blob: 18fcf9604094c5854aab2fa52cc8d53193c5f6e7 [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 Tarreau72e92272018-12-08 16:20:55 +01007 2018/12/08
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 Tarreau75c62c22018-11-22 11:02:09 +0100622 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200623 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200624 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200625 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000626 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000627 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200628 - tune.buffers.limit
629 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200630 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200631 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100632 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200633 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200634 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200635 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100636 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200637 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200638 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100639 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100640 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100641 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100642 - tune.lua.session-timeout
643 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200644 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100645 - tune.maxaccept
646 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200647 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200648 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200649 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100650 - tune.rcvbuf.client
651 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100652 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200653 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100654 - tune.sndbuf.client
655 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100656 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100657 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200658 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100659 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200660 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200661 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100662 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200663 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100664 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200665 - tune.vars.reqres-max-size
666 - tune.vars.sess-max-size
667 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100668 - tune.zlib.memlevel
669 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100670
Willy Tarreau6a06a402007-07-15 20:15:28 +0200671 * Debugging
672 - debug
673 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200674
675
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006763.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200677------------------------------------
678
Emeric Brunc8e8d122012-10-02 18:42:10 +0200679ca-base <dir>
680 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200681 relative path is used with "ca-file" or "crl-file" directives. Absolute
682 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200683
Willy Tarreau6a06a402007-07-15 20:15:28 +0200684chroot <jail dir>
685 Changes current directory to <jail dir> and performs a chroot() there before
686 dropping privileges. This increases the security level in case an unknown
687 vulnerability would be exploited, since it would make it very hard for the
688 attacker to exploit the system. This only works when the process is started
689 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100690 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100691
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100692cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
693 On Linux 2.6 and above, it is possible to bind a process or a thread to a
694 specific CPU set. This means that the process or the thread will never run on
695 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
696 sets. The first argument is a process set, eventually followed by a thread
697 set. These sets have the format
698
699 all | odd | even | number[-[number]]
700
701 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100702 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100703 ignored. It is possible to specify a range with two such number delimited by
704 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100705 "all", only odd numbers using "odd" or even numbers using "even", just like
706 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100707 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 +0100708 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100709 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100710 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100711 specified. Each "cpu-map" directive will replace the previous ones when they
712 overlap. A thread will be bound on the intersection of its mapping and the
713 one of the process on which it is attached. If the intersection is null, no
714 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100715
Christopher Fauletff4121f2017-11-22 16:38:49 +0100716 Ranges can be partially defined. The higher bound can be omitted. In such
717 case, it is replaced by the corresponding maximum value, 32 or 64 depending
718 on the machine's word size.
719
Christopher Faulet26028f62017-11-22 15:01:51 +0100720 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100721 automatically bind a process or a thread to a CPU by incrementing
722 process/thread and CPU sets. To be valid, both sets must have the same
723 size. No matter the declaration order of the CPU sets, it will be bound from
724 the lowest to the highest bound. Having a process and a thread range with the
725 "auto:" prefix is not supported. Only one range is supported, the other one
726 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100727
728 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100729 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
730
731 cpu-map 1/all 0-3 # bind all threads of the first process on the
732 # first 4 CPUs
733
734 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
735 # or "cpu-map 1-32 0-31" depending on the machine's
736 # word size.
737
Christopher Faulet26028f62017-11-22 15:01:51 +0100738 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100739 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100740 cpu-map auto:1-4 0-3
741 cpu-map auto:1-4 0-1 2-3
742 cpu-map auto:1-4 3 2 1 0
743
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100744 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
745 # and so on.
746 cpu-map auto:1/1-4 0-3
747 cpu-map auto:1/1-4 0-1 2-3
748 cpu-map auto:1/1-4 3 2 1 0
749
Davor Ocelice9ed2812017-12-25 17:49:28 +0100750 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100751 cpu-map auto:all 0-63
752 cpu-map auto:even 0-31
753 cpu-map auto:odd 32-63
754
755 # invalid cpu-map because process and CPU sets have different sizes.
756 cpu-map auto:1-4 0 # invalid
757 cpu-map auto:1 0-3 # invalid
758
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100759 # invalid cpu-map because automatic binding is used with a process range
760 # and a thread range.
761 cpu-map auto:all/all 0 # invalid
762 cpu-map auto:all/1-4 0 # invalid
763 cpu-map auto:1-4/all 0 # invalid
764
Emeric Brunc8e8d122012-10-02 18:42:10 +0200765crt-base <dir>
766 Assigns a default directory to fetch SSL certificates from when a relative
767 path is used with "crtfile" directives. Absolute locations specified after
768 "crtfile" prevail and ignore "crt-base".
769
Willy Tarreau6a06a402007-07-15 20:15:28 +0200770daemon
771 Makes the process fork into background. This is the recommended mode of
772 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100773 disabled by the command line "-db" argument. This option is ignored in
774 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200775
David Carlier8167f302015-06-01 13:50:06 +0200776deviceatlas-json-file <path>
777 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100778 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200779
780deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100781 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200782 optional and set to 0 by default if not set.
783
784deviceatlas-separator <char>
785 Sets the character separator for the API properties results. This directive
786 is optional and set to | by default if not set.
787
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100788deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200789 Sets the client cookie's name used for the detection if the DeviceAtlas
790 Client-side component was used during the request. This directive is optional
791 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100792
Simon Horman98637e52014-06-20 12:30:16 +0900793external-check
794 Allows the use of an external agent to perform health checks.
795 This is disabled by default as a security precaution.
796 See "option external-check".
797
Willy Tarreau6a06a402007-07-15 20:15:28 +0200798gid <number>
799 Changes the process' group ID to <number>. It is recommended that the group
800 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
801 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100802 Note that if haproxy is started from a user having supplementary groups, it
803 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200804 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100805
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100806hard-stop-after <time>
807 Defines the maximum time allowed to perform a clean soft-stop.
808
809 Arguments :
810 <time> is the maximum time (by default in milliseconds) for which the
811 instance will remain alive when a soft-stop is received via the
812 SIGUSR1 signal.
813
814 This may be used to ensure that the instance will quit even if connections
815 remain opened during a soft-stop (for example with long timeouts for a proxy
816 in tcp mode). It applies both in TCP and HTTP mode.
817
818 Example:
819 global
820 hard-stop-after 30s
821
Willy Tarreau6a06a402007-07-15 20:15:28 +0200822group <group name>
823 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
824 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100825
Dragan Dosen7ad31542015-09-28 17:16:47 +0200826log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100827 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100828 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100829 configured with "log global".
830
831 <address> can be one of:
832
Willy Tarreau2769aa02007-12-27 18:26:09 +0100833 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100834 no port is specified, 514 is used by default (the standard syslog
835 port).
836
David du Colombier24bb5f52011-03-17 10:40:23 +0100837 - An IPv6 address followed by a colon and optionally a UDP port. If
838 no port is specified, 514 is used by default (the standard syslog
839 port).
840
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100841 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100842 considerations for chroot (be sure the path is accessible inside
843 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100844 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100845
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100846 - A file descriptor number in the form "fd@<number>", which may point
847 to a pipe, terminal, or socket. In this case unbuffered logs are used
848 and one writev() call per log is performed. This is a bit expensive
849 but acceptable for most workloads. Messages sent this way will not be
850 truncated but may be dropped, in which case the DroppedLogs counter
851 will be incremented. The writev() call is atomic even on pipes for
852 messages up to PIPE_BUF size, which POSIX recommends to be at least
853 512 and which is 4096 bytes on most modern operating systems. Any
854 larger message may be interleaved with messages from other processes.
855 Exceptionally for debugging purposes the file descriptor may also be
856 directed to a file, but doing so will significantly slow haproxy down
857 as non-blocking calls will be ignored. Also there will be no way to
858 purge nor rotate this file without restarting the process. Note that
859 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100860 for use with a TCP syslog server. See also the "short" and "raw"
861 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100862
863 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
864 "fd@2", see above.
865
William Lallemandb2f07452015-05-12 14:27:13 +0200866 You may want to reference some environment variables in the address
867 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100868
Willy Tarreau18324f52014-06-27 18:10:07 +0200869 <length> is an optional maximum line length. Log lines larger than this value
870 will be truncated before being sent. The reason is that syslog
871 servers act differently on log line length. All servers support the
872 default value of 1024, but some servers simply drop larger lines
873 while others do log them. If a server supports long lines, it may
874 make sense to set this value here in order to avoid truncating long
875 lines. Similarly, if a server drops long lines, it is preferable to
876 truncate them before sending them. Accepted values are 80 to 65535
877 inclusive. The default value of 1024 is generally fine for all
878 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100879 JSON-formatted logs may require larger values. You may also need to
880 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200881
Dragan Dosen7ad31542015-09-28 17:16:47 +0200882 <format> is the log format used when generating syslog messages. It may be
883 one of the following :
884
885 rfc3164 The RFC3164 syslog message format. This is the default.
886 (https://tools.ietf.org/html/rfc3164)
887
888 rfc5424 The RFC5424 syslog message format.
889 (https://tools.ietf.org/html/rfc5424)
890
Willy Tarreaue8746a02018-11-12 08:45:00 +0100891 short A message containing only a level between angle brackets such as
892 '<3>', followed by the text. The PID, date, time, process name
893 and system name are omitted. This is designed to be used with a
894 local log server. This format is compatible with what the systemd
895 logger consumes.
896
Willy Tarreauc1b06452018-11-12 11:57:56 +0100897 raw A message containing only the text. The level, PID, date, time,
898 process name and system name are omitted. This is designed to be
899 used in containers or during development, where the severity only
900 depends on the file descriptor used (stdout/stderr).
901
Robert Tsai81ae1952007-12-05 10:47:29 +0100902 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200903
Willy Tarreaue8746a02018-11-12 08:45:00 +0100904 kern user mail daemon auth syslog lpr news
905 uucp cron auth2 ftp ntp audit alert cron2
906 local0 local1 local2 local3 local4 local5 local6 local7
907
Willy Tarreauc1b06452018-11-12 11:57:56 +0100908 Note that the facility is ignored for the "short" and "raw"
909 formats, but still required as a positional field. It is
910 recommended to use "daemon" in this case to make it clear that
911 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200912
913 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200914 all messages are sent. If a maximum level is specified, only messages with a
915 severity at least as important as this level will be sent. An optional minimum
916 level can be specified. If it is set, logs emitted with a more severe level
917 than this one will be capped to this level. This is used to avoid sending
918 "emerg" messages on all terminals on some default syslog configurations.
919 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200920
Cyril Bontédc4d9032012-04-08 21:57:39 +0200921 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200922
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100923log-send-hostname [<string>]
924 Sets the hostname field in the syslog header. If optional "string" parameter
925 is set the header is set to the string contents, otherwise uses the hostname
926 of the system. Generally used if one is not relaying logs through an
927 intermediate syslog server or for simply customizing the hostname printed in
928 the logs.
929
Kevinm48936af2010-12-22 16:08:21 +0000930log-tag <string>
931 Sets the tag field in the syslog header to this string. It defaults to the
932 program name as launched from the command line, which usually is "haproxy".
933 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100934 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000935
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100936lua-load <file>
937 This global directive loads and executes a Lua file. This directive can be
938 used multiple times.
939
William Lallemand4cfede82017-11-24 22:02:34 +0100940master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200941 Master-worker mode. It is equivalent to the command line "-W" argument.
942 This mode will launch a "master" which will monitor the "workers". Using
943 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100944 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200945 or daemon mode. It is recommended to use this mode with multiprocess and
946 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100947 By default, if a worker exits with a bad return code, in the case of a
948 segfault for example, all workers will be killed, and the master will leave.
949 It is convenient to combine this behavior with Restart=on-failure in a
950 systemd unit file in order to relaunch the whole process. If you don't want
951 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200952
William Lallemand4cfede82017-11-24 22:02:34 +0100953 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200954
Willy Tarreau6a06a402007-07-15 20:15:28 +0200955nbproc <number>
956 Creates <number> processes when going daemon. This requires the "daemon"
957 mode. By default, only one process is created, which is the recommended mode
958 of operation. For systems limited to small sets of file descriptors per
959 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
960 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
961
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200962nbthread <number>
963 This setting is only available when support for threads was built in. It
964 creates <number> threads for each created processes. It means if HAProxy is
965 started in foreground, it only creates <number> threads for the first
966 process. FOR NOW, THREADS SUPPORT IN HAPROXY IS HIGHLY EXPERIMENTAL AND IT
967 MUST BE ENABLED WITH CAUTION AND AT YOUR OWN RISK. See also "nbproc".
968
Willy Tarreau6a06a402007-07-15 20:15:28 +0200969pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100970 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200971 the "-p" command line argument. The file must be accessible to the user
972 starting the process. See also "daemon".
973
Willy Tarreau1d549722016-02-16 12:41:57 +0100974presetenv <name> <value>
975 Sets environment variable <name> to value <value>. If the variable exists, it
976 is NOT overwritten. The changes immediately take effect so that the next line
977 in the configuration file sees the new value. See also "setenv", "resetenv",
978 and "unsetenv".
979
980resetenv [<name> ...]
981 Removes all environment variables except the ones specified in argument. It
982 allows to use a clean controlled environment before setting new values with
983 setenv or unsetenv. Please note that some internal functions may make use of
984 some environment variables, such as time manipulation functions, but also
985 OpenSSL or even external checks. This must be used with extreme care and only
986 after complete validation. The changes immediately take effect so that the
987 next line in the configuration file sees the new environment. See also
988 "setenv", "presetenv", and "unsetenv".
989
Christopher Fauletff4121f2017-11-22 16:38:49 +0100990stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200991 Limits the stats socket to a certain set of processes numbers. By default the
992 stats socket is bound to all processes, causing a warning to be emitted when
993 nbproc is greater than 1 because there is no way to select the target process
994 when connecting. However, by using this setting, it becomes possible to pin
995 the stats socket to a specific set of processes, typically the first one. The
996 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100997 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +0100998 word size (32 or 64). Ranges can be partially defined. The higher bound can
999 be omitted. In such case, it is replaced by the corresponding maximum
1000 value. A better option consists in using the "process" setting of the "stats
1001 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001002
Baptiste Assmann5626f482015-08-23 10:00:10 +02001003server-state-base <directory>
1004 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001005 file names which do not start with a '/'. See also "server-state-file",
1006 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001007
1008server-state-file <file>
1009 Specifies the path to the file containing state of servers. If the path starts
1010 with a slash ('/'), it is considered absolute, otherwise it is considered
1011 relative to the directory specified using "server-state-base" (if set) or to
1012 the current directory. Before reloading HAProxy, it is possible to save the
1013 servers' current state using the stats command "show servers state". The
1014 output of this command must be written in the file pointed by <file>. When
1015 starting up, before handling traffic, HAProxy will read, load and apply state
1016 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001017 configuration. See also "server-state-base" and "show servers state",
1018 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001019
Willy Tarreau1d549722016-02-16 12:41:57 +01001020setenv <name> <value>
1021 Sets environment variable <name> to value <value>. If the variable exists, it
1022 is overwritten. The changes immediately take effect so that the next line in
1023 the configuration file sees the new value. See also "presetenv", "resetenv",
1024 and "unsetenv".
1025
Willy Tarreau610f04b2014-02-13 11:36:41 +01001026ssl-default-bind-ciphers <ciphers>
1027 This setting is only available when support for OpenSSL was built in. It sets
1028 the default string describing the list of cipher algorithms ("cipher suite")
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001029 that are negotiated during the SSL/TLS handshake except for TLSv1.3 for all
1030 "bind" lines which do not explicitly define theirs. The format of the string
1031 is defined in "man 1 ciphers" from OpenSSL man pages, and can be for instance
1032 a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). For
1033 TLSv1.3 cipher configuration, please check the "ssl-default-bind-ciphersuites"
1034 keyword. Please check the "bind" keyword for more information.
1035
1036ssl-default-bind-ciphersuites <ciphersuites>
1037 This setting is only available when support for OpenSSL was built in and
1038 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1039 describing the list of cipher algorithms ("cipher suite") that are negotiated
1040 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1041 theirs. The format of the string is defined in
1042 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites", and can
1043 be for instance a string such as
1044 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
1045 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
1046 the "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
1047 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001048
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001049ssl-default-bind-options [<option>]...
1050 This setting is only available when support for OpenSSL was built in. It sets
1051 default ssl-options to force on all "bind" lines. Please check the "bind"
1052 keyword to see available options.
1053
1054 Example:
1055 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001056 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001057
Willy Tarreau610f04b2014-02-13 11:36:41 +01001058ssl-default-server-ciphers <ciphers>
1059 This setting is only available when support for OpenSSL was built in. It
1060 sets the default string describing the list of cipher algorithms that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001061 negotiated during the SSL/TLS handshake except for TLSv1.3 with the server,
1062 for all "server" lines which do not explicitly define theirs. The format of
1063 the string is defined in "man 1 ciphers". For TLSv1.3 cipher configuration,
1064 please check the "ssl-default-server-ciphersuites" keyword. Please check the
1065 "server" keyword for more information.
1066
1067ssl-default-server-ciphersuites <ciphersuites>
1068 This setting is only available when support for OpenSSL was built in and
1069 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1070 string describing the list of cipher algorithms that are negotiated during
1071 the TLSv1.3 handshake with the server, for all "server" lines which do not
1072 explicitly define theirs. The format of the string is defined in
1073 "man 1 ciphers" under the "ciphersuites" section. For cipher configuration for
1074 TLSv1.2 and earlier, please check the "ssl-default-server-ciphers" keyword.
1075 Please check the "server" keyword for more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001076
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001077ssl-default-server-options [<option>]...
1078 This setting is only available when support for OpenSSL was built in. It sets
1079 default ssl-options to force on all "server" lines. Please check the "server"
1080 keyword to see available options.
1081
Remi Gacogne47783ef2015-05-29 15:53:22 +02001082ssl-dh-param-file <file>
1083 This setting is only available when support for OpenSSL was built in. It sets
1084 the default DH parameters that are used during the SSL/TLS handshake when
1085 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001086 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001087 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001088 are not specified either by using ssl-dh-param-file or by setting them
1089 directly in the certificate file, pre-generated DH parameters of the size
1090 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1091 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001092 Custom DH parameters may be generated by using the OpenSSL command
1093 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1094 parameters should not be considered secure anymore.
1095
Emeric Brun850efd52014-01-29 12:24:34 +01001096ssl-server-verify [none|required]
1097 The default behavior for SSL verify on servers side. If specified to 'none',
1098 servers certificates are not verified. The default is 'required' except if
1099 forced using cmdline option '-dV'.
1100
Willy Tarreauabb175f2012-09-24 12:43:26 +02001101stats socket [<address:port>|<path>] [param*]
1102 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1103 Connections to this socket will return various statistics outputs and even
1104 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001105 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001106 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001107
Willy Tarreauabb175f2012-09-24 12:43:26 +02001108 All parameters supported by "bind" lines are supported, for instance to
1109 restrict access to some users or their access rights. Please consult
1110 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001111
1112stats timeout <timeout, in milliseconds>
1113 The default timeout on the stats socket is set to 10 seconds. It is possible
1114 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001115 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001116
1117stats maxconn <connections>
1118 By default, the stats socket is limited to 10 concurrent connections. It is
1119 possible to change this value with "stats maxconn".
1120
Willy Tarreau6a06a402007-07-15 20:15:28 +02001121uid <number>
1122 Changes the process' user ID to <number>. It is recommended that the user ID
1123 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1124 be started with superuser privileges in order to be able to switch to another
1125 one. See also "gid" and "user".
1126
1127ulimit-n <number>
1128 Sets the maximum number of per-process file-descriptors to <number>. By
1129 default, it is automatically computed, so it is recommended not to use this
1130 option.
1131
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001132unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1133 [ group <group> ] [ gid <gid> ]
1134
1135 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1136 This is mainly used to simplify declaration of those UNIX sockets and reduce
1137 the risk of errors, since those settings are most commonly required but are
1138 also process-specific. The <prefix> setting can be used to force all socket
1139 path to be relative to that directory. This might be needed to access another
1140 component's chroot. Note that those paths are resolved before haproxy chroots
1141 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1142 all have the same meaning as their homonyms used by the "bind" statement. If
1143 both are specified, the "bind" statement has priority, meaning that the
1144 "unix-bind" settings may be seen as process-wide default settings.
1145
Willy Tarreau1d549722016-02-16 12:41:57 +01001146unsetenv [<name> ...]
1147 Removes environment variables specified in arguments. This can be useful to
1148 hide some sensitive information that are occasionally inherited from the
1149 user's environment during some operations. Variables which did not exist are
1150 silently ignored so that after the operation, it is certain that none of
1151 these variables remain. The changes immediately take effect so that the next
1152 line in the configuration file will not see these variables. See also
1153 "setenv", "presetenv", and "resetenv".
1154
Willy Tarreau6a06a402007-07-15 20:15:28 +02001155user <user name>
1156 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1157 See also "uid" and "group".
1158
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001159node <name>
1160 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1161
1162 This statement is useful in HA configurations where two or more processes or
1163 servers share the same IP address. By setting a different node-name on all
1164 nodes, it becomes easy to immediately spot what server is handling the
1165 traffic.
1166
1167description <text>
1168 Add a text that describes the instance.
1169
1170 Please note that it is required to escape certain characters (# for example)
1171 and this text is inserted into a html page so you should avoid using
1172 "<" and ">" characters.
1173
Thomas Holmesdb04f192015-05-18 13:21:39 +0100117451degrees-data-file <file path>
1175 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001176 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001177
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001178 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001179 compiled with USE_51DEGREES.
1180
Ben Shillitof25e8e52016-12-02 14:25:37 +0000118151degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001182 A list of 51Degrees property names to be load from the dataset. A full list
1183 of names is available on the 51Degrees website:
1184 https://51degrees.com/resources/property-dictionary
1185
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001186 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001187 compiled with USE_51DEGREES.
1188
Dragan Dosen93b38d92015-06-29 16:43:25 +0200118951degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001190 A char that will be appended to every property value in a response header
1191 containing 51Degrees results. If not set that will be set as ','.
1192
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001193 Please note that this option is only available when haproxy has been
1194 compiled with USE_51DEGREES.
1195
119651degrees-cache-size <number>
1197 Sets the size of the 51Degrees converter cache to <number> entries. This
1198 is an LRU cache which reminds previous device detections and their results.
1199 By default, this cache is disabled.
1200
1201 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001202 compiled with USE_51DEGREES.
1203
scientiamobiled0027ed2016-11-04 10:55:08 +01001204wurfl-data-file <file path>
1205 The path of the WURFL data file to provide device detection services. The
1206 file should be accessible by HAProxy with relevant permissions.
1207
1208 Please note that this option is only available when haproxy has been compiled
1209 with USE_WURFL=1.
1210
1211wurfl-information-list [<capability>]*
1212 A space-delimited list of WURFL capabilities, virtual capabilities, property
1213 names we plan to use in injected headers. A full list of capability and
1214 virtual capability names is available on the Scientiamobile website :
1215
1216 https://www.scientiamobile.com/wurflCapability
1217
1218 Valid WURFL properties are:
1219 - wurfl_id Contains the device ID of the matched device.
1220
1221 - wurfl_root_id Contains the device root ID of the matched
1222 device.
1223
1224 - wurfl_isdevroot Tells if the matched device is a root device.
1225 Possible values are "TRUE" or "FALSE".
1226
1227 - wurfl_useragent The original useragent coming with this
1228 particular web request.
1229
1230 - wurfl_api_version Contains a string representing the currently
1231 used Libwurfl API version.
1232
1233 - wurfl_engine_target Contains a string representing the currently
1234 set WURFL Engine Target. Possible values are
1235 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1236
1237 - wurfl_info A string containing information on the parsed
1238 wurfl.xml and its full path.
1239
1240 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1241 WURFL has been loaded successfully.
1242
1243 - wurfl_normalized_useragent The normalized useragent.
1244
1245 - wurfl_useragent_priority The user agent priority used by WURFL.
1246
1247 Please note that this option is only available when haproxy has been compiled
1248 with USE_WURFL=1.
1249
1250wurfl-information-list-separator <char>
1251 A char that will be used to separate values in a response header containing
1252 WURFL results. If not set that a comma (',') will be used by default.
1253
1254 Please note that this option is only available when haproxy has been compiled
1255 with USE_WURFL=1.
1256
1257wurfl-patch-file [<file path>]
1258 A list of WURFL patch file paths. Note that patches are loaded during startup
1259 thus before the chroot.
1260
1261 Please note that this option is only available when haproxy has been compiled
1262 with USE_WURFL=1.
1263
1264wurfl-engine-mode { accuracy | performance }
1265 Sets the WURFL engine target. You can choose between 'accuracy' or
1266 'performance' targets. In performance mode, desktop web browser detection is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001267 done programmatically without referencing the WURFL data. As a result, most
scientiamobiled0027ed2016-11-04 10:55:08 +01001268 desktop web browsers are returned as generic_web_browser WURFL ID for
Davor Ocelice9ed2812017-12-25 17:49:28 +01001269 performance. If either performance or accuracy are not defined, performance
scientiamobiled0027ed2016-11-04 10:55:08 +01001270 mode is enabled by default.
1271
1272 Please note that this option is only available when haproxy has been compiled
1273 with USE_WURFL=1.
1274
1275wurfl-cache-size <U>[,<D>]
1276 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1277 <D> is the internal device cache size. There are three possibilities here :
1278 - "0" : no cache is used.
1279 - <U> : the Single LRU cache is used, the size is expressed in elements.
1280 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1281 the highest performing option.
1282
1283 Please note that this option is only available when haproxy has been compiled
1284 with USE_WURFL=1.
1285
1286wurfl-useragent-priority { plain | sideloaded_browser }
1287 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1288 over the default sideloaded browser user agent ('sideloaded_browser').
1289
1290 Please note that this option is only available when haproxy has been compiled
1291 with USE_WURFL=1.
1292
Willy Tarreau6a06a402007-07-15 20:15:28 +02001293
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012943.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001295-----------------------
1296
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001297busy-polling
1298 In some situations, especially when dealing with low latency on processors
1299 supporting a variable frequency or when running inside virtual machines, each
1300 time the process waits for an I/O using the poller, the processor goes back
1301 to sleep or is offered to another VM for a long time, and it causes
1302 excessively high latencies. This option provides a solution preventing the
1303 processor from sleeping by always using a null timeout on the pollers. This
1304 results in a significant latency reduction (30 to 100 microseconds observed)
1305 at the expense of a risk to overheat the processor. It may even be used with
1306 threads, in which case improperly bound threads may heavily conflict,
1307 resulting in a worse performance and high values for the CPU stolen fields
1308 in "show info" output, indicating which threads are misconfigured. It is
1309 important not to let the process run on the same processor as the network
1310 interrupts when this option is used. It is also better to avoid using it on
1311 multiple CPU threads sharing the same core. This option is disabled by
1312 default. If it has been enabled, it may still be forcibly disabled by
1313 prefixing it with the "no" keyword. It is ignored by the "select" and
1314 "poll" pollers.
1315
Willy Tarreau1746eec2014-04-25 10:46:47 +02001316max-spread-checks <delay in milliseconds>
1317 By default, haproxy tries to spread the start of health checks across the
1318 smallest health check interval of all the servers in a farm. The principle is
1319 to avoid hammering services running on the same server. But when using large
1320 check intervals (10 seconds or more), the last servers in the farm take some
1321 time before starting to be tested, which can be a problem. This parameter is
1322 used to enforce an upper bound on delay between the first and the last check,
1323 even if the servers' check intervals are larger. When servers run with
1324 shorter intervals, their intervals will be respected though.
1325
Willy Tarreau6a06a402007-07-15 20:15:28 +02001326maxconn <number>
1327 Sets the maximum per-process number of concurrent connections to <number>. It
1328 is equivalent to the command-line argument "-n". Proxies will stop accepting
1329 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001330 automatically adjusted according to this value. See also "ulimit-n". Note:
1331 the "select" poller cannot reliably use more than 1024 file descriptors on
1332 some platforms. If your platform only supports select and reports "select
1333 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001334 below 500 in general). If this value is not set, it will default to the value
1335 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1336 limit is enforced, or will be computed based on the memory limit, the buffer
1337 size, memory allocated to compression, SSL cache size, and use or not of SSL
1338 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001339
Willy Tarreau81c25d02011-09-07 15:17:21 +02001340maxconnrate <number>
1341 Sets the maximum per-process number of connections per second to <number>.
1342 Proxies will stop accepting connections when this limit is reached. It can be
1343 used to limit the global capacity regardless of each frontend capacity. It is
1344 important to note that this can only be used as a service protection measure,
1345 as there will not necessarily be a fair share between frontends when the
1346 limit is reached, so it's a good idea to also limit each frontend to some
1347 value close to its expected share. Also, lowering tune.maxaccept can improve
1348 fairness.
1349
William Lallemandd85f9172012-11-09 17:05:39 +01001350maxcomprate <number>
1351 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001352 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001353 level will be decreased during the session. If the maximum is reached at the
1354 beginning of a session, the session will not compress at all. If the maximum
1355 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001356 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001357 default value.
1358
William Lallemand072a2bf2012-11-20 17:01:01 +01001359maxcompcpuusage <number>
1360 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1361 for new requests or decreasing the compression level of current requests.
1362 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1363 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1364 case of multiple processes (nbproc > 1), each process manages its individual
1365 usage. A value of 100 disable the limit. The default value is 100. Setting
1366 a lower value will prevent the compression work from slowing the whole
1367 process down and from introducing high latencies.
1368
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001369maxpipes <number>
1370 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1371 are only used by kernel-based tcp splicing. Since a pipe contains two file
1372 descriptors, the "ulimit-n" value will be increased accordingly. The default
1373 value is maxconn/4, which seems to be more than enough for most heavy usages.
1374 The splice code dynamically allocates and releases pipes, and can fall back
1375 to standard copy, so setting this value too low may only impact performance.
1376
Willy Tarreau93e7c002013-10-07 18:51:07 +02001377maxsessrate <number>
1378 Sets the maximum per-process number of sessions per second to <number>.
1379 Proxies will stop accepting connections when this limit is reached. It can be
1380 used to limit the global capacity regardless of each frontend capacity. It is
1381 important to note that this can only be used as a service protection measure,
1382 as there will not necessarily be a fair share between frontends when the
1383 limit is reached, so it's a good idea to also limit each frontend to some
1384 value close to its expected share. Also, lowering tune.maxaccept can improve
1385 fairness.
1386
Willy Tarreau403edff2012-09-06 11:58:37 +02001387maxsslconn <number>
1388 Sets the maximum per-process number of concurrent SSL connections to
1389 <number>. By default there is no SSL-specific limit, which means that the
1390 global maxconn setting will apply to all connections. Setting this limit
1391 avoids having openssl use too much memory and crash when malloc returns NULL
1392 (since it unfortunately does not reliably check for such conditions). Note
1393 that the limit applies both to incoming and outgoing connections, so one
1394 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001395 If this value is not set, but a memory limit is enforced, this value will be
1396 automatically computed based on the memory limit, maxconn, the buffer size,
1397 memory allocated to compression, SSL cache size, and use of SSL in either
1398 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1399 when there is a memory limit, haproxy will automatically adjust these values
1400 so that 100% of the connections can be made over SSL with no risk, and will
1401 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001402
Willy Tarreaue43d5322013-10-07 20:01:52 +02001403maxsslrate <number>
1404 Sets the maximum per-process number of SSL sessions per second to <number>.
1405 SSL listeners will stop accepting connections when this limit is reached. It
1406 can be used to limit the global SSL CPU usage regardless of each frontend
1407 capacity. It is important to note that this can only be used as a service
1408 protection measure, as there will not necessarily be a fair share between
1409 frontends when the limit is reached, so it's a good idea to also limit each
1410 frontend to some value close to its expected share. It is also important to
1411 note that the sessions are accounted before they enter the SSL stack and not
1412 after, which also protects the stack against bad handshakes. Also, lowering
1413 tune.maxaccept can improve fairness.
1414
William Lallemand9d5f5482012-11-07 16:12:57 +01001415maxzlibmem <number>
1416 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1417 When the maximum amount is reached, future sessions will not compress as long
1418 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001419 The default value is 0. The value is available in bytes on the UNIX socket
1420 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1421 "ZlibMemUsage" in bytes.
1422
Willy Tarreau6a06a402007-07-15 20:15:28 +02001423noepoll
1424 Disables the use of the "epoll" event polling system on Linux. It is
1425 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001426 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001427
1428nokqueue
1429 Disables the use of the "kqueue" event polling system on BSD. It is
1430 equivalent to the command-line argument "-dk". The next polling system
1431 used will generally be "poll". See also "nopoll".
1432
1433nopoll
1434 Disables the use of the "poll" event polling system. It is equivalent to the
1435 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001436 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001437 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001438
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001439nosplice
1440 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001441 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001442 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001443 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001444 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1445 be used. This option makes it easier to globally disable kernel splicing in
1446 case of doubt. See also "option splice-auto", "option splice-request" and
1447 "option splice-response".
1448
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001449nogetaddrinfo
1450 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1451 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1452
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001453noreuseport
1454 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1455 command line argument "-dR".
1456
Willy Tarreau75c62c22018-11-22 11:02:09 +01001457profiling.tasks { on | off }
1458 Enables ('on') or disables ('off') per-task CPU profiling. CPU profiling per
1459 task can be very convenient to report where the time is spent and which
1460 requests have what effect on which other request. It is not enabled by
1461 default as it may consume a little bit extra CPU. This requires a system
1462 supporting the clock_gettime(2) syscall with clock identifiers
1463 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1464 be zero. This option may be changed at run time using "set profiling" on the
1465 CLI.
1466
Willy Tarreaufe255b72007-10-14 23:09:26 +02001467spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001468 Sometimes it is desirable to avoid sending agent and health checks to
1469 servers at exact intervals, for instance when many logical servers are
1470 located on the same physical server. With the help of this parameter, it
1471 becomes possible to add some randomness in the check interval between 0
1472 and +/- 50%. A value between 2 and 5 seems to show good results. The
1473 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001474
Davor Ocelice9ed2812017-12-25 17:49:28 +01001475ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001476 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001477 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001478 multiple times, it will simply enable multiple crypto engines. Referencing an
1479 unsupported engine will prevent haproxy from starting. Note that many engines
1480 will lead to lower HTTPS performance than pure software with recent
1481 processors. The optional command "algo" sets the default algorithms an ENGINE
1482 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001483 of "ALL" uses the engine for all cryptographic operations. If no list of
1484 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001485 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1486 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1487 openssl configuration file uses:
1488 https://www.openssl.org/docs/man1.0.2/apps/config.html
1489
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001490ssl-mode-async
1491 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001492 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001493 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1494 doesn't support moving read/write buffers and is not compliant with
1495 haproxy's buffer management. So the asynchronous mode is disabled on
1496 read/write operations (it is only enabled during initial and reneg
1497 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001498
Willy Tarreau33cb0652014-12-23 22:52:37 +01001499tune.buffers.limit <number>
1500 Sets a hard limit on the number of buffers which may be allocated per process.
1501 The default value is zero which means unlimited. The minimum non-zero value
1502 will always be greater than "tune.buffers.reserve" and should ideally always
1503 be about twice as large. Forcing this value can be particularly useful to
1504 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001505 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001506 another one to be released by another session. Since buffers are dynamically
1507 allocated and released, the waiting time is very short and not perceptible
1508 provided that limits remain reasonable. In fact sometimes reducing the limit
1509 may even increase performance by increasing the CPU cache's efficiency. Tests
1510 have shown good results on average HTTP traffic with a limit to 1/10 of the
1511 expected global maxconn setting, which also significantly reduces memory
1512 usage. The memory savings come from the fact that a number of connections
1513 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1514 advised to do so by an haproxy core developer.
1515
Willy Tarreau1058ae72014-12-23 22:40:40 +01001516tune.buffers.reserve <number>
1517 Sets the number of buffers which are pre-allocated and reserved for use only
1518 during memory shortage conditions resulting in failed memory allocations. The
1519 minimum value is 2 and is also the default. There is no reason a user would
1520 want to change this value, it's mostly aimed at haproxy core developers.
1521
Willy Tarreau27a674e2009-08-17 07:23:33 +02001522tune.bufsize <number>
1523 Sets the buffer size to this size (in bytes). Lower values allow more
1524 sessions to coexist in the same amount of RAM, and higher values allow some
1525 applications with very large cookies to work. The default value is 16384 and
1526 can be changed at build time. It is strongly recommended not to change this
1527 from the default value, as very low values will break some services such as
1528 statistics, and values larger than default size will increase memory usage,
1529 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001530 parameter should be decreased by the same factor as this one is increased. In
1531 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1532 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001533 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
1534 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +02001535
Willy Tarreau43961d52010-10-04 20:39:20 +02001536tune.chksize <number>
1537 Sets the check buffer size to this size (in bytes). Higher values may help
1538 find string or regex patterns in very large pages, though doing so may imply
1539 more memory and CPU usage. The default value is 16384 and can be changed at
1540 build time. It is not recommended to change this value, but to use better
1541 checks whenever possible.
1542
William Lallemandf3747832012-11-09 12:33:10 +01001543tune.comp.maxlevel <number>
1544 Sets the maximum compression level. The compression level affects CPU
1545 usage during compression. This value affects CPU usage during compression.
1546 Each session using compression initializes the compression algorithm with
1547 this value. The default value is 1.
1548
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001549tune.h2.header-table-size <number>
1550 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1551 cannot be larger than 65536 bytes. A larger value may help certain clients
1552 send more compact requests, depending on their capabilities. This amount of
1553 memory is consumed for each HTTP/2 connection. It is recommended not to
1554 change it.
1555
Willy Tarreaue6baec02017-07-27 11:45:11 +02001556tune.h2.initial-window-size <number>
1557 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001558 can upload before waiting for an acknowledgment from haproxy. This setting
1559 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001560 The default value is 65535, which roughly allows up to 5 Mbps of upload
1561 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1562 over a 1-ms local network. It can make sense to increase this value to allow
1563 faster uploads, or to reduce it to increase fairness when dealing with many
1564 clients. It doesn't affect resource usage.
1565
Willy Tarreau5242ef82017-07-27 11:47:28 +02001566tune.h2.max-concurrent-streams <number>
1567 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1568 number of outstanding requests on a single connection). The default value is
1569 100. A larger one may slightly improve page load time for complex sites when
1570 visited over high latency networks, but increases the amount of resources a
1571 single client may allocate. A value of zero disables the limit so a single
1572 client may create as many streams as allocatable by haproxy. It is highly
1573 recommended not to change this value.
1574
Willy Tarreau193b8c62012-11-22 00:17:38 +01001575tune.http.cookielen <number>
1576 Sets the maximum length of captured cookies. This is the maximum value that
1577 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1578 will automatically be truncated to this one. It is important not to set too
1579 high a value because all cookie captures still allocate this size whatever
1580 their configured value (they share a same pool). This value is per request
1581 per response, so the memory allocated is twice this value per connection.
1582 When not specified, the limit is set to 63 characters. It is recommended not
1583 to change this value.
1584
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001585tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001586 Sets the maximum length of request URI in logs. This prevents truncating long
1587 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001588 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001589 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001590 configuration directives too.
1591 The default value is 1024.
1592
Willy Tarreauac1932d2011-10-24 19:14:41 +02001593tune.http.maxhdr <number>
1594 Sets the maximum number of headers in a request. When a request comes with a
1595 number of headers greater than this value (including the first line), it is
1596 rejected with a "400 Bad Request" status code. Similarly, too large responses
1597 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1598 for all usages, considering that the widely deployed Apache server uses the
1599 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001600 a buggy application to work by the time it gets fixed. The accepted range is
1601 1..32767. Keep in mind that each new header consumes 32bits of memory for
1602 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001603
Willy Tarreau7e312732014-02-12 16:35:14 +01001604tune.idletimer <timeout>
1605 Sets the duration after which haproxy will consider that an empty buffer is
1606 probably associated with an idle stream. This is used to optimally adjust
1607 some packet sizes while forwarding large and small data alternatively. The
1608 decision to use splice() or to send large buffers in SSL is modulated by this
1609 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1610 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001611 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001612 clicking). There should be not reason for changing this value. Please check
1613 tune.ssl.maxrecord below.
1614
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001615tune.lua.forced-yield <number>
1616 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001617 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001618 HAProxy scheduler to process other tasks like accepting connections or
1619 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001620 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001621 lowered. If the Lua code is quite long and its result is absolutely required
1622 to process the data, the <number> can be increased.
1623
Willy Tarreau32f61e22015-03-18 17:54:59 +01001624tune.lua.maxmem
1625 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1626 default it is zero which means unlimited. It is important to set a limit to
1627 ensure that a bug in a script will not result in the system running out of
1628 memory.
1629
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001630tune.lua.session-timeout <timeout>
1631 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001632 preventing infinite loops or spending too much time in Lua. This timeout
1633 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001634 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001635
1636tune.lua.task-timeout <timeout>
1637 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1638 dedicated to the tasks. By default, this timeout isn't set because a task may
1639 remain alive during of the lifetime of HAProxy. For example, a task used to
1640 check servers.
1641
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001642tune.lua.service-timeout <timeout>
1643 This is the execution timeout for the Lua services. This is useful for
1644 preventing infinite loops or spending too much time in Lua. This timeout
1645 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001646 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001647
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001648tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001649 Sets the maximum number of consecutive connections a process may accept in a
1650 row before switching to other work. In single process mode, higher numbers
1651 give better performance at high connection rates. However in multi-process
1652 modes, keeping a bit of fairness between processes generally is better to
1653 increase performance. This value applies individually to each listener, so
1654 that the number of processes a listener is bound to is taken into account.
1655 This value defaults to 64. In multi-process mode, it is divided by twice
1656 the number of processes the listener is bound to. Setting this value to -1
1657 completely disables the limitation. It should normally not be needed to tweak
1658 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001659
1660tune.maxpollevents <number>
1661 Sets the maximum amount of events that can be processed at once in a call to
1662 the polling system. The default value is adapted to the operating system. It
1663 has been noticed that reducing it below 200 tends to slightly decrease
1664 latency at the expense of network bandwidth, and increasing it above 200
1665 tends to trade latency for slightly increased bandwidth.
