using a telnet client, and aborted
16998 too early. The HTTP status code is likely a 400 here. Sometimes this
16999 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017000 and the client. "option http-ignore-probes" can be used to ignore
17001 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017002
17003 cR The "timeout http-request" stroke before the client sent a full HTTP
17004 request. This is sometimes caused by too large TCP MSS values on the
17005 client side for PPPoE networks which cannot transport full-sized
17006 packets, or by clients sending requests by hand and not typing fast
17007 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017008 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017009 some browsers started to implement a "pre-connect" feature consisting
17010 in speculatively connecting to some recently visited web sites just
17011 in case the user would like to visit them. This results in many
17012 connections being established to web sites, which end up in 408
17013 Request Timeout if the timeout strikes first, or 400 Bad Request when
17014 the browser decides to close them first. These ones pollute the log
17015 and feed the error counters. Some versions of some browsers have even
17016 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017017 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017018 http-ignore-probes" in the frontend, resulting in connections with
17019 zero data transfer to be totally ignored. This will definitely hide
17020 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017021
17022 CT The client aborted while its session was tarpitted. It is important to
17023 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017024 wrong tarpit rules have been written. If a lot of them happen, it
17025 might make sense to lower the "timeout tarpit" value to something
17026 closer to the average reported "Tw" timer, in order not to consume
17027 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017028
Willy Tarreau570f2212013-06-10 16:42:09 +020017029 LR The request was intercepted and locally handled by haproxy. Generally
17030 it means that this was a redirect or a stats request.
17031
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017032 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017033 the TCP connection (the proxy received a TCP RST or an ICMP message
17034 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017035 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017036 or no ARP response on local network). When this happens in HTTP mode,
17037 the status code is likely a 502 or 503 here.
17038
17039 sC The "timeout connect" stroke before a connection to the server could
17040 complete. When this happens in HTTP mode, the status code is likely a
17041 503 or 504 here.
17042
17043 SD The connection to the server died with an error during the data
17044 transfer. This usually means that haproxy has received an RST from
17045 the server or an ICMP message from an intermediate equipment while
17046 exchanging data with the server. This can be caused by a server crash
17047 or by a network issue on an intermediate equipment.
17048
17049 sD The server did not send nor acknowledge any data for as long as the
17050 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017051 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017052 load-balancers, ...), as well as keep-alive sessions maintained
17053 between the client and the server expiring first on haproxy.
17054
17055 SH The server aborted before sending its full HTTP response headers, or
17056 it crashed while processing the request. Since a server aborting at
17057 this moment is very rare, it would be wise to inspect its logs to
17058 control whether it crashed and why. The logged request may indicate a
17059 small set of faulty requests, demonstrating bugs in the application.
17060 Sometimes this might also be caused by an IDS killing the connection
17061 between haproxy and the server.
17062
17063 sH The "timeout server" stroke before the server could return its
17064 response headers. This is the most common anomaly, indicating too
17065 long transactions, probably caused by server or database saturation.
17066 The immediate workaround consists in increasing the "timeout server"
17067 setting, but it is important to keep in mind that the user experience
17068 will suffer from these long response times. The only long term
17069 solution is to fix the application.
17070
17071 sQ The session spent too much time in queue and has been expired. See
17072 the "timeout queue" and "timeout connect" settings to find out how to
17073 fix this if it happens too often. If it often happens massively in
17074 short periods, it may indicate general problems on the affected
17075 servers due to I/O or database congestion, or saturation caused by
17076 external attacks.
17077
17078 PC The proxy refused to establish a connection to the server because the
17079 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017080 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017081 so that it does not happen anymore. This status is very rare and
17082 might happen when the global "ulimit-n" parameter is forced by hand.
17083
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017084 PD The proxy blocked an incorrectly formatted chunked encoded message in
17085 a request or a response, after the server has emitted its headers. In
17086 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017087 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017088 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017089
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017090 PH The proxy blocked the server's response, because it was invalid,
17091 incomplete, dangerous (cache control), or matched a security filter.
17092 In any case, an HTTP 502 error is sent to the client. One possible
17093 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017094 containing unauthorized characters. It is also possible but quite
17095 rare, that the proxy blocked a chunked-encoding request from the
17096 client due to an invalid syntax, before the server responded. In this
17097 case, an HTTP 400 error is sent to the client and reported in the
17098 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017099
17100 PR The proxy blocked the client's HTTP request, either because of an
17101 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17102 the client, or because a deny filter matched, in which case it
17103 returned an HTTP 403 error.
17104
17105 PT The proxy blocked the client's request and has tarpitted its
17106 connection before returning it a 500 server error. Nothing was sent
17107 to the server. The connection was maintained open for as long as
17108 reported by the "Tw" timer field.