1666
Willy Tarreau27a674e2009-08-17 07:23:33 +02001667tune.maxrewrite <number>
1668 Sets the reserved buffer space to this size in bytes. The reserved space is
1669 used for header rewriting or appending. The first reads on sockets will never
1670 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1671 bufsize, though that does not make much sense since there are rarely large
1672 numbers of headers to add. Setting it too high prevents processing of large
1673 requests or responses. Setting it too low prevents addition of new headers
1674 to already large requests or to POST requests. It is generally wise to set it
1675 to about 1024. It is automatically readjusted to half of bufsize if it is
1676 larger than that. This means you don't have to worry about it when changing
1677 bufsize.
1678
Willy Tarreauf3045d22015-04-29 16:24:50 +02001679tune.pattern.cache-size <number>
1680 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1681 cache which reminds previous lookups and their results. It is used by ACLs
1682 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1683 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1684 strings. It applies to pattern expressions which means that it will be able
1685 to memorize the result of a lookup among all the patterns specified on a
1686 configuration line (including all those loaded from files). It automatically
1687 invalidates entries which are updated using HTTP actions or on the CLI. The
1688 default cache size is set to 10000 entries, which limits its footprint to
1689 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1690 risk of collision in this cache, which is in the order of the size of the
1691 cache divided by 2^64. Typically, at 10000 requests per second with the
1692 default cache size of 10000 entries, there's 1% chance that a brute force
1693 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1694 This is considered much lower than the risk of a memory corruption caused by
1695 aging components. If this is not acceptable, the cache can be disabled by
1696 setting this parameter to 0.
1697
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001698tune.pipesize <number>
1699 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1700 are the default size for the system. But sometimes when using TCP splicing,
1701 it can improve performance to increase pipe sizes, especially if it is
1702 suspected that pipes are not filled and that many calls to splice() are
1703 performed. This has an impact on the kernel's memory footprint, so this must
1704 not be changed if impacts are not understood.
1705
Willy Tarreaue803de22010-01-21 17:43:04 +01001706tune.rcvbuf.client <number>
1707tune.rcvbuf.server <number>
1708 Forces the kernel socket receive buffer size on the client or the server side
1709 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1710 and backends. It should normally never be set, and the default size (0) lets
1711 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001712 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001713 order to save kernel memory by preventing it from buffering too large amounts
1714 of received data. Lower values will significantly increase CPU usage though.
1715
Willy Tarreaub22fc302015-12-14 12:04:35 +01001716tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001717 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001718 socket buffers. One of them is that a read returns more than <recv_enough>
1719 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1720 may be changed by this setting to better deal with workloads involving lots
1721 of short messages such as telnet or SSH sessions.
1722
Olivier Houchard1599b802018-05-24 18:59:04 +02001723tune.runqueue-depth <number>
1724 Sets the maxinum amount of task that can be processed at once when running
1725 tasks. The default value is 200. Increasing it may incur latency when
1726 dealing with I/Os, making it too small can incur extra overhead.
1727
Willy Tarreaue803de22010-01-21 17:43:04 +01001728tune.sndbuf.client <number>
1729tune.sndbuf.server <number>
1730 Forces the kernel socket send buffer size on the client or the server side to
1731 the specified value in bytes. This value applies to all TCP/HTTP frontends
1732 and backends. It should normally never be set, and the default size (0) lets
1733 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001734 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001735 order to save kernel memory by preventing it from buffering too large amounts
1736 of received data. Lower values will significantly increase CPU usage though.
1737 Another use case is to prevent write timeouts with extremely slow clients due
1738 to the kernel waiting for a large part of the buffer to be read before
1739 notifying haproxy again.
1740
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001741tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001742 Sets the size of the global SSL session cache, in a number of blocks. A block
1743 is large enough to contain an encoded session without peer certificate.
1744 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001745 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001746 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001747 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001748 and reassigned. Higher values reduce the occurrence of such a purge, hence
1749 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1750 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001751 and are shared between all processes if "nbproc" is greater than 1. Setting
1752 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001753
Emeric Brun8dc60392014-05-09 13:52:00 +02001754tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001755 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001756 should normally not be used since it will force many renegotiations due to
1757 clients hitting a random process. But it may be required on some operating
1758 systems where none of the SSL cache synchronization method may be used. In
1759 this case, adding a first layer of hash-based load balancing before the SSL
1760 layer might limit the impact of the lack of session sharing.
1761
Emeric Brun4f65bff2012-11-16 15:11:00 +01001762tune.ssl.lifetime <timeout>
1763 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001764 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001765 does not guarantee that sessions will last that long, because if the cache is
1766 full, the longest idle sessions will be purged despite their configured
1767 lifetime. The real usefulness of this setting is to prevent sessions from
1768 being used for too long.
1769
Willy Tarreaubfd59462013-02-21 07:46:09 +01001770tune.ssl.maxrecord <number>
1771 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1772 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1773 data only once it has received a full record. With large records, it means
1774 that clients might have to download up to 16kB of data before starting to
1775 process them. Limiting the value can improve page load times on browsers
1776 located over high latency or low bandwidth networks. It is suggested to find
1777 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1778 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1779 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1780 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001781 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001782 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001783
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001784tune.ssl.default-dh-param <number>
1785 Sets the maximum size of the Diffie-Hellman parameters used for generating
1786 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1787 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1788 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1789 this maximum value. Default value if 1024. Only 1024 or higher values are
1790 allowed. Higher values will increase the CPU load, and values greater than
1791 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001792 used if static Diffie-Hellman parameters are supplied either directly
1793 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001794
Christopher Faulet31af49d2015-06-09 17:29:50 +02001795tune.ssl.ssl-ctx-cache-size <number>
1796 Sets the size of the cache used to store generated certificates to <number>
1797 entries. This is a LRU cache. Because generating a SSL certificate
1798 dynamically is expensive, they are cached. The default cache size is set to
1799 1000 entries.
1800
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001801tune.ssl.capture-cipherlist-size <number>
1802 Sets the maximum size of the buffer used for capturing client-hello cipher
1803 list. If the value is 0 (default value) the capture is disabled, otherwise
1804 a buffer is allocated for each SSL/TLS connection.
1805
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001806tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001807tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001808tune.vars.reqres-max-size <size>
1809tune.vars.sess-max-size <size>
1810tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001811 These five tunes help to manage the maximum amount of memory used by the
1812 variables system. "global" limits the overall amount of memory available for
1813 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1814 memory for the session scope, "txn" for the transaction scope, and "reqres"
1815 limits the memory for each request or response processing.
1816 Memory accounting is hierarchical, meaning more coarse grained limits include
1817 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1818 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001819
Daniel Schneller0b547052016-03-21 20:46:57 +01001820 For example, when "tune.vars.sess-max-size" is limited to 100,
1821 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1822 100 either. If we create a variable "txn.var" that contains 100 bytes,
1823 all available space is consumed.
1824 Notice that exceeding the limits at runtime will not result in an error
1825 message, but values might be cut off or corrupted. So make sure to accurately
1826 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001827
William Lallemanda509e4c2012-11-07 16:54:34 +01001828tune.zlib.memlevel <number>
1829 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001830 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001831 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001832 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001833 between 1 and 9. The default value is 8.
1834
1835tune.zlib.windowsize <number>
1836 Sets the window size (the size of the history buffer) as a parameter of the
1837 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001838 in better compression at the expense of memory usage. Can be a value between
1839 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001840
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018413.3. Debugging
1842--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001843
1844debug
1845 Enables debug mode which dumps to stdout all exchanges, and disables forking
1846 into background. It is the equivalent of the command-line argument "-d". It
1847 should never be used in a production configuration since it may prevent full
1848 system startup.
1849
1850quiet
1851 Do not display any message during startup. It is equivalent to the command-
1852 line argument "-q".
1853
Emeric Brunf099e792010-09-27 12:05:28 +02001854
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018553.4. Userlists
1856--------------
1857It is possible to control access to frontend/backend/listen sections or to
1858http stats by allowing only authenticated and authorized users. To do this,
1859it is required to create at least one userlist and to define users.
1860
1861userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001862 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001863 used to store authentication & authorization data for independent customers.
1864
1865group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001866 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001867 attach users to this group by using a comma separated list of names
1868 proceeded by "users" keyword.
1869
Cyril Bontéf0c60612010-02-06 14:44:47 +01001870user <username> [password|insecure-password <password>]
1871 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001872 Adds user <username> to the current userlist. Both secure (encrypted) and
1873 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001874 evaluated using the crypt(3) function, so depending on the system's
1875 capabilities, different algorithms are supported. For example, modern Glibc
1876 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1877 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001878
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001879 Attention: Be aware that using encrypted passwords might cause significantly
1880 increased CPU usage, depending on the number of requests, and the algorithm
1881 used. For any of the hashed variants, the password for each request must
1882 be processed through the chosen algorithm, before it can be compared to the
1883 value specified in the config file. Most current algorithms are deliberately
1884 designed to be expensive to compute to achieve resistance against brute
1885 force attacks. They do not simply salt/hash the clear text password once,
1886 but thousands of times. This can quickly become a major factor in haproxy's
1887 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001888
1889 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001890 userlist L1
1891 group G1 users tiger,scott
1892 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001893
Cyril Bontéf0c60612010-02-06 14:44:47 +01001894 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1895 user scott insecure-password elgato
1896 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001897
Cyril Bontéf0c60612010-02-06 14:44:47 +01001898 userlist L2
1899 group G1
1900 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001901
Cyril Bontéf0c60612010-02-06 14:44:47 +01001902 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1903 user scott insecure-password elgato groups G1,G2
1904 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001905
1906 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001907
Emeric Brunf099e792010-09-27 12:05:28 +02001908
19093.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001910----------
Emeric Brun94900952015-06-11 18:25:54 +02001911It is possible to propagate entries of any data-types in stick-tables between
1912several haproxy instances over TCP connections in a multi-master fashion. Each
1913instance pushes its local updates and insertions to remote peers. The pushed
1914values overwrite remote ones without aggregation. Interrupted exchanges are
1915automatically detected and recovered from the last known point.
1916In addition, during a soft restart, the old process connects to the new one
1917using such a TCP connection to push all its entries before the new process
1918tries to connect to other peers. That ensures very fast replication during a
1919reload, it typically takes a fraction of a second even for large tables.
1920Note that Server IDs are used to identify servers remotely, so it is important
1921that configurations look similar or at least that the same IDs are forced on
1922each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001923
1924peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001925 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001926 which is referenced by one or more stick-tables.
1927
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001928disabled
1929 Disables a peers section. It disables both listening and any synchronization
1930 related to this section. This is provided to disable synchronization of stick
1931 tables without having to comment out all "peers" references.
1932
1933enable
1934 This re-enables a disabled peers section which was previously disabled.
1935
Emeric Brunf099e792010-09-27 12:05:28 +02001936peer <peername> <ip>:<port>
1937 Defines a peer inside a peers section.
1938 If <peername> is set to the local peer name (by default hostname, or forced
1939 using "-L" command line option), haproxy will listen for incoming remote peer
1940 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1941 to join the remote peer, and <peername> is used at the protocol level to
1942 identify and validate the remote peer on the server side.
1943
1944 During a soft restart, local peer <ip>:<port> is used by the old instance to
1945 connect the new one and initiate a complete replication (teaching process).
1946
1947 It is strongly recommended to have the exact same peers declaration on all
1948 peers and to only rely on the "-L" command line argument to change the local
1949 peer name. This makes it easier to maintain coherent configuration files
1950 across all peers.
1951
William Lallemandb2f07452015-05-12 14:27:13 +02001952 You may want to reference some environment variables in the address
1953 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001954
Cyril Bontédc4d9032012-04-08 21:57:39 +02001955 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001956 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001957 peer haproxy1 192.168.0.1:1024
1958 peer haproxy2 192.168.0.2:1024
1959 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001960
1961 backend mybackend
1962 mode tcp
1963 balance roundrobin
1964 stick-table type ip size 20k peers mypeers
1965 stick on src
1966
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001967 server srv1 192.168.0.30:80
1968 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001969
1970
Simon Horman51a1cf62015-02-03 13:00:44 +090019713.6. Mailers
1972------------
1973It is possible to send email alerts when the state of servers changes.
1974If configured email alerts are sent to each mailer that is configured
1975in a mailers section. Email is sent to mailers using SMTP.
1976
Pieter Baauw386a1272015-08-16 15:26:24 +02001977mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001978 Creates a new mailer list with the name <mailersect>. It is an
1979 independent section which is referenced by one or more proxies.
1980
1981mailer <mailername> <ip>:<port>
1982 Defines a mailer inside a mailers section.
1983
1984 Example:
1985 mailers mymailers
1986 mailer smtp1 192.168.0.1:587
1987 mailer smtp2 192.168.0.2:587
1988
1989 backend mybackend
1990 mode tcp
1991 balance roundrobin
1992
1993 email-alert mailers mymailers
1994 email-alert from test1@horms.org
1995 email-alert to test2@horms.org
1996
1997 server srv1 192.168.0.30:80
1998 server srv2 192.168.0.31:80
1999
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002000timeout mail <time>
2001 Defines the time available for a mail/connection to be made and send to
2002 the mail-server. If not defined the default value is 10 seconds. To allow
2003 for at least two SYN-ACK packets to be send during initial TCP handshake it
2004 is advised to keep this value above 4 seconds.
2005
2006 Example:
2007 mailers mymailers
2008 timeout mail 20s
2009 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002010
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020114. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002012----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002013
Willy Tarreau6a06a402007-07-15 20:15:28 +02002014Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002015 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002016 - frontend <name>
2017 - backend <name>
2018 - listen <name>
2019
2020A "defaults" section sets default parameters for all other sections following
2021its declaration. Those default parameters are reset by the next "defaults"
2022section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002023section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002024
2025A "frontend" section describes a set of listening sockets accepting client
2026connections.
2027
2028A "backend" section describes a set of servers to which the proxy will connect
2029to forward incoming connections.
2030
2031A "listen" section defines a complete proxy with its frontend and backend
2032parts combined in one section. It is generally useful for TCP-only traffic.
2033
Willy Tarreau0ba27502007-12-24 16:55:16 +01002034All proxy names must be formed from upper and lower case letters, digits,
2035'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2036case-sensitive, which means that "www" and "WWW" are two different proxies.
2037
2038Historically, all proxy names could overlap, it just caused troubles in the
2039logs. Since the introduction of content switching, it is mandatory that two
2040proxies with overlapping capabilities (frontend/backend) have different names.
2041However, it is still permitted that a frontend and a backend share the same
2042name, as this configuration seems to be commonly encountered.
2043
2044Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2045and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002046bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002047protocol, and can interact with it by allowing, blocking, switching, adding,
2048modifying, or removing arbitrary contents in requests or responses, based on
2049arbitrary criteria.
2050
Willy Tarreau70dffda2014-01-30 03:07:23 +01002051In HTTP mode, the processing applied to requests and responses flowing over
2052a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002053the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002054
2055 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2056 requests and responses are processed, and connections remain open but idle
2057 between responses and new requests.
2058
2059 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2060 1.0 to 1.5-dev21 : only the first request and response are processed, and
2061 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet4212a302018-09-21 10:42:19 +02002062 be used as it creates lots of trouble with logging and HTTP processing. It
2063 is supported only on frontends.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002064
Willy Tarreau70dffda2014-01-30 03:07:23 +01002065 - SCL: server close ("option http-server-close") : the server-facing
2066 connection is closed after the end of the response is received, but the
2067 client-facing connection remains open.
2068
Christopher Faulet315b39c2018-09-21 16:26:19 +02002069 - CLO: close ("option httpclose"): the connection is closed after the end of
2070 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002071
2072The effective mode that will be applied to a connection passing through a
2073frontend and a backend can be determined by both proxy modes according to the
2074following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002075weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002076
Christopher Faulet315b39c2018-09-21 16:26:19 +02002077 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002078
Christopher Faulet315b39c2018-09-21 16:26:19 +02002079 | KAL | SCL | CLO
2080 ----+-----+-----+----
2081 KAL | KAL | SCL | CLO
2082 ----+-----+-----+----
2083 TUN | TUN | SCL | CLO
2084 Frontend ----+-----+-----+----
2085 mode SCL | SCL | SCL | CLO
2086 ----+-----+-----+----
2087 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002088
Willy Tarreau0ba27502007-12-24 16:55:16 +01002089
Willy Tarreau70dffda2014-01-30 03:07:23 +01002090
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020914.1. Proxy keywords matrix
2092--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002093
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002094The following list of keywords is supported. Most of them may only be used in a
2095limited set of section types. Some of them are marked as "deprecated" because
2096they are inherited from an old syntax which may be confusing or functionally
2097limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002098marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002099option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002100and must be disabled for a specific instance. Such options may also be prefixed
2101with "default" in order to restore default settings regardless of what has been
2102specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002103
Willy Tarreau6a06a402007-07-15 20:15:28 +02002104
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002105 keyword defaults frontend listen backend
2106------------------------------------+----------+----------+---------+---------
2107acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002108appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002109backlog X X X -
2110balance X - X X
2111bind - X X -
2112bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002113block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002114capture cookie - X X -
2115capture request header - X X -
2116capture response header - X X -
2117clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002118compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002119contimeout (deprecated) X - X X
2120cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002121declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002122default-server X - X X
2123default_backend X X X -
2124description - X X X
2125disabled X X X X
2126dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002127email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002128email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002129email-alert mailers X X X X
2130email-alert myhostname X X X X
2131email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002132enabled X X X X
2133errorfile X X X X
2134errorloc X X X X
2135errorloc302 X X X X
2136-- keyword -------------------------- defaults - frontend - listen -- backend -
2137errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002138force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002139filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002140fullconn X - X X
2141grace X X X X
2142hash-type X - X X
2143http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002144http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002145http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002146http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002147http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002148http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002149http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002150id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002151ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002152load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002153log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002154log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002155log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002156log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002157max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002158maxconn X X X -
2159mode X X X X
2160monitor fail - X X -
2161monitor-net X X X -
2162monitor-uri X X X -
2163option abortonclose (*) X - X X
2164option accept-invalid-http-request (*) X X X -
2165option accept-invalid-http-response (*) X - X X
2166option allbackups (*) X - X X
2167option checkcache (*) X - X X
2168option clitcpka (*) X X X -
2169option contstats (*) X X X -
2170option dontlog-normal (*) X X X -
2171option dontlognull (*) X X X -
Christopher Faulet315b39c2018-09-21 16:26:19 +02002172option forceclose (deprectated) (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002173-- keyword -------------------------- defaults - frontend - listen -- backend -
2174option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002175option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002176option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002177option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002178option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002179option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002180option http-server-close (*) X X X X
Christopher Faulet4212a302018-09-21 10:42:19 +02002181option http-tunnel (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002182option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002183option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002184option httpchk X - X X
2185option httpclose (*) X X X X
2186option httplog X X X X
2187option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002188option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002189option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002190option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002191option log-health-checks (*) X - X X
2192option log-separate-errors (*) X X X -
2193option logasap (*) X X X -
2194option mysql-check X - X X
2195option nolinger (*) X X X X
2196option originalto X X X X
2197option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002198option pgsql-check X - X X
2199option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002200option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002201option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002202option smtpchk X - X X
2203option socket-stats (*) X X X -
2204option splice-auto (*) X X X X
2205option splice-request (*) X X X X
2206option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002207option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002208option srvtcpka (*) X - X X
2209option ssl-hello-chk X - X X
2210-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002211option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002212option tcp-smart-accept (*) X X X -
2213option tcp-smart-connect (*) X - X X
2214option tcpka X X X X
2215option tcplog X X X X
2216option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002217external-check command X - X X
2218external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002219persist rdp-cookie X - X X
2220rate-limit sessions X X X -
2221redirect - X X X
2222redisp (deprecated) X - X X
2223redispatch (deprecated) X - X X
2224reqadd - X X X
2225reqallow - X X X
2226reqdel - X X X
2227reqdeny - X X X
2228reqiallow - X X X
2229reqidel - X X X
2230reqideny - X X X
2231reqipass - X X X
2232reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002233reqitarpit - X X X
2234reqpass - X X X
2235reqrep - X X X
2236-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002237reqtarpit - X X X
2238retries X - X X
2239rspadd - X X X
2240rspdel - X X X
2241rspdeny - X X X
2242rspidel - X X X
2243rspideny - X X X
2244rspirep - X X X
2245rsprep - X X X
2246server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002247server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002248server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002249source X - X X
2250srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002251stats admin - X X X
2252stats auth X X X X
2253stats enable X X X X
2254stats hide-version X X X X
2255stats http-request - X X X
2256stats realm X X X X
2257stats refresh X X X X
2258stats scope X X X X
2259stats show-desc X X X X
2260stats show-legends X X X X
2261stats show-node X X X X
2262stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002263-- keyword -------------------------- defaults - frontend - listen -- backend -
2264stick match - - X X
2265stick on - - X X
2266stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002267stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002268stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002269tcp-check connect - - X X
2270tcp-check expect - - X X
2271tcp-check send - - X X
2272tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002273tcp-request connection - X X -
2274tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002275tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002276tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002277tcp-response content - - X X
2278tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002279timeout check X - X X
2280timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002281timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002282timeout clitimeout (deprecated) X X X -
2283timeout connect X - X X
2284timeout contimeout (deprecated) X - X X
2285timeout http-keep-alive X X X X
2286timeout http-request X X X X
2287timeout queue X - X X
2288timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002289timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002290timeout srvtimeout (deprecated) X - X X
2291timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002292timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002293transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002294unique-id-format X X X -
2295unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002296use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002297use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002298------------------------------------+----------+----------+---------+---------
2299 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002300
Willy Tarreau0ba27502007-12-24 16:55:16 +01002301
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023024.2. Alphabetically sorted keywords reference
2303---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002304
2305This section provides a description of each keyword and its usage.
2306
2307
2308acl <aclname> <criterion> [flags] [operator] <value> ...
2309 Declare or complete an access list.
2310 May be used in sections : defaults | frontend | listen | backend
2311 no | yes | yes | yes
2312 Example:
2313 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2314 acl invalid_src src_port 0:1023
2315 acl local_dst hdr(host) -i localhost
2316
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002317 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002318
2319
Cyril Bontéb21570a2009-11-29 20:04:48 +01002320appsession <cookie> len <length> timeout <holdtime>
2321 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002322 Define session stickiness on an existing application cookie.
2323 May be used in sections : defaults | frontend | listen | backend
2324 no | no | yes | yes
2325 Arguments :
2326 <cookie> this is the name of the cookie used by the application and which
2327 HAProxy will have to learn for each new session.
2328
Cyril Bontéb21570a2009-11-29 20:04:48 +01002329 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002330 checked in each cookie value.
2331
2332 <holdtime> this is the time after which the cookie will be removed from
2333 memory if unused. If no unit is specified, this time is in
2334 milliseconds.
2335
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002336 request-learn
2337 If this option is specified, then haproxy will be able to learn
2338 the cookie found in the request in case the server does not
2339 specify any in response. This is typically what happens with
2340 PHPSESSID cookies, or when haproxy's session expires before
2341 the application's session and the correct server is selected.
2342 It is recommended to specify this option to improve reliability.
2343
Cyril Bontéb21570a2009-11-29 20:04:48 +01002344 prefix When this option is specified, haproxy will match on the cookie
2345 prefix (or URL parameter prefix). The appsession value is the
2346 data following this prefix.
2347
2348 Example :
2349 appsession ASPSESSIONID len 64 timeout 3h prefix
2350
Davor Ocelice9ed2812017-12-25 17:49:28 +01002351 This will match the cookie ASPSESSIONIDXXX=XXXX,
2352 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002353
2354 mode This option allows to change the URL parser mode.
2355 2 modes are currently supported :
2356 - path-parameters :
2357 The parser looks for the appsession in the path parameters
2358 part (each parameter is separated by a semi-colon), which is
2359 convenient for JSESSIONID for example.
2360 This is the default mode if the option is not set.
2361 - query-string :
2362 In this mode, the parser will look for the appsession in the
2363 query string.
2364
Willy Tarreau294d0f02015-08-10 19:40:12 +02002365 As of version 1.6, appsessions was removed. It is more flexible and more
2366 convenient to use stick-tables instead, and stick-tables support multi-master
2367 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002368
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002369 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2370 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002371
2372
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002373backlog <conns>
2374 Give hints to the system about the approximate listen backlog desired size
2375 May be used in sections : defaults | frontend | listen | backend
2376 yes | yes | yes | no
2377 Arguments :
2378 <conns> is the number of pending connections. Depending on the operating
2379 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002380 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002381
2382 In order to protect against SYN flood attacks, one solution is to increase
2383 the system's SYN backlog size. Depending on the system, sometimes it is just
2384 tunable via a system parameter, sometimes it is not adjustable at all, and
2385 sometimes the system relies on hints given by the application at the time of
2386 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2387 to the listen() syscall. On systems which can make use of this value, it can
2388 sometimes be useful to be able to specify a different value, hence this
2389 backlog parameter.
2390
2391 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2392 used as a hint and the system accepts up to the smallest greater power of
2393 two, and never more than some limits (usually 32768).
2394
2395 See also : "maxconn" and the target operating system's tuning guide.
2396
2397
Willy Tarreau0ba27502007-12-24 16:55:16 +01002398balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002399balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002400 Define the load balancing algorithm to be used in a backend.
2401 May be used in sections : defaults | frontend | listen | backend
2402 yes | no | yes | yes
2403 Arguments :
2404 <algorithm> is the algorithm used to select a server when doing load
2405 balancing. This only applies when no persistence information
2406 is available, or when a connection is redispatched to another
2407 server. <algorithm> may be one of the following :
2408
2409 roundrobin Each server is used in turns, according to their weights.
2410 This is the smoothest and fairest algorithm when the server's
2411 processing time remains equally distributed. This algorithm
2412 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002413 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002414 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002415 large farms, when a server becomes up after having been down
2416 for a very short time, it may sometimes take a few hundreds
2417 requests for it to be re-integrated into the farm and start
2418 receiving traffic. This is normal, though very rare. It is
2419 indicated here in case you would have the chance to observe
2420 it, so that you don't worry.
2421
2422 static-rr Each server is used in turns, according to their weights.
2423 This algorithm is as similar to roundrobin except that it is
2424 static, which means that changing a server's weight on the
2425 fly will have no effect. On the other hand, it has no design
2426 limitation on the number of servers, and when a server goes
2427 up, it is always immediately reintroduced into the farm, once
2428 the full map is recomputed. It also uses slightly less CPU to
2429 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002430
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002431 leastconn The server with the lowest number of connections receives the
2432 connection. Round-robin is performed within groups of servers
2433 of the same load to ensure that all servers will be used. Use
2434 of this algorithm is recommended where very long sessions are
2435 expected, such as LDAP, SQL, TSE, etc... but is not very well
2436 suited for protocols using short sessions such as HTTP. This
2437 algorithm is dynamic, which means that server weights may be
2438 adjusted on the fly for slow starts for instance.
2439
Willy Tarreauf09c6602012-02-13 17:12:08 +01002440 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002441 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002442 identifier to the highest (see server parameter "id"), which
2443 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002444 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002445 not make sense to use this algorithm without setting maxconn.
2446 The purpose of this algorithm is to always use the smallest
2447 number of servers so that extra servers can be powered off
2448 during non-intensive hours. This algorithm ignores the server
2449 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002450 or IMAP than HTTP, though it can be useful there too. In
2451 order to use this algorithm efficiently, it is recommended
2452 that a cloud controller regularly checks server usage to turn
2453 them off when unused, and regularly checks backend queue to
2454 turn new servers on when the queue inflates. Alternatively,
2455 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002456
Willy Tarreau0ba27502007-12-24 16:55:16 +01002457 source The source IP address is hashed and divided by the total
2458 weight of the running servers to designate which server will
2459 receive the request. This ensures that the same client IP
2460 address will always reach the same server as long as no
2461 server goes down or up. If the hash result changes due to the
2462 number of running servers changing, many clients will be
2463 directed to a different server. This algorithm is generally
2464 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002465 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002466 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002467 static by default, which means that changing a server's
2468 weight on the fly will have no effect, but this can be
2469 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002470
Oskar Stolc8dc41842012-05-19 10:19:54 +01002471 uri This algorithm hashes either the left part of the URI (before
2472 the question mark) or the whole URI (if the "whole" parameter
2473 is present) and divides the hash value by the total weight of
2474 the running servers. The result designates which server will
2475 receive the request. This ensures that the same URI will
2476 always be directed to the same server as long as no server
2477 goes up or down. This is used with proxy caches and
2478 anti-virus proxies in order to maximize the cache hit rate.
2479 Note that this algorithm may only be used in an HTTP backend.
2480 This algorithm is static by default, which means that
2481 changing a server's weight on the fly will have no effect,
2482 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002483
Oskar Stolc8dc41842012-05-19 10:19:54 +01002484 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002485 "depth", both followed by a positive integer number. These
2486 options may be helpful when it is needed to balance servers
2487 based on the beginning of the URI only. The "len" parameter
2488 indicates that the algorithm should only consider that many
2489 characters at the beginning of the URI to compute the hash.
2490 Note that having "len" set to 1 rarely makes sense since most
2491 URIs start with a leading "/".
2492
2493 The "depth" parameter indicates the maximum directory depth
2494 to be used to compute the hash. One level is counted for each
2495 slash in the request. If both parameters are specified, the
2496 evaluation stops when either is reached.
2497
Willy Tarreau0ba27502007-12-24 16:55:16 +01002498 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002499 the query string of each HTTP GET request.
2500
2501 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002502 request entity will be searched for the parameter argument,
2503 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002504 ('?') in the URL. The message body will only start to be
2505 analyzed once either the advertised amount of data has been
2506 received or the request buffer is full. In the unlikely event
2507 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002508 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002509 be randomly balanced if at all. This keyword used to support
2510 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002511
2512 If the parameter is found followed by an equal sign ('=') and
2513 a value, then the value is hashed and divided by the total
2514 weight of the running servers. The result designates which
2515 server will receive the request.
2516
2517 This is used to track user identifiers in requests and ensure
2518 that a same user ID will always be sent to the same server as
2519 long as no server goes up or down. If no value is found or if
2520 the parameter is not found, then a round robin algorithm is
2521 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002522 backend. This algorithm is static by default, which means
2523 that changing a server's weight on the fly will have no
2524 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002525
Cyril Bontédc4d9032012-04-08 21:57:39 +02002526 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2527 request. Just as with the equivalent ACL 'hdr()' function,
2528 the header name in parenthesis is not case sensitive. If the
2529 header is absent or if it does not contain any value, the
2530 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002531
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002532 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002533 reducing the hash algorithm to the main domain part with some
2534 specific headers such as 'Host'. For instance, in the Host
2535 value "haproxy.1wt.eu", only "1wt" will be considered.
2536
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002537 This algorithm is static by default, which means that
2538 changing a server's weight on the fly will have no effect,
2539 but this can be changed using "hash-type".
2540
Willy Tarreau760e81d2018-05-03 07:20:40 +02002541 random A random number will be used as the key for the consistent
2542 hashing function. This means that the servers' weights are
2543 respected, dynamic weight changes immediately take effect, as
2544 well as new server additions. Random load balancing can be
2545 useful with large farms or when servers are frequently added
2546 or removed. The hash-balance-factor directive can be used to
2547 further improve fairness of the load balancing, especially
2548 in situations where servers show highly variable response
2549 times.
2550
Emeric Brun736aa232009-06-30 17:56:00 +02002551 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002552 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002553 The RDP cookie <name> (or "mstshash" if omitted) will be
2554 looked up and hashed for each incoming TCP request. Just as
2555 with the equivalent ACL 'req_rdp_cookie()' function, the name
2556 is not case-sensitive. This mechanism is useful as a degraded
2557 persistence mode, as it makes it possible to always send the
2558 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002559 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002560 used instead.
2561
2562 Note that for this to work, the frontend must ensure that an
2563 RDP cookie is already present in the request buffer. For this
2564 you must use 'tcp-request content accept' rule combined with
2565 a 'req_rdp_cookie_cnt' ACL.
2566
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002567 This algorithm is static by default, which means that
2568 changing a server's weight on the fly will have no effect,
2569 but this can be changed using "hash-type".
2570
Cyril Bontédc4d9032012-04-08 21:57:39 +02002571 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002572
Willy Tarreau0ba27502007-12-24 16:55:16 +01002573 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002574 algorithms. Right now, only "url_param" and "uri" support an
2575 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002576
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002577 The load balancing algorithm of a backend is set to roundrobin when no other
2578 algorithm, mode nor option have been set. The algorithm may only be set once
2579 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002580
Lukas Tribus80512b12018-10-27 20:07:40 +02002581 With authentication schemes that require the same connection like NTLM, URI
2582 based alghoritms must not be used, as they would cause subsequent requests
2583 to be routed to different backend servers, breaking the invalid assumptions
2584 NTLM relies on.
2585
Willy Tarreau0ba27502007-12-24 16:55:16 +01002586 Examples :
2587 balance roundrobin
2588 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002589 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002590 balance hdr(User-Agent)
2591 balance hdr(host)
2592 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002593
2594 Note: the following caveats and limitations on using the "check_post"
2595 extension with "url_param" must be considered :
2596
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002597 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002598 to determine if the parameters will be found in the body or entity which
2599 may contain binary data. Therefore another method may be required to
2600 restrict consideration of POST requests that have no URL parameters in
2601 the body. (see acl reqideny http_end)
2602
2603 - using a <max_wait> value larger than the request buffer size does not
2604 make sense and is useless. The buffer size is set at build time, and
2605 defaults to 16 kB.
2606
2607 - Content-Encoding is not supported, the parameter search will probably
2608 fail; and load balancing will fall back to Round Robin.
2609
2610 - Expect: 100-continue is not supported, load balancing will fall back to
2611 Round Robin.
2612
Lukas Tribus23953682017-04-28 13:24:30 +00002613 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002614 If the entire parameter value is not present in the first chunk, the
2615 selection of server is undefined (actually, defined by how little
2616 actually appeared in the first chunk).
2617
2618 - This feature does not support generation of a 100, 411 or 501 response.
2619
2620 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002621 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002622 white space or control characters are found, indicating the end of what
2623 might be a URL parameter list. This is probably not a concern with SGML
2624 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002625
Willy Tarreau294d0f02015-08-10 19:40:12 +02002626 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002627
2628
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002629bind [<address>]:<port_range> [, ...] [param*]
2630bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002631 Define one or several listening addresses and/or ports in a frontend.
2632 May be used in sections : defaults | frontend | listen | backend
2633 no | yes | yes | no
2634 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002635 <address> is optional and can be a host name, an IPv4 address, an IPv6
2636 address, or '*'. It designates the address the frontend will
2637 listen on. If unset, all IPv4 addresses of the system will be
2638 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002639 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002640 Optionally, an address family prefix may be used before the
2641 address to force the family regardless of the address format,
2642 which can be useful to specify a path to a unix socket with
2643 no slash ('/'). Currently supported prefixes are :
2644 - 'ipv4@' -> address is always IPv4
2645 - 'ipv6@' -> address is always IPv6
2646 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002647 - 'abns@' -> address is in abstract namespace (Linux only).
2648 Note: since abstract sockets are not "rebindable", they
2649 do not cope well with multi-process mode during
2650 soft-restart, so it is better to avoid them if
2651 nbproc is greater than 1. The effect is that if the
2652 new process fails to start, only one of the old ones
2653 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002654 - 'fd@<n>' -> use file descriptor <n> inherited from the
2655 parent. The fd must be bound and may or may not already
2656 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002657 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2658 connected unix socket or of a socketpair. The bind waits
2659 to receive a FD over the unix socket and uses it as if it
2660 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002661 You may want to reference some environment variables in the
2662 address parameter, see section 2.3 about environment
2663 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002664
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002665 <port_range> is either a unique TCP port, or a port range for which the
2666 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002667 above. The port is mandatory for TCP listeners. Note that in
2668 the case of an IPv6 address, the port is always the number
2669 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002670 - a numerical port (ex: '80')
2671 - a dash-delimited ports range explicitly stating the lower
2672 and upper bounds (ex: '2000-2100') which are included in
2673 the range.
2674
2675 Particular care must be taken against port ranges, because
2676 every <address:port> couple consumes one socket (= a file
2677 descriptor), so it's easy to consume lots of descriptors
2678 with a simple range, and to run out of sockets. Also, each
2679 <address:port> couple must be used only once among all
2680 instances running on a same system. Please note that binding
2681 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002682 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002683 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002684
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002685 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002686 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002687 receive UNIX connections on the socket located at this place.
2688 The path must begin with a slash and by default is absolute.
2689 It can be relative to the prefix defined by "unix-bind" in
2690 the global section. Note that the total length of the prefix
2691 followed by the socket path cannot exceed some system limits
2692 for UNIX sockets, which commonly are set to 107 characters.
2693
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002694 <param*> is a list of parameters common to all sockets declared on the
2695 same line. These numerous parameters depend on OS and build
2696 options and have a complete section dedicated to them. Please
2697 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002698
Willy Tarreau0ba27502007-12-24 16:55:16 +01002699 It is possible to specify a list of address:port combinations delimited by
2700 commas. The frontend will then listen on all of these addresses. There is no
2701 fixed limit to the number of addresses and ports which can be listened on in
2702 a frontend, as well as there is no limit to the number of "bind" statements
2703 in a frontend.
2704
2705 Example :
2706 listen http_proxy
2707 bind :80,:443
2708 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002709 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002710
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002711 listen http_https_proxy
2712 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002713 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002714
Willy Tarreau24709282013-03-10 21:32:12 +01002715 listen http_https_proxy_explicit
2716 bind ipv6@:80
2717 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2718 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2719
Willy Tarreaudad36a32013-03-11 01:20:04 +01002720 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002721 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002722
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002723 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2724 sun_path length is used for the address length. Some other programs
2725 such as socat use the string length only by default. Pass the option
2726 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2727 make it compatible with HAProxy's.
2728
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002729 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002730 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002731
2732
Christopher Fauletff4121f2017-11-22 16:38:49 +01002733bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002734 Limit visibility of an instance to a certain set of processes numbers.
2735 May be used in sections : defaults | frontend | listen | backend
2736 yes | yes | yes | yes
2737 Arguments :
2738 all All process will see this instance. This is the default. It
2739 may be used to override a default value.
2740
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002741 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002742 option may be combined with other numbers.
2743
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002744 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002745 option may be combined with other numbers. Do not use it
2746 with less than 2 processes otherwise some instances might be
2747 missing from all processes.
2748
Christopher Fauletff4121f2017-11-22 16:38:49 +01002749 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002750 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002751 the machine's word size. Ranges can be partially defined. The
2752 higher bound can be omitted. In such case, it is replaced by
2753 the corresponding maximum value. If a proxy is bound to
2754 process numbers greater than the configured global.nbproc, it
2755 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002756 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002757
2758 This keyword limits binding of certain instances to certain processes. This
2759 is useful in order not to have too many processes listening to the same
2760 ports. For instance, on a dual-core machine, it might make sense to set
2761 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2762 and 'even' instances.
2763
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002764 At the moment, it is not possible to reference more than 32 or 64 processes
2765 using this keyword, but this should be more than enough for most setups.
2766 Please note that 'all' really means all processes regardless of the machine's
2767 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002768
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002769 Each "bind" line may further be limited to a subset of the proxy's processes,
2770 please consult the "process" bind keyword in section 5.1.
2771
Willy Tarreaub369a042014-09-16 13:21:03 +02002772 When a frontend has no explicit "bind-process" line, it tries to bind to all
2773 the processes referenced by its "bind" lines. That means that frontends can
2774 easily adapt to their listeners' processes.
2775
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002776 If some backends are referenced by frontends bound to other processes, the
2777 backend automatically inherits the frontend's processes.
2778
2779 Example :
2780 listen app_ip1
2781 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002782 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002783
2784 listen app_ip2
2785 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002786 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002787
2788 listen management
2789 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002790 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002791
Willy Tarreau110ecc12012-11-15 17:50:01 +01002792 listen management
2793 bind 10.0.0.4:80
2794 bind-process 1-4
2795
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002796 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002797
2798
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002799block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002800 Block a layer 7 request if/unless a condition is matched
2801 May be used in sections : defaults | frontend | listen | backend
2802 no | yes | yes | yes
2803
2804 The HTTP request will be blocked very early in the layer 7 processing
2805 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002806 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002807 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002808 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002809 "block" statements per instance. To block connections at layer 4 (without
2810 sending a 403 error) see "tcp-request connection reject" and
2811 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002812
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002813 This form is deprecated, do not use it in any new configuration, use the new
2814 "http-request deny" instead.
2815
Willy Tarreau0ba27502007-12-24 16:55:16 +01002816 Example:
2817 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2818 acl invalid_src src_port 0:1023
2819 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002820 # block is deprecated. Use http-request deny instead:
2821 #block if invalid_src || local_dst
2822 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002823
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002824 See also : section 7 about ACL usage, "http-request deny",
2825 "http-response deny", "tcp-request connection reject" and
2826 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002827
2828capture cookie <name> len <length>
2829 Capture and log a cookie in the request and in the response.
2830 May be used in sections : defaults | frontend | listen | backend
2831 no | yes | yes | no
2832 Arguments :
2833 <name> is the beginning of the name of the cookie to capture. In order
2834 to match the exact name, simply suffix the name with an equal
2835 sign ('='). The full name will appear in the logs, which is
2836 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002837 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002838
2839 <length> is the maximum number of characters to report in the logs, which
2840 include the cookie name, the equal sign and the value, all in the
2841 standard "name=value" form. The string will be truncated on the
2842 right if it exceeds <length>.
2843
2844 Only the first cookie is captured. Both the "cookie" request headers and the
2845 "set-cookie" response headers are monitored. This is particularly useful to
2846 check for application bugs causing session crossing or stealing between
2847 users, because generally the user's cookies can only change on a login page.
2848
2849 When the cookie was not presented by the client, the associated log column
2850 will report "-". When a request does not cause a cookie to be assigned by the
2851 server, a "-" is reported in the response column.
2852
2853 The capture is performed in the frontend only because it is necessary that
2854 the log format does not change for a given frontend depending on the
2855 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002856 "capture cookie" statement in a frontend. The maximum capture length is set
2857 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2858 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002859
2860 Example:
2861 capture cookie ASPSESSION len 32
2862
2863 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002864 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002865
2866
2867capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002868 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002869 May be used in sections : defaults | frontend | listen | backend
2870 no | yes | yes | no
2871 Arguments :
2872 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002873 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002874 appear in the requests, with the first letter of each word in
2875 upper case. The header name will not appear in the logs, only the
2876 value is reported, but the position in the logs is respected.
2877
2878 <length> is the maximum number of characters to extract from the value and
2879 report in the logs. The string will be truncated on the right if
2880 it exceeds <length>.
2881
Willy Tarreau4460d032012-11-21 23:37:37 +01002882 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002883 value will be added to the logs between braces ('{}'). If multiple headers
2884 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002885 in the same order they were declared in the configuration. Non-existent
2886 headers will be logged just as an empty string. Common uses for request
2887 header captures include the "Host" field in virtual hosting environments, the
2888 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002889 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002890 environments to find where the request came from.
2891
2892 Note that when capturing headers such as "User-agent", some spaces may be
2893 logged, making the log analysis more difficult. Thus be careful about what
2894 you log if you know your log parser is not smart enough to rely on the
2895 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002896
Willy Tarreau0900abb2012-11-22 00:21:46 +01002897 There is no limit to the number of captured request headers nor to their
2898 length, though it is wise to keep them low to limit memory usage per session.
2899 In order to keep log format consistent for a same frontend, header captures
2900 can only be declared in a frontend. It is not possible to specify a capture
2901 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002902
2903 Example:
2904 capture request header Host len 15
2905 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002906 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002907
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002908 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002909 about logging.
2910
2911
2912capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002913 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002914 May be used in sections : defaults | frontend | listen | backend
2915 no | yes | yes | no
2916 Arguments :
2917 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002918 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002919 appear in the response, with the first letter of each word in
2920 upper case. The header name will not appear in the logs, only the
2921 value is reported, but the position in the logs is respected.
2922
2923 <length> is the maximum number of characters to extract from the value and
2924 report in the logs. The string will be truncated on the right if
2925 it exceeds <length>.
2926
Willy Tarreau4460d032012-11-21 23:37:37 +01002927 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002928 result will be added to the logs between braces ('{}') after the captured
2929 request headers. If multiple headers are captured, they will be delimited by
2930 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002931 the configuration. Non-existent headers will be logged just as an empty
2932 string. Common uses for response header captures include the "Content-length"
2933 header which indicates how many bytes are expected to be returned, the
2934 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002935
Willy Tarreau0900abb2012-11-22 00:21:46 +01002936 There is no limit to the number of captured response headers nor to their
2937 length, though it is wise to keep them low to limit memory usage per session.
2938 In order to keep log format consistent for a same frontend, header captures
2939 can only be declared in a frontend. It is not possible to specify a capture
2940 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002941
2942 Example:
2943 capture response header Content-length len 9
2944 capture response header Location len 15
2945
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002946 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002947 about logging.
2948
2949
Cyril Bontéf0c60612010-02-06 14:44:47 +01002950clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002951 Set the maximum inactivity time on the client side.
2952 May be used in sections : defaults | frontend | listen | backend
2953 yes | yes | yes | no
2954 Arguments :
2955 <timeout> is the timeout value is specified in milliseconds by default, but
2956 can be in any other unit if the number is suffixed by the unit,
2957 as explained at the top of this document.
2958
2959 The inactivity timeout applies when the client is expected to acknowledge or
2960 send data. In HTTP mode, this timeout is particularly important to consider
2961 during the first phase, when the client sends the request, and during the
2962 response while it is reading data sent by the server. The value is specified
2963 in milliseconds by default, but can be in any other unit if the number is
2964 suffixed by the unit, as specified at the top of this document. In TCP mode
2965 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2966 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002967 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002968 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002969 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002970
2971 This parameter is specific to frontends, but can be specified once for all in
2972 "defaults" sections. This is in fact one of the easiest solutions not to
2973 forget about it. An unspecified timeout results in an infinite timeout, which
2974 is not recommended. Such a usage is accepted and works but reports a warning
2975 during startup because it may results in accumulation of expired sessions in
2976 the system if the system's timeouts are not configured either.
2977
2978 This parameter is provided for compatibility but is currently deprecated.
2979 Please use "timeout client" instead.
2980
Willy Tarreau036fae02008-01-06 13:24:40 +01002981 See also : "timeout client", "timeout http-request", "timeout server", and
2982 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002983
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002984compression algo <algorithm> ...
2985compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002986compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002987 Enable HTTP compression.
2988 May be used in sections : defaults | frontend | listen | backend
2989 yes | yes | yes | yes
2990 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002991 algo is followed by the list of supported compression algorithms.
2992 type is followed by the list of MIME types that will be compressed.
2993 offload makes haproxy work as a compression offloader only (see notes).
2994
2995 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002996 identity this is mostly for debugging, and it was useful for developing
2997 the compression feature. Identity does not apply any change on
2998 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002999
Willy Tarreauc91840a2015-03-28 17:00:39 +01003000 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003001 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003002
3003 deflate same as "gzip", but with deflate algorithm and zlib format.
3004 Note that this algorithm has ambiguous support on many
3005 browsers and no support at all from recent ones. It is
3006 strongly recommended not to use it for anything else than
3007 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003008 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003009
Willy Tarreauc91840a2015-03-28 17:00:39 +01003010 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3011 alternative when the browser wants "deflate". All major
3012 browsers understand it and despite violating the standards,
3013 it is known to work better than "deflate", at least on MSIE
3014 and some versions of Safari. Do not use it in conjunction
3015 with "deflate", use either one or the other since both react
3016 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003017 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003018
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003019 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003020 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003021 If backend servers support HTTP compression, these directives
3022 will be no-op: haproxy will see the compressed response and will not
3023 compress again. If backend servers do not support HTTP compression and
3024 there is Accept-Encoding header in request, haproxy will compress the
3025 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003026
3027 The "offload" setting makes haproxy remove the Accept-Encoding header to
3028 prevent backend servers from compressing responses. It is strongly
3029 recommended not to do this because this means that all the compression work
3030 will be done on the single point where haproxy is located. However in some
3031 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003032 with broken HTTP compression implementation which can't be turned off.
3033 In that case haproxy can be used to prevent that gateway from emitting
3034 invalid payloads. In this case, simply removing the header in the
3035 configuration does not work because it applies before the header is parsed,
3036 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003037 then be used for such scenarios. Note: for now, the "offload" setting is
3038 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003039
William Lallemand05097442012-11-20 12:14:28 +01003040 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003041 * the request does not advertise a supported compression algorithm in the
3042 "Accept-Encoding" header
3043 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01003044 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01003045 * response header "Transfer-Encoding" contains "chunked" (Temporary
3046 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003047 * response contain neither a "Content-Length" header nor a
3048 "Transfer-Encoding" whose last value is "chunked"
3049 * response contains a "Content-Type" header whose first value starts with
3050 "multipart"
3051 * the response contains the "no-transform" value in the "Cache-control"
3052 header
3053 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3054 and later
3055 * The response contains a "Content-Encoding" header, indicating that the
3056 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01003057
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003058 Note: The compression does not rewrite Etag headers, and does not emit the
3059 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003060
William Lallemand82fe75c2012-10-23 10:25:10 +02003061 Examples :
3062 compression algo gzip
3063 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003064
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003065
Cyril Bontéf0c60612010-02-06 14:44:47 +01003066contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003067 Set the maximum time to wait for a connection attempt to a server to succeed.
3068 May be used in sections : defaults | frontend | listen | backend
3069 yes | no | yes | yes
3070 Arguments :
3071 <timeout> is the timeout value is specified in milliseconds by default, but
3072 can be in any other unit if the number is suffixed by the unit,
3073 as explained at the top of this document.
3074
3075 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003076 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003077 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003078 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003079 connect timeout also presets the queue timeout to the same value if this one
3080 has not been specified. Historically, the contimeout was also used to set the
3081 tarpit timeout in a listen section, which is not possible in a pure frontend.
3082
3083 This parameter is specific to backends, but can be specified once for all in
3084 "defaults" sections. This is in fact one of the easiest solutions not to
3085 forget about it. An unspecified timeout results in an infinite timeout, which
3086 is not recommended. Such a usage is accepted and works but reports a warning
3087 during startup because it may results in accumulation of failed sessions in
3088 the system if the system's timeouts are not configured either.
3089
3090 This parameter is provided for backwards compatibility but is currently
3091 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3092 instead.
3093
3094 See also : "timeout connect", "timeout queue", "timeout tarpit",
3095 "timeout server", "contimeout".
3096
3097
Willy Tarreau55165fe2009-05-10 12:02:55 +02003098cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003099 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3100 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003101 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003102 Enable cookie-based persistence in a backend.
3103 May be used in sections : defaults | frontend | listen | backend
3104 yes | no | yes | yes
3105 Arguments :
3106 <name> is the name of the cookie which will be monitored, modified or
3107 inserted in order to bring persistence. This cookie is sent to
3108 the client via a "Set-Cookie" header in the response, and is
3109 brought back by the client in a "Cookie" header in all requests.
3110 Special care should be taken to choose a name which does not
3111 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003112 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003113 HTTP/HTTPS), care should be taken to use different cookie names
3114 between all backends if persistence between them is not desired.
3115
3116 rewrite This keyword indicates that the cookie will be provided by the
3117 server and that haproxy will have to modify its value to set the
3118 server's identifier in it. This mode is handy when the management
3119 of complex combinations of "Set-cookie" and "Cache-control"
3120 headers is left to the application. The application can then
3121 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003122 cookie. Since all responses should be monitored, this mode
3123 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003124 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003125 start with this mode for new deployments. This keyword is
3126 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003127
3128 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003129 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003130
Willy Tarreaua79094d2010-08-31 22:54:15 +02003131 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003132 server. When used without the "preserve" option, if the server
3133 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003134 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003135 existing configurations running in the "rewrite" mode. The cookie
3136 will only be a session cookie and will not be stored on the
3137 client's disk. By default, unless the "indirect" option is added,
3138 the server will see the cookies emitted by the client. Due to
3139 caching effects, it is generally wise to add the "nocache" or
3140 "postonly" keywords (see below). The "insert" keyword is not
3141 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003142
3143 prefix This keyword indicates that instead of relying on a dedicated
3144 cookie for the persistence, an existing one will be completed.
3145 This may be needed in some specific environments where the client
3146 does not support more than one single cookie and the application
3147 already needs it. In this case, whenever the server sets a cookie
3148 named <name>, it will be prefixed with the server's identifier
3149 and a delimiter. The prefix will be removed from all client
3150 requests so that the server still finds the cookie it emitted.
3151 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003152 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003153 not compatible with "rewrite" and "insert". Note: it is highly
3154 recommended not to use "indirect" with "prefix", otherwise server
3155 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003156
Willy Tarreaua79094d2010-08-31 22:54:15 +02003157 indirect When this option is specified, no cookie will be emitted to a
3158 client which already has a valid one for the server which has
3159 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003160 it will be removed, unless the "preserve" option is also set. In
3161 "insert" mode, this will additionally remove cookies from the
3162 requests transmitted to the server, making the persistence
3163 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003164 Note: it is highly recommended not to use "indirect" with
3165 "prefix", otherwise server cookie updates would not be sent to
3166 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003167
3168 nocache This option is recommended in conjunction with the insert mode
3169 when there is a cache between the client and HAProxy, as it
3170 ensures that a cacheable response will be tagged non-cacheable if
3171 a cookie needs to be inserted. This is important because if all
3172 persistence cookies are added on a cacheable home page for
3173 instance, then all customers will then fetch the page from an
3174 outer cache and will all share the same persistence cookie,
3175 leading to one server receiving much more traffic than others.
3176 See also the "insert" and "postonly" options.
3177
3178 postonly This option ensures that cookie insertion will only be performed
3179 on responses to POST requests. It is an alternative to the
3180 "nocache" option, because POST responses are not cacheable, so
3181 this ensures that the persistence cookie will never get cached.
3182 Since most sites do not need any sort of persistence before the
3183 first POST which generally is a login request, this is a very
3184 efficient method to optimize caching without risking to find a
3185 persistence cookie in the cache.
3186 See also the "insert" and "nocache" options.
3187
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003188 preserve This option may only be used with "insert" and/or "indirect". It
3189 allows the server to emit the persistence cookie itself. In this
3190 case, if a cookie is found in the response, haproxy will leave it
3191 untouched. This is useful in order to end persistence after a
3192 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003193 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003194 the past. By combining this mechanism with the "disable-on-404"
3195 check option, it is possible to perform a completely graceful
3196 shutdown because users will definitely leave the server after
3197 they logout.
3198
Willy Tarreau4992dd22012-05-31 21:02:17 +02003199 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3200 when a cookie is inserted. This attribute is used so that a
3201 user agent doesn't share the cookie with non-HTTP components.
3202 Please check RFC6265 for more information on this attribute.
3203
3204 secure This option tells haproxy to add a "Secure" cookie attribute when
3205 a cookie is inserted. This attribute is used so that a user agent
3206 never emits this cookie over non-secure channels, which means
3207 that a cookie learned with this flag will be presented only over
3208 SSL/TLS connections. Please check RFC6265 for more information on
3209 this attribute.
3210
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003211 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003212 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003213 name. If the domain begins with a dot, the browser is allowed to
3214 use it for any host ending with that name. It is also possible to
3215 specify several domain names by invoking this option multiple
3216 times. Some browsers might have small limits on the number of
3217 domains, so be careful when doing that. For the record, sending
3218 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003219
Willy Tarreau996a92c2010-10-13 19:30:47 +02003220 maxidle This option allows inserted cookies to be ignored after some idle
3221 time. It only works with insert-mode cookies. When a cookie is
3222 sent to the client, the date this cookie was emitted is sent too.
3223 Upon further presentations of this cookie, if the date is older
3224 than the delay indicated by the parameter (in seconds), it will
3225 be ignored. Otherwise, it will be refreshed if needed when the
3226 response is sent to the client. This is particularly useful to
3227 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003228 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003229 this option is set and a cookie has no date, it is always
3230 accepted, but gets refreshed in the response. This maintains the
3231 ability for admins to access their sites. Cookies that have a
3232 date in the future further than 24 hours are ignored. Doing so
3233 lets admins fix timezone issues without risking kicking users off
3234 the site.
3235
3236 maxlife This option allows inserted cookies to be ignored after some life
3237 time, whether they're in use or not. It only works with insert
3238 mode cookies. When a cookie is first sent to the client, the date
3239 this cookie was emitted is sent too. Upon further presentations
3240 of this cookie, if the date is older than the delay indicated by
3241 the parameter (in seconds), it will be ignored. If the cookie in
3242 the request has no date, it is accepted and a date will be set.
3243 Cookies that have a date in the future further than 24 hours are
3244 ignored. Doing so lets admins fix timezone issues without risking
3245 kicking users off the site. Contrary to maxidle, this value is
3246 not refreshed, only the first visit date counts. Both maxidle and
3247 maxlife may be used at the time. This is particularly useful to
3248 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003249 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003250 is stronger than the maxidle method in that it forces a
3251 redispatch after some absolute delay.
3252
Olivier Houchard4e694042017-03-14 20:01:29 +01003253 dynamic Activate dynamic cookies. When used, a session cookie is
3254 dynamically created for each server, based on the IP and port
3255 of the server, and a secret key, specified in the
3256 "dynamic-cookie-key" backend directive.
3257 The cookie will be regenerated each time the IP address change,
3258 and is only generated for IPv4/IPv6.
3259
Willy Tarreau0ba27502007-12-24 16:55:16 +01003260 There can be only one persistence cookie per HTTP backend, and it can be
3261 declared in a defaults section. The value of the cookie will be the value
3262 indicated after the "cookie" keyword in a "server" statement. If no cookie
3263 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003264
Willy Tarreau0ba27502007-12-24 16:55:16 +01003265 Examples :
3266 cookie JSESSIONID prefix
3267 cookie SRV insert indirect nocache
3268 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003269 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003270
Willy Tarreau294d0f02015-08-10 19:40:12 +02003271 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003272
Willy Tarreau983e01e2010-01-11 18:42:06 +01003273
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003274declare capture [ request | response ] len <length>
3275 Declares a capture slot.
3276 May be used in sections : defaults | frontend | listen | backend
3277 no | yes | yes | no
3278 Arguments:
3279 <length> is the length allowed for the capture.
3280
3281 This declaration is only available in the frontend or listen section, but the
3282 reserved slot can be used in the backends. The "request" keyword allocates a
3283 capture slot for use in the request, and "response" allocates a capture slot
3284 for use in the response.
3285
3286 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003287 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003288 "http-request capture" and "http-response capture".
3289
3290
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003291default-server [param*]
3292 Change default options for a server in a backend
3293 May be used in sections : defaults | frontend | listen | backend
3294 yes | no | yes | yes
3295 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003296 <param*> is a list of parameters for this server. The "default-server"
3297 keyword accepts an important number of options and has a complete
3298 section dedicated to it. Please refer to section 5 for more
3299 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003300
Willy Tarreau983e01e2010-01-11 18:42:06 +01003301 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003302 default-server inter 1000 weight 13
3303
3304 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003305
Willy Tarreau983e01e2010-01-11 18:42:06 +01003306
Willy Tarreau0ba27502007-12-24 16:55:16 +01003307default_backend <backend>
3308 Specify the backend to use when no "use_backend" rule has been matched.
3309 May be used in sections : defaults | frontend | listen | backend
3310 yes | yes | yes | no
3311 Arguments :
3312 <backend> is the name of the backend to use.
3313
3314 When doing content-switching between frontend and backends using the
3315 "use_backend" keyword, it is often useful to indicate which backend will be
3316 used when no rule has matched. It generally is the dynamic backend which
3317 will catch all undetermined requests.
3318
Willy Tarreau0ba27502007-12-24 16:55:16 +01003319 Example :
3320
3321 use_backend dynamic if url_dyn
3322 use_backend static if url_css url_img extension_img
3323 default_backend dynamic
3324
Willy Tarreau98d04852015-05-26 12:18:29 +02003325 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003326
Willy Tarreau0ba27502007-12-24 16:55:16 +01003327
Baptiste Assmann27f51342013-10-09 06:51:49 +02003328description <string>
3329 Describe a listen, frontend or backend.
3330 May be used in sections : defaults | frontend | listen | backend
3331 no | yes | yes | yes
3332 Arguments : string
3333
3334 Allows to add a sentence to describe the related object in the HAProxy HTML
3335 stats page. The description will be printed on the right of the object name
3336 it describes.
3337 No need to backslash spaces in the <string> arguments.
3338
3339
Willy Tarreau0ba27502007-12-24 16:55:16 +01003340disabled
3341 Disable a proxy, frontend or backend.
3342 May be used in sections : defaults | frontend | listen | backend
3343 yes | yes | yes | yes
3344 Arguments : none
3345
3346 The "disabled" keyword is used to disable an instance, mainly in order to
3347 liberate a listening port or to temporarily disable a service. The instance
3348 will still be created and its configuration will be checked, but it will be
3349 created in the "stopped" state and will appear as such in the statistics. It
3350 will not receive any traffic nor will it send any health-checks or logs. It
3351 is possible to disable many instances at once by adding the "disabled"
3352 keyword in a "defaults" section.
3353
3354 See also : "enabled"
3355
3356
Willy Tarreau5ce94572010-06-07 14:35:41 +02003357dispatch <address>:<port>
3358 Set a default server address
3359 May be used in sections : defaults | frontend | listen | backend
3360 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003361 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003362
3363 <address> is the IPv4 address of the default server. Alternatively, a
3364 resolvable hostname is supported, but this name will be resolved
3365 during start-up.
3366
3367 <ports> is a mandatory port specification. All connections will be sent
3368 to this port, and it is not permitted to use port offsets as is
3369 possible with normal servers.
3370
Willy Tarreau787aed52011-04-15 06:45:37 +02003371 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003372 server can take the connection. In the past it was used to forward non
3373 persistent connections to an auxiliary load balancer. Due to its simple
3374 syntax, it has also been used for simple TCP relays. It is recommended not to
3375 use it for more clarity, and to use the "server" directive instead.
3376
3377 See also : "server"
3378
Olivier Houchard4e694042017-03-14 20:01:29 +01003379
3380dynamic-cookie-key <string>
3381 Set the dynamic cookie secret key for a backend.
3382 May be used in sections : defaults | frontend | listen | backend
3383 yes | no | yes | yes
3384 Arguments : The secret key to be used.
3385
3386 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003387 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003388 specified on the "server" line), using a hash of the IP address of the
3389 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003390 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003391 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003392
Willy Tarreau0ba27502007-12-24 16:55:16 +01003393enabled
3394 Enable a proxy, frontend or backend.
3395 May be used in sections : defaults | frontend | listen | backend
3396 yes | yes | yes | yes
3397 Arguments : none
3398
3399 The "enabled" keyword is used to explicitly enable an instance, when the
3400 defaults has been set to "disabled". This is very rarely used.
3401
3402 See also : "disabled"
3403
3404
3405errorfile <code> <file>
3406 Return a file contents instead of errors generated by HAProxy
3407 May be used in sections : defaults | frontend | listen | backend
3408 yes | yes | yes | yes
3409 Arguments :
3410 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003411 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3412 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003413
3414 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003415 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003416 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003417 error pages, and to use absolute paths, since files are read
3418 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003419
3420 It is important to understand that this keyword is not meant to rewrite
3421 errors returned by the server, but errors detected and returned by HAProxy.
3422 This is why the list of supported errors is limited to a small set.
3423
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003424 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3425
Willy Tarreau0ba27502007-12-24 16:55:16 +01003426 The files are returned verbatim on the TCP socket. This allows any trick such
3427 as redirections to another URL or site, as well as tricks to clean cookies,
3428 force enable or disable caching, etc... The package provides default error
3429 files returning the same contents as default errors.
3430
Willy Tarreau59140a22009-02-22 12:02:30 +01003431 The files should not exceed the configured buffer size (BUFSIZE), which
3432 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003433 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003434 loops between the client and HAProxy when all servers are down, causing an
3435 error to be returned instead of an image. For better HTTP compliance, it is
3436 recommended that all header lines end with CR-LF and not LF alone.
3437
Willy Tarreau0ba27502007-12-24 16:55:16 +01003438 The files are read at the same time as the configuration and kept in memory.
3439 For this reason, the errors continue to be returned even when the process is
3440 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003441 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003442 403 status code and interrogating a blocked URL.
3443
3444 See also : "errorloc", "errorloc302", "errorloc303"
3445
Willy Tarreau59140a22009-02-22 12:02:30 +01003446 Example :
3447 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003448 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003449 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3450 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3451
Willy Tarreau2769aa02007-12-27 18:26:09 +01003452
3453errorloc <code> <url>
3454errorloc302 <code> <url>
3455 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3456 May be used in sections : defaults | frontend | listen | backend
3457 yes | yes | yes | yes
3458 Arguments :
3459 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003460 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3461 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003462
3463 <url> it is the exact contents of the "Location" header. It may contain
3464 either a relative URI to an error page hosted on the same site,
3465 or an absolute URI designating an error page on another site.
3466 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003467 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003468
3469 It is important to understand that this keyword is not meant to rewrite
3470 errors returned by the server, but errors detected and returned by HAProxy.
3471 This is why the list of supported errors is limited to a small set.
3472
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003473 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3474
Willy Tarreau2769aa02007-12-27 18:26:09 +01003475 Note that both keyword return the HTTP 302 status code, which tells the
3476 client to fetch the designated URL using the same HTTP method. This can be
3477 quite problematic in case of non-GET methods such as POST, because the URL
3478 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003479 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003480 status code, indicating to the client that the URL must be fetched with a GET
3481 request.
3482
3483 See also : "errorfile", "errorloc303"
3484
3485
3486errorloc303 <code> <url>
3487 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3488 May be used in sections : defaults | frontend | listen | backend
3489 yes | yes | yes | yes
3490 Arguments :
3491 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003492 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3493 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003494
3495 <url> it is the exact contents of the "Location" header. It may contain
3496 either a relative URI to an error page hosted on the same site,
3497 or an absolute URI designating an error page on another site.
3498 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003499 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003500
3501 It is important to understand that this keyword is not meant to rewrite
3502 errors returned by the server, but errors detected and returned by HAProxy.
3503 This is why the list of supported errors is limited to a small set.
3504
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003505 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3506
Willy Tarreau2769aa02007-12-27 18:26:09 +01003507 Note that both keyword return the HTTP 303 status code, which tells the
3508 client to fetch the designated URL using the same HTTP GET method. This
3509 solves the usual problems associated with "errorloc" and the 302 code. It is
3510 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003511 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003512
3513 See also : "errorfile", "errorloc", "errorloc302"
3514
3515
Simon Horman51a1cf62015-02-03 13:00:44 +09003516email-alert from <emailaddr>
3517 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003518 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003519 May be used in sections: defaults | frontend | listen | backend
3520 yes | yes | yes | yes
3521
3522 Arguments :
3523
3524 <emailaddr> is the from email address to use when sending email alerts
3525
3526 Also requires "email-alert mailers" and "email-alert to" to be set
3527 and if so sending email alerts is enabled for the proxy.
3528
Simon Horman64e34162015-02-06 11:11:57 +09003529 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003530 "email-alert myhostname", "email-alert to", section 3.6 about
3531 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003532
3533
3534email-alert level <level>
3535 Declare the maximum log level of messages for which email alerts will be
3536 sent. This acts as a filter on the sending of email alerts.
3537 May be used in sections: defaults | frontend | listen | backend
3538 yes | yes | yes | yes
3539
3540 Arguments :
3541
3542 <level> One of the 8 syslog levels:
3543 emerg alert crit err warning notice info debug
3544 The above syslog levels are ordered from lowest to highest.
3545
3546 By default level is alert
3547
3548 Also requires "email-alert from", "email-alert mailers" and
3549 "email-alert to" to be set and if so sending email alerts is enabled
3550 for the proxy.
3551
Simon Horman1421e212015-04-30 13:10:35 +09003552 Alerts are sent when :
3553
3554 * An un-paused server is marked as down and <level> is alert or lower
3555 * A paused server is marked as down and <level> is notice or lower
3556 * A server is marked as up or enters the drain state and <level>
3557 is notice or lower
3558 * "option log-health-checks" is enabled, <level> is info or lower,
3559 and a health check status update occurs
3560
Simon Horman64e34162015-02-06 11:11:57 +09003561 See also : "email-alert from", "email-alert mailers",
3562 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003563 section 3.6 about mailers.
3564
3565
3566email-alert mailers <mailersect>
3567 Declare the mailers to be used when sending email alerts
3568 May be used in sections: defaults | frontend | listen | backend
3569 yes | yes | yes | yes
3570
3571 Arguments :
3572
3573 <mailersect> is the name of the mailers section to send email alerts.
3574
3575 Also requires "email-alert from" and "email-alert to" to be set
3576 and if so sending email alerts is enabled for the proxy.
3577
Simon Horman64e34162015-02-06 11:11:57 +09003578 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3579 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003580
3581
3582email-alert myhostname <hostname>
3583 Declare the to hostname address to be used when communicating with
3584 mailers.
3585 May be used in sections: defaults | frontend | listen | backend
3586 yes | yes | yes | yes
3587
3588 Arguments :
3589
Baptiste Assmann738bad92015-12-21 15:27:53 +01003590 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003591
3592 By default the systems hostname is used.
3593
3594 Also requires "email-alert from", "email-alert mailers" and
3595 "email-alert to" to be set and if so sending email alerts is enabled
3596 for the proxy.
3597
Simon Horman64e34162015-02-06 11:11:57 +09003598 See also : "email-alert from", "email-alert level", "email-alert mailers",
3599 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003600
3601
3602email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003603 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003604 header of email alerts. This is the address that email alerts are sent to.
3605 May be used in sections: defaults | frontend | listen | backend
3606 yes | yes | yes | yes
3607
3608 Arguments :
3609
3610 <emailaddr> is the to email address to use when sending email alerts
3611
3612 Also requires "email-alert mailers" and "email-alert to" to be set
3613 and if so sending email alerts is enabled for the proxy.
3614
Simon Horman64e34162015-02-06 11:11:57 +09003615 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003616 "email-alert myhostname", section 3.6 about mailers.
3617
3618
Willy Tarreau4de91492010-01-22 19:10:05 +01003619force-persist { if | unless } <condition>
3620 Declare a condition to force persistence on down servers
3621 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003622 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003623
3624 By default, requests are not dispatched to down servers. It is possible to
3625 force this using "option persist", but it is unconditional and redispatches
3626 to a valid server if "option redispatch" is set. That leaves with very little
3627 possibilities to force some requests to reach a server which is artificially
3628 marked down for maintenance operations.
3629
3630 The "force-persist" statement allows one to declare various ACL-based
3631 conditions which, when met, will cause a request to ignore the down status of
3632 a server and still try to connect to it. That makes it possible to start a
3633 server, still replying an error to the health checks, and run a specially
3634 configured browser to test the service. Among the handy methods, one could
3635 use a specific source IP address, or a specific cookie. The cookie also has
3636 the advantage that it can easily be added/removed on the browser from a test
3637 page. Once the service is validated, it is then possible to open the service
3638 to the world by returning a valid response to health checks.
3639
3640 The forced persistence is enabled when an "if" condition is met, or unless an
3641 "unless" condition is met. The final redispatch is always disabled when this
3642 is used.
3643
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003644 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003645 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003646
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003647
3648filter <name> [param*]
3649 Add the filter <name> in the filter list attached to the proxy.
3650 May be used in sections : defaults | frontend | listen | backend
3651 no | yes | yes | yes
3652 Arguments :
3653 <name> is the name of the filter. Officially supported filters are
3654 referenced in section 9.
3655
Tim Düsterhus4896c442016-11-29 02:15:19 +01003656 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003657 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003658 filter. Please refer to the documentation of the corresponding
3659 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003660
3661 Multiple occurrences of the filter line can be used for the same proxy. The
3662 same filter can be referenced many times if needed.
3663
3664 Example:
3665 listen
3666 bind *:80
3667
3668 filter trace name BEFORE-HTTP-COMP
3669 filter compression
3670 filter trace name AFTER-HTTP-COMP
3671
3672 compression algo gzip
3673 compression offload
3674
3675 server srv1 192.168.0.1:80
3676
3677 See also : section 9.
3678
Willy Tarreau4de91492010-01-22 19:10:05 +01003679
Willy Tarreau2769aa02007-12-27 18:26:09 +01003680fullconn <conns>
3681 Specify at what backend load the servers will reach their maxconn
3682 May be used in sections : defaults | frontend | listen | backend
3683 yes | no | yes | yes
3684 Arguments :
3685 <conns> is the number of connections on the backend which will make the
3686 servers use the maximal number of connections.
3687
Willy Tarreau198a7442008-01-17 12:05:32 +01003688 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003689 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003690 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003691 load. The server will then always accept at least <minconn> connections,
3692 never more than <maxconn>, and the limit will be on the ramp between both
3693 values when the backend has less than <conns> concurrent connections. This
3694 makes it possible to limit the load on the servers during normal loads, but
3695 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003696 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003697
Willy Tarreaufbb78422011-06-05 15:38:35 +02003698 Since it's hard to get this value right, haproxy automatically sets it to
3699 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003700 backend (based on "use_backend" and "default_backend" rules). That way it's
3701 safe to leave it unset. However, "use_backend" involving dynamic names are
3702 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003703
Willy Tarreau2769aa02007-12-27 18:26:09 +01003704 Example :
3705 # The servers will accept between 100 and 1000 concurrent connections each
3706 # and the maximum of 1000 will be reached when the backend reaches 10000
3707 # connections.
3708 backend dynamic
3709 fullconn 10000
3710 server srv1 dyn1:80 minconn 100 maxconn 1000
3711 server srv2 dyn2:80 minconn 100 maxconn 1000
3712
3713 See also : "maxconn", "server"
3714
3715
3716grace <time>
3717 Maintain a proxy operational for some time after a soft stop
3718 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003719 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003720 Arguments :
3721 <time> is the time (by default in milliseconds) for which the instance
3722 will remain operational with the frontend sockets still listening
3723 when a soft-stop is received via the SIGUSR1 signal.
3724
3725 This may be used to ensure that the services disappear in a certain order.
3726 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003727 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003728 needed by the equipment to detect the failure.
3729
3730 Note that currently, there is very little benefit in using this parameter,
3731 and it may in fact complicate the soft-reconfiguration process more than
3732 simplify it.
3733
Willy Tarreau0ba27502007-12-24 16:55:16 +01003734
Andrew Rodland17be45e2016-10-25 17:04:12 -04003735hash-balance-factor <factor>
3736 Specify the balancing factor for bounded-load consistent hashing
3737 May be used in sections : defaults | frontend | listen | backend
3738 yes | no | no | yes
3739 Arguments :
3740 <factor> is the control for the maximum number of concurrent requests to
3741 send to a server, expressed as a percentage of the average number
3742 of concurrent requests across all of the active servers.
3743
3744 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3745 enables an algorithm that prevents any one server from getting too many
3746 requests at once, even if some hash buckets receive many more requests than
3747 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3748 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3749 then no server will be allowed to have a load more than 1.5 times the average.
3750 If server weights are used, they will be respected.
3751
3752 If the first-choice server is disqualified, the algorithm will choose another
3753 server based on the request hash, until a server with additional capacity is
3754 found. A higher <factor> allows more imbalance between the servers, while a
3755 lower <factor> means that more servers will be checked on average, affecting
3756 performance. Reasonable values are from 125 to 200.
3757
Willy Tarreau760e81d2018-05-03 07:20:40 +02003758 This setting is also used by "balance random" which internally relies on the
3759 consistent hashing mechanism.
3760
Andrew Rodland17be45e2016-10-25 17:04:12 -04003761 See also : "balance" and "hash-type".
3762
3763
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003764hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003765 Specify a method to use for mapping hashes to servers
3766 May be used in sections : defaults | frontend | listen | backend
3767 yes | no | yes | yes
3768 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003769 <method> is the method used to select a server from the hash computed by
3770 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003771
Bhaskar98634f02013-10-29 23:30:51 -04003772 map-based the hash table is a static array containing all alive servers.
3773 The hashes will be very smooth, will consider weights, but
3774 will be static in that weight changes while a server is up
3775 will be ignored. This means that there will be no slow start.
3776 Also, since a server is selected by its position in the array,
3777 most mappings are changed when the server count changes. This
3778 means that when a server goes up or down, or when a server is
3779 added to a farm, most connections will be redistributed to
3780 different servers. This can be inconvenient with caches for
3781 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003782
Bhaskar98634f02013-10-29 23:30:51 -04003783 consistent the hash table is a tree filled with many occurrences of each
3784 server. The hash key is looked up in the tree and the closest
3785 server is chosen. This hash is dynamic, it supports changing
3786 weights while the servers are up, so it is compatible with the
3787 slow start feature. It has the advantage that when a server
3788 goes up or down, only its associations are moved. When a
3789 server is added to the farm, only a few part of the mappings
3790 are redistributed, making it an ideal method for caches.
3791 However, due to its principle, the distribution will never be
3792 very smooth and it may sometimes be necessary to adjust a
3793 server's weight or its ID to get a more balanced distribution.
3794 In order to get the same distribution on multiple load
3795 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003796 same IDs. Note: consistent hash uses sdbm and avalanche if no
3797 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003798
3799 <function> is the hash function to be used :
3800
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003801 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003802 reimplementation of ndbm) database library. It was found to do
3803 well in scrambling bits, causing better distribution of the keys
3804 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003805 function with good distribution, unless the total server weight
3806 is a multiple of 64, in which case applying the avalanche
3807 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003808
3809 djb2 this function was first proposed by Dan Bernstein many years ago
3810 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003811 function provides a better distribution than sdbm. It generally
3812 works well with text-based inputs though it can perform extremely
3813 poorly with numeric-only input or when the total server weight is
3814 a multiple of 33, unless the avalanche modifier is also used.
3815
Willy Tarreaua0f42712013-11-14 14:30:35 +01003816 wt6 this function was designed for haproxy while testing other
3817 functions in the past. It is not as smooth as the other ones, but
3818 is much less sensible to the input data set or to the number of
3819 servers. It can make sense as an alternative to sdbm+avalanche or
3820 djb2+avalanche for consistent hashing or when hashing on numeric
3821 data such as a source IP address or a visitor identifier in a URL
3822 parameter.
3823
Willy Tarreau324f07f2015-01-20 19:44:50 +01003824 crc32 this is the most common CRC32 implementation as used in Ethernet,
3825 gzip, PNG, etc. It is slower than the other ones but may provide
3826 a better distribution or less predictable results especially when
3827 used on strings.
3828
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003829 <modifier> indicates an optional method applied after hashing the key :
3830
3831 avalanche This directive indicates that the result from the hash
3832 function above should not be used in its raw form but that
3833 a 4-byte full avalanche hash must be applied first. The
3834 purpose of this step is to mix the resulting bits from the
3835 previous hash in order to avoid any undesired effect when
3836 the input contains some limited values or when the number of
3837 servers is a multiple of one of the hash's components (64
3838 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3839 result less predictable, but it's also not as smooth as when
3840 using the original function. Some testing might be needed
3841 with some workloads. This hash is one of the many proposed
3842 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003843
Bhaskar98634f02013-10-29 23:30:51 -04003844 The default hash type is "map-based" and is recommended for most usages. The
3845 default function is "sdbm", the selection of a function should be based on
3846 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003847
Andrew Rodland17be45e2016-10-25 17:04:12 -04003848 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003849
3850
Willy Tarreau0ba27502007-12-24 16:55:16 +01003851http-check disable-on-404
3852 Enable a maintenance mode upon HTTP/404 response to health-checks
3853 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003854 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003855 Arguments : none
3856
3857 When this option is set, a server which returns an HTTP code 404 will be
3858 excluded from further load-balancing, but will still receive persistent
3859 connections. This provides a very convenient method for Web administrators
3860 to perform a graceful shutdown of their servers. It is also important to note
3861 that a server which is detected as failed while it was in this mode will not
3862 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3863 will immediately be reinserted into the farm. The status on the stats page
3864 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003865 option only works in conjunction with the "httpchk" option. If this option
3866 is used with "http-check expect", then it has precedence over it so that 404
3867 responses will still be considered as soft-stop.
3868
3869 See also : "option httpchk", "http-check expect"
3870
3871
3872http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003873 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003874 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003875 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003876 Arguments :
3877 <match> is a keyword indicating how to look for a specific pattern in the
3878 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003879 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003880 exclamation mark ("!") to negate the match. Spaces are allowed
3881 between the exclamation mark and the keyword. See below for more
3882 details on the supported keywords.
3883
3884 <pattern> is the pattern to look for. It may be a string or a regular
3885 expression. If the pattern contains spaces, they must be escaped
3886 with the usual backslash ('\').
3887
3888 By default, "option httpchk" considers that response statuses 2xx and 3xx
3889 are valid, and that others are invalid. When "http-check expect" is used,
3890 it defines what is considered valid or invalid. Only one "http-check"
3891 statement is supported in a backend. If a server fails to respond or times
3892 out, the check obviously fails. The available matches are :
3893
3894 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003895 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003896 response's status code is exactly this string. If the
3897 "status" keyword is prefixed with "!", then the response
3898 will be considered invalid if the status code matches.
3899
3900 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003901 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003902 response's status code matches the expression. If the
3903 "rstatus" keyword is prefixed with "!", then the response
3904 will be considered invalid if the status code matches.
3905 This is mostly used to check for multiple codes.
3906
3907 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003908 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003909 response's body contains this exact string. If the
3910 "string" keyword is prefixed with "!", then the response
3911 will be considered invalid if the body contains this
3912 string. This can be used to look for a mandatory word at
3913 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003914 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003915 trace).
3916
3917 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003918 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003919 response's body matches this expression. If the "rstring"
3920 keyword is prefixed with "!", then the response will be
3921 considered invalid if the body matches the expression.
3922 This can be used to look for a mandatory word at the end
3923 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003924 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003925
3926 It is important to note that the responses will be limited to a certain size
3927 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3928 Thus, too large responses may not contain the mandatory pattern when using
3929 "string" or "rstring". If a large response is absolutely required, it is
3930 possible to change the default max size by setting the global variable.
3931 However, it is worth keeping in mind that parsing very large responses can
3932 waste some CPU cycles, especially when regular expressions are used, and that
3933 it is always better to focus the checks on smaller resources.
3934
Cyril Bonté32602d22015-01-30 00:07:07 +01003935 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3936 will automatically append a "Connection: close" header, meaning that this
3937 header should not be present in the request provided by "option httpchk".
3938
Willy Tarreaubd741542010-03-16 18:46:54 +01003939 Last, if "http-check expect" is combined with "http-check disable-on-404",
3940 then this last one has precedence when the server responds with 404.
3941
3942 Examples :
3943 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003944 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003945
3946 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003947 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003948
3949 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003950 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003951
3952 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003953 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003954
Willy Tarreaubd741542010-03-16 18:46:54 +01003955 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003956
3957
Willy Tarreauef781042010-01-27 11:53:01 +01003958http-check send-state
3959 Enable emission of a state header with HTTP health checks
3960 May be used in sections : defaults | frontend | listen | backend
3961 yes | no | yes | yes
3962 Arguments : none
3963
3964 When this option is set, haproxy will systematically send a special header
3965 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3966 how they are seen by haproxy. This can be used for instance when a server is
3967 manipulated without access to haproxy and the operator needs to know whether
3968 haproxy still sees it up or not, or if the server is the last one in a farm.
3969
3970 The header is composed of fields delimited by semi-colons, the first of which
3971 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3972 checks on the total number before transition, just as appears in the stats
3973 interface. Next headers are in the form "<variable>=<value>", indicating in
3974 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003975 - a variable "address", containing the address of the backend server.
3976 This corresponds to the <address> field in the server declaration. For
3977 unix domain sockets, it will read "unix".