17109
17110 RC A local resource has been exhausted (memory, sockets, source ports)
17111 preventing the connection to the server from establishing. The error
17112 logs will tell precisely what was missing. This is very rare and can
17113 only be solved by proper system tuning.
17114
Willy Tarreau996a92c2010-10-13 19:30:47 +020017115The combination of the two last flags gives a lot of information about how
17116persistence was handled by the client, the server and by haproxy. This is very
17117important to troubleshoot disconnections, when users complain they have to
17118re-authenticate. The commonly encountered flags are :
17119
17120 -- Persistence cookie is not enabled.
17121
17122 NN No cookie was provided by the client, none was inserted in the
17123 response. For instance, this can be in insert mode with "postonly"
17124 set on a GET request.
17125
17126 II A cookie designating an invalid server was provided by the client,
17127 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017128 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017129 value can be presented by a client when no other server knows it.
17130
17131 NI No cookie was provided by the client, one was inserted in the
17132 response. This typically happens for first requests from every user
17133 in "insert" mode, which makes it an easy way to count real users.
17134
17135 VN A cookie was provided by the client, none was inserted in the
17136 response. This happens for most responses for which the client has
17137 already got a cookie.
17138
17139 VU A cookie was provided by the client, with a last visit date which is
17140 not completely up-to-date, so an updated cookie was provided in
17141 response. This can also happen if there was no date at all, or if
17142 there was a date but the "maxidle" parameter was not set, so that the
17143 cookie can be switched to unlimited time.
17144
17145 EI A cookie was provided by the client, with a last visit date which is
17146 too old for the "maxidle" parameter, so the cookie was ignored and a
17147 new cookie was inserted in the response.
17148
17149 OI A cookie was provided by the client, with a first visit date which is
17150 too old for the "maxlife" parameter, so the cookie was ignored and a
17151 new cookie was inserted in the response.
17152
17153 DI The server designated by the cookie was down, a new server was
17154 selected and a new cookie was emitted in the response.
17155
17156 VI The server designated by the cookie was not marked dead but could not
17157 be reached. A redispatch happened and selected another one, which was
17158 then advertised in the response.
17159
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017160
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171618.6. Non-printable characters
17162-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017163
17164In order not to cause trouble to log analysis tools or terminals during log
17165consulting, non-printable characters are not sent as-is into log files, but are
17166converted to the two-digits hexadecimal representation of their ASCII code,
17167prefixed by the character '#'. The only characters that can be logged without
17168being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17169escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17170is the same for the character '"' which becomes "#22", as well as '{', '|' and
17171'}' when logging headers.
17172
17173Note that the space character (' ') is not encoded in headers, which can cause
17174issues for tools relying on space count to locate fields. A typical header
17175containing spaces is "User-Agent".
17176
17177Last, it has been observed that some syslog daemons such as syslog-ng escape
17178the quote ('"') with a backslash ('\'). The reverse operation can safely be
17179performed since no quote may appear anywhere else in the logs.
17180
17181
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171828.7. Capturing HTTP cookies
17183---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017184
17185Cookie capture simplifies the tracking a complete user session. This can be
17186achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017187section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017188cookie will simultaneously be checked in the request ("Cookie:" header) and in
17189the response ("Set-Cookie:" header). The respective values will be reported in
17190the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017191locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017192not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17193user switches to a new session for example, because the server will reassign it
17194a new cookie. It is also possible to detect if a server unexpectedly sets a
17195wrong cookie to a client, leading to session crossing.
17196
17197 Examples :
17198 # capture the first cookie whose name starts with "ASPSESSION"
17199 capture cookie ASPSESSION len 32
17200
17201 # capture the first cookie whose name is exactly "vgnvisitor"
17202 capture cookie vgnvisitor= len 32
17203
17204
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172058.8. Capturing HTTP headers
17206---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017207
17208Header captures are useful to track unique request identifiers set by an upper
17209proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17210the response, one can search for information about the response length, how the
17211server asked the cache to behave, or an object location during a redirection.
17212
17213Header captures are performed using the "capture request header" and "capture
17214response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017215section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017216
17217It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017218time. Non-existent headers are logged as empty strings, and if one header
17219appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017220are grouped within braces '{' and '}' in the same order as they were declared,
17221and delimited with a vertical bar '|' without any space. Response headers
17222follow the same representation, but are displayed after a space following the
17223request headers block. These blocks are displayed just before the HTTP request
17224in the logs.
17225
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017226As a special case, it is possible to specify an HTTP header capture in a TCP
17227frontend. The purpose is to enable logging of headers which will be parsed in
17228an HTTP backend if the request is then switched to this HTTP backend.