3978
3979 - a variable "port", containing the port of the backend server. This
3980 corresponds to the <port> field in the server declaration. For unix
3981 domain sockets, it will read "unix".
3982
Willy Tarreauef781042010-01-27 11:53:01 +01003983 - a variable "name", containing the name of the backend followed by a slash
3984 ("/") then the name of the server. This can be used when a server is
3985 checked in multiple backends.
3986
3987 - a variable "node" containing the name of the haproxy node, as set in the
3988 global "node" variable, otherwise the system's hostname if unspecified.
3989
3990 - a variable "weight" indicating the weight of the server, a slash ("/")
3991 and the total weight of the farm (just counting usable servers). This
3992 helps to know if other servers are available to handle the load when this
3993 one fails.
3994
3995 - a variable "scur" indicating the current number of concurrent connections
3996 on the server, followed by a slash ("/") then the total number of
3997 connections on all servers of the same backend.
3998
3999 - a variable "qcur" indicating the current number of requests in the
4000 server's queue.
4001
4002 Example of a header received by the application server :
4003 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4004 scur=13/22; qcur=0
4005
4006 See also : "option httpchk", "http-check disable-on-404"
4007
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004008
4009http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004010 Access control for Layer 7 requests
4011
4012 May be used in sections: defaults | frontend | listen | backend
4013 no | yes | yes | yes
4014
Willy Tarreau20b0de52012-12-24 15:45:22 +01004015 The http-request statement defines a set of rules which apply to layer 7
4016 processing. The rules are evaluated in their declaration order when they are
4017 met in a frontend, listen or backend section. Any rule may optionally be
4018 followed by an ACL-based condition, in which case it will only be evaluated
4019 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004020
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004021 The first keyword is the rule's action. The supported actions are described
4022 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004023
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004024 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004025
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004026 It is important to know that http-request rules are processed very early in
4027 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
4028 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4029 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004030
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004031 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4032 in newer versions (>= 1.5). But if you need to use regular expression to
4033 delete headers, you can still use "reqdel". Also please use
4034 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004035
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004036 Example:
4037 acl nagios src 192.168.129.3
4038 acl local_net src 192.168.0.0/16
4039 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004040
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004041 http-request allow if nagios
4042 http-request allow if local_net auth_ok
4043 http-request auth realm Gimme if local_net auth_ok
4044 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004045
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004046 Example:
4047 acl key req.hdr(X-Add-Acl-Key) -m found
4048 acl add path /addacl
4049 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004050
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004051 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004052
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004053 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4054 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004055
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004056 Example:
4057 acl value req.hdr(X-Value) -m found
4058 acl setmap path /setmap
4059 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004060
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004061 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004062
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004063 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4064 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004065
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004066 See also : "stats http-request", section 3.4 about userlists and section 7
4067 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004068
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004069http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004070
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004071 This is used to add a new entry into an ACL. The ACL must be loaded from a
4072 file (even a dummy empty file). The file name of the ACL to be updated is
4073 passed between parentheses. It takes one argument: <key fmt>, which follows
4074 log-format rules, to collect content of the new entry. It performs a lookup
4075 in the ACL before insertion, to avoid duplicated (or more) values. This
4076 lookup is done by a linear search and can be expensive with large lists!
4077 It is the equivalent of the "add acl" command from the stats socket, but can
4078 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004079
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004080http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004081
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004082 This appends an HTTP header field whose name is specified in <name> and
4083 whose value is defined by <fmt> which follows the log-format rules (see
4084 Custom Log Format in section 8.2.4). This is particularly useful to pass
4085 connection-specific information to the server (e.g. the client's SSL
4086 certificate), or to combine several headers into one. This rule is not
4087 final, so it is possible to add other similar rules. Note that header
4088 addition is performed immediately, so one rule might reuse the resulting
4089 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004090
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004091http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004092
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004093 This stops the evaluation of the rules and lets the request pass the check.
4094 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004095
Sasha Pachev218f0642014-06-16 12:05:59 -06004096
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004097http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004098
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004099 This stops the evaluation of the rules and immediately responds with an
4100 HTTP 401 or 407 error code to invite the user to present a valid user name
4101 and password. No further "http-request" rules are evaluated. An optional
4102 "realm" parameter is supported, it sets the authentication realm that is
4103 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004104
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004105 Example:
4106 acl auth_ok http_auth_group(L1) G1
4107 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004108
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004109http-request cache-use [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004110
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004111 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004112
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004113http-request capture <sample> [ len <length> | id <id> ]
4114 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004115
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004116 This captures sample expression <sample> from the request buffer, and
4117 converts it to a string of at most <len> characters. The resulting string is
4118 stored into the next request "capture" slot, so it will possibly appear next
4119 to some captured HTTP headers. It will then automatically appear in the logs,
4120 and it will be possible to extract it using sample fetch rules to feed it
4121 into headers or anything. The length should be limited given that this size
4122 will be allocated for each capture during the whole session life.
4123 Please check section 7.3 (Fetching samples) and "capture request header" for
4124 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004125
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004126 If the keyword "id" is used instead of "len", the action tries to store the
4127 captured string in a previously declared capture slot. This is useful to run
4128 captures in backends. The slot id can be declared by a previous directive
4129 "http-request capture" or with the "declare capture" keyword. If the slot
4130 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4131 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004132
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004133http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004134
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004135 This is used to delete an entry from an ACL. The ACL must be loaded from a
4136 file (even a dummy empty file). The file name of the ACL to be updated is
4137 passed between parentheses. It takes one argument: <key fmt>, which follows
4138 log-format rules, to collect content of the entry to delete.
4139 It is the equivalent of the "del acl" command from the stats socket, but can
4140 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004141
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004142http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004143
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004144 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004145
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004146http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004147
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004148 This is used to delete an entry from a MAP. The MAP must be loaded from a
4149 file (even a dummy empty file). The file name of the MAP to be updated is
4150 passed between parentheses. It takes one argument: <key fmt>, which follows
4151 log-format rules, to collect content of the entry to delete.
4152 It takes one argument: "file name" It is the equivalent of the "del map"
4153 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004154
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004155http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004156
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004157 This stops the evaluation of the rules and immediately rejects the request
4158 and emits an HTTP 403 error, or optionally the status code specified as an
4159 argument to "deny_status". The list of permitted status codes is limited to
4160 those that can be overridden by the "errorfile" directive.
4161 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004162
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004163http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4164
4165 This is used to build an HTTP 103 Early Hints response prior to any other one.
4166 This appends an HTTP header field to this response whose name is specified in
4167 <name> and whose value is defined by <fmt> which follows the log-format rules
4168 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004169 to the client some Link headers to preload resources required to render the
4170 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004171
4172 See RFC 8297 for more information.
4173
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004174http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004175
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004176 This performs an HTTP redirection based on a redirect rule. This is exactly
4177 the same as the "redirect" statement except that it inserts a redirect rule
4178 which can be processed in the middle of other "http-request" rules and that
4179 these rules use the "log-format" strings. See the "redirect" keyword for the
4180 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004181
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004182http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004183
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004184 This stops the evaluation of the rules and immediately closes the connection
4185 without sending any response. It acts similarly to the
4186 "tcp-request content reject" rules. It can be useful to force an immediate
4187 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004188
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004189http-request replace-header <name> <match-regex> <replace-fmt>
4190 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004191
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004192 This matches the regular expression in all occurrences of header field
4193 <name> according to <match-regex>, and replaces them with the
4194 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4195 work like in <fmt> arguments in "http-request add-header". The match is
4196 only case-sensitive. It is important to understand that this action only
4197 considers whole header lines, regardless of the number of values they may
4198 contain. This usage is suited to headers naturally containing commas in
4199 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004200
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004201 Example:
4202 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004203
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004204 # applied to:
4205 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004206
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004207 # outputs:
4208 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004209
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004210 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004211
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004212http-request replace-value <name> <match-regex> <replace-fmt>
4213 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004214
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004215 This works like "replace-header" except that it matches the regex against
4216 every comma-delimited value of the header field <name> instead of the
4217 entire header. This is suited for all headers which are allowed to carry
4218 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004219
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004220 Example:
4221 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004222
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004223 # applied to:
4224 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004225
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004226 # outputs:
4227 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004228
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004229http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4230http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004231
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004232 This actions increments the GPC0 or GPC1 counter according with the sticky
4233 counter designated by <sc-id>. If an error occurs, this action silently fails
4234 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004235
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004236http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004237
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004238 This action sets the GPT0 tag according to the sticky counter designated by
4239 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4240 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004241
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004242http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004243
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004244 This is used to set the destination IP address to the value of specified
4245 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4246 but provides the correct IP in a HTTP header; or you want to mask the IP for
4247 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4248 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004249
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004250 Arguments:
4251 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4252 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004253
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004254 Example:
4255 http-request set-dst hdr(x-dst)
4256 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004257
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004258 When possible, set-dst preserves the original destination port as long as the
4259 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004260
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004261http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004262
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004263 This is used to set the destination port address to the value of specified
4264 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4265 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004266
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004267 Arguments:
4268 <expr> Is a standard HAProxy expression formed by a sample-fetch
4269 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004270
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004271 Example:
4272 http-request set-dst-port hdr(x-port)
4273 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004274
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004275 When possible, set-dst-port preserves the original destination address as
4276 long as the address family supports a port, otherwise it forces the
4277 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004278
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004279http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004280
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004281 This does the same as "http-request add-header" except that the header name
4282 is first removed if it existed. This is useful when passing security
4283 information to the server, where the header must not be manipulated by
4284 external users. Note that the new value is computed before the removal so it
4285 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004286
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004287 Example:
4288 http-request set-header X-Haproxy-Current-Date %T
4289 http-request set-header X-SSL %[ssl_fc]
4290 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4291 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4292 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4293 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4294 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4295 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4296 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004297
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004298http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004299
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004300 This is used to change the log level of the current request when a certain
4301 condition is met. Valid levels are the 8 syslog levels (see the "log"
4302 keyword) plus the special level "silent" which disables logging for this
4303 request. This rule is not final so the last matching rule wins. This rule
4304 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004305
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004306http-request set-map(<file-name>) <key fmt> <value fmt>
4307 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004308
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004309 This is used to add a new entry into a MAP. The MAP must be loaded from a
4310 file (even a dummy empty file). The file name of the MAP to be updated is
4311 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4312 log-format rules, used to collect MAP key, and <value fmt>, which follows
4313 log-format rules, used to collect content for the new entry.
4314 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4315 more) values. This lookup is done by a linear search and can be expensive
4316 with large lists! It is the equivalent of the "set map" command from the
4317 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004318
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004319http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004320
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004321 This is used to set the Netfilter MARK on all packets sent to the client to
4322 the value passed in <mark> on platforms which support it. This value is an
4323 unsigned 32 bit value which can be matched by netfilter and by the routing
4324 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4325 "0x"). This can be useful to force certain packets to take a different route
4326 (for example a cheaper network path for bulk downloads). This works on Linux
4327 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004328
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004329http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004331 This rewrites the request method with the result of the evaluation of format
4332 string <fmt>. There should be very few valid reasons for having to do so as
4333 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004334
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004335http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004336
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004337 This sets the "nice" factor of the current request being processed. It only
4338 has effect against the other requests being processed at the same time.
4339 The default value is 0, unless altered by the "nice" setting on the "bind"
4340 line. The accepted range is -1024..1024. The higher the value, the nicest
4341 the request will be. Lower values will make the request more important than
4342 other ones. This can be useful to improve the speed of some requests, or
4343 lower the priority of non-important requests. Using this setting without
4344 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004345
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004346http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004347
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004348 This rewrites the request path with the result of the evaluation of format
4349 string <fmt>. The query string, if any, is left intact. If a scheme and
4350 authority is found before the path, they are left intact as well. If the
4351 request doesn't have a path ("*"), this one is replaced with the format.
4352 This can be used to prepend a directory component in front of a path for
4353 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004354
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004355 Example :
4356 # prepend the host name before the path
4357 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004358
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004359http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004360
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004361 This is used to set the queue priority class of the current request.
4362 The value must be a sample expression which converts to an integer in the
4363 range -2047..2047. Results outside this range will be truncated.
4364 The priority class determines the order in which queued requests are
4365 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004366
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004367http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004368
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004369 This is used to set the queue priority timestamp offset of the current
4370 request. The value must be a sample expression which converts to an integer
4371 in the range -524287..524287. Results outside this range will be truncated.
4372 When a request is queued, it is ordered first by the priority class, then by
4373 the current timestamp adjusted by the given offset in milliseconds. Lower
4374 values have higher priority.
4375 Note that the resulting timestamp is is only tracked with enough precision
4376 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4377 adjusted timestamp exceeds this value, it will be misidentified as highest
4378 priority. Thus it is important to set "timeout queue" to a value, where when
4379 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004380
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004381http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004382
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004383 This rewrites the request's query string which appears after the first
4384 question mark ("?") with the result of the evaluation of format string <fmt>.
4385 The part prior to the question mark is left intact. If the request doesn't
4386 contain a question mark and the new value is not empty, then one is added at
4387 the end of the URI, followed by the new value. If a question mark was
4388 present, it will never be removed even if the value is empty. This can be
4389 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004390
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004391 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004392
4393 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004394 # replace "%3D" with "=" in the query string
4395 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004396
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004397http-request set-src <expr> [ { if | unless } <condition> ]
4398 This is used to set the source IP address to the value of specified
4399 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4400 provides the correct IP in a HTTP header; or you want to mask source IP for
4401 privacy.
4402
4403 Arguments :
4404 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4405 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004406
Cyril Bonté78caf842010-03-10 22:41:43 +01004407 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004408 http-request set-src hdr(x-forwarded-for)
4409 http-request set-src src,ipmask(24)
4410
4411 When possible, set-src preserves the original source port as long as the
4412 address family allows it, otherwise the source port is set to 0.
4413
4414http-request set-src-port <expr> [ { if | unless } <condition> ]
4415
4416 This is used to set the source port address to the value of specified
4417 expression.
4418
4419 Arguments:
4420 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4421 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004422
Willy Tarreau20b0de52012-12-24 15:45:22 +01004423 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004424 http-request set-src-port hdr(x-port)
4425 http-request set-src-port int(4000)
4426
4427 When possible, set-src-port preserves the original source address as long as
4428 the address family supports a port, otherwise it forces the source address to
4429 IPv4 "0.0.0.0" before rewriting the port.
4430
4431http-request set-tos <tos> [ { if | unless } <condition> ]
4432
4433 This is used to set the TOS or DSCP field value of packets sent to the client
4434 to the value passed in <tos> on platforms which support this. This value
4435 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4436 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4437 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4438 be used to adjust some routing behavior on border routers based on some
4439 information from the request.
4440
4441 See RFC 2474, 2597, 3260 and 4594 for more information.
4442
4443http-request set-uri <fmt> [ { if | unless } <condition> ]
4444
4445 This rewrites the request URI with the result of the evaluation of format
4446 string <fmt>. The scheme, authority, path and query string are all replaced
4447 at once. This can be used to rewrite hosts in front of proxies, or to
4448 perform complex modifications to the URI such as moving parts between the
4449 path and the query string.
4450 See also "http-request set-path" and "http-request set-query".
4451
4452http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4453
4454 This is used to set the contents of a variable. The variable is declared
4455 inline.
4456
4457 Arguments:
4458 <var-name> The name of the variable starts with an indication about its
4459 scope. The scopes allowed are:
4460 "proc" : the variable is shared with the whole process
4461 "sess" : the variable is shared with the whole session
4462 "txn" : the variable is shared with the transaction
4463 (request and response)
4464 "req" : the variable is shared only during request
4465 processing
4466 "res" : the variable is shared only during response
4467 processing
4468 This prefix is followed by a name. The separator is a '.'.
4469 The name may only contain characters 'a-z', 'A-Z', '0-9'
4470 and '_'.
4471
4472 <expr> Is a standard HAProxy expression formed by a sample-fetch
4473 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004474
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004475 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004476 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004477
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004478http-request send-spoe-group <engine-name> <group-name>
4479 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004480
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004481 This action is used to trigger sending of a group of SPOE messages. To do so,
4482 the SPOE engine used to send messages must be defined, as well as the SPOE
4483 group to send. Of course, the SPOE engine must refer to an existing SPOE
4484 filter. If not engine name is provided on the SPOE filter line, the SPOE
4485 agent name must be used.
4486
4487 Arguments:
4488 <engine-name> The SPOE engine name.
4489
4490 <group-name> The SPOE group name as specified in the engine
4491 configuration.
4492
4493http-request silent-drop [ { if | unless } <condition> ]
4494
4495 This stops the evaluation of the rules and makes the client-facing connection
4496 suddenly disappear using a system-dependent way that tries to prevent the
4497 client from being notified. The effect it then that the client still sees an
4498 established connection while there's none on HAProxy. The purpose is to
4499 achieve a comparable effect to "tarpit" except that it doesn't use any local
4500 resource at all on the machine running HAProxy. It can resist much higher
4501 loads than "tarpit", and slow down stronger attackers. It is important to
4502 understand the impact of using this mechanism. All stateful equipment placed
4503 between the client and HAProxy (firewalls, proxies, load balancers) will also
4504 keep the established connection for a long time and may suffer from this
4505 action.
4506 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4507 option is used to block the emission of a TCP reset. On other systems, the
4508 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4509 router, though it's still delivered to local networks. Do not use it unless
4510 you fully understand how it works.
4511
4512http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4513
4514 This stops the evaluation of the rules and immediately blocks the request
4515 without responding for a delay specified by "timeout tarpit" or
4516 "timeout connect" if the former is not set. After that delay, if the client
4517 is still connected, an HTTP error 500 (or optionally the status code
4518 specified as an argument to "deny_status") is returned so that the client
4519 does not suspect it has been tarpitted. Logs will report the flags "PT".
4520 The goal of the tarpit rule is to slow down robots during an attack when
4521 they're limited on the number of concurrent requests. It can be very
4522 efficient against very dumb robots, and will significantly reduce the load
4523 on firewalls compared to a "deny" rule. But when facing "correctly"
4524 developed robots, it can make things worse by forcing haproxy and the front
4525 firewall to support insane number of concurrent connections.
4526 See also the "silent-drop" action.
4527
4528http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4529http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4530http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4531
4532 This enables tracking of sticky counters from current request. These rules do
4533 not stop evaluation and do not change default action. The number of counters
4534 that may be simultaneously tracked by the same connection is set in
4535 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4536 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4537 "track-sc0" rule executed enables tracking of the counters of the specified
4538 table as the first set. The first "track-sc1" rule executed enables tracking
4539 of the counters of the specified table as the second set. The first
4540 "track-sc2" rule executed enables tracking of the counters of the specified
4541 table as the third set. It is a recommended practice to use the first set of
4542 counters for the per-frontend counters and the second set for the per-backend
4543 ones. But this is just a guideline, all may be used everywhere.
4544
4545 Arguments :
4546 <key> is mandatory, and is a sample expression rule as described in
4547 section 7.3. It describes what elements of the incoming request or
4548 connection will be analyzed, extracted, combined, and used to
4549 select which table entry to update the counters.
4550
4551 <table> is an optional table to be used instead of the default one, which
4552 is the stick-table declared in the current proxy. All the counters
4553 for the matches and updates for the key will then be performed in
4554 that table until the session ends.
4555
4556 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4557 it is not found, an entry is allocated for it. Then a pointer to that entry
4558 is kept during all the session's life, and this entry's counters are updated
4559 as often as possible, every time the session's counters are updated, and also
4560 systematically when the session ends. Counters are only updated for events
4561 that happen after the tracking has been started. As an exception, connection
4562 counters and request counters are systematically updated so that they reflect
4563 useful information.
4564
4565 If the entry tracks concurrent connection counters, one connection is counted
4566 for as long as the entry is tracked, and the entry will not expire during
4567 that time. Tracking counters also provides a performance advantage over just
4568 checking the keys, because only one table lookup is performed for all ACL
4569 checks that make use of it.
4570
4571http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4572
4573 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004574
4575 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004576 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004577
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004578http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004579
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004580 This will delay the processing of the request until the SSL handshake
4581 happened. This is mostly useful to delay processing early data until we're
4582 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004583
Willy Tarreauef781042010-01-27 11:53:01 +01004584
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004585http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004586 Access control for Layer 7 responses
4587
4588 May be used in sections: defaults | frontend | listen | backend
4589 no | yes | yes | yes
4590
4591 The http-response statement defines a set of rules which apply to layer 7
4592 processing. The rules are evaluated in their declaration order when they are
4593 met in a frontend, listen or backend section. Any rule may optionally be
4594 followed by an ACL-based condition, in which case it will only be evaluated
4595 if the condition is true. Since these rules apply on responses, the backend
4596 rules are applied first, followed by the frontend's rules.
4597
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004598 The first keyword is the rule's action. The supported actions are described
4599 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004600
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004601 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004602
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004603 It is important to know that http-response rules are processed very early in
4604 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4605 headers added by "add-header"/"set-header" are visible by almost all further
4606 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004607
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004608 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4609 in newer versions (>= 1.5). But if you need to use regular expression to
4610 delete headers, you can still use "rspdel". Also please use
4611 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004612
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004613 Example:
4614 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004615
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004616 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004617
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004618 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4619 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004620
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004621 Example:
4622 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004623
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004624 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004625
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004626 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4627 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004628
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004629 See also : "http-request", section 3.4 about userlists and section 7 about
4630 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004631
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004632http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004633
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004634 This is used to add a new entry into an ACL. The ACL must be loaded from a
4635 file (even a dummy empty file). The file name of the ACL to be updated is
4636 passed between parentheses. It takes one argument: <key fmt>, which follows
4637 log-format rules, to collect content of the new entry. It performs a lookup
4638 in the ACL before insertion, to avoid duplicated (or more) values.
4639 This lookup is done by a linear search and can be expensive with large lists!
4640 It is the equivalent of the "add acl" command from the stats socket, but can
4641 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004642
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004643http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004644
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004645 This appends an HTTP header field whose name is specified in <name> and whose
4646 value is defined by <fmt> which follows the log-format rules (see Custom Log
4647 Format in section 8.2.4). This may be used to send a cookie to a client for
4648 example, or to pass some internal information.
4649 This rule is not final, so it is possible to add other similar rules.
4650 Note that header addition is performed immediately, so one rule might reuse
4651 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004652
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004653http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004654
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004655 This stops the evaluation of the rules and lets the response pass the check.
4656 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004657
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004658http-response cache-store [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004659
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004660 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004661
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004662http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004663
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004664 This captures sample expression <sample> from the response buffer, and
4665 converts it to a string. The resulting string is stored into the next request
4666 "capture" slot, so it will possibly appear next to some captured HTTP
4667 headers. It will then automatically appear in the logs, and it will be
4668 possible to extract it using sample fetch rules to feed it into headers or
4669 anything. Please check section 7.3 (Fetching samples) and
4670 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004671
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004672 The keyword "id" is the id of the capture slot which is used for storing the
4673 string. The capture slot must be defined in an associated frontend.
4674 This is useful to run captures in backends. The slot id can be declared by a
4675 previous directive "http-response capture" or with the "declare capture"
4676 keyword.
4677 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4678 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004679
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004680http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004681
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004682 This is used to delete an entry from an ACL. The ACL must be loaded from a
4683 file (even a dummy empty file). The file name of the ACL to be updated is
4684 passed between parentheses. It takes one argument: <key fmt>, which follows
4685 log-format rules, to collect content of the entry to delete.
4686 It is the equivalent of the "del acl" command from the stats socket, but can
4687 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004688
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004689http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004690
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004691 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004692
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004693http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004694
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004695 This is used to delete an entry from a MAP. The MAP must be loaded from a
4696 file (even a dummy empty file). The file name of the MAP to be updated is
4697 passed between parentheses. It takes one argument: <key fmt>, which follows
4698 log-format rules, to collect content of the entry to delete.
4699 It takes one argument: "file name" It is the equivalent of the "del map"
4700 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004701
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004702http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004703
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004704 This stops the evaluation of the rules and immediately rejects the response
4705 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004706
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004707http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004708
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004709 This performs an HTTP redirection based on a redirect rule.
4710 This supports a format string similarly to "http-request redirect" rules,
4711 with the exception that only the "location" type of redirect is possible on
4712 the response. See the "redirect" keyword for the rule's syntax. When a
4713 redirect rule is applied during a response, connections to the server are
4714 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004715
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004716http-response replace-header <name> <regex-match> <replace-fmt>
4717 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004718
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004719 This matches the regular expression in all occurrences of header field <name>
4720 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4721 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4722 in "add-header". The match is only case-sensitive. It is important to
4723 understand that this action only considers whole header lines, regardless of
4724 the number of values they may contain. This usage is suited to headers
4725 naturally containing commas in their value, such as Set-Cookie, Expires and
4726 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004727
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004728 Example:
4729 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004730
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004731 # applied to:
4732 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004733
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004734 # outputs:
4735 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004736
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004737 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004738
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004739http-response replace-value <name> <regex-match> <replace-fmt>
4740 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004741
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004742 This works like "replace-header" except that it matches the regex against
4743 every comma-delimited value of the header field <name> instead of the entire
4744 header. This is suited for all headers which are allowed to carry more than
4745 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004746
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004747 Example:
4748 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004749
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004750 # applied to:
4751 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004752
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004753 # outputs:
4754 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004755
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004756http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4757http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004759 This action increments the GPC0 or GPC1 counter according with the sticky
4760 counter designated by <sc-id>. If an error occurs, this action silently fails
4761 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004762
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004763http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004764
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004765 This action sets the GPT0 tag according to the sticky counter designated by
4766 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4767 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004768
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004769http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004770
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004771 This action is used to trigger sending of a group of SPOE messages. To do so,
4772 the SPOE engine used to send messages must be defined, as well as the SPOE
4773 group to send. Of course, the SPOE engine must refer to an existing SPOE
4774 filter. If not engine name is provided on the SPOE filter line, the SPOE
4775 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004776
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004777 Arguments:
4778 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004779
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004780 <group-name> The SPOE group name as specified in the engine
4781 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004782
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004783http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004784
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004785 This does the same as "add-header" except that the header name is first
4786 removed if it existed. This is useful when passing security information to
4787 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004788
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004789http-response set-log-level <level> [ { if | unless } <condition> ]
4790
4791 This is used to change the log level of the current request when a certain
4792 condition is met. Valid levels are the 8 syslog levels (see the "log"
4793 keyword) plus the special level "silent" which disables logging for this
4794 request. This rule is not final so the last matching rule wins. This rule can
4795 be useful to disable health checks coming from another equipment.
4796
4797http-response set-map(<file-name>) <key fmt> <value fmt>
4798
4799 This is used to add a new entry into a MAP. The MAP must be loaded from a
4800 file (even a dummy empty file). The file name of the MAP to be updated is
4801 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4802 log-format rules, used to collect MAP key, and <value fmt>, which follows
4803 log-format rules, used to collect content for the new entry. It performs a
4804 lookup in the MAP before insertion, to avoid duplicated (or more) values.
4805 This lookup is done by a linear search and can be expensive with large lists!
4806 It is the equivalent of the "set map" command from the stats socket, but can
4807 be triggered by an HTTP response.
4808
4809http-response set-mark <mark> [ { if | unless } <condition> ]
4810
4811 This is used to set the Netfilter MARK on all packets sent to the client to
4812 the value passed in <mark> on platforms which support it. This value is an
4813 unsigned 32 bit value which can be matched by netfilter and by the routing
4814 table. It can be expressed both in decimal or hexadecimal format (prefixed
4815 by "0x"). This can be useful to force certain packets to take a different
4816 route (for example a cheaper network path for bulk downloads). This works on
4817 Linux kernels 2.6.32 and above and requires admin privileges.
4818
4819http-response set-nice <nice> [ { if | unless } <condition> ]
4820
4821 This sets the "nice" factor of the current request being processed.
4822 It only has effect against the other requests being processed at the same
4823 time. The default value is 0, unless altered by the "nice" setting on the
4824 "bind" line. The accepted range is -1024..1024. The higher the value, the
4825 nicest the request will be. Lower values will make the request more important
4826 than other ones. This can be useful to improve the speed of some requests, or
4827 lower the priority of non-important requests. Using this setting without
4828 prior experimentation can cause some major slowdown.
4829
4830http-response set-status <status> [reason <str>]
4831 [ { if | unless } <condition> ]
4832
4833 This replaces the response status code with <status> which must be an integer
4834 between 100 and 999. Optionally, a custom reason text can be provided defined
4835 by <str>, or the default reason for the specified code will be used as a
4836 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004837
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004838 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004839 # return "431 Request Header Fields Too Large"
4840 http-response set-status 431
4841 # return "503 Slow Down", custom reason
4842 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004843
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004844http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004845
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004846 This is used to set the TOS or DSCP field value of packets sent to the client
4847 to the value passed in <tos> on platforms which support this.
4848 This value represents the whole 8 bits of the IP TOS field, and can be
4849 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
4850 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
4851 always 0. This can be used to adjust some routing behavior on border routers
4852 based on some information from the request.
4853
4854 See RFC 2474, 2597, 3260 and 4594 for more information.
4855
4856http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4857
4858 This is used to set the contents of a variable. The variable is declared
4859 inline.
4860
4861 Arguments:
4862 <var-name> The name of the variable starts with an indication about its
4863 scope. The scopes allowed are:
4864 "proc" : the variable is shared with the whole process
4865 "sess" : the variable is shared with the whole session
4866 "txn" : the variable is shared with the transaction
4867 (request and response)
4868 "req" : the variable is shared only during request
4869 processing
4870 "res" : the variable is shared only during response
4871 processing
4872 This prefix is followed by a name. The separator is a '.'.
4873 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4874 and '_'.
4875
4876 <expr> Is a standard HAProxy expression formed by a sample-fetch
4877 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004878
4879 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004880 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004881
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004882http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004883
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004884 This stops the evaluation of the rules and makes the client-facing connection
4885 suddenly disappear using a system-dependent way that tries to prevent the
4886 client from being notified. The effect it then that the client still sees an
4887 established connection while there's none on HAProxy. The purpose is to
4888 achieve a comparable effect to "tarpit" except that it doesn't use any local
4889 resource at all on the machine running HAProxy. It can resist much higher
4890 loads than "tarpit", and slow down stronger attackers. It is important to
4891 understand the impact of using this mechanism. All stateful equipment placed
4892 between the client and HAProxy (firewalls, proxies, load balancers) will also
4893 keep the established connection for a long time and may suffer from this
4894 action.
4895 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4896 option is used to block the emission of a TCP reset. On other systems, the
4897 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4898 router, though it's still delivered to local networks. Do not use it unless
4899 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004900
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004901http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4902http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4903http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004904
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004905 This enables tracking of sticky counters from current response. Please refer
4906 to "http-request track-sc" for a complete description. The only difference
4907 from "http-request track-sc" is the <key> sample expression can only make use
4908 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
4909 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
4910 supported, haproxy will fail and warn while parsing the config.
4911
4912http-response unset-var(<var-name>) [ { if | unless } <condition> ]
4913
4914 This is used to unset a variable. See "http-response set-var" for details
4915 about <var-name>.
4916
4917 Example:
4918 http-response unset-var(sess.last_redir)
4919
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004920
Willy Tarreau30631952015-08-06 15:05:24 +02004921http-reuse { never | safe | aggressive | always }
4922 Declare how idle HTTP connections may be shared between requests
4923
4924 May be used in sections: defaults | frontend | listen | backend
4925 yes | no | yes | yes
4926
4927 By default, a connection established between haproxy and the backend server
4928 belongs to the session that initiated it. The downside is that between the
4929 response and the next request, the connection remains idle and is not used.
4930 In many cases for performance reasons it is desirable to make it possible to
4931 reuse these idle connections to serve other requests from different sessions.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004932 This directive allows to tune this behavior.
Willy Tarreau30631952015-08-06 15:05:24 +02004933
4934 The argument indicates the desired connection reuse strategy :
4935
4936 - "never" : idle connections are never shared between sessions. This is
4937 the default choice. It may be enforced to cancel a different
4938 strategy inherited from a defaults section or for
4939 troubleshooting. For example, if an old bogus application
4940 considers that multiple requests over the same connection come
4941 from the same client and it is not possible to fix the
4942 application, it may be desirable to disable connection sharing
4943 in a single backend. An example of such an application could
4944 be an old haproxy using cookie insertion in tunnel mode and
4945 not checking any request past the first one.
4946
4947 - "safe" : this is the recommended strategy. The first request of a
4948 session is always sent over its own connection, and only
4949 subsequent requests may be dispatched over other existing
4950 connections. This ensures that in case the server closes the
4951 connection when the request is being sent, the browser can
4952 decide to silently retry it. Since it is exactly equivalent to
4953 regular keep-alive, there should be no side effects.
4954
4955 - "aggressive" : this mode may be useful in webservices environments where
4956 all servers are not necessarily known and where it would be
4957 appreciable to deliver most first requests over existing
4958 connections. In this case, first requests are only delivered
4959 over existing connections that have been reused at least once,
4960 proving that the server correctly supports connection reuse.
4961 It should only be used when it's sure that the client can
4962 retry a failed request once in a while and where the benefit
4963 of aggressive connection reuse significantly outweights the
4964 downsides of rare connection failures.
4965
4966 - "always" : this mode is only recommended when the path to the server is
4967 known for never breaking existing connections quickly after
4968 releasing them. It allows the first request of a session to be
4969 sent to an existing connection. This can provide a significant
4970 performance increase over the "safe" strategy when the backend
4971 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004972 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004973 sharing. It is recommended that the "http-keep-alive" timeout
4974 remains low in this mode so that no dead connections remain
4975 usable. In most cases, this will lead to the same performance
4976 gains as "aggressive" but with more risks. It should only be
4977 used when it improves the situation over "aggressive".
4978
4979 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004980 connection properties and compatibility. Specifically :
4981 - connections made with "usesrc" followed by a client-dependent value
4982 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004983
4984 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004985 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004986
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02004987 - connections with certain bogus authentication schemes (relying on the
4988 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004989
4990 No connection pool is involved, once a session dies, the last idle connection
4991 it was attached to is deleted at the same time. This ensures that connections
4992 may not last after all sessions are closed.
4993
4994 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4995 because almost no new connection will be established while idle connections
4996 remain available. This is particularly true with the "always" strategy.
4997
4998 See also : "option http-keep-alive", "server maxconn"
4999
5000
Mark Lamourinec2247f02012-01-04 13:02:01 -05005001http-send-name-header [<header>]
5002 Add the server name to a request. Use the header string given by <header>
5003
5004 May be used in sections: defaults | frontend | listen | backend
5005 yes | no | yes | yes
5006
5007 Arguments :
5008
5009 <header> The header string to use to send the server name
5010
5011 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01005012 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05005013 is added with the header string proved.
5014
5015 See also : "server"
5016
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005017id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005018 Set a persistent ID to a proxy.
5019 May be used in sections : defaults | frontend | listen | backend
5020 no | yes | yes | yes
5021 Arguments : none
5022
5023 Set a persistent ID for the proxy. This ID must be unique and positive.
5024 An unused ID will automatically be assigned if unset. The first assigned
5025 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005026
5027
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005028ignore-persist { if | unless } <condition>
5029 Declare a condition to ignore persistence
5030 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005031 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005032
5033 By default, when cookie persistence is enabled, every requests containing
5034 the cookie are unconditionally persistent (assuming the target server is up
5035 and running).
5036
5037 The "ignore-persist" statement allows one to declare various ACL-based
5038 conditions which, when met, will cause a request to ignore persistence.
5039 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005040 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005041 persistence for a specific User-Agent (for example, some web crawler bots).
5042
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005043 The persistence is ignored when an "if" condition is met, or unless an
5044 "unless" condition is met.
5045
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005046 Example:
5047 acl url_static path_beg /static /images /img /css
5048 acl url_static path_end .gif .png .jpg .css .js
5049 ignore-persist if url_static
5050
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005051 See also : "force-persist", "cookie", and section 7 about ACL usage.
5052
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005053load-server-state-from-file { global | local | none }
5054 Allow seamless reload of HAProxy
5055 May be used in sections: defaults | frontend | listen | backend
5056 yes | no | yes | yes
5057
5058 This directive points HAProxy to a file where server state from previous
5059 running process has been saved. That way, when starting up, before handling
5060 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005061 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005062 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5063 loading state or load states from a file containing all backends and servers.
5064 The state file can be generated by running the command "show servers state"
5065 over the stats socket and redirect output.
5066
Davor Ocelice9ed2812017-12-25 17:49:28 +01005067 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005068 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005069 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005070
5071 Arguments:
5072 global load the content of the file pointed by the global directive
5073 named "server-state-file".
5074
5075 local load the content of the file pointed by the directive
5076 "server-state-file-name" if set. If not set, then the backend
5077 name is used as a file name.
5078
5079 none don't load any stat for this backend
5080
5081 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005082 - server's IP address is preserved across reloads by default, but the
5083 order can be changed thanks to the server's "init-addr" setting. This
5084 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005085 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005086 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005087
5088 - server's weight is applied from previous running process unless it has
5089 has changed between previous and new configuration files.
5090
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005091 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005092
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005093 global
5094 stats socket /tmp/socket
5095 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005096
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005097 defaults
5098 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005099
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005100 backend bk
5101 server s1 127.0.0.1:22 check weight 11
5102 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005103
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005104
5105 Then one can run :
5106
5107 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5108
5109 Content of the file /tmp/server_state would be like this:
5110
5111 1
5112 # <field names skipped for the doc example>
5113 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5114 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5115
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005116 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005117
5118 global
5119 stats socket /tmp/socket
5120 server-state-base /etc/haproxy/states
5121
5122 defaults
5123 load-server-state-from-file local
5124
5125 backend bk
5126 server s1 127.0.0.1:22 check weight 11
5127 server s2 127.0.0.1:22 check weight 12
5128
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005129
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005130 Then one can run :
5131
5132 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5133
5134 Content of the file /etc/haproxy/states/bk would be like this:
5135
5136 1
5137 # <field names skipped for the doc example>
5138 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5139 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5140
5141 See also: "server-state-file", "server-state-file-name", and
5142 "show servers state"
5143
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005144
Willy Tarreau2769aa02007-12-27 18:26:09 +01005145log global
Willy Tarreauadb345d2018-11-12 07:56:13 +01005146log <address> [len <length>] [format <format>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005147no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005148 Enable per-instance logging of events and traffic.
5149 May be used in sections : defaults | frontend | listen | backend
5150 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005151
5152 Prefix :
5153 no should be used when the logger list must be flushed. For example,
5154 if you don't want to inherit from the default logger list. This
5155 prefix does not allow arguments.
5156
Willy Tarreau2769aa02007-12-27 18:26:09 +01005157 Arguments :
5158 global should be used when the instance's logging parameters are the
5159 same as the global ones. This is the most common usage. "global"
5160 replaces <address>, <facility> and <level> with those of the log
5161 entries found in the "global" section. Only one "log global"
5162 statement may be used per instance, and this form takes no other
5163 parameter.
5164
5165 <address> indicates where to send the logs. It takes the same format as
5166 for the "global" section's logs, and can be one of :
5167
5168 - An IPv4 address optionally followed by a colon (':') and a UDP
5169 port. If no port is specified, 514 is used by default (the
5170 standard syslog port).
5171
David du Colombier24bb5f52011-03-17 10:40:23 +01005172 - An IPv6 address followed by a colon (':') and optionally a UDP
5173 port. If no port is specified, 514 is used by default (the
5174 standard syslog port).
5175
Willy Tarreau2769aa02007-12-27 18:26:09 +01005176 - A filesystem path to a UNIX domain socket, keeping in mind
5177 considerations for chroot (be sure the path is accessible
5178 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005179 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005180
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005181 - A file descriptor number in the form "fd@<number>", which may
5182 point to a pipe, terminal, or socket. In this case unbuffered
5183 logs are used and one writev() call per log is performed. This
5184 is a bit expensive but acceptable for most workloads. Messages
5185 sent this way will not be truncated but may be dropped, in
5186 which case the DroppedLogs counter will be incremented. The
5187 writev() call is atomic even on pipes for messages up to
5188 PIPE_BUF size, which POSIX recommends to be at least 512 and
5189 which is 4096 bytes on most modern operating systems. Any
5190 larger message may be interleaved with messages from other
5191 processes. Exceptionally for debugging purposes the file
5192 descriptor may also be directed to a file, but doing so will
5193 significantly slow haproxy down as non-blocking calls will be
5194 ignored. Also there will be no way to purge nor rotate this
5195 file without restarting the process. Note that the configured
5196 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005197 with a TCP syslog server. See also the "short" and "raw"
5198 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005199
5200 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5201 and "fd@2", see above.
5202
5203 You may want to reference some environment variables in the
5204 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005205
Willy Tarreau18324f52014-06-27 18:10:07 +02005206 <length> is an optional maximum line length. Log lines larger than this
5207 value will be truncated before being sent. The reason is that
5208 syslog servers act differently on log line length. All servers
5209 support the default value of 1024, but some servers simply drop
5210 larger lines while others do log them. If a server supports long
5211 lines, it may make sense to set this value here in order to avoid
5212 truncating long lines. Similarly, if a server drops long lines,
5213 it is preferable to truncate them before sending them. Accepted
5214 values are 80 to 65535 inclusive. The default value of 1024 is
5215 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005216 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005217
Willy Tarreauadb345d2018-11-12 07:56:13 +01005218 <format> is the log format used when generating syslog messages. It may be
5219 one of the following :
5220
5221 rfc3164 The RFC3164 syslog message format. This is the default.
5222 (https://tools.ietf.org/html/rfc3164)
5223
5224 rfc5424 The RFC5424 syslog message format.
5225 (https://tools.ietf.org/html/rfc5424)
5226
Willy Tarreaue8746a02018-11-12 08:45:00 +01005227 short A message containing only a level between angle brackets such as
5228 '<3>', followed by the text. The PID, date, time, process name
5229 and system name are omitted. This is designed to be used with a
5230 local log server. This format is compatible with what the
5231 systemd logger consumes.
5232
Willy Tarreauc1b06452018-11-12 11:57:56 +01005233 raw A message containing only the text. The level, PID, date, time,
5234 process name and system name are omitted. This is designed to
5235 be used in containers or during development, where the severity
5236 only depends on the file descriptor used (stdout/stderr).
5237
Willy Tarreau2769aa02007-12-27 18:26:09 +01005238 <facility> must be one of the 24 standard syslog facilities :
5239
Willy Tarreaue8746a02018-11-12 08:45:00 +01005240 kern user mail daemon auth syslog lpr news
5241 uucp cron auth2 ftp ntp audit alert cron2
5242 local0 local1 local2 local3 local4 local5 local6 local7
5243
Willy Tarreauc1b06452018-11-12 11:57:56 +01005244 Note that the facility is ignored for the "short" and "raw"
5245 formats, but still required as a positional field. It is
5246 recommended to use "daemon" in this case to make it clear that
5247 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005248
5249 <level> is optional and can be specified to filter outgoing messages. By
5250 default, all messages are sent. If a level is specified, only
5251 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005252 will be sent. An optional minimum level can be specified. If it
5253 is set, logs emitted with a more severe level than this one will
5254 be capped to this level. This is used to avoid sending "emerg"
5255 messages on all terminals on some default syslog configurations.
5256 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005257
5258 emerg alert crit err warning notice info debug
5259
William Lallemand0f99e342011-10-12 17:50:54 +02005260 It is important to keep in mind that it is the frontend which decides what to
5261 log from a connection, and that in case of content switching, the log entries
5262 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005263
5264 However, backend log declaration define how and where servers status changes
5265 will be logged. Level "notice" will be used to indicate a server going up,
5266 "warning" will be used for termination signals and definitive service
5267 termination, and "alert" will be used for when a server goes down.
5268
5269 Note : According to RFC3164, messages are truncated to 1024 bytes before
5270 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005271
5272 Example :
5273 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005274 log stdout format short daemon # send log to systemd
5275 log stdout format raw daemon # send everything to stdout
5276 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005277 log 127.0.0.1:514 local0 notice # only send important events
5278 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005279 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005280
Willy Tarreau2769aa02007-12-27 18:26:09 +01005281
William Lallemand48940402012-01-30 16:47:22 +01005282log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005283 Specifies the log format string to use for traffic logs
5284 May be used in sections: defaults | frontend | listen | backend
5285 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005286
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005287 This directive specifies the log format string that will be used for all logs
5288 resulting from traffic passing through the frontend using this line. If the
5289 directive is used in a defaults section, all subsequent frontends will use
5290 the same log format. Please see section 8.2.4 which covers the log format
5291 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005292
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005293 "log-format" directive overrides previous "option tcplog", "log-format" and
5294 "option httplog" directives.
5295
Dragan Dosen7ad31542015-09-28 17:16:47 +02005296log-format-sd <string>
5297 Specifies the RFC5424 structured-data log format string
5298 May be used in sections: defaults | frontend | listen | backend
5299 yes | yes | yes | no
5300
5301 This directive specifies the RFC5424 structured-data log format string that
5302 will be used for all logs resulting from traffic passing through the frontend
5303 using this line. If the directive is used in a defaults section, all
5304 subsequent frontends will use the same log format. Please see section 8.2.4
5305 which covers the log format string in depth.
5306
5307 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5308 about the RFC5424 structured-data part.
5309
5310 Note : This log format string will be used only for loggers that have set
5311 log format to "rfc5424".
5312
5313 Example :
5314 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5315
5316
Willy Tarreau094af4e2015-01-07 15:03:42 +01005317log-tag <string>
5318 Specifies the log tag to use for all outgoing logs
5319 May be used in sections: defaults | frontend | listen | backend
5320 yes | yes | yes | yes
5321
5322 Sets the tag field in the syslog header to this string. It defaults to the
5323 log-tag set in the global section, otherwise the program name as launched
5324 from the command line, which usually is "haproxy". Sometimes it can be useful
5325 to differentiate between multiple processes running on the same host, or to
5326 differentiate customer instances running in the same process. In the backend,
5327 logs about servers up/down will use this tag. As a hint, it can be convenient
5328 to set a log-tag related to a hosted customer in a defaults section then put
5329 all the frontends and backends for that customer, then start another customer
5330 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005331
Willy Tarreauc35362a2014-04-25 13:58:37 +02005332max-keep-alive-queue <value>
5333 Set the maximum server queue size for maintaining keep-alive connections
5334 May be used in sections: defaults | frontend | listen | backend
5335 yes | no | yes | yes
5336
5337 HTTP keep-alive tries to reuse the same server connection whenever possible,
5338 but sometimes it can be counter-productive, for example if a server has a lot
5339 of connections while other ones are idle. This is especially true for static
5340 servers.
5341
5342 The purpose of this setting is to set a threshold on the number of queued
5343 connections at which haproxy stops trying to reuse the same server and prefers
5344 to find another one. The default value, -1, means there is no limit. A value
5345 of zero means that keep-alive requests will never be queued. For very close
5346 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005347 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005348 use a value of 10 or less). For remote servers suffering from a high latency,
5349 higher values might be needed to cover for the latency and/or the cost of
5350 picking a different server.
5351
5352 Note that this has no impact on responses which are maintained to the same
5353 server consecutively to a 401 response. They will still go to the same server
5354 even if they have to be queued.
5355
5356 See also : "option http-server-close", "option prefer-last-server", server
5357 "maxconn" and cookie persistence.
5358
5359
Willy Tarreau2769aa02007-12-27 18:26:09 +01005360maxconn <conns>
5361 Fix the maximum number of concurrent connections on a frontend
5362 May be used in sections : defaults | frontend | listen | backend
5363 yes | yes | yes | no
5364 Arguments :
5365 <conns> is the maximum number of concurrent connections the frontend will
5366 accept to serve. Excess connections will be queued by the system
5367 in the socket's listen queue and will be served once a connection
5368 closes.
5369
5370 If the system supports it, it can be useful on big sites to raise this limit
5371 very high so that haproxy manages connection queues, instead of leaving the
5372 clients with unanswered connection attempts. This value should not exceed the
5373 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005374 of tune.bufsize (16kB by default) each, as well as some other data resulting
5375 in about 33 kB of RAM being consumed per established connection. That means
5376 that a medium system equipped with 1GB of RAM can withstand around
5377 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005378
5379 Also, when <conns> is set to large values, it is possible that the servers
5380 are not sized to accept such loads, and for this reason it is generally wise
5381 to assign them some reasonable connection limits.
5382
Vincent Bernat6341be52012-06-27 17:18:30 +02005383 By default, this value is set to 2000.
5384
Willy Tarreau2769aa02007-12-27 18:26:09 +01005385 See also : "server", global section's "maxconn", "fullconn"
5386
5387
5388mode { tcp|http|health }
5389 Set the running mode or protocol of the instance
5390 May be used in sections : defaults | frontend | listen | backend
5391 yes | yes | yes | yes
5392 Arguments :
5393 tcp The instance will work in pure TCP mode. A full-duplex connection
5394 will be established between clients and servers, and no layer 7
5395 examination will be performed. This is the default mode. It
5396 should be used for SSL, SSH, SMTP, ...
5397
5398 http The instance will work in HTTP mode. The client request will be
5399 analyzed in depth before connecting to any server. Any request
5400 which is not RFC-compliant will be rejected. Layer 7 filtering,
5401 processing and switching will be possible. This is the mode which
5402 brings HAProxy most of its value.
5403
5404 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005405 to incoming connections and close the connection. Alternatively,
5406 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5407 instead. Nothing will be logged in either case. This mode is used
5408 to reply to external components health checks. This mode is
5409 deprecated and should not be used anymore as it is possible to do
5410 the same and even better by combining TCP or HTTP modes with the
5411 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005412
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005413 When doing content switching, it is mandatory that the frontend and the
5414 backend are in the same mode (generally HTTP), otherwise the configuration
5415 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005416
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005417 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005418 defaults http_instances
5419 mode http
5420
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005421 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005422
Willy Tarreau0ba27502007-12-24 16:55:16 +01005423
Cyril Bontéf0c60612010-02-06 14:44:47 +01005424monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005425 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005426 May be used in sections : defaults | frontend | listen | backend
5427 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005428 Arguments :
5429 if <cond> the monitor request will fail if the condition is satisfied,
5430 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005431 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005432 are met, for instance a low number of servers both in a
5433 backend and its backup.
5434
5435 unless <cond> the monitor request will succeed only if the condition is
5436 satisfied, and will fail otherwise. Such a condition may be
5437 based on a test on the presence of a minimum number of active
5438 servers in a list of backends.
5439
5440 This statement adds a condition which can force the response to a monitor
5441 request to report a failure. By default, when an external component queries
5442 the URI dedicated to monitoring, a 200 response is returned. When one of the
5443 conditions above is met, haproxy will return 503 instead of 200. This is
5444 very useful to report a site failure to an external component which may base
5445 routing advertisements between multiple sites on the availability reported by
5446 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005447 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5448 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005449
5450 Example:
5451 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005452 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005453 acl site_dead nbsrv(dynamic) lt 2
5454 acl site_dead nbsrv(static) lt 2
5455 monitor-uri /site_alive
5456 monitor fail if site_dead
5457
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005458 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005459
5460
5461monitor-net <source>
5462 Declare a source network which is limited to monitor requests
5463 May be used in sections : defaults | frontend | listen | backend
5464 yes | yes | yes | no
5465 Arguments :
5466 <source> is the source IPv4 address or network which will only be able to
5467 get monitor responses to any request. It can be either an IPv4
5468 address, a host name, or an address followed by a slash ('/')
5469 followed by a mask.
5470
5471 In TCP mode, any connection coming from a source matching <source> will cause
5472 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005473 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005474 forwarding the connection to a remote server.
5475
5476 In HTTP mode, a connection coming from a source matching <source> will be
5477 accepted, the following response will be sent without waiting for a request,
5478 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5479 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005480 running without forwarding the request to a backend server. Note that this
5481 response is sent in raw format, without any transformation. This is important
5482 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005483
Willy Tarreau82569f92012-09-27 23:48:56 +02005484 Monitor requests are processed very early, just after tcp-request connection
5485 ACLs which are the only ones able to block them. These connections are short
5486 lived and never wait for any data from the client. They cannot be logged, and
5487 it is the intended purpose. They are only used to report HAProxy's health to
5488 an upper component, nothing more. Please note that "monitor fail" rules do
5489 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005490
Willy Tarreau95cd2832010-03-04 23:36:33 +01005491 Last, please note that only one "monitor-net" statement can be specified in
5492 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005493
Willy Tarreau2769aa02007-12-27 18:26:09 +01005494 Example :
5495 # addresses .252 and .253 are just probing us.
5496 frontend www
5497 monitor-net 192.168.0.252/31
5498
5499 See also : "monitor fail", "monitor-uri"
5500
5501
5502monitor-uri <uri>
5503 Intercept a URI used by external components' monitor requests
5504 May be used in sections : defaults | frontend | listen | backend
5505 yes | yes | yes | no
5506 Arguments :
5507 <uri> is the exact URI which we want to intercept to return HAProxy's
5508 health status instead of forwarding the request.
5509
5510 When an HTTP request referencing <uri> will be received on a frontend,
5511 HAProxy will not forward it nor log it, but instead will return either
5512 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5513 conditions defined with "monitor fail". This is normally enough for any
5514 front-end HTTP probe to detect that the service is UP and running without
5515 forwarding the request to a backend server. Note that the HTTP method, the
5516 version and all headers are ignored, but the request must at least be valid
5517 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5518
Willy Tarreau721d8e02017-12-01 18:25:08 +01005519 Monitor requests are processed very early, just after the request is parsed
5520 and even before any "http-request" or "block" rulesets. The only rulesets
5521 applied before are the tcp-request ones. They cannot be logged either, and it
5522 is the intended purpose. They are only used to report HAProxy's health to an
5523 upper component, nothing more. However, it is possible to add any number of
5524 conditions using "monitor fail" and ACLs so that the result can be adjusted
5525 to whatever check can be imagined (most often the number of available servers
5526 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005527
5528 Example :
5529 # Use /haproxy_test to report haproxy's status
5530 frontend www
5531 mode http
5532 monitor-uri /haproxy_test
5533
5534 See also : "monitor fail", "monitor-net"
5535
Willy Tarreau0ba27502007-12-24 16:55:16 +01005536
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005537option abortonclose
5538no option abortonclose
5539 Enable or disable early dropping of aborted requests pending in queues.
5540 May be used in sections : defaults | frontend | listen | backend
5541 yes | no | yes | yes
5542 Arguments : none
5543
5544 In presence of very high loads, the servers will take some time to respond.
5545 The per-instance connection queue will inflate, and the response time will
5546 increase respective to the size of the queue times the average per-session
5547 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005548 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005549 the queue, and slowing down other users, and the servers as well, because the
5550 request will eventually be served, then aborted at the first error
5551 encountered while delivering the response.
5552
5553 As there is no way to distinguish between a full STOP and a simple output
5554 close on the client side, HTTP agents should be conservative and consider
5555 that the client might only have closed its output channel while waiting for
5556 the response. However, this introduces risks of congestion when lots of users
5557 do the same, and is completely useless nowadays because probably no client at
5558 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005559 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005560 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005561 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005562 of being the single component to break rare but valid traffic is extremely
5563 low, which adds to the temptation to be able to abort a session early while
5564 still not served and not pollute the servers.
5565
Davor Ocelice9ed2812017-12-25 17:49:28 +01005566 In HAProxy, the user can choose the desired behavior using the option
5567 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005568 compliant and aborted requests will be served. But when the option is
5569 specified, a session with an incoming channel closed will be aborted while
5570 it is still possible, either pending in the queue for a connection slot, or
5571 during the connection establishment if the server has not yet acknowledged
5572 the connection request. This considerably reduces the queue size and the load
5573 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005574 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005575
5576 If this option has been enabled in a "defaults" section, it can be disabled
5577 in a specific instance by prepending the "no" keyword before it.
5578
5579 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5580
5581
Willy Tarreau4076a152009-04-02 15:18:36 +02005582option accept-invalid-http-request
5583no option accept-invalid-http-request
5584 Enable or disable relaxing of HTTP request parsing
5585 May be used in sections : defaults | frontend | listen | backend
5586 yes | yes | yes | no
5587 Arguments : none
5588
Willy Tarreau91852eb2015-05-01 13:26:00 +02005589 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005590 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005591 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005592 forbidden characters are essentially used to build attacks exploiting server
5593 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5594 server will emit invalid header names for whatever reason (configuration,
5595 implementation) and the issue will not be immediately fixed. In such a case,
5596 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005597 even if that does not make sense, by specifying this option. Similarly, the
5598 list of characters allowed to appear in a URI is well defined by RFC3986, and
5599 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5600 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005601 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005602 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005603 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5604 to pass through (no version specified) and multiple digits for both the major
5605 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005606
5607 This option should never be enabled by default as it hides application bugs
5608 and open security breaches. It should only be deployed after a problem has
5609 been confirmed.
5610
5611 When this option is enabled, erroneous header names will still be accepted in
5612 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005613 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5614 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005615 also helps confirming that the issue has been solved.
5616
5617 If this option has been enabled in a "defaults" section, it can be disabled
5618 in a specific instance by prepending the "no" keyword before it.
5619
5620 See also : "option accept-invalid-http-response" and "show errors" on the
5621 stats socket.
5622
5623
5624option accept-invalid-http-response
5625no option accept-invalid-http-response
5626 Enable or disable relaxing of HTTP response parsing
5627 May be used in sections : defaults | frontend | listen | backend
5628 yes | no | yes | yes
5629 Arguments : none
5630
Willy Tarreau91852eb2015-05-01 13:26:00 +02005631 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005632 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005633 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005634 forbidden characters are essentially used to build attacks exploiting server
5635 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5636 server will emit invalid header names for whatever reason (configuration,
5637 implementation) and the issue will not be immediately fixed. In such a case,
5638 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005639 even if that does not make sense, by specifying this option. This option also
5640 relaxes the test on the HTTP version format, it allows multiple digits for
5641 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005642
5643 This option should never be enabled by default as it hides application bugs
5644 and open security breaches. It should only be deployed after a problem has
5645 been confirmed.
5646
5647 When this option is enabled, erroneous header names will still be accepted in
5648 responses, but the complete response will be captured in order to permit
5649 later analysis using the "show errors" request on the UNIX stats socket.
5650 Doing this also helps confirming that the issue has been solved.
5651
5652 If this option has been enabled in a "defaults" section, it can be disabled
5653 in a specific instance by prepending the "no" keyword before it.
5654
5655 See also : "option accept-invalid-http-request" and "show errors" on the
5656 stats socket.
5657
5658
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005659option allbackups
5660no option allbackups
5661 Use either all backup servers at a time or only the first one
5662 May be used in sections : defaults | frontend | listen | backend
5663 yes | no | yes | yes
5664 Arguments : none
5665
5666 By default, the first operational backup server gets all traffic when normal
5667 servers are all down. Sometimes, it may be preferred to use multiple backups
5668 at once, because one will not be enough. When "option allbackups" is enabled,
5669 the load balancing will be performed among all backup servers when all normal
5670 ones are unavailable. The same load balancing algorithm will be used and the
5671 servers' weights will be respected. Thus, there will not be any priority
5672 order between the backup servers anymore.
5673
5674 This option is mostly used with static server farms dedicated to return a
5675 "sorry" page when an application is completely offline.
5676
5677 If this option has been enabled in a "defaults" section, it can be disabled
5678 in a specific instance by prepending the "no" keyword before it.
5679
5680
5681option checkcache
5682no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005683 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005684 May be used in sections : defaults | frontend | listen | backend
5685 yes | no | yes | yes
5686 Arguments : none
5687
5688 Some high-level frameworks set application cookies everywhere and do not
5689 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005690 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005691 high risk of session crossing or stealing between users traversing the same
5692 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005693 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005694
5695 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005696 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005697 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005698 response to check if there's a risk of caching a cookie on a client-side
5699 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005700 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005701 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005702 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5703 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005704 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005705 - all those that result from a request using a method other than GET, HEAD,
5706 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005707 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005708 - those with a 'Pragma: no-cache' header
5709 - those with a 'Cache-control: private' header
5710 - those with a 'Cache-control: no-store' header
5711 - those with a 'Cache-control: max-age=0' header
5712 - those with a 'Cache-control: s-maxage=0' header
5713 - those with a 'Cache-control: no-cache' header
5714 - those with a 'Cache-control: no-cache="set-cookie"' header
5715 - those with a 'Cache-control: no-cache="set-cookie,' header
5716 (allowing other fields after set-cookie)
5717
5718 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005719 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005720 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005721 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005722 that admins are informed that there's something to be fixed.
5723
5724 Due to the high impact on the application, the application should be tested
5725 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005726 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005727 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005728
5729 If this option has been enabled in a "defaults" section, it can be disabled
5730 in a specific instance by prepending the "no" keyword before it.
5731
5732
5733option clitcpka
5734no option clitcpka
5735 Enable or disable the sending of TCP keepalive packets on the client side
5736 May be used in sections : defaults | frontend | listen | backend
5737 yes | yes | yes | no
5738 Arguments : none
5739
5740 When there is a firewall or any session-aware component between a client and
5741 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005742 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005743 components decides to expire a session which has remained idle for too long.
5744
5745 Enabling socket-level TCP keep-alives makes the system regularly send packets
5746 to the other end of the connection, leaving it active. The delay between
5747 keep-alive probes is controlled by the system only and depends both on the
5748 operating system and its tuning parameters.
5749
5750 It is important to understand that keep-alive packets are neither emitted nor
5751 received at the application level. It is only the network stacks which sees
5752 them. For this reason, even if one side of the proxy already uses keep-alives
5753 to maintain its connection alive, those keep-alive packets will not be
5754 forwarded to the other side of the proxy.
5755
5756 Please note that this has nothing to do with HTTP keep-alive.
5757
5758 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5759 client side of a connection, which should help when session expirations are
5760 noticed between HAProxy and a client.
5761
5762 If this option has been enabled in a "defaults" section, it can be disabled
5763 in a specific instance by prepending the "no" keyword before it.
5764
5765 See also : "option srvtcpka", "option tcpka"
5766
5767
Willy Tarreau0ba27502007-12-24 16:55:16 +01005768option contstats
5769 Enable continuous traffic statistics updates
5770 May be used in sections : defaults | frontend | listen | backend
5771 yes | yes | yes | no
5772 Arguments : none
5773
5774 By default, counters used for statistics calculation are incremented
5775 only when a session finishes. It works quite well when serving small
5776 objects, but with big ones (for example large images or archives) or
5777 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005778 a hedgehog. With this option enabled counters get incremented frequently
5779 along the session, typically every 5 seconds, which is often enough to
5780 produce clean graphs. Recounting touches a hotpath directly so it is not
5781 not enabled by default, as it can cause a lot of wakeups for very large
5782 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005783
5784
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005785option dontlog-normal
5786no option dontlog-normal
5787 Enable or disable logging of normal, successful connections
5788 May be used in sections : defaults | frontend | listen | backend
5789 yes | yes | yes | no
5790 Arguments : none
5791
5792 There are large sites dealing with several thousand connections per second
5793 and for which logging is a major pain. Some of them are even forced to turn
5794 logs off and cannot debug production issues. Setting this option ensures that
5795 normal connections, those which experience no error, no timeout, no retry nor
5796 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5797 mode, the response status code is checked and return codes 5xx will still be
5798 logged.
5799
5800 It is strongly discouraged to use this option as most of the time, the key to
5801 complex issues is in the normal logs which will not be logged here. If you
5802 need to separate logs, see the "log-separate-errors" option instead.
5803
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005804 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005805 logging.
5806
5807
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005808option dontlognull
5809no option dontlognull
5810 Enable or disable logging of null connections
5811 May be used in sections : defaults | frontend | listen | backend
5812 yes | yes | yes | no
5813 Arguments : none
5814
5815 In certain environments, there are components which will regularly connect to
5816 various systems to ensure that they are still alive. It can be the case from
5817 another load balancer as well as from monitoring systems. By default, even a
5818 simple port probe or scan will produce a log. If those connections pollute
5819 the logs too much, it is possible to enable option "dontlognull" to indicate
5820 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005821 which typically corresponds to those probes. Note that errors will still be
5822 returned to the client and accounted for in the stats. If this is not what is
5823 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005824
5825 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005826 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005827 would not be logged.
5828
5829 If this option has been enabled in a "defaults" section, it can be disabled
5830 in a specific instance by prepending the "no" keyword before it.
5831
Willy Tarreau0f228a02015-05-01 15:37:53 +02005832 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5833 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005834
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005835
Christopher Faulet315b39c2018-09-21 16:26:19 +02005836option forceclose (deprecated)
5837no option forceclose (deprecated)
5838 This is an alias for "option httpclose". Thus this option is deprecated.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005839
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005840 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005841
5842
Willy Tarreau87cf5142011-08-19 22:57:24 +02005843option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005844 Enable insertion of the X-Forwarded-For header to requests sent to servers
5845 May be used in sections : defaults | frontend | listen | backend
5846 yes | yes | yes | yes
5847 Arguments :
5848 <network> is an optional argument used to disable this option for sources
5849 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005850 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005851 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005852
5853 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5854 their client address. This is sometimes annoying when the client's IP address
5855 is expected in server logs. To solve this problem, the well-known HTTP header
5856 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5857 This header contains a value representing the client's IP address. Since this
5858 header is always appended at the end of the existing header list, the server
5859 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005860 the server's manual to find how to enable use of this standard header. Note
5861 that only the last occurrence of the header must be used, since it is really
5862 possible that the client has already brought one.
5863
Willy Tarreaud72758d2010-01-12 10:42:19 +01005864 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005865 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005866 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005867 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005868 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005869 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005870
5871 Sometimes, a same HAProxy instance may be shared between a direct client
5872 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5873 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5874 header for a known source address or network by adding the "except" keyword
5875 followed by the network address. In this case, any source IP matching the
5876 network will not cause an addition of this header. Most common uses are with
5877 private networks or 127.0.0.1.
5878
Willy Tarreau87cf5142011-08-19 22:57:24 +02005879 Alternatively, the keyword "if-none" states that the header will only be
5880 added if it is not present. This should only be used in perfectly trusted
5881 environment, as this might cause a security issue if headers reaching haproxy
5882 are under the control of the end-user.
5883
Willy Tarreauc27debf2008-01-06 08:57:02 +01005884 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005885 least one of them uses it, the header will be added. Note that the backend's
5886 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005887 both are defined. In the case of the "if-none" argument, if at least one of
5888 the frontend or the backend does not specify it, it wants the addition to be
5889 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005890
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005891 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005892 # Public HTTP address also used by stunnel on the same machine
5893 frontend www
5894 mode http
5895 option forwardfor except 127.0.0.1 # stunnel already adds the header
5896
Ross Westaf72a1d2008-08-03 10:51:45 +02005897 # Those servers want the IP Address in X-Client
5898 backend www
5899 mode http
5900 option forwardfor header X-Client
5901
Willy Tarreau87cf5142011-08-19 22:57:24 +02005902 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005903 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005904
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005905
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005906option http-buffer-request
5907no option http-buffer-request
5908 Enable or disable waiting for whole HTTP request body before proceeding
5909 May be used in sections : defaults | frontend | listen | backend
5910 yes | yes | yes | yes
5911 Arguments : none
5912
5913 It is sometimes desirable to wait for the body of an HTTP request before
5914 taking a decision. This is what is being done by "balance url_param" for
5915 example. The first use case is to buffer requests from slow clients before
5916 connecting to the server. Another use case consists in taking the routing
5917 decision based on the request body's contents. This option placed in a
5918 frontend or backend forces the HTTP processing to wait until either the whole
5919 body is received, or the request buffer is full, or the first chunk is
5920 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005921 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005922 the frontend and the backend, so this should definitely not be used by
5923 default.
5924
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005925 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005926
5927
Willy Tarreau0f228a02015-05-01 15:37:53 +02005928option http-ignore-probes
5929no option http-ignore-probes
5930 Enable or disable logging of null connections and request timeouts
5931 May be used in sections : defaults | frontend | listen | backend
5932 yes | yes | yes | no
5933 Arguments : none
5934
5935 Recently some browsers started to implement a "pre-connect" feature
5936 consisting in speculatively connecting to some recently visited web sites
5937 just in case the user would like to visit them. This results in many
5938 connections being established to web sites, which end up in 408 Request
5939 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5940 decides to close them first. These ones pollute the log and feed the error
5941 counters. There was already "option dontlognull" but it's insufficient in
5942 this case. Instead, this option does the following things :
5943 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005944 was received over a connection before it was closed;
5945 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005946 - prevent any error counter from being incremented
5947
5948 That way the empty connection is silently ignored. Note that it is better
5949 not to use this unless it is clear that it is needed, because it will hide
5950 real problems. The most common reason for not receiving a request and seeing
5951 a 408 is due to an MTU inconsistency between the client and an intermediary
5952 element such as a VPN, which blocks too large packets. These issues are
5953 generally seen with POST requests as well as GET with large cookies. The logs
5954 are often the only way to detect them.
5955
5956 If this option has been enabled in a "defaults" section, it can be disabled
5957 in a specific instance by prepending the "no" keyword before it.
5958
5959 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5960
5961
Willy Tarreau16bfb022010-01-16 19:48:41 +01005962option http-keep-alive
5963no option http-keep-alive
5964 Enable or disable HTTP keep-alive from client to server
5965 May be used in sections : defaults | frontend | listen | backend
5966 yes | yes | yes | yes
5967 Arguments : none
5968
Willy Tarreau70dffda2014-01-30 03:07:23 +01005969 By default HAProxy operates in keep-alive mode with regards to persistent
5970 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02005971 leaves the connection idle on both sides between the end of a response and
5972 the start of a new request. This mode may be changed by several options such
5973 as "option http-server-close", "option httpclose" or "option http-tunnel".
5974 This option allows to set back the keep-alive mode, which can be useful when
5975 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01005976
5977 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5978 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005979 network) and the fastest session reuse on the server side at the expense
5980 of maintaining idle connections to the servers. In general, it is possible
5981 with this option to achieve approximately twice the request rate that the
5982 "http-server-close" option achieves on small objects. There are mainly two
5983 situations where this option may be useful :
5984
5985 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005986 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005987
5988 - when the cost of establishing the connection to the server is significant
5989 compared to the cost of retrieving the associated object from the server.
5990
5991 This last case can happen when the server is a fast static server of cache.
5992 In this case, the server will need to be properly tuned to support high enough
5993 connection counts because connections will last until the client sends another
5994 request.
5995
5996 If the client request has to go to another backend or another server due to
5997 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01005998 immediately be closed and a new one re-opened. Option "prefer-last-server" is
5999 available to try optimize server selection so that if the server currently
6000 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006001
Willy Tarreau16bfb022010-01-16 19:48:41 +01006002 At the moment, logs will not indicate whether requests came from the same
6003 session or not. The accept date reported in the logs corresponds to the end
6004 of the previous request, and the request time corresponds to the time spent
6005 waiting for a new request. The keep-alive request time is still bound to the
6006 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6007 not set.
6008
Cyril Bonté653dcd62014-02-20 00:13:15 +01006009 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02006010 http-server-close" or "option http-tunnel". When backend and frontend options
6011 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006012
Christopher Faulet315b39c2018-09-21 16:26:19 +02006013 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006014 "option prefer-last-server", "option http-pretend-keepalive",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006015 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006016
6017
Willy Tarreau96e31212011-05-30 18:10:30 +02006018option http-no-delay
6019no option http-no-delay
6020 Instruct the system to favor low interactive delays over performance in HTTP
6021 May be used in sections : defaults | frontend | listen | backend
6022 yes | yes | yes | yes
6023 Arguments : none
6024
6025 In HTTP, each payload is unidirectional and has no notion of interactivity.
6026 Any agent is expected to queue data somewhat for a reasonably low delay.
6027 There are some very rare server-to-server applications that abuse the HTTP
6028 protocol and expect the payload phase to be highly interactive, with many
6029 interleaved data chunks in both directions within a single request. This is
6030 absolutely not supported by the HTTP specification and will not work across
6031 most proxies or servers. When such applications attempt to do this through
6032 haproxy, it works but they will experience high delays due to the network
6033 optimizations which favor performance by instructing the system to wait for
6034 enough data to be available in order to only send full packets. Typical
6035 delays are around 200 ms per round trip. Note that this only happens with
6036 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6037 affected.
6038
6039 When "option http-no-delay" is present in either the frontend or the backend
6040 used by a connection, all such optimizations will be disabled in order to
6041 make the exchanges as fast as possible. Of course this offers no guarantee on
6042 the functionality, as it may break at any other place. But if it works via
6043 HAProxy, it will work as fast as possible. This option should never be used
6044 by default, and should never be used at all unless such a buggy application
6045 is discovered. The impact of using this option is an increase of bandwidth
6046 usage and CPU usage, which may significantly lower performance in high
6047 latency environments.
6048
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006049 See also : "option http-buffer-request"
6050
Willy Tarreau96e31212011-05-30 18:10:30 +02006051
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006052option http-pretend-keepalive
6053no option http-pretend-keepalive
6054 Define whether haproxy will announce keepalive to the server or not
6055 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006056 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006057 Arguments : none
6058
Christopher Faulet315b39c2018-09-21 16:26:19 +02006059 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006060 adds a "Connection: close" header to the request forwarded to the server.
6061 Unfortunately, when some servers see this header, they automatically refrain
6062 from using the chunked encoding for responses of unknown length, while this
6063 is totally unrelated. The immediate effect is that this prevents haproxy from
6064 maintaining the client connection alive. A second effect is that a client or
6065 a cache could receive an incomplete response without being aware of it, and
6066 consider the response complete.
6067
6068 By setting "option http-pretend-keepalive", haproxy will make the server
6069 believe it will keep the connection alive. The server will then not fall back
6070 to the abnormal undesired above. When haproxy gets the whole response, it
6071 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006072 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006073 connection is correctly closed on the server side.
6074
6075 It is recommended not to enable this option by default, because most servers
6076 will more efficiently close the connection themselves after the last packet,
6077 and release its buffers slightly earlier. Also, the added packet on the
6078 network could slightly reduce the overall peak performance. However it is
6079 worth noting that when this option is enabled, haproxy will have slightly
6080 less work to do. So if haproxy is the bottleneck on the whole architecture,
6081 enabling this option might save a few CPU cycles.
6082
Christopher Faulet98db9762018-09-21 10:25:19 +02006083 This option may be set in backend and listen sections. Using it in a frontend
6084 section will be ignored and a warning will be reported during startup. It is
6085 a backend related option, so there is no real reason to set it on a
6086 frontend. This option may be combined with "option httpclose", which will
6087 cause keepalive to be announced to the server and close to be announced to
6088 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006089
6090 If this option has been enabled in a "defaults" section, it can be disabled
6091 in a specific instance by prepending the "no" keyword before it.
6092
Christopher Faulet315b39c2018-09-21 16:26:19 +02006093 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006094 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006095
Willy Tarreauc27debf2008-01-06 08:57:02 +01006096
Willy Tarreaub608feb2010-01-02 22:47:18 +01006097option http-server-close
6098no option http-server-close
6099 Enable or disable HTTP connection closing on the server side
6100 May be used in sections : defaults | frontend | listen | backend
6101 yes | yes | yes | yes
6102 Arguments : none
6103
Willy Tarreau70dffda2014-01-30 03:07:23 +01006104 By default HAProxy operates in keep-alive mode with regards to persistent
6105 connections: for each connection it processes each request and response, and
6106 leaves the connection idle on both sides between the end of a response and
6107 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006108 as "option http-server-close", "option httpclose" or "option http-tunnel".
6109 Setting "option http-server-close" enables HTTP connection-close mode on the
6110 server side while keeping the ability to support HTTP keep-alive and
6111 pipelining on the client side. This provides the lowest latency on the client
6112 side (slow network) and the fastest session reuse on the server side to save
6113 server resources, similarly to "option httpclose". It also permits
6114 non-keepalive capable servers to be served in keep-alive mode to the clients
6115 if they conform to the requirements of RFC7230. Please note that some servers
6116 do not always conform to those requirements when they see "Connection: close"
6117 in the request. The effect will be that keep-alive will never be used. A
6118 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006119
6120 At the moment, logs will not indicate whether requests came from the same
6121 session or not. The accept date reported in the logs corresponds to the end
6122 of the previous request, and the request time corresponds to the time spent
6123 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006124 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6125 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006126
6127 This option may be set both in a frontend and in a backend. It is enabled if
6128 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006129 It disables and replaces any previous "option httpclose", "option http-tunnel"
6130 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6131 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006132
6133 If this option has been enabled in a "defaults" section, it can be disabled
6134 in a specific instance by prepending the "no" keyword before it.
6135
Christopher Faulet315b39c2018-09-21 16:26:19 +02006136 See also : "option httpclose", "option http-pretend-keepalive",
6137 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006138
6139
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006140option http-tunnel
6141no option http-tunnel
6142 Disable or enable HTTP connection processing after first transaction
6143 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006144 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006145 Arguments : none
6146
Willy Tarreau70dffda2014-01-30 03:07:23 +01006147 By default HAProxy operates in keep-alive mode with regards to persistent
6148 connections: for each connection it processes each request and response, and
6149 leaves the connection idle on both sides between the end of a response and
6150 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006151 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006152
6153 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006154 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006155 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6156 is normally not needed anymore unless in very specific cases such as when
6157 using an in-house protocol that looks like HTTP but is not compatible, or
6158 just to log one request per client in order to reduce log size. Note that
6159 everything which works at the HTTP level, including header parsing/addition,
6160 cookie processing or content switching will only work for the first request
6161 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006162
Christopher Faulet4212a302018-09-21 10:42:19 +02006163 This option may be set on frontend and listen sections. Using it on a backend
6164 section will be ignored and a warning will be reported during the startup. It
6165 is a frontend related option, so there is no real reason to set it on a
6166 backend.
6167
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006168 If this option has been enabled in a "defaults" section, it can be disabled
6169 in a specific instance by prepending the "no" keyword before it.
6170
Christopher Faulet315b39c2018-09-21 16:26:19 +02006171 See also : "option httpclose", "option http-server-close",
6172 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006173
6174
Willy Tarreau88d349d2010-01-25 12:15:43 +01006175option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006176no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006177 Make use of non-standard Proxy-Connection header instead of Connection
6178 May be used in sections : defaults | frontend | listen | backend
6179 yes | yes | yes | no
6180 Arguments : none
6181
Lukas Tribus23953682017-04-28 13:24:30 +00006182 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006183 Connection header to indicate their wish of persistent or non-persistent
6184 connections, both browsers and proxies ignore this header for proxied
6185 connections and make use of the undocumented, non-standard Proxy-Connection
6186 header instead. The issue begins when trying to put a load balancer between
6187 browsers and such proxies, because there will be a difference between what
6188 haproxy understands and what the client and the proxy agree on.
6189
6190 By setting this option in a frontend, haproxy can automatically switch to use
6191 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006192 defined here as one where the URI begins with neither a '/' nor a '*'. This
6193 is incompatible with the HTTP tunnel mode. Note that this option can only be
6194 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006195
Willy Tarreau844a7e72010-01-31 21:46:18 +01006196 Also, when this option is set, a request which requires authentication will
6197 automatically switch to use proxy authentication headers if it is itself a
6198 proxied request. That makes it possible to check or enforce authentication in
6199 front of an existing proxy.
6200
Willy Tarreau88d349d2010-01-25 12:15:43 +01006201 This option should normally never be used, except in front of a proxy.
6202
Christopher Faulet315b39c2018-09-21 16:26:19 +02006203 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006204
6205
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006206option http-use-htx
6207no option http-use-htx
6208 Switch to the new HTX internal representation for HTTP protocol elements
6209 May be used in sections : defaults | frontend | listen | backend
6210 yes | yes | yes | yes
6211 Arguments : none
6212
6213 By default, the HTTP protocol is processed as-is. Inserting, deleting, or
6214 modifying a header field requires to rewrite the affected part in the buffer
6215 and to move the buffer's tail accordingly. Since this principle has deep
6216 roots in haproxy, the HTTP/2 protocol is converted to HTTP/1.1 before being
6217 processed this way. It also results in the inability to establish HTTP/2
6218 connections to servers because of the loss of HTTP/2 semantics in the HTTP/1
6219 representation.
6220
6221 HTX is the name of a totally new native internal representation for the HTTP
6222 protocol, that is agnostic to the version and aims at preserving semantics
6223 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6224 the protocol elements so that no more memory moves are necessary and that
6225 most elements are directly accessed. This mechanism is still limited to the
6226 most basic operations (no compression, filters, Lua, applets, cache, etc).
6227 But it supports using either HTTP/1 or HTTP/2 on any side regardless of the
6228 other side's version.
6229
6230 This option indicates that HTX needs to be used. It will cause errors to be
6231 emitted if incompatible features are used, but will allow H2 to be selected
6232 as a server protocol. It is recommended to use this option on new reasonably
6233 simple configurations, but since the feature still has incomplete functional
6234 coverage, it is not enabled by default.
6235
6236 See also : "mode http"
6237
6238
Willy Tarreaud63335a2010-02-26 12:56:52 +01006239option httpchk
6240option httpchk <uri>
6241option httpchk <method> <uri>
6242option httpchk <method> <uri> <version>
6243 Enable HTTP protocol to check on the servers health
6244 May be used in sections : defaults | frontend | listen | backend
6245 yes | no | yes | yes
6246 Arguments :
6247 <method> is the optional HTTP method used with the requests. When not set,
6248 the "OPTIONS" method is used, as it generally requires low server
6249 processing and is easy to filter out from the logs. Any method
6250 may be used, though it is not recommended to invent non-standard
6251 ones.
6252
6253 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6254 which is accessible by default on almost any server, but may be
6255 changed to any other URI. Query strings are permitted.
6256
6257 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6258 but some servers might behave incorrectly in HTTP 1.0, so turning
6259 it to HTTP/1.1 may sometimes help. Note that the Host field is
6260 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6261 after "\r\n" following the version string.
6262
6263 By default, server health checks only consist in trying to establish a TCP
6264 connection. When "option httpchk" is specified, a complete HTTP request is
6265 sent once the TCP connection is established, and responses 2xx and 3xx are
6266 considered valid, while all other ones indicate a server failure, including
6267 the lack of any response.
6268
6269 The port and interval are specified in the server configuration.
6270
6271 This option does not necessarily require an HTTP backend, it also works with
6272 plain TCP backends. This is particularly useful to check simple scripts bound
6273 to some dedicated ports using the inetd daemon.
6274
6275 Examples :
6276 # Relay HTTPS traffic to Apache instance and check service availability
6277 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6278 backend https_relay
6279 mode tcp
6280 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6281 server apache1 192.168.1.1:443 check port 80
6282
Simon Hormanafc47ee2013-11-25 10:46:35 +09006283 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6284 "option pgsql-check", "http-check" and the "check", "port" and
6285 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006286
6287
Willy Tarreauc27debf2008-01-06 08:57:02 +01006288option httpclose
6289no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006290 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006291 May be used in sections : defaults | frontend | listen | backend
6292 yes | yes | yes | yes
6293 Arguments : none
6294
Willy Tarreau70dffda2014-01-30 03:07:23 +01006295 By default HAProxy operates in keep-alive mode with regards to persistent
6296 connections: for each connection it processes each request and response, and
6297 leaves the connection idle on both sides between the end of a response and
6298 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006299 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006300
Christopher Faulet315b39c2018-09-21 16:26:19 +02006301 If "option httpclose" is set, HAProxy will close connections with the server
6302 and the client as soon as the request and the response are received. It will
6303 alos check if a "Connection: close" header is already set in each direction,
6304 and will add one if missing. Any "Connection" header different from "close"
6305 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006306
Christopher Faulet315b39c2018-09-21 16:26:19 +02006307 This option may also be combined with "option http-pretend-keepalive", which
6308 will disable sending of the "Connection: close" header, but will still cause
6309 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006310
6311 This option may be set both in a frontend and in a backend. It is enabled if
6312 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006313 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006314 "option http-keep-alive" or "option http-tunnel". Please check section 4
6315 ("Proxies") to see how this option combines with others when frontend and
6316 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006317
6318 If this option has been enabled in a "defaults" section, it can be disabled
6319 in a specific instance by prepending the "no" keyword before it.