17229
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017230 Example :
17231 # This instance chains to the outgoing proxy
17232 listen proxy-out
17233 mode http
17234 option httplog
17235 option logasap
17236 log global
17237 server cache1 192.168.1.1:3128
17238
17239 # log the name of the virtual server
17240 capture request header Host len 20
17241
17242 # log the amount of data uploaded during a POST
17243 capture request header Content-Length len 10
17244
17245 # log the beginning of the referrer
17246 capture request header Referer len 20
17247
17248 # server name (useful for outgoing proxies only)
17249 capture response header Server len 20
17250
17251 # logging the content-length is useful with "option logasap"
17252 capture response header Content-Length len 10
17253
Davor Ocelice9ed2812017-12-25 17:49:28 +010017254 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017255 capture response header Cache-Control len 8
17256
17257 # the Via header will report the next proxy's name
17258 capture response header Via len 20
17259
17260 # log the URL location during a redirection
17261 capture response header Location len 20
17262
17263 >>> Aug 9 20:26:09 localhost \
17264 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17265 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17266 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17267 "GET http://fr.adserver.yahoo.com/"
17268
17269 >>> Aug 9 20:30:46 localhost \
17270 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17271 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17272 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017273 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017274
17275 >>> Aug 9 20:30:46 localhost \
17276 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17277 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17278 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17279 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017280 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017281
17282
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172838.9. Examples of logs
17284---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017285
17286These are real-world examples of logs accompanied with an explanation. Some of
17287them have been made up by hand. The syslog part has been removed for better
17288reading. Their sole purpose is to explain how to decipher them.
17289
17290 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17291 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17292 "HEAD / HTTP/1.0"
17293
17294 => long request (6.5s) entered by hand through 'telnet'. The server replied
17295 in 147 ms, and the session ended normally ('----')
17296
17297 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17298 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17299 0/9 "HEAD / HTTP/1.0"
17300
17301 => Idem, but the request was queued in the global queue behind 9 other
17302 requests, and waited there for 1230 ms.
17303
17304 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17305 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17306 "GET /image.iso HTTP/1.0"
17307
17308 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017309 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017310 14 ms, 243 bytes of headers were sent to the client, and total time from
17311 accept to first data byte is 30 ms.
17312
17313 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17314 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17315 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17316
17317 => the proxy blocked a server response either because of an "rspdeny" or
17318 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017319 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017320 risked being cached. In this case, the response is replaced with a "502
17321 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17322 to return the 502 and not the server.
17323
17324 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017325 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 +010017326
17327 => the client never completed its request and aborted itself ("C---") after
17328 8.5s, while the proxy was waiting for the request headers ("-R--").
17329 Nothing was sent to any server.
17330
17331 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17332 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17333
17334 => The client never completed its request, which was aborted by the
17335 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017336 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017337 send a 408 return code to the client.
17338
17339 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17340 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17341
17342 => This log was produced with "option tcplog". The client timed out after
17343 5 seconds ("c----").
17344
17345 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17346 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017347 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017348
17349 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017350 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017351 (config says 'retries 3'), and no redispatch (otherwise we would have
17352 seen "/+3"). Status code 503 was returned to the client. There were 115
17353 connections on this server, 202 connections on this proxy, and 205 on
17354 the global process. It is possible that the server refused the
17355 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017356
Willy Tarreau52b2d222011-09-07 23:48:48 +020017357
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200173589. Supported filters
17359--------------------
17360
17361Here are listed officially supported filters with the list of parameters they
17362accept. Depending on compile options, some of these filters might be
17363unavailable. The list of available filters is reported in haproxy -vv.
17364
17365See also : "filter"
17366
173679.1. Trace
17368----------
17369
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017370filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017371
17372 Arguments:
17373 <name> is an arbitrary name that will be reported in
17374 messages. If no name is provided, "TRACE" is used.
17375
17376 <random-parsing> enables the random parsing of data exchanged between
17377 the client and the server. By default, this filter
17378 parses all available data. With this parameter, it
17379 only parses a random amount of the available data.
17380
Davor Ocelice9ed2812017-12-25 17:49:28 +010017381 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017382 default, this filter forwards all previously parsed
17383 data. With this parameter, it only forwards a random
17384 amount of the parsed data.
17385
Davor Ocelice9ed2812017-12-25 17:49:28 +010017386 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017387
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017388This filter can be used as a base to develop new filters. It defines all
17389callbacks and print a message on the standard error stream (stderr) with useful
17390information for all of them. It may be useful to debug the activity of other
17391filters or, quite simply, HAProxy's activity.
17392
17393Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17394tests the behavior of a filter that parses data exchanged between a client and
17395a server by adding some latencies in the processing.
17396
17397
173989.2. HTTP compression
17399---------------------
17400
17401filter compression
17402
17403The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17404keyword must still be used to enable and configure the HTTP compression. And
17405when no other filter is used, it is enough. But it is mandatory to explicitly
17406use a filter line to enable the HTTP compression when two or more filters are
17407used for the same listener/frontend/backend. This is important to know the
17408filters evaluation order.