6320
Christopher Faulet315b39c2018-09-21 16:26:19 +02006321 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006322
6323
Emeric Brun3a058f32009-06-30 18:26:00 +02006324option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006325 Enable logging of HTTP request, session state and timers
6326 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006327 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006328 Arguments :
6329 clf if the "clf" argument is added, then the output format will be
6330 the CLF format instead of HAProxy's default HTTP format. You can
6331 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006332 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006333 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006334
6335 By default, the log output format is very poor, as it only contains the
6336 source and destination addresses, and the instance name. By specifying
6337 "option httplog", each log line turns into a much richer format including,
6338 but not limited to, the HTTP request, the connection timers, the session
6339 status, the connections numbers, the captured headers and cookies, the
6340 frontend, backend and server name, and of course the source address and
6341 ports.
6342
PiBa-NLbd556bf2014-12-11 21:31:54 +01006343 Specifying only "option httplog" will automatically clear the 'clf' mode
6344 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006345
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006346 "option httplog" overrides any previous "log-format" directive.
6347
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006348 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006349
Willy Tarreau55165fe2009-05-10 12:02:55 +02006350
6351option http_proxy
6352no option http_proxy
6353 Enable or disable plain HTTP proxy mode
6354 May be used in sections : defaults | frontend | listen | backend
6355 yes | yes | yes | yes
6356 Arguments : none
6357
6358 It sometimes happens that people need a pure HTTP proxy which understands
6359 basic proxy requests without caching nor any fancy feature. In this case,
6360 it may be worth setting up an HAProxy instance with the "option http_proxy"
6361 set. In this mode, no server is declared, and the connection is forwarded to
6362 the IP address and port found in the URL after the "http://" scheme.
6363
6364 No host address resolution is performed, so this only works when pure IP
6365 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006366 it will probably be used only by experts who know they need exactly it. This
6367 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006368
6369 If this option has been enabled in a "defaults" section, it can be disabled
6370 in a specific instance by prepending the "no" keyword before it.
6371
6372 Example :
6373 # this backend understands HTTP proxy requests and forwards them directly.
6374 backend direct_forward
6375 option httpclose
6376 option http_proxy
6377
6378 See also : "option httpclose"
6379
Willy Tarreau211ad242009-10-03 21:45:07 +02006380
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006381option independent-streams
6382no option independent-streams
6383 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006384 May be used in sections : defaults | frontend | listen | backend
6385 yes | yes | yes | yes
6386 Arguments : none
6387
6388 By default, when data is sent over a socket, both the write timeout and the
6389 read timeout for that socket are refreshed, because we consider that there is
6390 activity on that socket, and we have no other means of guessing if we should
6391 receive data or not.
6392
Davor Ocelice9ed2812017-12-25 17:49:28 +01006393 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006394 exists a situation where it is desirable to disable it, and only refresh the
6395 read timeout if there are incoming data. This happens on sessions with large
6396 timeouts and low amounts of exchanged data such as telnet session. If the
6397 server suddenly disappears, the output data accumulates in the system's
6398 socket buffers, both timeouts are correctly refreshed, and there is no way
6399 to know the server does not receive them, so we don't timeout. However, when
6400 the underlying protocol always echoes sent data, it would be enough by itself
6401 to detect the issue using the read timeout. Note that this problem does not
6402 happen with more verbose protocols because data won't accumulate long in the
6403 socket buffers.
6404
6405 When this option is set on the frontend, it will disable read timeout updates
6406 on data sent to the client. There probably is little use of this case. When
6407 the option is set on the backend, it will disable read timeout updates on
6408 data sent to the server. Doing so will typically break large HTTP posts from
6409 slow lines, so use it with caution.
6410
Lukas Tribus745f15e2018-11-08 12:41:42 +01006411 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006412 with a spelling mistake. This spelling is still supported but
6413 deprecated.
6414
Willy Tarreauce887fd2012-05-12 12:50:00 +02006415 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006416
6417
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006418option ldap-check
6419 Use LDAPv3 health checks for server testing
6420 May be used in sections : defaults | frontend | listen | backend
6421 yes | no | yes | yes
6422 Arguments : none
6423
6424 It is possible to test that the server correctly talks LDAPv3 instead of just
6425 testing that it accepts the TCP connection. When this option is set, an
6426 LDAPv3 anonymous simple bind message is sent to the server, and the response
6427 is analyzed to find an LDAPv3 bind response message.
6428
6429 The server is considered valid only when the LDAP response contains success
6430 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6431
6432 Logging of bind requests is server dependent see your documentation how to
6433 configure it.
6434
6435 Example :
6436 option ldap-check
6437
6438 See also : "option httpchk"
6439
6440
Simon Horman98637e52014-06-20 12:30:16 +09006441option external-check
6442 Use external processes for server health checks
6443 May be used in sections : defaults | frontend | listen | backend
6444 yes | no | yes | yes
6445
6446 It is possible to test the health of a server using an external command.
6447 This is achieved by running the executable set using "external-check
6448 command".
6449
6450 Requires the "external-check" global to be set.
6451
6452 See also : "external-check", "external-check command", "external-check path"
6453
6454
Willy Tarreau211ad242009-10-03 21:45:07 +02006455option log-health-checks
6456no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006457 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006458 May be used in sections : defaults | frontend | listen | backend
6459 yes | no | yes | yes
6460 Arguments : none
6461
Willy Tarreaubef1b322014-05-13 21:01:39 +02006462 By default, failed health check are logged if server is UP and successful
6463 health checks are logged if server is DOWN, so the amount of additional
6464 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006465
Willy Tarreaubef1b322014-05-13 21:01:39 +02006466 When this option is enabled, any change of the health check status or to
6467 the server's health will be logged, so that it becomes possible to know
6468 that a server was failing occasional checks before crashing, or exactly when
6469 it failed to respond a valid HTTP status, then when the port started to
6470 reject connections, then when the server stopped responding at all.
6471
Davor Ocelice9ed2812017-12-25 17:49:28 +01006472 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006473 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006474
Willy Tarreaubef1b322014-05-13 21:01:39 +02006475 See also: "option httpchk", "option ldap-check", "option mysql-check",
6476 "option pgsql-check", "option redis-check", "option smtpchk",
6477 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006478
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006479
6480option log-separate-errors
6481no option log-separate-errors
6482 Change log level for non-completely successful connections
6483 May be used in sections : defaults | frontend | listen | backend
6484 yes | yes | yes | no
6485 Arguments : none
6486
6487 Sometimes looking for errors in logs is not easy. This option makes haproxy
6488 raise the level of logs containing potentially interesting information such
6489 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6490 level changes from "info" to "err". This makes it possible to log them
6491 separately to a different file with most syslog daemons. Be careful not to
6492 remove them from the original file, otherwise you would lose ordering which
6493 provides very important information.
6494
6495 Using this option, large sites dealing with several thousand connections per
6496 second may log normal traffic to a rotating buffer and only archive smaller
6497 error logs.
6498
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006499 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006500 logging.
6501
Willy Tarreauc27debf2008-01-06 08:57:02 +01006502
6503option logasap
6504no option logasap
6505 Enable or disable early logging of HTTP requests
6506 May be used in sections : defaults | frontend | listen | backend
6507 yes | yes | yes | no
6508 Arguments : none
6509
6510 By default, HTTP requests are logged upon termination so that the total
6511 transfer time and the number of bytes appear in the logs. When large objects
6512 are being transferred, it may take a while before the request appears in the
6513 logs. Using "option logasap", the request gets logged as soon as the server
6514 sends the complete headers. The only missing information in the logs will be
6515 the total number of bytes which will indicate everything except the amount
6516 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006517 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006518 "Content-Length" response header so that the logs at least indicate how many
6519 bytes are expected to be transferred.
6520
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006521 Examples :
6522 listen http_proxy 0.0.0.0:80
6523 mode http
6524 option httplog
6525 option logasap
6526 log 192.168.2.200 local3
6527
6528 >>> Feb 6 12:14:14 localhost \
6529 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6530 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6531 "GET /image.iso HTTP/1.0"
6532
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006533 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006534 logging.
6535
6536
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006537option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006538 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006539 May be used in sections : defaults | frontend | listen | backend
6540 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006541 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006542 <username> This is the username which will be used when connecting to MySQL
6543 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006544 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006545
6546 If you specify a username, the check consists of sending two MySQL packet,
6547 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006548 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006549 Error packet. It is a basic but useful test which does not produce error nor
6550 aborted connect on the server. However, it requires adding an authorization
6551 in the MySQL table, like this :
6552
6553 USE mysql;
6554 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6555 FLUSH PRIVILEGES;
6556
6557 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006558 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006559 Error packet, we don't send anything in this mode. It was reported that it
6560 can generate lockout if check is too frequent and/or if there is not enough
6561 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6562 value as if a connection is established successfully within fewer than MySQL
6563 "max_connect_errors" attempts after a previous connection was interrupted,
6564 the error count for the host is cleared to zero. If HAProxy's server get
6565 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6566
6567 Remember that this does not check database presence nor database consistency.
6568 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006569
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006570 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006571
6572 Most often, an incoming MySQL server needs to see the client's IP address for
6573 various purposes, including IP privilege matching and connection logging.
6574 When possible, it is often wise to masquerade the client's IP address when
6575 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006576 which requires the transparent proxy feature to be compiled in, and the MySQL
6577 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006578
6579 See also: "option httpchk"
6580
6581
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006582option nolinger
6583no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006584 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006585 May be used in sections: defaults | frontend | listen | backend
6586 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006587 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006588
Davor Ocelice9ed2812017-12-25 17:49:28 +01006589 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006590 physically disconnected), the session timeouts triggers and the session is
6591 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6592 using some resources and possibly limiting the ability to establish newer
6593 connections.
6594
6595 When this happens, it is possible to activate "option nolinger" which forces
6596 the system to immediately remove any socket's pending data on close. Thus,
6597 the session is instantly purged from the system's tables. This usually has
6598 side effects such as increased number of TCP resets due to old retransmits
6599 getting immediately rejected. Some firewalls may sometimes complain about
6600 this too.
6601
6602 For this reason, it is not recommended to use this option when not absolutely
6603 needed. You know that you need it when you have thousands of FIN_WAIT1
6604 sessions on your system (TIME_WAIT ones do not count).
6605
6606 This option may be used both on frontends and backends, depending on the side
6607 where it is required. Use it on the frontend for clients, and on the backend
6608 for servers.
6609
6610 If this option has been enabled in a "defaults" section, it can be disabled
6611 in a specific instance by prepending the "no" keyword before it.
6612
6613
Willy Tarreau55165fe2009-05-10 12:02:55 +02006614option originalto [ except <network> ] [ header <name> ]
6615 Enable insertion of the X-Original-To header to requests sent to servers
6616 May be used in sections : defaults | frontend | listen | backend
6617 yes | yes | yes | yes
6618 Arguments :
6619 <network> is an optional argument used to disable this option for sources
6620 matching <network>
6621 <name> an optional argument to specify a different "X-Original-To"
6622 header name.
6623
6624 Since HAProxy can work in transparent mode, every request from a client can
6625 be redirected to the proxy and HAProxy itself can proxy every request to a
6626 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6627 be lost. This is annoying when you want access rules based on destination ip
6628 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6629 added by HAProxy to all requests sent to the server. This header contains a
6630 value representing the original destination IP address. Since this must be
6631 configured to always use the last occurrence of this header only. Note that
6632 only the last occurrence of the header must be used, since it is really
6633 possible that the client has already brought one.
6634
6635 The keyword "header" may be used to supply a different header name to replace
6636 the default "X-Original-To". This can be useful where you might already
6637 have a "X-Original-To" header from a different application, and you need
6638 preserve it. Also if your backend server doesn't use the "X-Original-To"
6639 header and requires different one.
6640
6641 Sometimes, a same HAProxy instance may be shared between a direct client
6642 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6643 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6644 header for a known source address or network by adding the "except" keyword
6645 followed by the network address. In this case, any source IP matching the
6646 network will not cause an addition of this header. Most common uses are with
6647 private networks or 127.0.0.1.
6648
6649 This option may be specified either in the frontend or in the backend. If at
6650 least one of them uses it, the header will be added. Note that the backend's
6651 setting of the header subargument takes precedence over the frontend's if
6652 both are defined.
6653
Willy Tarreau55165fe2009-05-10 12:02:55 +02006654 Examples :
6655 # Original Destination address
6656 frontend www
6657 mode http
6658 option originalto except 127.0.0.1
6659
6660 # Those servers want the IP Address in X-Client-Dst
6661 backend www
6662 mode http
6663 option originalto header X-Client-Dst
6664
Christopher Faulet315b39c2018-09-21 16:26:19 +02006665 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006666
6667
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006668option persist
6669no option persist
6670 Enable or disable forced persistence on down servers
6671 May be used in sections: defaults | frontend | listen | backend
6672 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006673 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006674
6675 When an HTTP request reaches a backend with a cookie which references a dead
6676 server, by default it is redispatched to another server. It is possible to
6677 force the request to be sent to the dead server first using "option persist"
6678 if absolutely needed. A common use case is when servers are under extreme
6679 load and spend their time flapping. In this case, the users would still be
6680 directed to the server they opened the session on, in the hope they would be
6681 correctly served. It is recommended to use "option redispatch" in conjunction
6682 with this option so that in the event it would not be possible to connect to
6683 the server at all (server definitely dead), the client would finally be
6684 redirected to another valid server.
6685
6686 If this option has been enabled in a "defaults" section, it can be disabled
6687 in a specific instance by prepending the "no" keyword before it.
6688
Willy Tarreau4de91492010-01-22 19:10:05 +01006689 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006690
6691
Willy Tarreau0c122822013-12-15 18:49:01 +01006692option pgsql-check [ user <username> ]
6693 Use PostgreSQL health checks for server testing
6694 May be used in sections : defaults | frontend | listen | backend
6695 yes | no | yes | yes
6696 Arguments :
6697 <username> This is the username which will be used when connecting to
6698 PostgreSQL server.
6699
6700 The check sends a PostgreSQL StartupMessage and waits for either
6701 Authentication request or ErrorResponse message. It is a basic but useful
6702 test which does not produce error nor aborted connect on the server.
6703 This check is identical with the "mysql-check".
6704
6705 See also: "option httpchk"
6706
6707
Willy Tarreau9420b122013-12-15 18:58:25 +01006708option prefer-last-server
6709no option prefer-last-server
6710 Allow multiple load balanced requests to remain on the same server
6711 May be used in sections: defaults | frontend | listen | backend
6712 yes | no | yes | yes
6713 Arguments : none
6714
6715 When the load balancing algorithm in use is not deterministic, and a previous
6716 request was sent to a server to which haproxy still holds a connection, it is
6717 sometimes desirable that subsequent requests on a same session go to the same
6718 server as much as possible. Note that this is different from persistence, as
6719 we only indicate a preference which haproxy tries to apply without any form
6720 of warranty. The real use is for keep-alive connections sent to servers. When
6721 this option is used, haproxy will try to reuse the same connection that is
6722 attached to the server instead of rebalancing to another server, causing a
6723 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006724 not make much sense to use this in combination with hashing algorithms. Note,
6725 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006726 to a proxy which sends a 407 (authentication required), when the load
6727 balancing algorithm is not deterministic. This is mandatory for use with the
6728 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006729 troubleshooting some faulty applications. Option prefer-last-server might be
6730 desirable in these environments as well, to avoid redistributing the traffic
6731 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006732
6733 If this option has been enabled in a "defaults" section, it can be disabled
6734 in a specific instance by prepending the "no" keyword before it.
6735
6736 See also: "option http-keep-alive"
6737
6738
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006739option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006740option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006741no option redispatch
6742 Enable or disable session redistribution in case of connection failure
6743 May be used in sections: defaults | frontend | listen | backend
6744 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006745 Arguments :
6746 <interval> The optional integer value that controls how often redispatches
6747 occur when retrying connections. Positive value P indicates a
6748 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006749 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006750 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006751 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006752 positive value of 1 would indicate a redispatch on every retry,
6753 and a positive value of 3 would indicate a redispatch on every
6754 third retry. You can disable redispatches with a value of 0.
6755
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006756
6757 In HTTP mode, if a server designated by a cookie is down, clients may
6758 definitely stick to it because they cannot flush the cookie, so they will not
6759 be able to access the service anymore.
6760
6761 Specifying "option redispatch" will allow the proxy to break their
6762 persistence and redistribute them to a working server.
6763
Joseph Lynch726ab712015-05-11 23:25:34 -07006764 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006765 connection failures. Of course, it requires having "retries" set to a nonzero
6766 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006767
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006768 This form is the preferred form, which replaces both the "redispatch" and
6769 "redisp" keywords.
6770
6771 If this option has been enabled in a "defaults" section, it can be disabled
6772 in a specific instance by prepending the "no" keyword before it.
6773
Willy Tarreau4de91492010-01-22 19:10:05 +01006774 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006775
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006776
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006777option redis-check
6778 Use redis health checks for server testing
6779 May be used in sections : defaults | frontend | listen | backend
6780 yes | no | yes | yes
6781 Arguments : none
6782
6783 It is possible to test that the server correctly talks REDIS protocol instead
6784 of just testing that it accepts the TCP connection. When this option is set,
6785 a PING redis command is sent to the server, and the response is analyzed to
6786 find the "+PONG" response message.
6787
6788 Example :
6789 option redis-check
6790
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006791 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006792
6793
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006794option smtpchk
6795option smtpchk <hello> <domain>
6796 Use SMTP health checks for server testing
6797 May be used in sections : defaults | frontend | listen | backend
6798 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006799 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006800 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006801 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006802 values will be turned into the default command ("HELO").
6803
6804 <domain> is the domain name to present to the server. It may only be
6805 specified (and is mandatory) if the hello command has been
6806 specified. By default, "localhost" is used.
6807
6808 When "option smtpchk" is set, the health checks will consist in TCP
6809 connections followed by an SMTP command. By default, this command is
6810 "HELO localhost". The server's return code is analyzed and only return codes
6811 starting with a "2" will be considered as valid. All other responses,
6812 including a lack of response will constitute an error and will indicate a
6813 dead server.
6814
6815 This test is meant to be used with SMTP servers or relays. Depending on the
6816 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006817 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006818 25 is often easier than adjusting the configuration.
6819
6820 Most often, an incoming SMTP server needs to see the client's IP address for
6821 various purposes, including spam filtering, anti-spoofing and logging. When
6822 possible, it is often wise to masquerade the client's IP address when
6823 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006824 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006825
6826 Example :
6827 option smtpchk HELO mydomain.org
6828
6829 See also : "option httpchk", "source"
6830
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006831
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006832option socket-stats
6833no option socket-stats
6834
6835 Enable or disable collecting & providing separate statistics for each socket.
6836 May be used in sections : defaults | frontend | listen | backend
6837 yes | yes | yes | no
6838
6839 Arguments : none
6840
6841
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006842option splice-auto
6843no option splice-auto
6844 Enable or disable automatic kernel acceleration on sockets in both directions
6845 May be used in sections : defaults | frontend | listen | backend
6846 yes | yes | yes | yes
6847 Arguments : none
6848
6849 When this option is enabled either on a frontend or on a backend, haproxy
6850 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006851 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006852 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006853 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006854 are not much aggressive in order to limit excessive use of splicing. This
6855 option requires splicing to be enabled at compile time, and may be globally
6856 disabled with the global option "nosplice". Since splice uses pipes, using it
6857 requires that there are enough spare pipes.
6858
6859 Important note: kernel-based TCP splicing is a Linux-specific feature which
6860 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6861 transfer data between sockets without copying these data to user-space, thus
6862 providing noticeable performance gains and CPU cycles savings. Since many
6863 early implementations are buggy, corrupt data and/or are inefficient, this
6864 feature is not enabled by default, and it should be used with extreme care.
6865 While it is not possible to detect the correctness of an implementation,
6866 2.6.29 is the first version offering a properly working implementation. In
6867 case of doubt, splicing may be globally disabled using the global "nosplice"
6868 keyword.
6869
6870 Example :
6871 option splice-auto
6872
6873 If this option has been enabled in a "defaults" section, it can be disabled
6874 in a specific instance by prepending the "no" keyword before it.
6875
6876 See also : "option splice-request", "option splice-response", and global
6877 options "nosplice" and "maxpipes"
6878
6879
6880option splice-request
6881no option splice-request
6882 Enable or disable automatic kernel acceleration on sockets for requests
6883 May be used in sections : defaults | frontend | listen | backend
6884 yes | yes | yes | yes
6885 Arguments : none
6886
6887 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006888 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006889 the client to the server. It might still use the recv/send scheme if there
6890 are no spare pipes left. This option requires splicing to be enabled at
6891 compile time, and may be globally disabled with the global option "nosplice".
6892 Since splice uses pipes, using it requires that there are enough spare pipes.
6893
6894 Important note: see "option splice-auto" for usage limitations.
6895
6896 Example :
6897 option splice-request
6898
6899 If this option has been enabled in a "defaults" section, it can be disabled
6900 in a specific instance by prepending the "no" keyword before it.
6901
6902 See also : "option splice-auto", "option splice-response", and global options
6903 "nosplice" and "maxpipes"
6904
6905
6906option splice-response
6907no option splice-response
6908 Enable or disable automatic kernel acceleration on sockets for responses
6909 May be used in sections : defaults | frontend | listen | backend
6910 yes | yes | yes | yes
6911 Arguments : none
6912
6913 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006914 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006915 the server to the client. It might still use the recv/send scheme if there
6916 are no spare pipes left. This option requires splicing to be enabled at
6917 compile time, and may be globally disabled with the global option "nosplice".
6918 Since splice uses pipes, using it requires that there are enough spare pipes.
6919
6920 Important note: see "option splice-auto" for usage limitations.
6921
6922 Example :
6923 option splice-response
6924
6925 If this option has been enabled in a "defaults" section, it can be disabled
6926 in a specific instance by prepending the "no" keyword before it.
6927
6928 See also : "option splice-auto", "option splice-request", and global options
6929 "nosplice" and "maxpipes"
6930
6931
Christopher Fauletba7bc162016-11-07 21:07:38 +01006932option spop-check
6933 Use SPOP health checks for server testing
6934 May be used in sections : defaults | frontend | listen | backend
6935 no | no | no | yes
6936 Arguments : none
6937
6938 It is possible to test that the server correctly talks SPOP protocol instead
6939 of just testing that it accepts the TCP connection. When this option is set,
6940 a HELLO handshake is performed between HAProxy and the server, and the
6941 response is analyzed to check no error is reported.
6942
6943 Example :
6944 option spop-check
6945
6946 See also : "option httpchk"
6947
6948
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006949option srvtcpka
6950no option srvtcpka
6951 Enable or disable the sending of TCP keepalive packets on the server side
6952 May be used in sections : defaults | frontend | listen | backend
6953 yes | no | yes | yes
6954 Arguments : none
6955
6956 When there is a firewall or any session-aware component between a client and
6957 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006958 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006959 components decides to expire a session which has remained idle for too long.
6960
6961 Enabling socket-level TCP keep-alives makes the system regularly send packets
6962 to the other end of the connection, leaving it active. The delay between
6963 keep-alive probes is controlled by the system only and depends both on the
6964 operating system and its tuning parameters.
6965
6966 It is important to understand that keep-alive packets are neither emitted nor
6967 received at the application level. It is only the network stacks which sees
6968 them. For this reason, even if one side of the proxy already uses keep-alives
6969 to maintain its connection alive, those keep-alive packets will not be
6970 forwarded to the other side of the proxy.
6971
6972 Please note that this has nothing to do with HTTP keep-alive.
6973
6974 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6975 server side of a connection, which should help when session expirations are
6976 noticed between HAProxy and a server.
6977
6978 If this option has been enabled in a "defaults" section, it can be disabled
6979 in a specific instance by prepending the "no" keyword before it.
6980
6981 See also : "option clitcpka", "option tcpka"
6982
6983
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006984option ssl-hello-chk
6985 Use SSLv3 client hello health checks for server testing
6986 May be used in sections : defaults | frontend | listen | backend
6987 yes | no | yes | yes
6988 Arguments : none
6989
6990 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6991 possible to test that the server correctly talks SSL instead of just testing
6992 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6993 SSLv3 client hello messages are sent once the connection is established to
6994 the server, and the response is analyzed to find an SSL server hello message.
6995 The server is considered valid only when the response contains this server
6996 hello message.
6997
6998 All servers tested till there correctly reply to SSLv3 client hello messages,
6999 and most servers tested do not even log the requests containing only hello
7000 messages, which is appreciable.
7001
Willy Tarreau763a95b2012-10-04 23:15:39 +02007002 Note that this check works even when SSL support was not built into haproxy
7003 because it forges the SSL message. When SSL support is available, it is best
7004 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007005
Willy Tarreau763a95b2012-10-04 23:15:39 +02007006 See also: "option httpchk", "check-ssl"
7007
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007008
Willy Tarreaued179852013-12-16 01:07:00 +01007009option tcp-check
7010 Perform health checks using tcp-check send/expect sequences
7011 May be used in sections: defaults | frontend | listen | backend
7012 yes | no | yes | yes
7013
7014 This health check method is intended to be combined with "tcp-check" command
7015 lists in order to support send/expect types of health check sequences.
7016
7017 TCP checks currently support 4 modes of operations :
7018 - no "tcp-check" directive : the health check only consists in a connection
7019 attempt, which remains the default mode.
7020
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007021 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007022 used to send a string along with a connection opening. With some
7023 protocols, it helps sending a "QUIT" message for example that prevents
7024 the server from logging a connection error for each health check. The
7025 check result will still be based on the ability to open the connection
7026 only.
7027
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007028 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007029 The connection is opened and haproxy waits for the server to present some
7030 contents which must validate some rules. The check result will be based
7031 on the matching between the contents and the rules. This is suited for
7032 POP, IMAP, SMTP, FTP, SSH, TELNET.
7033
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007034 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007035 used to test a hello-type protocol. HAProxy sends a message, the server
7036 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007037 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007038 suited for protocols which require a binding or a request/response model.
7039 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7040 already all have their dedicated checks with a deeper understanding of
7041 the respective protocols.
7042 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007043 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007044
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007045 A fifth mode can be used to insert comments in different steps of the
7046 script.
7047
7048 For each tcp-check rule you create, you can add a "comment" directive,
7049 followed by a string. This string will be reported in the log and stderr
7050 in debug mode. It is useful to make user-friendly error reporting.
7051 The "comment" is of course optional.
7052
7053
Willy Tarreaued179852013-12-16 01:07:00 +01007054 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007055 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007056 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007057 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007058
Davor Ocelice9ed2812017-12-25 17:49:28 +01007059 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007060 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007061 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007062
7063 # look for the redis master server after ensuring it speaks well
7064 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007065 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007066 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007067 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007068 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007069 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007070 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007071 tcp-check send info\ replication\r\n
7072 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007073 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007074 tcp-check send QUIT\r\n
7075 tcp-check expect string +OK
7076
Davor Ocelice9ed2812017-12-25 17:49:28 +01007077 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007078 (send many headers before analyzing)
7079 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007080 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007081 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7082 tcp-check send Host:\ www.mydomain.com\r\n
7083 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7084 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007085 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007086
7087
7088 See also : "tcp-check expect", "tcp-check send"
7089
7090
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007091option tcp-smart-accept
7092no option tcp-smart-accept
7093 Enable or disable the saving of one ACK packet during the accept sequence
7094 May be used in sections : defaults | frontend | listen | backend
7095 yes | yes | yes | no
7096 Arguments : none
7097
7098 When an HTTP connection request comes in, the system acknowledges it on
7099 behalf of HAProxy, then the client immediately sends its request, and the
7100 system acknowledges it too while it is notifying HAProxy about the new
7101 connection. HAProxy then reads the request and responds. This means that we
7102 have one TCP ACK sent by the system for nothing, because the request could
7103 very well be acknowledged by HAProxy when it sends its response.
7104
7105 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7106 sending this useless ACK on platforms which support it (currently at least
7107 Linux). It must not cause any problem, because the system will send it anyway
7108 after 40 ms if the response takes more time than expected to come.
7109
7110 During complex network debugging sessions, it may be desirable to disable
7111 this optimization because delayed ACKs can make troubleshooting more complex
7112 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007113 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007114
7115 It is also possible to force it for non-HTTP proxies by simply specifying
7116 "option tcp-smart-accept". For instance, it can make sense with some services
7117 such as SMTP where the server speaks first.
7118
7119 It is recommended to avoid forcing this option in a defaults section. In case
7120 of doubt, consider setting it back to automatic values by prepending the
7121 "default" keyword before it, or disabling it using the "no" keyword.
7122
Willy Tarreaud88edf22009-06-14 15:48:17 +02007123 See also : "option tcp-smart-connect"
7124
7125
7126option tcp-smart-connect
7127no option tcp-smart-connect
7128 Enable or disable the saving of one ACK packet during the connect sequence
7129 May be used in sections : defaults | frontend | listen | backend
7130 yes | no | yes | yes
7131 Arguments : none
7132
7133 On certain systems (at least Linux), HAProxy can ask the kernel not to
7134 immediately send an empty ACK upon a connection request, but to directly
7135 send the buffer request instead. This saves one packet on the network and
7136 thus boosts performance. It can also be useful for some servers, because they
7137 immediately get the request along with the incoming connection.
7138
7139 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7140 It is not enabled by default because it makes network troubleshooting more
7141 complex.
7142
7143 It only makes sense to enable it with protocols where the client speaks first
7144 such as HTTP. In other situations, if there is no data to send in place of
7145 the ACK, a normal ACK is sent.
7146
7147 If this option has been enabled in a "defaults" section, it can be disabled
7148 in a specific instance by prepending the "no" keyword before it.
7149
7150 See also : "option tcp-smart-accept"
7151
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007152
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007153option tcpka
7154 Enable or disable the sending of TCP keepalive packets on both sides
7155 May be used in sections : defaults | frontend | listen | backend
7156 yes | yes | yes | yes
7157 Arguments : none
7158
7159 When there is a firewall or any session-aware component between a client and
7160 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007161 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007162 components decides to expire a session which has remained idle for too long.
7163
7164 Enabling socket-level TCP keep-alives makes the system regularly send packets
7165 to the other end of the connection, leaving it active. The delay between
7166 keep-alive probes is controlled by the system only and depends both on the
7167 operating system and its tuning parameters.
7168
7169 It is important to understand that keep-alive packets are neither emitted nor
7170 received at the application level. It is only the network stacks which sees
7171 them. For this reason, even if one side of the proxy already uses keep-alives
7172 to maintain its connection alive, those keep-alive packets will not be
7173 forwarded to the other side of the proxy.
7174
7175 Please note that this has nothing to do with HTTP keep-alive.
7176
7177 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7178 the client and server sides of a connection. Note that this is meaningful
7179 only in "defaults" or "listen" sections. If this option is used in a
7180 frontend, only the client side will get keep-alives, and if this option is
7181 used in a backend, only the server side will get keep-alives. For this
7182 reason, it is strongly recommended to explicitly use "option clitcpka" and
7183 "option srvtcpka" when the configuration is split between frontends and
7184 backends.
7185
7186 See also : "option clitcpka", "option srvtcpka"
7187
Willy Tarreau844e3c52008-01-11 16:28:18 +01007188
7189option tcplog
7190 Enable advanced logging of TCP connections with session state and timers
7191 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007192 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007193 Arguments : none
7194
7195 By default, the log output format is very poor, as it only contains the
7196 source and destination addresses, and the instance name. By specifying
7197 "option tcplog", each log line turns into a much richer format including, but
7198 not limited to, the connection timers, the session status, the connections
7199 numbers, the frontend, backend and server name, and of course the source
7200 address and ports. This option is useful for pure TCP proxies in order to
7201 find which of the client or server disconnects or times out. For normal HTTP
7202 proxies, it's better to use "option httplog" which is even more complete.
7203
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007204 "option tcplog" overrides any previous "log-format" directive.
7205
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007206 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007207
7208
Willy Tarreau844e3c52008-01-11 16:28:18 +01007209option transparent
7210no option transparent
7211 Enable client-side transparent proxying
7212 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007213 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007214 Arguments : none
7215
7216 This option was introduced in order to provide layer 7 persistence to layer 3
7217 load balancers. The idea is to use the OS's ability to redirect an incoming
7218 connection for a remote address to a local process (here HAProxy), and let
7219 this process know what address was initially requested. When this option is
7220 used, sessions without cookies will be forwarded to the original destination
7221 IP address of the incoming request (which should match that of another
7222 equipment), while requests with cookies will still be forwarded to the
7223 appropriate server.
7224
7225 Note that contrary to a common belief, this option does NOT make HAProxy
7226 present the client's IP to the server when establishing the connection.
7227
Willy Tarreaua1146052011-03-01 09:51:54 +01007228 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007229 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007230
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007231
Simon Horman98637e52014-06-20 12:30:16 +09007232external-check command <command>
7233 Executable to run when performing an external-check
7234 May be used in sections : defaults | frontend | listen | backend
7235 yes | no | yes | yes
7236
7237 Arguments :
7238 <command> is the external command to run
7239
Simon Horman98637e52014-06-20 12:30:16 +09007240 The arguments passed to the to the command are:
7241
Cyril Bonté777be862014-12-02 21:21:35 +01007242 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007243
Cyril Bonté777be862014-12-02 21:21:35 +01007244 The <proxy_address> and <proxy_port> are derived from the first listener
7245 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7246 listener the proxy_address will be the path of the socket and the
7247 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7248 possible to determine a listener, and both <proxy_address> and <proxy_port>
7249 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007250
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007251 Some values are also provided through environment variables.
7252
7253 Environment variables :
7254 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7255 applicable, for example in a "backend" section).
7256
7257 HAPROXY_PROXY_ID The backend id.
7258
7259 HAPROXY_PROXY_NAME The backend name.
7260
7261 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7262 applicable, for example in a "backend" section or
7263 for a UNIX socket).
7264
7265 HAPROXY_SERVER_ADDR The server address.
7266
7267 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7268
7269 HAPROXY_SERVER_ID The server id.
7270
7271 HAPROXY_SERVER_MAXCONN The server max connections.
7272
7273 HAPROXY_SERVER_NAME The server name.
7274
7275 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7276 socket).
7277
7278 PATH The PATH environment variable used when executing
7279 the command may be set using "external-check path".
7280
Simon Horman98637e52014-06-20 12:30:16 +09007281 If the command executed and exits with a zero status then the check is
7282 considered to have passed, otherwise the check is considered to have
7283 failed.
7284
7285 Example :
7286 external-check command /bin/true
7287
7288 See also : "external-check", "option external-check", "external-check path"
7289
7290
7291external-check path <path>
7292 The value of the PATH environment variable used when running an external-check
7293 May be used in sections : defaults | frontend | listen | backend
7294 yes | no | yes | yes
7295
7296 Arguments :
7297 <path> is the path used when executing external command to run
7298
7299 The default path is "".
7300
7301 Example :
7302 external-check path "/usr/bin:/bin"
7303
7304 See also : "external-check", "option external-check",
7305 "external-check command"
7306
7307
Emeric Brun647caf12009-06-30 17:57:00 +02007308persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007309persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007310 Enable RDP cookie-based persistence
7311 May be used in sections : defaults | frontend | listen | backend
7312 yes | no | yes | yes
7313 Arguments :
7314 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007315 default cookie name "msts" will be used. There currently is no
7316 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007317
7318 This statement enables persistence based on an RDP cookie. The RDP cookie
7319 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007320 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007321 and if an RDP cookie is found, it is decoded. If it matches a known server
7322 which is still UP (or if "option persist" is set), then the connection is
7323 forwarded to this server.
7324
7325 Note that this only makes sense in a TCP backend, but for this to work, the
7326 frontend must have waited long enough to ensure that an RDP cookie is present
7327 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007328 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007329 a single "listen" section.
7330
Willy Tarreau61e28f22010-05-16 22:31:05 +02007331 Also, it is important to understand that the terminal server will emit this
7332 RDP cookie only if it is configured for "token redirection mode", which means
7333 that the "IP address redirection" option is disabled.
7334
Emeric Brun647caf12009-06-30 17:57:00 +02007335 Example :
7336 listen tse-farm
7337 bind :3389
7338 # wait up to 5s for an RDP cookie in the request
7339 tcp-request inspect-delay 5s
7340 tcp-request content accept if RDP_COOKIE
7341 # apply RDP cookie persistence
7342 persist rdp-cookie
7343 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007344 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007345 balance rdp-cookie
7346 server srv1 1.1.1.1:3389
7347 server srv2 1.1.1.2:3389
7348
Simon Hormanab814e02011-06-24 14:50:20 +09007349 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7350 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007351
7352
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007353rate-limit sessions <rate>
7354 Set a limit on the number of new sessions accepted per second on a frontend
7355 May be used in sections : defaults | frontend | listen | backend
7356 yes | yes | yes | no
7357 Arguments :
7358 <rate> The <rate> parameter is an integer designating the maximum number
7359 of new sessions per second to accept on the frontend.
7360
7361 When the frontend reaches the specified number of new sessions per second, it
7362 stops accepting new connections until the rate drops below the limit again.
7363 During this time, the pending sessions will be kept in the socket's backlog
7364 (in system buffers) and haproxy will not even be aware that sessions are
7365 pending. When applying very low limit on a highly loaded service, it may make
7366 sense to increase the socket's backlog using the "backlog" keyword.
7367
7368 This feature is particularly efficient at blocking connection-based attacks
7369 or service abuse on fragile servers. Since the session rate is measured every
7370 millisecond, it is extremely accurate. Also, the limit applies immediately,
7371 no delay is needed at all to detect the threshold.
7372
7373 Example : limit the connection rate on SMTP to 10 per second max
7374 listen smtp
7375 mode tcp
7376 bind :25
7377 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007378 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007379
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007380 Note : when the maximum rate is reached, the frontend's status is not changed
7381 but its sockets appear as "WAITING" in the statistics if the
7382 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007383
7384 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7385
7386
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007387redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7388redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7389redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007390 Return an HTTP redirection if/unless a condition is matched
7391 May be used in sections : defaults | frontend | listen | backend
7392 no | yes | yes | yes
7393
7394 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007395 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007396
Willy Tarreau0140f252008-11-19 21:07:09 +01007397 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007398 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007399 the HTTP "Location" header. When used in an "http-request" rule,
7400 <loc> value follows the log-format rules and can include some
7401 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007402
7403 <pfx> With "redirect prefix", the "Location" header is built from the
7404 concatenation of <pfx> and the complete URI path, including the
7405 query string, unless the "drop-query" option is specified (see
7406 below). As a special case, if <pfx> equals exactly "/", then
7407 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007408 redirect to the same URL (for instance, to insert a cookie). When
7409 used in an "http-request" rule, <pfx> value follows the log-format
7410 rules and can include some dynamic values (see Custom Log Format
7411 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007412
7413 <sch> With "redirect scheme", then the "Location" header is built by
7414 concatenating <sch> with "://" then the first occurrence of the
7415 "Host" header, and then the URI path, including the query string
7416 unless the "drop-query" option is specified (see below). If no
7417 path is found or if the path is "*", then "/" is used instead. If
7418 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007419 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007420 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007421 HTTPS. When used in an "http-request" rule, <sch> value follows
7422 the log-format rules and can include some dynamic values (see
7423 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007424
7425 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007426 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7427 with 302 used by default if no code is specified. 301 means
7428 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007429 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007430 cache the redirection. 303 is equivalent to 302 except that the
7431 browser will fetch the location with a GET method. 307 is just
7432 like 302 but makes it clear that the same method must be reused.
7433 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007434
7435 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007436 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007437
7438 - "drop-query"
7439 When this keyword is used in a prefix-based redirection, then the
7440 location will be set without any possible query-string, which is useful
7441 for directing users to a non-secure page for instance. It has no effect
7442 with a location-type redirect.
7443
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007444 - "append-slash"
7445 This keyword may be used in conjunction with "drop-query" to redirect
7446 users who use a URL not ending with a '/' to the same one with the '/'.
7447 It can be useful to ensure that search engines will only see one URL.
7448 For this, a return code 301 is preferred.
7449
Willy Tarreau0140f252008-11-19 21:07:09 +01007450 - "set-cookie NAME[=value]"
7451 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7452 to the response. This is sometimes used to indicate that a user has
7453 been seen, for instance to protect against some types of DoS. No other
7454 cookie option is added, so the cookie will be a session cookie. Note
7455 that for a browser, a sole cookie name without an equal sign is
7456 different from a cookie with an equal sign.
7457
7458 - "clear-cookie NAME[=]"
7459 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7460 with the "Max-Age" attribute set to zero. This will tell the browser to
7461 delete this cookie. It is useful for instance on logout pages. It is
7462 important to note that clearing the cookie "NAME" will not remove a
7463 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7464 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007465
7466 Example: move the login URL only to HTTPS.
7467 acl clear dst_port 80
7468 acl secure dst_port 8080
7469 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007470 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007471 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007472 acl cookie_set hdr_sub(cookie) SEEN=1
7473
7474 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007475 redirect prefix https://mysite.com if login_page !secure
7476 redirect prefix http://mysite.com drop-query if login_page !uid_given
7477 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007478 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007479
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007480 Example: send redirects for request for articles without a '/'.
7481 acl missing_slash path_reg ^/article/[^/]*$
7482 redirect code 301 prefix / drop-query append-slash if missing_slash
7483
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007484 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007485 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007486
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007487 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007488 http-request redirect code 301 location \
7489 http://www.%[hdr(host)]%[capture.req.uri] \
7490 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007491
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007492 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007493
7494
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007495redisp (deprecated)
7496redispatch (deprecated)
7497 Enable or disable session redistribution in case of connection failure
7498 May be used in sections: defaults | frontend | listen | backend
7499 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007500 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007501
7502 In HTTP mode, if a server designated by a cookie is down, clients may
7503 definitely stick to it because they cannot flush the cookie, so they will not
7504 be able to access the service anymore.
7505
7506 Specifying "redispatch" will allow the proxy to break their persistence and
7507 redistribute them to a working server.
7508
7509 It also allows to retry last connection to another server in case of multiple
7510 connection failures. Of course, it requires having "retries" set to a nonzero
7511 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007512
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007513 This form is deprecated, do not use it in any new configuration, use the new
7514 "option redispatch" instead.
7515
7516 See also : "option redispatch"
7517
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007518
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007519reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007520 Add a header at the end of the HTTP request
7521 May be used in sections : defaults | frontend | listen | backend
7522 no | yes | yes | yes
7523 Arguments :
7524 <string> is the complete line to be added. Any space or known delimiter
7525 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007526 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007527
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007528 <cond> is an optional matching condition built from ACLs. It makes it
7529 possible to ignore this rule when other conditions are not met.
7530
Willy Tarreau303c0352008-01-17 19:01:39 +01007531 A new line consisting in <string> followed by a line feed will be added after
7532 the last header of an HTTP request.
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.
7537
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007538 Example : add "X-Proto: SSL" to requests coming via port 81
7539 acl is-ssl dst_port 81
7540 reqadd X-Proto:\ SSL if is-ssl
7541
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007542 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7543 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007544
7545
Willy Tarreau5321c422010-01-28 20:35:13 +01007546reqallow <search> [{if | unless} <cond>]
7547reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007548 Definitely allow an HTTP request if a line matches a regular expression
7549 May be used in sections : defaults | frontend | listen | backend
7550 no | yes | yes | yes
7551 Arguments :
7552 <search> is the regular expression applied to HTTP headers and to the
7553 request line. This is an extended regular expression. Parenthesis
7554 grouping is supported and no preliminary backslash is required.
7555 Any space or known delimiter must be escaped using a backslash
7556 ('\'). The pattern applies to a full line at a time. The
7557 "reqallow" keyword strictly matches case while "reqiallow"
7558 ignores case.
7559
Willy Tarreau5321c422010-01-28 20:35:13 +01007560 <cond> is an optional matching condition built from ACLs. It makes it
7561 possible to ignore this rule when other conditions are not met.
7562
Willy Tarreau303c0352008-01-17 19:01:39 +01007563 A request containing any line which matches extended regular expression
7564 <search> will mark the request as allowed, even if any later test would
7565 result in a deny. The test applies both to the request line and to request
7566 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007567 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007568
7569 It is easier, faster and more powerful to use ACLs to write access policies.
7570 Reqdeny, reqallow and reqpass should be avoided in new designs.
7571
7572 Example :
7573 # allow www.* but refuse *.local
7574 reqiallow ^Host:\ www\.
7575 reqideny ^Host:\ .*\.local
7576
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007577 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7578 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007579
7580
Willy Tarreau5321c422010-01-28 20:35:13 +01007581reqdel <search> [{if | unless} <cond>]
7582reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007583 Delete all headers matching a regular expression in an HTTP request
7584 May be used in sections : defaults | frontend | listen | backend
7585 no | yes | yes | yes
7586 Arguments :
7587 <search> is the regular expression applied to HTTP headers and to the
7588 request line. This is an extended regular expression. Parenthesis
7589 grouping is supported and no preliminary backslash is required.
7590 Any space or known delimiter must be escaped using a backslash
7591 ('\'). The pattern applies to a full line at a time. The "reqdel"
7592 keyword strictly matches case while "reqidel" ignores case.
7593
Willy Tarreau5321c422010-01-28 20:35:13 +01007594 <cond> is an optional matching condition built from ACLs. It makes it
7595 possible to ignore this rule when other conditions are not met.
7596
Willy Tarreau303c0352008-01-17 19:01:39 +01007597 Any header line matching extended regular expression <search> in the request
7598 will be completely deleted. Most common use of this is to remove unwanted
7599 and/or dangerous headers or cookies from a request before passing it to the
7600 next servers.
7601
7602 Header transformations only apply to traffic which passes through HAProxy,
7603 and not to traffic generated by HAProxy, such as health-checks or error
7604 responses. Keep in mind that header names are not case-sensitive.
7605
7606 Example :
7607 # remove X-Forwarded-For header and SERVER cookie
7608 reqidel ^X-Forwarded-For:.*
7609 reqidel ^Cookie:.*SERVER=
7610
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007611 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7612 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007613
7614
Willy Tarreau5321c422010-01-28 20:35:13 +01007615reqdeny <search> [{if | unless} <cond>]
7616reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007617 Deny an HTTP request if a line matches a regular expression
7618 May be used in sections : defaults | frontend | listen | backend
7619 no | yes | yes | yes
7620 Arguments :
7621 <search> is the regular expression applied to HTTP headers and to the
7622 request line. This is an extended regular expression. Parenthesis
7623 grouping is supported and no preliminary backslash is required.
7624 Any space or known delimiter must be escaped using a backslash
7625 ('\'). The pattern applies to a full line at a time. The
7626 "reqdeny" keyword strictly matches case while "reqideny" ignores
7627 case.
7628
Willy Tarreau5321c422010-01-28 20:35:13 +01007629 <cond> is an optional matching condition built from ACLs. It makes it
7630 possible to ignore this rule when other conditions are not met.
7631
Willy Tarreau303c0352008-01-17 19:01:39 +01007632 A request containing any line which matches extended regular expression
7633 <search> will mark the request as denied, even if any later test would
7634 result in an allow. The test applies both to the request line and to request
7635 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007636 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007637
Willy Tarreauced27012008-01-17 20:35:34 +01007638 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007639 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007640 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007641
Willy Tarreau303c0352008-01-17 19:01:39 +01007642 It is easier, faster and more powerful to use ACLs to write access policies.
7643 Reqdeny, reqallow and reqpass should be avoided in new designs.
7644
7645 Example :
7646 # refuse *.local, then allow www.*
7647 reqideny ^Host:\ .*\.local
7648 reqiallow ^Host:\ www\.
7649
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007650 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7651 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007652
7653
Willy Tarreau5321c422010-01-28 20:35:13 +01007654reqpass <search> [{if | unless} <cond>]
7655reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007656 Ignore any HTTP request line matching a regular expression in next rules
7657 May be used in sections : defaults | frontend | listen | backend
7658 no | yes | yes | yes
7659 Arguments :
7660 <search> is the regular expression applied to HTTP headers and to the
7661 request line. This is an extended regular expression. Parenthesis
7662 grouping is supported and no preliminary backslash is required.
7663 Any space or known delimiter must be escaped using a backslash
7664 ('\'). The pattern applies to a full line at a time. The
7665 "reqpass" keyword strictly matches case while "reqipass" ignores
7666 case.
7667
Willy Tarreau5321c422010-01-28 20:35:13 +01007668 <cond> is an optional matching condition built from ACLs. It makes it
7669 possible to ignore this rule when other conditions are not met.
7670
Willy Tarreau303c0352008-01-17 19:01:39 +01007671 A request containing any line which matches extended regular expression
7672 <search> will skip next rules, without assigning any deny or allow verdict.
7673 The test applies both to the request line and to request headers. Keep in
7674 mind that URLs in request line are case-sensitive while header names are not.
7675
7676 It is easier, faster and more powerful to use ACLs to write access policies.
7677 Reqdeny, reqallow and reqpass should be avoided in new designs.
7678
7679 Example :
7680 # refuse *.local, then allow www.*, but ignore "www.private.local"
7681 reqipass ^Host:\ www.private\.local
7682 reqideny ^Host:\ .*\.local
7683 reqiallow ^Host:\ www\.
7684
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007685 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7686 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007687
7688
Willy Tarreau5321c422010-01-28 20:35:13 +01007689reqrep <search> <string> [{if | unless} <cond>]
7690reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007691 Replace a regular expression with a string in an HTTP request line
7692 May be used in sections : defaults | frontend | listen | backend
7693 no | yes | yes | yes
7694 Arguments :
7695 <search> is the regular expression applied to HTTP headers and to the
7696 request line. This is an extended regular expression. Parenthesis
7697 grouping is supported and no preliminary backslash is required.
7698 Any space or known delimiter must be escaped using a backslash
7699 ('\'). The pattern applies to a full line at a time. The "reqrep"
7700 keyword strictly matches case while "reqirep" ignores case.
7701
7702 <string> is the complete line to be added. Any space or known delimiter
7703 must be escaped using a backslash ('\'). References to matched
7704 pattern groups are possible using the common \N form, with N
7705 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007706 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007707
Willy Tarreau5321c422010-01-28 20:35:13 +01007708 <cond> is an optional matching condition built from ACLs. It makes it
7709 possible to ignore this rule when other conditions are not met.
7710
Willy Tarreau303c0352008-01-17 19:01:39 +01007711 Any line matching extended regular expression <search> in the request (both
7712 the request line and header lines) will be completely replaced with <string>.
7713 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7714
7715 Header transformations only apply to traffic which passes through HAProxy,
7716 and not to traffic generated by HAProxy, such as health-checks or error
7717 responses. Note that for increased readability, it is suggested to add enough
7718 spaces between the request and the response. Keep in mind that URLs in
7719 request line are case-sensitive while header names are not.
7720
7721 Example :
7722 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007723 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007724 # replace "www.mydomain.com" with "www" in the host name.
7725 reqirep ^Host:\ www.mydomain.com Host:\ www
7726
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007727 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7728 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007729
7730
Willy Tarreau5321c422010-01-28 20:35:13 +01007731reqtarpit <search> [{if | unless} <cond>]
7732reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007733 Tarpit an HTTP request containing a line matching a regular expression
7734 May be used in sections : defaults | frontend | listen | backend
7735 no | yes | yes | yes
7736 Arguments :
7737 <search> is the regular expression applied to HTTP headers and to the
7738 request line. This is an extended regular expression. Parenthesis
7739 grouping is supported and no preliminary backslash is required.
7740 Any space or known delimiter must be escaped using a backslash
7741 ('\'). The pattern applies to a full line at a time. The
7742 "reqtarpit" keyword strictly matches case while "reqitarpit"
7743 ignores case.
7744
Willy Tarreau5321c422010-01-28 20:35:13 +01007745 <cond> is an optional matching condition built from ACLs. It makes it
7746 possible to ignore this rule when other conditions are not met.
7747
Willy Tarreau303c0352008-01-17 19:01:39 +01007748 A request containing any line which matches extended regular expression
7749 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007750 be kept open for a pre-defined time, then will return an HTTP error 500 so
7751 that the attacker does not suspect it has been tarpitted. The status 500 will
7752 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007753 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7754 not set.
7755
7756 The goal of the tarpit is to slow down robots attacking servers with
7757 identifiable requests. Many robots limit their outgoing number of connections
7758 and stay connected waiting for a reply which can take several minutes to
7759 come. Depending on the environment and attack, it may be particularly
7760 efficient at reducing the load on the network and firewalls.
7761
Willy Tarreau5321c422010-01-28 20:35:13 +01007762 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007763 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007764 # block all others.
7765 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7766 reqitarpit ^User-Agent:
7767
Willy Tarreau5321c422010-01-28 20:35:13 +01007768 # block bad guys
7769 acl badguys src 10.1.0.3 172.16.13.20/28
7770 reqitarpit . if badguys
7771
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007772 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7773 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007774
7775
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007776retries <value>
7777 Set the number of retries to perform on a server after a connection failure
7778 May be used in sections: defaults | frontend | listen | backend
7779 yes | no | yes | yes
7780 Arguments :
7781 <value> is the number of times a connection attempt should be retried on
7782 a server when a connection either is refused or times out. The
7783 default value is 3.
7784
7785 It is important to understand that this value applies to the number of
7786 connection attempts, not full requests. When a connection has effectively
7787 been established to a server, there will be no more retry.
7788
7789 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007790 a turn-around timer of min("timeout connect", one second) is applied before
7791 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007792
7793 When "option redispatch" is set, the last retry may be performed on another
7794 server even if a cookie references a different server.
7795
7796 See also : "option redispatch"
7797
7798
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007799rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007800 Add a header at the end of the HTTP response
7801 May be used in sections : defaults | frontend | listen | backend
7802 no | yes | yes | yes
7803 Arguments :
7804 <string> is the complete line to be added. Any space or known delimiter
7805 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007806 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007807
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007808 <cond> is an optional matching condition built from ACLs. It makes it
7809 possible to ignore this rule when other conditions are not met.
7810
Willy Tarreau303c0352008-01-17 19:01:39 +01007811 A new line consisting in <string> followed by a line feed will be added after
7812 the last header of an HTTP response.
7813
7814 Header transformations only apply to traffic which passes through HAProxy,
7815 and not to traffic generated by HAProxy, such as health-checks or error
7816 responses.
7817
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007818 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7819 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007820
7821
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007822rspdel <search> [{if | unless} <cond>]
7823rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007824 Delete all headers matching a regular expression in an HTTP response
7825 May be used in sections : defaults | frontend | listen | backend
7826 no | yes | yes | yes
7827 Arguments :
7828 <search> is the regular expression applied to HTTP headers and to the
7829 response line. This is an extended regular expression, so
7830 parenthesis grouping is supported and no preliminary backslash
7831 is required. Any space or known delimiter must be escaped using
7832 a backslash ('\'). The pattern applies to a full line at a time.
7833 The "rspdel" keyword strictly matches case while "rspidel"
7834 ignores case.
7835
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007836 <cond> is an optional matching condition built from ACLs. It makes it
7837 possible to ignore this rule when other conditions are not met.
7838
Willy Tarreau303c0352008-01-17 19:01:39 +01007839 Any header line matching extended regular expression <search> in the response
7840 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007841 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007842 client.
7843
7844 Header transformations only apply to traffic which passes through HAProxy,
7845 and not to traffic generated by HAProxy, such as health-checks or error
7846 responses. Keep in mind that header names are not case-sensitive.
7847
7848 Example :
7849 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007850 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007851
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007852 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7853 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007854
7855
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007856rspdeny <search> [{if | unless} <cond>]
7857rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007858 Block an HTTP response if a line matches a regular expression
7859 May be used in sections : defaults | frontend | listen | backend
7860 no | yes | yes | yes
7861 Arguments :
7862 <search> is the regular expression applied to HTTP headers and to the
7863 response line. This is an extended regular expression, so
7864 parenthesis grouping is supported and no preliminary backslash
7865 is required. Any space or known delimiter must be escaped using
7866 a backslash ('\'). The pattern applies to a full line at a time.
7867 The "rspdeny" keyword strictly matches case while "rspideny"
7868 ignores case.
7869
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007870 <cond> is an optional matching condition built from ACLs. It makes it
7871 possible to ignore this rule when other conditions are not met.
7872
Willy Tarreau303c0352008-01-17 19:01:39 +01007873 A response containing any line which matches extended regular expression
7874 <search> will mark the request as denied. The test applies both to the
7875 response line and to response headers. Keep in mind that header names are not
7876 case-sensitive.
7877
7878 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007879 block the response before it reaches the client. If a response is denied, it
7880 will be replaced with an HTTP 502 error so that the client never retrieves
7881 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007882
7883 It is easier, faster and more powerful to use ACLs to write access policies.
7884 Rspdeny should be avoided in new designs.
7885
7886 Example :
7887 # Ensure that no content type matching ms-word will leak
7888 rspideny ^Content-type:\.*/ms-word
7889
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007890 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7891 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007892
7893
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007894rsprep <search> <string> [{if | unless} <cond>]
7895rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007896 Replace a regular expression with a string in an HTTP response line
7897 May be used in sections : defaults | frontend | listen | backend
7898 no | yes | yes | yes
7899 Arguments :
7900 <search> is the regular expression applied to HTTP headers and to the
7901 response line. This is an extended regular expression, so
7902 parenthesis grouping is supported and no preliminary backslash
7903 is required. Any space or known delimiter must be escaped using
7904 a backslash ('\'). The pattern applies to a full line at a time.
7905 The "rsprep" keyword strictly matches case while "rspirep"
7906 ignores case.
7907
7908 <string> is the complete line to be added. Any space or known delimiter
7909 must be escaped using a backslash ('\'). References to matched
7910 pattern groups are possible using the common \N form, with N
7911 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007912 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007913
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007914 <cond> is an optional matching condition built from ACLs. It makes it
7915 possible to ignore this rule when other conditions are not met.
7916
Willy Tarreau303c0352008-01-17 19:01:39 +01007917 Any line matching extended regular expression <search> in the response (both
7918 the response line and header lines) will be completely replaced with
7919 <string>. Most common use of this is to rewrite Location headers.
7920
7921 Header transformations only apply to traffic which passes through HAProxy,
7922 and not to traffic generated by HAProxy, such as health-checks or error
7923 responses. Note that for increased readability, it is suggested to add enough
7924 spaces between the request and the response. Keep in mind that header names
7925 are not case-sensitive.
7926
7927 Example :
7928 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7929 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7930
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007931 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7932 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007933
7934
David du Colombier486df472011-03-17 10:40:26 +01007935server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007936 Declare a server in a backend
7937 May be used in sections : defaults | frontend | listen | backend
7938 no | no | yes | yes
7939 Arguments :
7940 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007941 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007942 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007943
David du Colombier486df472011-03-17 10:40:26 +01007944 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7945 resolvable hostname is supported, but this name will be resolved
7946 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7947 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007948 address as the one from the client connection. This is useful in
7949 transparent proxy architectures where the client's connection is
7950 intercepted and haproxy must forward to the original destination
7951 address. This is more or less what the "transparent" keyword does
7952 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007953 to report statistics. Optionally, an address family prefix may be
7954 used before the address to force the family regardless of the
7955 address format, which can be useful to specify a path to a unix
7956 socket with no slash ('/'). Currently supported prefixes are :
7957 - 'ipv4@' -> address is always IPv4
7958 - 'ipv6@' -> address is always IPv6
7959 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007960 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007961 - 'sockpair@' -> address is the FD of a connected unix
7962 socket or of a socketpair. During a connection, the
7963 backend creates a pair of connected sockets, and passes
7964 one of them over the FD. The bind part will use the
7965 received socket as the client FD. Should be used
7966 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007967 You may want to reference some environment variables in the
7968 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007969 variables. The "init-addr" setting can be used to modify the way
7970 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007971
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007972 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007973 be sent to this port. If unset, the same port the client
7974 connected to will be used. The port may also be prefixed by a "+"
7975 or a "-". In this case, the server's port will be determined by
7976 adding this value to the client's port.
7977
7978 <param*> is a list of parameters for this server. The "server" keywords
7979 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007980 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007981
7982 Examples :
7983 server first 10.1.1.1:1080 cookie first check inter 1000
7984 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007985 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007986 server backup "${SRV_BACKUP}:1080" backup
7987 server www1_dc1 "${LAN_DC1}.101:80"
7988 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007989
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007990 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7991 sun_path length is used for the address length. Some other programs
7992 such as socat use the string length only by default. Pass the option
7993 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7994 make it compatible with HAProxy's.
7995
Mark Lamourinec2247f02012-01-04 13:02:01 -05007996 See also: "default-server", "http-send-name-header" and section 5 about
7997 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007998
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007999server-state-file-name [<file>]
8000 Set the server state file to read, load and apply to servers available in
8001 this backend. It only applies when the directive "load-server-state-from-file"
8002 is set to "local". When <file> is not provided or if this directive is not
8003 set, then backend name is used. If <file> starts with a slash '/', then it is
8004 considered as an absolute path. Otherwise, <file> is concatenated to the
8005 global directive "server-state-file-base".
8006
8007 Example: the minimal configuration below would make HAProxy look for the
8008 state server file '/etc/haproxy/states/bk':
8009
8010 global
8011 server-state-file-base /etc/haproxy/states
8012
8013 backend bk
8014 load-server-state-from-file
8015
8016 See also: "server-state-file-base", "load-server-state-from-file", and
8017 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008018
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008019server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8020 Set a template to initialize servers with shared parameters.
8021 The names of these servers are built from <prefix> and <num | range> parameters.
8022 May be used in sections : defaults | frontend | listen | backend
8023 no | no | yes | yes
8024
8025 Arguments:
8026 <prefix> A prefix for the server names to be built.
8027
8028 <num | range>
8029 If <num> is provided, this template initializes <num> servers
8030 with 1 up to <num> as server name suffixes. A range of numbers
8031 <num_low>-<num_high> may also be used to use <num_low> up to
8032 <num_high> as server name suffixes.
8033
8034 <fqdn> A FQDN for all the servers this template initializes.
8035
8036 <port> Same meaning as "server" <port> argument (see "server" keyword).
8037
8038 <params*>
8039 Remaining server parameters among all those supported by "server"
8040 keyword.
8041
8042 Examples:
8043 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8044 # google.com as FQDN, and health-check enabled.
8045 server-template srv 1-3 google.com:80 check
8046
8047 # or
8048 server-template srv 3 google.com:80 check
8049
8050 # would be equivalent to:
8051 server srv1 google.com:80 check
8052 server srv2 google.com:80 check
8053 server srv3 google.com:80 check
8054
8055
8056
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008057source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008058source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008059source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008060 Set the source address for outgoing connections
8061 May be used in sections : defaults | frontend | listen | backend
8062 yes | no | yes | yes
8063 Arguments :
8064 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8065 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008066
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008067 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008068 the most appropriate address to reach its destination. Optionally
8069 an address family prefix may be used before the address to force
8070 the family regardless of the address format, which can be useful
8071 to specify a path to a unix socket with no slash ('/'). Currently
8072 supported prefixes are :
8073 - 'ipv4@' -> address is always IPv4
8074 - 'ipv6@' -> address is always IPv6
8075 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008076 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008077 You may want to reference some environment variables in the
8078 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008079
8080 <port> is an optional port. It is normally not needed but may be useful
8081 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008082 the system will select a free port. Note that port ranges are not
8083 supported in the backend. If you want to force port ranges, you
8084 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008085
8086 <addr2> is the IP address to present to the server when connections are
8087 forwarded in full transparent proxy mode. This is currently only
8088 supported on some patched Linux kernels. When this address is
8089 specified, clients connecting to the server will be presented
8090 with this address, while health checks will still use the address
8091 <addr>.
8092
8093 <port2> is the optional port to present to the server when connections
8094 are forwarded in full transparent proxy mode (see <addr2> above).
8095 The default value of zero means the system will select a free
8096 port.
8097
Willy Tarreaubce70882009-09-07 11:51:47 +02008098 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8099 This is the name of a comma-separated header list which can
8100 contain multiple IP addresses. By default, the last occurrence is
8101 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008102 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008103 by previous proxy, typically Stunnel. In order to use another
8104 occurrence from the last one, please see the <occ> parameter
8105 below. When the header (or occurrence) is not found, no binding
8106 is performed so that the proxy's default IP address is used. Also
8107 keep in mind that the header name is case insensitive, as for any
8108 HTTP header.
8109
8110 <occ> is the occurrence number of a value to be used in a multi-value
8111 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008112 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008113 address. Positive values indicate a position from the first
8114 occurrence, 1 being the first one. Negative values indicate
8115 positions relative to the last one, -1 being the last one. This
8116 is helpful for situations where an X-Forwarded-For header is set
8117 at the entry point of an infrastructure and must be used several
8118 proxy layers away. When this value is not specified, -1 is
8119 assumed. Passing a zero here disables the feature.
8120
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008121 <name> is an optional interface name to which to bind to for outgoing
8122 traffic. On systems supporting this features (currently, only
8123 Linux), this allows one to bind all traffic to the server to
8124 this interface even if it is not the one the system would select
8125 based on routing tables. This should be used with extreme care.
8126 Note that using this option requires root privileges.
8127
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008128 The "source" keyword is useful in complex environments where a specific
8129 address only is allowed to connect to the servers. It may be needed when a
8130 private address must be used through a public gateway for instance, and it is
8131 known that the system cannot determine the adequate source address by itself.
8132
8133 An extension which is available on certain patched Linux kernels may be used
8134 through the "usesrc" optional keyword. It makes it possible to connect to the
8135 servers with an IP address which does not belong to the system itself. This
8136 is called "full transparent proxy mode". For this to work, the destination
8137 servers have to route their traffic back to this address through the machine
8138 running HAProxy, and IP forwarding must generally be enabled on this machine.
8139
8140 In this "full transparent proxy" mode, it is possible to force a specific IP
8141 address to be presented to the servers. This is not much used in fact. A more
8142 common use is to tell HAProxy to present the client's IP address. For this,
8143 there are two methods :
8144
8145 - present the client's IP and port addresses. This is the most transparent
8146 mode, but it can cause problems when IP connection tracking is enabled on
8147 the machine, because a same connection may be seen twice with different
8148 states. However, this solution presents the huge advantage of not
8149 limiting the system to the 64k outgoing address+port couples, because all
8150 of the client ranges may be used.
8151
8152 - present only the client's IP address and select a spare port. This
8153 solution is still quite elegant but slightly less transparent (downstream
8154 firewalls logs will not match upstream's). It also presents the downside
8155 of limiting the number of concurrent connections to the usual 64k ports.
8156 However, since the upstream and downstream ports are different, local IP
8157 connection tracking on the machine will not be upset by the reuse of the
8158 same session.
8159
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008160 This option sets the default source for all servers in the backend. It may
8161 also be specified in a "defaults" section. Finer source address specification
8162 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008163 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008164
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008165 In order to work, "usesrc" requires root privileges.
8166
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008167 Examples :
8168 backend private
8169 # Connect to the servers using our 192.168.1.200 source address
8170 source 192.168.1.200
8171
8172 backend transparent_ssl1
8173 # Connect to the SSL farm from the client's source address
8174 source 192.168.1.200 usesrc clientip
8175
8176 backend transparent_ssl2
8177 # Connect to the SSL farm from the client's source address and port
8178 # not recommended if IP conntrack is present on the local machine.
8179 source 192.168.1.200 usesrc client
8180
8181 backend transparent_ssl3
8182 # Connect to the SSL farm from the client's source address. It
8183 # is more conntrack-friendly.
8184 source 192.168.1.200 usesrc clientip
8185
8186 backend transparent_smtp
8187 # Connect to the SMTP farm from the client's source address/port
8188 # with Tproxy version 4.
8189 source 0.0.0.0 usesrc clientip
8190
Willy Tarreaubce70882009-09-07 11:51:47 +02008191 backend transparent_http
8192 # Connect to the servers using the client's IP as seen by previous
8193 # proxy.
8194 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8195
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008196 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008197 the Linux kernel on www.balabit.com, the "bind" keyword.
8198
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008199
Willy Tarreau844e3c52008-01-11 16:28:18 +01008200srvtimeout <timeout> (deprecated)
8201 Set the maximum inactivity time on the server side.
8202 May be used in sections : defaults | frontend | listen | backend
8203 yes | no | yes | yes
8204 Arguments :
8205 <timeout> is the timeout value specified in milliseconds by default, but
8206 can be in any other unit if the number is suffixed by the unit,
8207 as explained at the top of this document.
8208
8209 The inactivity timeout applies when the server is expected to acknowledge or
8210 send data. In HTTP mode, this timeout is particularly important to consider
8211 during the first phase of the server's response, when it has to send the
8212 headers, as it directly represents the server's processing time for the
8213 request. To find out what value to put there, it's often good to start with
8214 what would be considered as unacceptable response times, then check the logs
8215 to observe the response time distribution, and adjust the value accordingly.
8216
8217 The value is specified in milliseconds by default, but can be in any other
8218 unit if the number is suffixed by the unit, as specified at the top of this
8219 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8220 recommended that the client timeout remains equal to the server timeout in
8221 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008222 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008223 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008224 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008225
8226 This parameter is specific to backends, but can be specified once for all in
8227 "defaults" sections. This is in fact one of the easiest solutions not to
8228 forget about it. An unspecified timeout results in an infinite timeout, which
8229 is not recommended. Such a usage is accepted and works but reports a warning
8230 during startup because it may results in accumulation of expired sessions in
8231 the system if the system's timeouts are not configured either.
8232
8233 This parameter is provided for compatibility but is currently deprecated.
8234 Please use "timeout server" instead.
8235
Willy Tarreauce887fd2012-05-12 12:50:00 +02008236 See also : "timeout server", "timeout tunnel", "timeout client" and
8237 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008238
8239
Cyril Bonté66c327d2010-10-12 00:14:37 +02008240stats admin { if | unless } <cond>
8241 Enable statistics admin level if/unless a condition is matched
8242 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008243 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008244
8245 This statement enables the statistics admin level if/unless a condition is
8246 matched.
8247
8248 The admin level allows to enable/disable servers from the web interface. By
8249 default, statistics page is read-only for security reasons.
8250
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008251 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8252 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008253 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008254
Cyril Bonté23b39d92011-02-10 22:54:44 +01008255 Currently, the POST request is limited to the buffer size minus the reserved
8256 buffer space, which means that if the list of servers is too long, the
8257 request won't be processed. It is recommended to alter few servers at a
8258 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008259
8260 Example :
8261 # statistics admin level only for localhost
8262 backend stats_localhost
8263 stats enable
8264 stats admin if LOCALHOST
8265
8266 Example :
8267 # statistics admin level always enabled because of the authentication
8268 backend stats_auth
8269 stats enable
8270 stats auth admin:AdMiN123
8271 stats admin if TRUE
8272
8273 Example :
8274 # statistics admin level depends on the authenticated user
8275 userlist stats-auth
8276 group admin users admin
8277 user admin insecure-password AdMiN123
8278 group readonly users haproxy
8279 user haproxy insecure-password haproxy
8280
8281 backend stats_auth
8282 stats enable
8283 acl AUTH http_auth(stats-auth)
8284 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8285 stats http-request auth unless AUTH
8286 stats admin if AUTH_ADMIN
8287
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008288 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8289 "bind-process", section 3.4 about userlists and section 7 about
8290 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008291
8292
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008293stats auth <user>:<passwd>
8294 Enable statistics with authentication and grant access to an account
8295 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008296 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008297 Arguments :
8298 <user> is a user name to grant access to
8299
8300 <passwd> is the cleartext password associated to this user
8301
8302 This statement enables statistics with default settings, and restricts access
8303 to declared users only. It may be repeated as many times as necessary to
8304 allow as many users as desired. When a user tries to access the statistics
8305 without a valid account, a "401 Forbidden" response will be returned so that
8306 the browser asks the user to provide a valid user and password. The real
8307 which will be returned to the browser is configurable using "stats realm".
8308
8309 Since the authentication method is HTTP Basic Authentication, the passwords
8310 circulate in cleartext on the network. Thus, it was decided that the
8311 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008312 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008313
8314 It is also possible to reduce the scope of the proxies which appear in the
8315 report using "stats scope".
8316
8317 Though this statement alone is enough to enable statistics reporting, it is
8318 recommended to set all other settings in order to avoid relying on default
8319 unobvious parameters.
8320
8321 Example :
8322 # public access (limited to this backend only)
8323 backend public_www
8324 server srv1 192.168.0.1:80
8325 stats enable
8326 stats hide-version
8327 stats scope .
8328 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008329 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008330 stats auth admin1:AdMiN123
8331 stats auth admin2:AdMiN321
8332
8333 # internal monitoring access (unlimited)
8334 backend private_monitoring
8335 stats enable
8336 stats uri /admin?stats
8337 stats refresh 5s
8338
8339 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8340
8341
8342stats enable
8343 Enable statistics reporting with default settings
8344 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008345 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008346 Arguments : none
8347
8348 This statement enables statistics reporting with default settings defined
8349 at build time. Unless stated otherwise, these settings are used :
8350 - stats uri : /haproxy?stats
8351 - stats realm : "HAProxy Statistics"
8352 - stats auth : no authentication
8353 - stats scope : no restriction
8354
8355 Though this statement alone is enough to enable statistics reporting, it is
8356 recommended to set all other settings in order to avoid relying on default
8357 unobvious parameters.
8358
8359 Example :
8360 # public access (limited to this backend only)
8361 backend public_www
8362 server srv1 192.168.0.1:80
8363 stats enable
8364 stats hide-version
8365 stats scope .
8366 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008367 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008368 stats auth admin1:AdMiN123
8369 stats auth admin2:AdMiN321
8370
8371 # internal monitoring access (unlimited)
8372 backend private_monitoring
8373 stats enable
8374 stats uri /admin?stats
8375 stats refresh 5s
8376
8377 See also : "stats auth", "stats realm", "stats uri"
8378
8379
Willy Tarreaud63335a2010-02-26 12:56:52 +01008380stats hide-version
8381 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008382 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008383 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008384 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008385
Willy Tarreaud63335a2010-02-26 12:56:52 +01008386 By default, the stats page reports some useful status information along with
8387 the statistics. Among them is HAProxy's version. However, it is generally
8388 considered dangerous to report precise version to anyone, as it can help them
8389 target known weaknesses with specific attacks. The "stats hide-version"
8390 statement removes the version from the statistics report. This is recommended
8391 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008392
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008393 Though this statement alone is enough to enable statistics reporting, it is
8394 recommended to set all other settings in order to avoid relying on default
8395 unobvious parameters.
8396
Willy Tarreaud63335a2010-02-26 12:56:52 +01008397 Example :
8398 # public access (limited to this backend only)
8399 backend public_www
8400 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008401 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008402 stats hide-version
8403 stats scope .
8404 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008405 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008406 stats auth admin1:AdMiN123
8407 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008408
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008409 # internal monitoring access (unlimited)
8410 backend private_monitoring
8411 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008412 stats uri /admin?stats
8413 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008414
Willy Tarreaud63335a2010-02-26 12:56:52 +01008415 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008416
Willy Tarreau983e01e2010-01-11 18:42:06 +01008417
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008418stats http-request { allow | deny | auth [realm <realm>] }
8419 [ { if | unless } <condition> ]
8420 Access control for statistics
8421
8422 May be used in sections: defaults | frontend | listen | backend
8423 no | no | yes | yes
8424
8425 As "http-request", these set of options allow to fine control access to
8426 statistics. Each option may be followed by if/unless and acl.
8427 First option with matched condition (or option without condition) is final.
8428 For "deny" a 403 error will be returned, for "allow" normal processing is
8429 performed, for "auth" a 401/407 error code is returned so the client
8430 should be asked to enter a username and password.
8431
8432 There is no fixed limit to the number of http-request statements per
8433 instance.
8434
8435 See also : "http-request", section 3.4 about userlists and section 7
8436 about ACL usage.
8437
8438
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008439stats realm <realm>
8440 Enable statistics and set authentication realm
8441 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008442 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008443 Arguments :
8444 <realm> is the name of the HTTP Basic Authentication realm reported to
8445 the browser. The browser uses it to display it in the pop-up
8446 inviting the user to enter a valid username and password.
8447
8448 The realm is read as a single word, so any spaces in it should be escaped
8449 using a backslash ('\').
8450
8451 This statement is useful only in conjunction with "stats auth" since it is
8452 only related to authentication.
8453
8454 Though this statement alone is enough to enable statistics reporting, it is
8455 recommended to set all other settings in order to avoid relying on default
8456 unobvious parameters.
8457
8458 Example :
8459 # public access (limited to this backend only)
8460 backend public_www
8461 server srv1 192.168.0.1:80
8462 stats enable
8463 stats hide-version
8464 stats scope .
8465 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008466 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008467 stats auth admin1:AdMiN123
8468 stats auth admin2:AdMiN321
8469
8470 # internal monitoring access (unlimited)
8471 backend private_monitoring
8472 stats enable
8473 stats uri /admin?stats
8474 stats refresh 5s
8475
8476 See also : "stats auth", "stats enable", "stats uri"
8477
8478
8479stats refresh <delay>
8480 Enable statistics with automatic refresh
8481 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008482 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008483 Arguments :
8484 <delay> is the suggested refresh delay, specified in seconds, which will
8485 be returned to the browser consulting the report page. While the
8486 browser is free to apply any delay, it will generally respect it
8487 and refresh the page this every seconds. The refresh interval may
8488 be specified in any other non-default time unit, by suffixing the
8489 unit after the value, as explained at the top of this document.
8490
8491 This statement is useful on monitoring displays with a permanent page
8492 reporting the load balancer's activity. When set, the HTML report page will
8493 include a link "refresh"/"stop refresh" so that the user can select whether
8494 he wants automatic refresh of the page or not.
8495
8496 Though this statement alone is enough to enable statistics reporting, it is
8497 recommended to set all other settings in order to avoid relying on default
8498 unobvious parameters.
8499
8500 Example :
8501 # public access (limited to this backend only)
8502 backend public_www
8503 server srv1 192.168.0.1:80
8504 stats enable
8505 stats hide-version
8506 stats scope .
8507 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008508 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008509 stats auth admin1:AdMiN123
8510 stats auth admin2:AdMiN321
8511
8512 # internal monitoring access (unlimited)
8513 backend private_monitoring
8514 stats enable
8515 stats uri /admin?stats
8516 stats refresh 5s
8517
8518 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8519
8520
8521stats scope { <name> | "." }
8522 Enable statistics and limit access scope
8523 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008524 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008525 Arguments :
8526 <name> is the name of a listen, frontend or backend section to be
8527 reported. The special name "." (a single dot) designates the
8528 section in which the statement appears.
8529
8530 When this statement is specified, only the sections enumerated with this
8531 statement will appear in the report. All other ones will be hidden. This
8532 statement may appear as many times as needed if multiple sections need to be
8533 reported. Please note that the name checking is performed as simple string
8534 comparisons, and that it is never checked that a give section name really
8535 exists.
8536
8537 Though this statement alone is enough to enable statistics reporting, it is
8538 recommended to set all other settings in order to avoid relying on default
8539 unobvious parameters.
8540
8541 Example :
8542 # public access (limited to this backend only)
8543 backend public_www
8544 server srv1 192.168.0.1:80
8545 stats enable
8546 stats hide-version
8547 stats scope .
8548 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008549 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008550 stats auth admin1:AdMiN123
8551 stats auth admin2:AdMiN321
8552
8553 # internal monitoring access (unlimited)
8554 backend private_monitoring
8555 stats enable
8556 stats uri /admin?stats
8557 stats refresh 5s
8558
8559 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8560
Willy Tarreaud63335a2010-02-26 12:56:52 +01008561
Willy Tarreauc9705a12010-07-27 20:05:50 +02008562stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008563 Enable reporting of a description on the statistics page.
8564 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008565 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008566
Willy Tarreauc9705a12010-07-27 20:05:50 +02008567 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008568 description from global section is automatically used instead.
8569
8570 This statement is useful for users that offer shared services to their
8571 customers, where node or description should be different for each customer.
8572
8573 Though this statement alone is enough to enable statistics reporting, it is
8574 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008575 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008576
8577 Example :
8578 # internal monitoring access (unlimited)
8579 backend private_monitoring
8580 stats enable
8581 stats show-desc Master node for Europe, Asia, Africa
8582 stats uri /admin?stats
8583 stats refresh 5s
8584
8585 See also: "show-node", "stats enable", "stats uri" and "description" in
8586 global section.
8587
8588
8589stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008590 Enable reporting additional information on the statistics page
8591 May be used in sections : defaults | frontend | listen | backend
8592 yes | yes | yes | yes
8593 Arguments : none
8594
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008595 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008596 - cap: capabilities (proxy)
8597 - mode: one of tcp, http or health (proxy)
8598 - id: SNMP ID (proxy, socket, server)
8599 - IP (socket, server)
8600 - cookie (backend, server)
8601
8602 Though this statement alone is enough to enable statistics reporting, it is
8603 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008604 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008605
8606 See also: "stats enable", "stats uri".
8607
8608
8609stats show-node [ <name> ]
8610 Enable reporting of a host name on the statistics page.
8611 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008612 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008613 Arguments:
8614 <name> is an optional name to be reported. If unspecified, the
8615 node name from global section is automatically used instead.
8616
8617 This statement is useful for users that offer shared services to their
8618 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008619 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008620
8621 Though this statement alone is enough to enable statistics reporting, it is
8622 recommended to set all other settings in order to avoid relying on default
8623 unobvious parameters.
8624
8625 Example:
8626 # internal monitoring access (unlimited)
8627 backend private_monitoring
8628 stats enable
8629 stats show-node Europe-1
8630 stats uri /admin?stats
8631 stats refresh 5s
8632
8633 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8634 section.
8635
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008636
8637stats uri <prefix>
8638 Enable statistics and define the URI prefix to access them
8639 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008640 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008641 Arguments :
8642 <prefix> is the prefix of any URI which will be redirected to stats. This
8643 prefix may contain a question mark ('?') to indicate part of a
8644 query string.
8645
8646 The statistics URI is intercepted on the relayed traffic, so it appears as a
8647 page within the normal application. It is strongly advised to ensure that the
8648 selected URI will never appear in the application, otherwise it will never be
8649 possible to reach it in the application.
8650
8651 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008652 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008653 It is generally a good idea to include a question mark in the URI so that
8654 intermediate proxies refrain from caching the results. Also, since any string
8655 beginning with the prefix will be accepted as a stats request, the question
8656 mark helps ensuring that no valid URI will begin with the same words.
8657
8658 It is sometimes very convenient to use "/" as the URI prefix, and put that
8659 statement in a "listen" instance of its own. That makes it easy to dedicate
8660 an address or a port to statistics only.
8661
8662 Though this statement alone is enough to enable statistics reporting, it is
8663 recommended to set all other settings in order to avoid relying on default
8664 unobvious parameters.
8665
8666 Example :
8667 # public access (limited to this backend only)
8668 backend public_www
8669 server srv1 192.168.0.1:80
8670 stats enable
8671 stats hide-version
8672 stats scope .
8673 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008674 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008675 stats auth admin1:AdMiN123
8676 stats auth admin2:AdMiN321
8677
8678 # internal monitoring access (unlimited)
8679 backend private_monitoring
8680 stats enable
8681 stats uri /admin?stats
8682 stats refresh 5s
8683
8684 See also : "stats auth", "stats enable", "stats realm"
8685
8686
Willy Tarreaud63335a2010-02-26 12:56:52 +01008687stick match <pattern> [table <table>] [{if | unless} <cond>]
8688 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008689 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008690 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008691
8692 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008693 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008694 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008695 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008696 stickiness table. This rule is mandatory.
8697
8698 <table> is an optional stickiness table name. If unspecified, the same
8699 backend's table is used. A stickiness table is declared using
8700 the "stick-table" statement.
8701
8702 <cond> is an optional matching condition. It makes it possible to match
8703 on a certain criterion only when other conditions are met (or
8704 not met). For instance, it could be used to match on a source IP
8705 address except when a request passes through a known proxy, in
8706 which case we'd match on a header containing that IP address.