17409
17410See also : "compression"
17411
17412
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200174139.3. Stream Processing Offload Engine (SPOE)
17414--------------------------------------------
17415
17416filter spoe [engine <name>] config <file>
17417
17418 Arguments :
17419
17420 <name> is the engine name that will be used to find the right scope in
17421 the configuration file. If not provided, all the file will be
17422 parsed.
17423
17424 <file> is the path of the engine configuration file. This file can
17425 contain configuration of several engines. In this case, each
17426 part must be placed in its own scope.
17427
17428The Stream Processing Offload Engine (SPOE) is a filter communicating with
17429external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017430streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017431exchanged with them are configured in dedicated files, for the main part. It
17432also requires dedicated backends, defined in HAProxy configuration.
17433
17434SPOE communicates with external components using an in-house binary protocol,
17435the Stream Processing Offload Protocol (SPOP).
17436
Tim Düsterhus4896c442016-11-29 02:15:19 +010017437For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017438"doc/SPOE.txt".
17439
17440Important note:
17441 The SPOE filter is highly experimental for now and was not heavily
17442 tested. It is really not production ready. So use it carefully.
17443
William Lallemand86d0df02017-11-24 21:36:45 +01001744410. Cache
17445---------
17446
17447HAProxy provides a cache, which was designed to perform cache on small objects
17448(favicon, css...). This is a minimalist low-maintenance cache which runs in
17449RAM.
17450
17451The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017452this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017453
17454If an object is not used anymore, it can be deleted to store a new object
17455independently of its expiration date. The oldest objects are deleted first
17456when we try to allocate a new one.
17457
Cyril Bonté7b888f12017-11-26 22:24:31 +010017458The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017459
17460It's possible to view the status of a cache using the Unix socket command
17461"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17462for more details.
17463
17464When an object is delivered from the cache, the server name in the log is
17465replaced by "<CACHE>".
17466
Cyril Bonté7b888f12017-11-26 22:24:31 +01001746710.1. Limitation
17468----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017469
17470The cache won't store and won't deliver objects in these cases:
17471
17472- If the response is not a 200
17473- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017474- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010017475- If the response is not cacheable
17476
17477- If the request is not a GET
17478- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020017479- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010017480
17481Caution!: Due to the current limitation of the filters, it is not recommended
17482to use the cache with other filters. Using them can cause undefined behavior
Cyril Bonté7b888f12017-11-26 22:24:31 +010017483if they modify the response (compression for example).
William Lallemand86d0df02017-11-24 21:36:45 +010017484
Cyril Bonté7b888f12017-11-26 22:24:31 +01001748510.2. Setup
17486-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017487
17488To setup a cache, you must define a cache section and use it in a proxy with
17489the corresponding http-request and response actions.
17490
Cyril Bonté7b888f12017-11-26 22:24:31 +01001749110.2.1. Cache section
17492---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017493
17494cache <name>
17495 Declare a cache section, allocate a shared cache memory named <name>, the
17496 size of cache is mandatory.
17497
17498total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017499 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017500 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010017501
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017502max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017503 Define the maximum size of the objects to be cached. Must not be greater than
17504 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
17505 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017506
William Lallemand86d0df02017-11-24 21:36:45 +010017507max-age <seconds>
17508 Define the maximum expiration duration. The expiration is set has the lowest
17509 value between the s-maxage or max-age (in this order) directive in the
17510 Cache-Control response header and this value. The default value is 60
17511 seconds, which means that you can't cache an object more than 60 seconds by
17512 default.
17513
Cyril Bonté7b888f12017-11-26 22:24:31 +01001751410.2.2. Proxy section
17515---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017516
17517http-request cache-use <name>
17518 Try to deliver a cached object from the cache <name>. This directive is also
17519 mandatory to store the cache as it calculates the cache hash. If you want to
17520 use a condition for both storage and delivering that's a good idea to put it
17521 after this one.
17522
17523http-response cache-store <name>
17524 Store an http-response within the cache. The storage of the response headers
17525 is done at this step, which means you can use others http-response actions
17526 to modify headers before or after the storage of the response. This action
17527 is responsible for the setup of the cache storage filter.
17528
17529
17530Example:
17531
17532 backend bck1
17533 mode http
17534
17535 http-request cache-use foobar
17536 http-response cache-store foobar
17537 server srv1 127.0.0.1:80
17538
17539 cache foobar
17540 total-max-size 4
17541 max-age 240
17542
Willy Tarreau0ba27502007-12-24 16:55:16 +010017543/*
17544 * Local variables:
17545 * fill-column: 79
17546 * End:
17547 */