8707
8708 Some protocols or applications require complex stickiness rules and cannot
8709 always simply rely on cookies nor hashing. The "stick match" statement
8710 describes a rule to extract the stickiness criterion from an incoming request
8711 or connection. See section 7 for a complete list of possible patterns and
8712 transformation rules.
8713
8714 The table has to be declared using the "stick-table" statement. It must be of
8715 a type compatible with the pattern. By default it is the one which is present
8716 in the same backend. It is possible to share a table with other backends by
8717 referencing it using the "table" keyword. If another table is referenced,
8718 the server's ID inside the backends are used. By default, all server IDs
8719 start at 1 in each backend, so the server ordering is enough. But in case of
8720 doubt, it is highly recommended to force server IDs using their "id" setting.
8721
8722 It is possible to restrict the conditions where a "stick match" statement
8723 will apply, using "if" or "unless" followed by a condition. See section 7 for
8724 ACL based conditions.
8725
8726 There is no limit on the number of "stick match" statements. The first that
8727 applies and matches will cause the request to be directed to the same server
8728 as was used for the request which created the entry. That way, multiple
8729 matches can be used as fallbacks.
8730
8731 The stick rules are checked after the persistence cookies, so they will not
8732 affect stickiness if a cookie has already been used to select a server. That
8733 way, it becomes very easy to insert cookies and match on IP addresses in
8734 order to maintain stickiness between HTTP and HTTPS.
8735
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008736 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8737 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008738 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008739
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008740 Example :
8741 # forward SMTP users to the same server they just used for POP in the
8742 # last 30 minutes
8743 backend pop
8744 mode tcp
8745 balance roundrobin
8746 stick store-request src
8747 stick-table type ip size 200k expire 30m
8748 server s1 192.168.1.1:110
8749 server s2 192.168.1.1:110
8750
8751 backend smtp
8752 mode tcp
8753 balance roundrobin
8754 stick match src table pop
8755 server s1 192.168.1.1:25
8756 server s2 192.168.1.1:25
8757
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008758 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008759 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008760
8761
8762stick on <pattern> [table <table>] [{if | unless} <condition>]
8763 Define a request pattern to associate a user to a server
8764 May be used in sections : defaults | frontend | listen | backend
8765 no | no | yes | yes
8766
8767 Note : This form is exactly equivalent to "stick match" followed by
8768 "stick store-request", all with the same arguments. Please refer
8769 to both keywords for details. It is only provided as a convenience
8770 for writing more maintainable configurations.
8771
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008772 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8773 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008774 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008775
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008776 Examples :
8777 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008778 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008779
8780 # ...is strictly equivalent to this one :
8781 stick match src table pop if !localhost
8782 stick store-request src table pop if !localhost
8783
8784
8785 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8786 # well as HTTP without cookie. Share the same table between both accesses.
8787 backend http
8788 mode http
8789 balance roundrobin
8790 stick on src table https
8791 cookie SRV insert indirect nocache
8792 server s1 192.168.1.1:80 cookie s1
8793 server s2 192.168.1.1:80 cookie s2
8794
8795 backend https
8796 mode tcp
8797 balance roundrobin
8798 stick-table type ip size 200k expire 30m
8799 stick on src
8800 server s1 192.168.1.1:443
8801 server s2 192.168.1.1:443
8802
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008803 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008804
8805
8806stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8807 Define a request pattern used to create an entry in a stickiness table
8808 May be used in sections : defaults | frontend | listen | backend
8809 no | no | yes | yes
8810
8811 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008812 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008813 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008814 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008815 server is selected.
8816
8817 <table> is an optional stickiness table name. If unspecified, the same
8818 backend's table is used. A stickiness table is declared using
8819 the "stick-table" statement.
8820
8821 <cond> is an optional storage condition. It makes it possible to store
8822 certain criteria only when some conditions are met (or not met).
8823 For instance, it could be used to store the source IP address
8824 except when the request passes through a known proxy, in which
8825 case we'd store a converted form of a header containing that IP
8826 address.
8827
8828 Some protocols or applications require complex stickiness rules and cannot
8829 always simply rely on cookies nor hashing. The "stick store-request" statement
8830 describes a rule to decide what to extract from the request and when to do
8831 it, in order to store it into a stickiness table for further requests to
8832 match it using the "stick match" statement. Obviously the extracted part must
8833 make sense and have a chance to be matched in a further request. Storing a
8834 client's IP address for instance often makes sense. Storing an ID found in a
8835 URL parameter also makes sense. Storing a source port will almost never make
8836 any sense because it will be randomly matched. See section 7 for a complete
8837 list of possible patterns and transformation rules.
8838
8839 The table has to be declared using the "stick-table" statement. It must be of
8840 a type compatible with the pattern. By default it is the one which is present
8841 in the same backend. It is possible to share a table with other backends by
8842 referencing it using the "table" keyword. If another table is referenced,
8843 the server's ID inside the backends are used. By default, all server IDs
8844 start at 1 in each backend, so the server ordering is enough. But in case of
8845 doubt, it is highly recommended to force server IDs using their "id" setting.
8846
8847 It is possible to restrict the conditions where a "stick store-request"
8848 statement will apply, using "if" or "unless" followed by a condition. This
8849 condition will be evaluated while parsing the request, so any criteria can be
8850 used. See section 7 for ACL based conditions.
8851
8852 There is no limit on the number of "stick store-request" statements, but
8853 there is a limit of 8 simultaneous stores per request or response. This
8854 makes it possible to store up to 8 criteria, all extracted from either the
8855 request or the response, regardless of the number of rules. Only the 8 first
8856 ones which match will be kept. Using this, it is possible to feed multiple
8857 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008858 another protocol or access method. Using multiple store-request rules with
8859 the same table is possible and may be used to find the best criterion to rely
8860 on, by arranging the rules by decreasing preference order. Only the first
8861 extracted criterion for a given table will be stored. All subsequent store-
8862 request rules referencing the same table will be skipped and their ACLs will
8863 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008864
8865 The "store-request" rules are evaluated once the server connection has been
8866 established, so that the table will contain the real server that processed
8867 the request.
8868
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008869 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8870 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008871 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008872
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008873 Example :
8874 # forward SMTP users to the same server they just used for POP in the
8875 # last 30 minutes
8876 backend pop
8877 mode tcp
8878 balance roundrobin
8879 stick store-request src
8880 stick-table type ip size 200k expire 30m
8881 server s1 192.168.1.1:110
8882 server s2 192.168.1.1:110
8883
8884 backend smtp
8885 mode tcp
8886 balance roundrobin
8887 stick match src table pop
8888 server s1 192.168.1.1:25
8889 server s2 192.168.1.1:25
8890
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008891 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008892 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008893
8894
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008895stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008896 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8897 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008898 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008899 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008900 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008901
8902 Arguments :
8903 ip a table declared with "type ip" will only store IPv4 addresses.
8904 This form is very compact (about 50 bytes per entry) and allows
8905 very fast entry lookup and stores with almost no overhead. This
8906 is mainly used to store client source IP addresses.
8907
David du Colombier9a6d3c92011-03-17 10:40:24 +01008908 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8909 This form is very compact (about 60 bytes per entry) and allows
8910 very fast entry lookup and stores with almost no overhead. This
8911 is mainly used to store client source IP addresses.
8912
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008913 integer a table declared with "type integer" will store 32bit integers
8914 which can represent a client identifier found in a request for
8915 instance.
8916
8917 string a table declared with "type string" will store substrings of up
8918 to <len> characters. If the string provided by the pattern
8919 extractor is larger than <len>, it will be truncated before
8920 being stored. During matching, at most <len> characters will be
8921 compared between the string in the table and the extracted
8922 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008923 to 32 characters.
8924
8925 binary a table declared with "type binary" will store binary blocks
8926 of <len> bytes. If the block provided by the pattern
8927 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008928 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008929 is shorter than <len>, it will be padded by 0. When not
8930 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008931
8932 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008933 "string" type table (See type "string" above). Or the number
8934 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008935 changing this parameter as memory usage will proportionally
8936 increase.
8937
8938 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008939 value directly impacts memory usage. Count approximately
8940 50 bytes per entry, plus the size of a string if any. The size
8941 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008942
8943 [nopurge] indicates that we refuse to purge older entries when the table
8944 is full. When not specified and the table is full when haproxy
8945 wants to store an entry in it, it will flush a few of the oldest
8946 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008947 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008948 be desirable to refuse new entries instead of purging the older
8949 ones. That may be the case when the amount of data to store is
8950 far above the hardware limits and we prefer not to offer access
8951 to new clients than to reject the ones already connected. When
8952 using this parameter, be sure to properly set the "expire"
8953 parameter (see below).
8954
Emeric Brunf099e792010-09-27 12:05:28 +02008955 <peersect> is the name of the peers section to use for replication. Entries
8956 which associate keys to server IDs are kept synchronized with
8957 the remote peers declared in this section. All entries are also
8958 automatically learned from the local peer (old process) during a
8959 soft restart.
8960
Willy Tarreau1abc6732015-05-01 19:21:02 +02008961 NOTE : each peers section may be referenced only by tables
8962 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008963
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008964 <expire> defines the maximum duration of an entry in the table since it
8965 was last created, refreshed or matched. The expiration delay is
8966 defined using the standard time format, similarly as the various
8967 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008968 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008969 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008970 be removed once full. Be sure not to use the "nopurge" parameter
8971 if not expiration delay is specified.
8972
Willy Tarreau08d5f982010-06-06 13:34:54 +02008973 <data_type> is used to store additional information in the stick-table. This
8974 may be used by ACLs in order to control various criteria related
8975 to the activity of the client matching the stick-table. For each
8976 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008977 that the additional data can fit. Several data types may be
8978 stored with an entry. Multiple data types may be specified after
8979 the "store" keyword, as a comma-separated list. Alternatively,
8980 it is possible to repeat the "store" keyword followed by one or
8981 several data types. Except for the "server_id" type which is
8982 automatically detected and enabled, all data types must be
8983 explicitly declared to be stored. If an ACL references a data
8984 type which is not stored, the ACL will simply not match. Some
8985 data types require an argument which must be passed just after
8986 the type between parenthesis. See below for the supported data
8987 types and their arguments.
8988
8989 The data types that can be stored with an entry are the following :
8990 - server_id : this is an integer which holds the numeric ID of the server a
8991 request was assigned to. It is used by the "stick match", "stick store",
8992 and "stick on" rules. It is automatically enabled when referenced.
8993
8994 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8995 integer which may be used for anything. Most of the time it will be used
8996 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008997 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008998
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008999 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9000 over a period. It is a positive 32-bit integer integer which may be used
9001 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009002 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009003 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009004 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009005
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009006 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9007 integer which may be used for anything. Most of the time it will be used
9008 to put a special tag on some entries, for instance to note that a
9009 specific behavior was detected and must be known for future matches.
9010
9011 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9012 over a period. It is a positive 32-bit integer integer which may be used
9013 for anything. Just like <gpc1>, it counts events, but instead of keeping
9014 a cumulative number, it maintains the rate at which the counter is
9015 incremented. Most of the time it will be used to measure the frequency of
9016 occurrence of certain events (e.g. requests to a specific URL).
9017
Willy Tarreauc9705a12010-07-27 20:05:50 +02009018 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9019 the absolute number of connections received from clients which matched
9020 this entry. It does not mean the connections were accepted, just that
9021 they were received.
9022
9023 - conn_cur : Current Connections. It is a positive 32-bit integer which
9024 stores the concurrent connection counts for the entry. It is incremented
9025 once an incoming connection matches the entry, and decremented once the
9026 connection leaves. That way it is possible to know at any time the exact
9027 number of concurrent connections for an entry.
9028
9029 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9030 integer parameter <period> which indicates in milliseconds the length
9031 of the period over which the average is measured. It reports the average
9032 incoming connection rate over that period, in connections per period. The
9033 result is an integer which can be matched using ACLs.
9034
9035 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9036 the absolute number of sessions received from clients which matched this
9037 entry. A session is a connection that was accepted by the layer 4 rules.
9038
9039 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9040 integer parameter <period> which indicates in milliseconds the length
9041 of the period over which the average is measured. It reports the average
9042 incoming session rate over that period, in sessions per period. The
9043 result is an integer which can be matched using ACLs.
9044
9045 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9046 counts the absolute number of HTTP requests received from clients which
9047 matched this entry. It does not matter whether they are valid requests or
9048 not. Note that this is different from sessions when keep-alive is used on
9049 the client side.
9050
9051 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9052 integer parameter <period> which indicates in milliseconds the length
9053 of the period over which the average is measured. It reports the average
9054 HTTP request rate over that period, in requests per period. The result is
9055 an integer which can be matched using ACLs. It does not matter whether
9056 they are valid requests or not. Note that this is different from sessions
9057 when keep-alive is used on the client side.
9058
9059 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9060 counts the absolute number of HTTP requests errors induced by clients
9061 which matched this entry. Errors are counted on invalid and truncated
9062 requests, as well as on denied or tarpitted requests, and on failed
9063 authentications. If the server responds with 4xx, then the request is
9064 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009065 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009066
9067 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9068 integer parameter <period> which indicates in milliseconds the length
9069 of the period over which the average is measured. It reports the average
9070 HTTP request error rate over that period, in requests per period (see
9071 http_err_cnt above for what is accounted as an error). The result is an
9072 integer which can be matched using ACLs.
9073
9074 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009075 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009076 which matched this entry. Headers are included in the count. This may be
9077 used to limit abuse of upload features on photo or video servers.
9078
9079 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9080 integer parameter <period> which indicates in milliseconds the length
9081 of the period over which the average is measured. It reports the average
9082 incoming bytes rate over that period, in bytes per period. It may be used
9083 to detect users which upload too much and too fast. Warning: with large
9084 uploads, it is possible that the amount of uploaded data will be counted
9085 once upon termination, thus causing spikes in the average transfer speed
9086 instead of having a smooth one. This may partially be smoothed with
9087 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9088 recommended for better fairness.
9089
9090 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009091 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009092 matched this entry. Headers are included in the count. This may be used
9093 to limit abuse of bots sucking the whole site.
9094
9095 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9096 an integer parameter <period> which indicates in milliseconds the length
9097 of the period over which the average is measured. It reports the average
9098 outgoing bytes rate over that period, in bytes per period. It may be used
9099 to detect users which download too much and too fast. Warning: with large
9100 transfers, it is possible that the amount of transferred data will be
9101 counted once upon termination, thus causing spikes in the average
9102 transfer speed instead of having a smooth one. This may partially be
9103 smoothed with "option contstats" though this is not perfect yet. Use of
9104 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009105
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009106 There is only one stick-table per proxy. At the moment of writing this doc,
9107 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009108 to be required, simply create a dummy backend with a stick-table in it and
9109 reference it.
9110
9111 It is important to understand that stickiness based on learning information
9112 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009113 lost upon restart unless peers are properly configured to transfer such
9114 information upon restart (recommended). In general it can be good as a
9115 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009116
Willy Tarreauc9705a12010-07-27 20:05:50 +02009117 Last, memory requirements may be important when storing many data types.
9118 Indeed, storing all indicators above at once in each entry requires 116 bytes
9119 per entry, or 116 MB for a 1-million entries table. This is definitely not
9120 something that can be ignored.
9121
9122 Example:
9123 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9124 # and store a general purpose counter and the average connection rate
9125 # computed over a sliding window of 30 seconds.
9126 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9127
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009128 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009129 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009130
9131
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009132stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009133 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009134 May be used in sections : defaults | frontend | listen | backend
9135 no | no | yes | yes
9136
9137 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009138 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009139 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009140 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009141 server is selected.
9142
9143 <table> is an optional stickiness table name. If unspecified, the same
9144 backend's table is used. A stickiness table is declared using
9145 the "stick-table" statement.
9146
9147 <cond> is an optional storage condition. It makes it possible to store
9148 certain criteria only when some conditions are met (or not met).
9149 For instance, it could be used to store the SSL session ID only
9150 when the response is a SSL server hello.
9151
9152 Some protocols or applications require complex stickiness rules and cannot
9153 always simply rely on cookies nor hashing. The "stick store-response"
9154 statement describes a rule to decide what to extract from the response and
9155 when to do it, in order to store it into a stickiness table for further
9156 requests to match it using the "stick match" statement. Obviously the
9157 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009158 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009159 See section 7 for a complete list of possible patterns and transformation
9160 rules.
9161
9162 The table has to be declared using the "stick-table" statement. It must be of
9163 a type compatible with the pattern. By default it is the one which is present
9164 in the same backend. It is possible to share a table with other backends by
9165 referencing it using the "table" keyword. If another table is referenced,
9166 the server's ID inside the backends are used. By default, all server IDs
9167 start at 1 in each backend, so the server ordering is enough. But in case of
9168 doubt, it is highly recommended to force server IDs using their "id" setting.
9169
9170 It is possible to restrict the conditions where a "stick store-response"
9171 statement will apply, using "if" or "unless" followed by a condition. This
9172 condition will be evaluated while parsing the response, so any criteria can
9173 be used. See section 7 for ACL based conditions.
9174
9175 There is no limit on the number of "stick store-response" statements, but
9176 there is a limit of 8 simultaneous stores per request or response. This
9177 makes it possible to store up to 8 criteria, all extracted from either the
9178 request or the response, regardless of the number of rules. Only the 8 first
9179 ones which match will be kept. Using this, it is possible to feed multiple
9180 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009181 another protocol or access method. Using multiple store-response rules with
9182 the same table is possible and may be used to find the best criterion to rely
9183 on, by arranging the rules by decreasing preference order. Only the first
9184 extracted criterion for a given table will be stored. All subsequent store-
9185 response rules referencing the same table will be skipped and their ACLs will
9186 not be evaluated. However, even if a store-request rule references a table, a
9187 store-response rule may also use the same table. This means that each table
9188 may learn exactly one element from the request and one element from the
9189 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009190
9191 The table will contain the real server that processed the request.
9192
9193 Example :
9194 # Learn SSL session ID from both request and response and create affinity.
9195 backend https
9196 mode tcp
9197 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009198 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009199 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009200
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009201 acl clienthello req_ssl_hello_type 1
9202 acl serverhello rep_ssl_hello_type 2
9203
9204 # use tcp content accepts to detects ssl client and server hello.
9205 tcp-request inspect-delay 5s
9206 tcp-request content accept if clienthello
9207
9208 # no timeout on response inspect delay by default.
9209 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009210
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009211 # SSL session ID (SSLID) may be present on a client or server hello.
9212 # Its length is coded on 1 byte at offset 43 and its value starts
9213 # at offset 44.
9214
9215 # Match and learn on request if client hello.
9216 stick on payload_lv(43,1) if clienthello
9217
9218 # Learn on response if server hello.
9219 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009220
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009221 server s1 192.168.1.1:443
9222 server s2 192.168.1.1:443
9223
9224 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9225 extraction.
9226
9227
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009228tcp-check connect [params*]
9229 Opens a new connection
9230 May be used in sections: defaults | frontend | listen | backend
9231 no | no | yes | yes
9232
9233 When an application lies on more than a single TCP port or when HAProxy
9234 load-balance many services in a single backend, it makes sense to probe all
9235 the services individually before considering a server as operational.
9236
9237 When there are no TCP port configured on the server line neither server port
9238 directive, then the 'tcp-check connect port <port>' must be the first step
9239 of the sequence.
9240
9241 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9242 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9243 do.
9244
9245 Parameters :
9246 They are optional and can be used to describe how HAProxy should open and
9247 use the TCP connection.
9248
9249 port if not set, check port or server port is used.
9250 It tells HAProxy where to open the connection to.
9251 <port> must be a valid TCP port source integer, from 1 to 65535.
9252
9253 send-proxy send a PROXY protocol string
9254
9255 ssl opens a ciphered connection
9256
9257 Examples:
9258 # check HTTP and HTTPs services on a server.
9259 # first open port 80 thanks to server line port directive, then
9260 # tcp-check opens port 443, ciphered and run a request on it:
9261 option tcp-check
9262 tcp-check connect
9263 tcp-check send GET\ /\ HTTP/1.0\r\n
9264 tcp-check send Host:\ haproxy.1wt.eu\r\n
9265 tcp-check send \r\n
9266 tcp-check expect rstring (2..|3..)
9267 tcp-check connect port 443 ssl
9268 tcp-check send GET\ /\ HTTP/1.0\r\n
9269 tcp-check send Host:\ haproxy.1wt.eu\r\n
9270 tcp-check send \r\n
9271 tcp-check expect rstring (2..|3..)
9272 server www 10.0.0.1 check port 80
9273
9274 # check both POP and IMAP from a single server:
9275 option tcp-check
9276 tcp-check connect port 110
9277 tcp-check expect string +OK\ POP3\ ready
9278 tcp-check connect port 143
9279 tcp-check expect string *\ OK\ IMAP4\ ready
9280 server mail 10.0.0.1 check
9281
9282 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9283
9284
9285tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009286 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009287 May be used in sections: defaults | frontend | listen | backend
9288 no | no | yes | yes
9289
9290 Arguments :
9291 <match> is a keyword indicating how to look for a specific pattern in the
9292 response. The keyword may be one of "string", "rstring" or
9293 binary.
9294 The keyword may be preceded by an exclamation mark ("!") to negate
9295 the match. Spaces are allowed between the exclamation mark and the
9296 keyword. See below for more details on the supported keywords.
9297
9298 <pattern> is the pattern to look for. It may be a string or a regular
9299 expression. If the pattern contains spaces, they must be escaped
9300 with the usual backslash ('\').
9301 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009302 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009303 two digits will represent a byte. The hexadecimal digits may be
9304 used upper or lower case.
9305
9306
9307 The available matches are intentionally similar to their http-check cousins :
9308
9309 string <string> : test the exact string matches in the response buffer.
9310 A health check response will be considered valid if the
9311 response's buffer contains this exact string. If the
9312 "string" keyword is prefixed with "!", then the response
9313 will be considered invalid if the body contains this
9314 string. This can be used to look for a mandatory pattern
9315 in a protocol response, or to detect a failure when a
9316 specific error appears in a protocol banner.
9317
9318 rstring <regex> : test a regular expression on the response buffer.
9319 A health check response will be considered valid if the
9320 response's buffer matches this expression. If the
9321 "rstring" keyword is prefixed with "!", then the response
9322 will be considered invalid if the body matches the
9323 expression.
9324
9325 binary <hexstring> : test the exact string in its hexadecimal form matches
9326 in the response buffer. A health check response will
9327 be considered valid if the response's buffer contains
9328 this exact hexadecimal string.
9329 Purpose is to match data on binary protocols.
9330
9331 It is important to note that the responses will be limited to a certain size
9332 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9333 Thus, too large responses may not contain the mandatory pattern when using
9334 "string", "rstring" or binary. If a large response is absolutely required, it
9335 is possible to change the default max size by setting the global variable.
9336 However, it is worth keeping in mind that parsing very large responses can
9337 waste some CPU cycles, especially when regular expressions are used, and that
9338 it is always better to focus the checks on smaller resources. Also, in its
9339 current state, the check will not find any string nor regex past a null
9340 character in the response. Similarly it is not possible to request matching
9341 the null character.
9342
9343 Examples :
9344 # perform a POP check
9345 option tcp-check
9346 tcp-check expect string +OK\ POP3\ ready
9347
9348 # perform an IMAP check
9349 option tcp-check
9350 tcp-check expect string *\ OK\ IMAP4\ ready
9351
9352 # look for the redis master server
9353 option tcp-check
9354 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009355 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009356 tcp-check send info\ replication\r\n
9357 tcp-check expect string role:master
9358 tcp-check send QUIT\r\n
9359 tcp-check expect string +OK
9360
9361
9362 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9363 "tcp-check send-binary", "http-check expect", tune.chksize
9364
9365
9366tcp-check send <data>
9367 Specify a string to be sent as a question during a generic health check
9368 May be used in sections: defaults | frontend | listen | backend
9369 no | no | yes | yes
9370
9371 <data> : the data to be sent as a question during a generic health check
9372 session. For now, <data> must be a string.
9373
9374 Examples :
9375 # look for the redis master server
9376 option tcp-check
9377 tcp-check send info\ replication\r\n
9378 tcp-check expect string role:master
9379
9380 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9381 "tcp-check send-binary", tune.chksize
9382
9383
Davor Ocelice9ed2812017-12-25 17:49:28 +01009384tcp-check send-binary <hexstring>
9385 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009386 tcp health check
9387 May be used in sections: defaults | frontend | listen | backend
9388 no | no | yes | yes
9389
9390 <data> : the data to be sent as a question during a generic health check
9391 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009392 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009393 response buffer. A health check response will be considered
9394 valid if the response's buffer contains this exact
9395 hexadecimal string.
9396 Purpose is to send binary data to ask on binary protocols.
9397
9398 Examples :
9399 # redis check in binary
9400 option tcp-check
9401 tcp-check send-binary 50494e470d0a # PING\r\n
9402 tcp-check expect binary 2b504F4e47 # +PONG
9403
9404
9405 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9406 "tcp-check send", tune.chksize
9407
9408
Willy Tarreaue9656522010-08-17 15:40:09 +02009409tcp-request connection <action> [{if | unless} <condition>]
9410 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009411 May be used in sections : defaults | frontend | listen | backend
9412 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009413 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009414 <action> defines the action to perform if the condition applies. See
9415 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009416
Willy Tarreaue9656522010-08-17 15:40:09 +02009417 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009418
9419 Immediately after acceptance of a new incoming connection, it is possible to
9420 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009421 or dropped or have its counters tracked. Those conditions cannot make use of
9422 any data contents because the connection has not been read from yet, and the
9423 buffers are not yet allocated. This is used to selectively and very quickly
9424 accept or drop connections from various sources with a very low overhead. If
9425 some contents need to be inspected in order to take the decision, the
9426 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009427
Willy Tarreaue9656522010-08-17 15:40:09 +02009428 The "tcp-request connection" rules are evaluated in their exact declaration
9429 order. If no rule matches or if there is no rule, the default action is to
9430 accept the incoming connection. There is no specific limit to the number of
9431 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009432
Willy Tarreaua9083d02015-05-08 15:27:59 +02009433 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009434 - accept :
9435 accepts the connection if the condition is true (when used with "if")
9436 or false (when used with "unless"). The first such rule executed ends
9437 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009438
Willy Tarreaue9656522010-08-17 15:40:09 +02009439 - reject :
9440 rejects the connection if the condition is true (when used with "if")
9441 or false (when used with "unless"). The first such rule executed ends
9442 the rules evaluation. Rejected connections do not even become a
9443 session, which is why they are accounted separately for in the stats,
9444 as "denied connections". They are not considered for the session
9445 rate-limit and are not logged either. The reason is that these rules
9446 should only be used to filter extremely high connection rates such as
9447 the ones encountered during a massive DDoS attack. Under these extreme
9448 conditions, the simple action of logging each event would make the
9449 system collapse and would considerably lower the filtering capacity. If
9450 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009451 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009452
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009453 - expect-proxy layer4 :
9454 configures the client-facing connection to receive a PROXY protocol
9455 header before any byte is read from the socket. This is equivalent to
9456 having the "accept-proxy" keyword on the "bind" line, except that using
9457 the TCP rule allows the PROXY protocol to be accepted only for certain
9458 IP address ranges using an ACL. This is convenient when multiple layers
9459 of load balancers are passed through by traffic coming from public
9460 hosts.
9461
Bertrand Jacquin90759682016-06-06 15:35:39 +01009462 - expect-netscaler-cip layer4 :
9463 configures the client-facing connection to receive a NetScaler Client
9464 IP insertion protocol header before any byte is read from the socket.
9465 This is equivalent to having the "accept-netscaler-cip" keyword on the
9466 "bind" line, except that using the TCP rule allows the PROXY protocol
9467 to be accepted only for certain IP address ranges using an ACL. This
9468 is convenient when multiple layers of load balancers are passed
9469 through by traffic coming from public hosts.
9470
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009471 - capture <sample> len <length> :
9472 This only applies to "tcp-request content" rules. It captures sample
9473 expression <sample> from the request buffer, and converts it to a
9474 string of at most <len> characters. The resulting string is stored into
9475 the next request "capture" slot, so it will possibly appear next to
9476 some captured HTTP headers. It will then automatically appear in the
9477 logs, and it will be possible to extract it using sample fetch rules to
9478 feed it into headers or anything. The length should be limited given
9479 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009480 session life. Please check section 7.3 (Fetching samples) and "capture
9481 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009482
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009483 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009484 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009485 rules do not stop evaluation and do not change default action. The
9486 number of counters that may be simultaneously tracked by the same
9487 connection is set in MAX_SESS_STKCTR at build time (reported in
9488 haproxy -vv) whichs defaults to 3, so the track-sc number is between 0
9489 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9490 tracking of the counters of the specified table as the first set. The
9491 first "track-sc1" rule executed enables tracking of the counters of the
9492 specified table as the second set. The first "track-sc2" rule executed
9493 enables tracking of the counters of the specified table as the third
9494 set. It is a recommended practice to use the first set of counters for
9495 the per-frontend counters and the second set for the per-backend ones.
9496 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009497
Willy Tarreaue9656522010-08-17 15:40:09 +02009498 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009499 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009500 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009501 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009502 and used to select which table entry to update the counters.
9503 Note that "tcp-request connection" cannot use content-based
9504 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009505
Willy Tarreaue9656522010-08-17 15:40:09 +02009506 <table> is an optional table to be used instead of the default one,
9507 which is the stick-table declared in the current proxy. All
9508 the counters for the matches and updates for the key will
9509 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009510
Willy Tarreaue9656522010-08-17 15:40:09 +02009511 Once a "track-sc*" rule is executed, the key is looked up in the table
9512 and if it is not found, an entry is allocated for it. Then a pointer to
9513 that entry is kept during all the session's life, and this entry's
9514 counters are updated as often as possible, every time the session's
9515 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009516 Counters are only updated for events that happen after the tracking has
9517 been started. For example, connection counters will not be updated when
9518 tracking layer 7 information, since the connection event happens before
9519 layer7 information is extracted.
9520
Willy Tarreaue9656522010-08-17 15:40:09 +02009521 If the entry tracks concurrent connection counters, one connection is
9522 counted for as long as the entry is tracked, and the entry will not
9523 expire during that time. Tracking counters also provides a performance
9524 advantage over just checking the keys, because only one table lookup is
9525 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009526
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009527 - sc-inc-gpc0(<sc-id>):
9528 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9529 counter designated by <sc-id>. If an error occurs, this action silently
9530 fails and the actions evaluation continues.
9531
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009532 - sc-inc-gpc1(<sc-id>):
9533 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9534 counter designated by <sc-id>. If an error occurs, this action silently
9535 fails and the actions evaluation continues.
9536
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009537 - sc-set-gpt0(<sc-id>) <int>:
9538 This action sets the GPT0 tag according to the sticky counter designated
9539 by <sc-id> and the value of <int>. The expected result is a boolean. If
9540 an error occurs, this action silently fails and the actions evaluation
9541 continues.
9542
William Lallemand2e785f22016-05-25 01:48:42 +02009543 - set-src <expr> :
9544 Is used to set the source IP address to the value of specified
9545 expression. Useful if you want to mask source IP for privacy.
9546 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009547 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009548
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009549 Arguments:
9550 <expr> Is a standard HAProxy expression formed by a sample-fetch
9551 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009552
9553 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009554 tcp-request connection set-src src,ipmask(24)
9555
Willy Tarreau0c630532016-10-21 17:52:58 +02009556 When possible, set-src preserves the original source port as long as the
9557 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009558
William Lallemand44be6402016-05-25 01:51:35 +02009559 - set-src-port <expr> :
9560 Is used to set the source port address to the value of specified
9561 expression.
9562
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009563 Arguments:
9564 <expr> Is a standard HAProxy expression formed by a sample-fetch
9565 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009566
9567 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009568 tcp-request connection set-src-port int(4000)
9569
Willy Tarreau0c630532016-10-21 17:52:58 +02009570 When possible, set-src-port preserves the original source address as long
9571 as the address family supports a port, otherwise it forces the source
9572 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009573
William Lallemand13e9b0c2016-05-25 02:34:07 +02009574 - set-dst <expr> :
9575 Is used to set the destination IP address to the value of specified
9576 expression. Useful if you want to mask IP for privacy in log.
9577 If you want to provide an IP from a HTTP header use "http-request
9578 set-dst". If you want to connect to the new address/port, use
9579 '0.0.0.0:0' as a server address in the backend.
9580
9581 <expr> Is a standard HAProxy expression formed by a sample-fetch
9582 followed by some converters.
9583
9584 Example:
9585
9586 tcp-request connection set-dst dst,ipmask(24)
9587 tcp-request connection set-dst ipv4(10.0.0.1)
9588
Willy Tarreau0c630532016-10-21 17:52:58 +02009589 When possible, set-dst preserves the original destination port as long as
9590 the address family allows it, otherwise the destination port is set to 0.
9591
William Lallemand13e9b0c2016-05-25 02:34:07 +02009592 - set-dst-port <expr> :
9593 Is used to set the destination port address to the value of specified
9594 expression. If you want to connect to the new address/port, use
9595 '0.0.0.0:0' as a server address in the backend.
9596
9597
9598 <expr> Is a standard HAProxy expression formed by a sample-fetch
9599 followed by some converters.
9600
9601 Example:
9602
9603 tcp-request connection set-dst-port int(4000)
9604
Willy Tarreau0c630532016-10-21 17:52:58 +02009605 When possible, set-dst-port preserves the original destination address as
9606 long as the address family supports a port, otherwise it forces the
9607 destination address to IPv4 "0.0.0.0" before rewriting the port.
9608
Willy Tarreau2d392c22015-08-24 01:43:45 +02009609 - "silent-drop" :
9610 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009611 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009612 to prevent the client from being notified. The effect it then that the
9613 client still sees an established connection while there's none on
9614 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9615 except that it doesn't use any local resource at all on the machine
9616 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009617 slow down stronger attackers. It is important to understand the impact
9618 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009619 client and HAProxy (firewalls, proxies, load balancers) will also keep
9620 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009621 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009622 TCP_REPAIR socket option is used to block the emission of a TCP
9623 reset. On other systems, the socket's TTL is reduced to 1 so that the
9624 TCP reset doesn't pass the first router, though it's still delivered to
9625 local networks. Do not use it unless you fully understand how it works.
9626
Willy Tarreaue9656522010-08-17 15:40:09 +02009627 Note that the "if/unless" condition is optional. If no condition is set on
9628 the action, it is simply performed unconditionally. That can be useful for
9629 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009630
Willy Tarreaue9656522010-08-17 15:40:09 +02009631 Example: accept all connections from white-listed hosts, reject too fast
9632 connection without counting them, and track accepted connections.
9633 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009634
Willy Tarreaue9656522010-08-17 15:40:09 +02009635 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009636 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009637 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009638
Willy Tarreaue9656522010-08-17 15:40:09 +02009639 Example: accept all connections from white-listed hosts, count all other
9640 connections and reject too fast ones. This results in abusive ones
9641 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009642
Willy Tarreaue9656522010-08-17 15:40:09 +02009643 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009644 tcp-request connection track-sc0 src
9645 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009646
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009647 Example: enable the PROXY protocol for traffic coming from all known proxies.
9648
9649 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9650
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009651 See section 7 about ACL usage.
9652
Willy Tarreau4f614292016-10-21 17:49:36 +02009653 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009654
9655
Willy Tarreaue9656522010-08-17 15:40:09 +02009656tcp-request content <action> [{if | unless} <condition>]
9657 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009658 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009659 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009660 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009661 <action> defines the action to perform if the condition applies. See
9662 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009663
Willy Tarreaue9656522010-08-17 15:40:09 +02009664 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009665
Davor Ocelice9ed2812017-12-25 17:49:28 +01009666 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009667 called "TCP content inspection". During this stage, ACL-based rules are
9668 evaluated every time the request contents are updated, until either an
9669 "accept" or a "reject" rule matches, or the TCP request inspection delay
9670 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009671
Willy Tarreaue9656522010-08-17 15:40:09 +02009672 The first difference between these rules and "tcp-request connection" rules
9673 is that "tcp-request content" rules can make use of contents to take a
9674 decision. Most often, these decisions will consider a protocol recognition or
9675 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009676 both frontends and backends. In case of HTTP keep-alive with the client, all
9677 tcp-request content rules are evaluated again, so haproxy keeps a record of
9678 what sticky counters were assigned by a "tcp-request connection" versus a
9679 "tcp-request content" rule, and flushes all the content-related ones after
9680 processing an HTTP request, so that they may be evaluated again by the rules
9681 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009682 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009683 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009684
Willy Tarreaue9656522010-08-17 15:40:09 +02009685 Content-based rules are evaluated in their exact declaration order. If no
9686 rule matches or if there is no rule, the default action is to accept the
9687 contents. There is no specific limit to the number of rules which may be
9688 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009689
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009690 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009691 - accept : the request is accepted
9692 - reject : the request is rejected and the connection is closed
9693 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009694 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009695 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009696 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009697 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009698 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009699 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009700 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009701 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009702 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009703
Willy Tarreaue9656522010-08-17 15:40:09 +02009704 They have the same meaning as their counter-parts in "tcp-request connection"
9705 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009706
Willy Tarreauf3338342014-01-28 21:40:28 +01009707 While there is nothing mandatory about it, it is recommended to use the
9708 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9709 content" rules in the frontend, and track-sc2 for "tcp-request content"
9710 rules in the backend, because that makes the configuration more readable
9711 and easier to troubleshoot, but this is just a guideline and all counters
9712 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009713
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009714 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009715 the action, it is simply performed unconditionally. That can be useful for
9716 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009717
Willy Tarreaue9656522010-08-17 15:40:09 +02009718 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009719 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9720 contents of a buffer before extracting the required data. If the buffered
9721 contents do not parse as a valid HTTP message, then the ACL does not match.
9722 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009723 processing, so there is no risk of parsing something differently. In an HTTP
9724 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9725 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009726
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009727 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009728 are present when the rule is processed. The rule processing engine is able to
9729 wait until the inspect delay expires when the data to be tracked is not yet
9730 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009731
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009732 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009733 declared inline. For "tcp-request session" rules, only session-level
9734 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009735
Daniel Schneller0b547052016-03-21 20:46:57 +01009736 <var-name> The name of the variable starts with an indication about
9737 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009738 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009739 "sess" : the variable is shared with the whole session
9740 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009741 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009742 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009743 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009744 "res" : the variable is shared only during response
9745 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009746 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009747 The name may only contain characters 'a-z', 'A-Z', '0-9',
9748 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009749
9750 <expr> Is a standard HAProxy expression formed by a sample-fetch
9751 followed by some converters.
9752
Christopher Faulet85d79c92016-11-09 16:54:56 +01009753 The "unset-var" is used to unset a variable. See above for details about
9754 <var-name>.
9755
Patrick Hemmer268a7072018-05-11 12:52:31 -04009756 The "set-priority-class" is used to set the queue priority class of the
9757 current request. The value must be a sample expression which converts to an
9758 integer in the range -2047..2047. Results outside this range will be
9759 truncated. The priority class determines the order in which queued requests
9760 are processed. Lower values have higher priority.
9761
9762 The "set-priority-offset" is used to set the queue priority timestamp offset
9763 of the current request. The value must be a sample expression which converts
9764 to an integer in the range -524287..524287. Results outside this range will be
9765 truncated. When a request is queued, it is ordered first by the priority
9766 class, then by the current timestamp adjusted by the given offset in
9767 milliseconds. Lower values have higher priority.
9768 Note that the resulting timestamp is is only tracked with enough precision for
9769 524,287ms (8m44s287ms). If the request is queued long enough to where the
9770 adjusted timestamp exceeds this value, it will be misidentified as highest
9771 priority. Thus it is important to set "timeout queue" to a value, where when
9772 combined with the offset, does not exceed this limit.
9773
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009774 The "send-spoe-group" is used to trigger sending of a group of SPOE
9775 messages. To do so, the SPOE engine used to send messages must be defined, as
9776 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9777 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9778 the SPOE agent name must be used.
9779
9780 <engine-name> The SPOE engine name.
9781
9782 <group-name> The SPOE group name as specified in the engine configuration.
9783
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009784 Example:
9785
9786 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009787 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009788
Willy Tarreau62644772008-07-16 18:36:06 +02009789 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009790 # Accept HTTP requests containing a Host header saying "example.com"
9791 # and reject everything else.
9792 acl is_host_com hdr(Host) -i example.com
9793 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009794 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009795 tcp-request content reject
9796
9797 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009798 # reject SMTP connection if client speaks first
9799 tcp-request inspect-delay 30s
9800 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009801 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009802
9803 # Forward HTTPS connection only if client speaks
9804 tcp-request inspect-delay 30s
9805 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009806 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009807 tcp-request content reject
9808
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009809 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009810 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009811 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009812 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009813 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9814 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009815
9816 Example:
9817 # track request counts per "base" (concatenation of Host+URL)
9818 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009819 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009820
Willy Tarreaue9656522010-08-17 15:40:09 +02009821 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009822 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009823
9824 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009825 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009826 # protecting all our sites
9827 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009828 tcp-request connection track-sc0 src
9829 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009830 ...
9831 use_backend http_dynamic if { path_end .php }
9832
9833 backend http_dynamic
9834 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009835 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009836 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009837 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009838 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009839 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009840 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009841
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009842 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009843
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009844 See also : "tcp-request connection", "tcp-request session",
9845 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009846
9847
9848tcp-request inspect-delay <timeout>
9849 Set the maximum allowed time to wait for data during content inspection
9850 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009851 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009852 Arguments :
9853 <timeout> is the timeout value specified in milliseconds by default, but
9854 can be in any other unit if the number is suffixed by the unit,
9855 as explained at the top of this document.
9856
9857 People using haproxy primarily as a TCP relay are often worried about the
9858 risk of passing any type of protocol to a server without any analysis. In
9859 order to be able to analyze the request contents, we must first withhold
9860 the data then analyze them. This statement simply enables withholding of
9861 data for at most the specified amount of time.
9862
Willy Tarreaufb356202010-08-03 14:02:05 +02009863 TCP content inspection applies very early when a connection reaches a
9864 frontend, then very early when the connection is forwarded to a backend. This
9865 means that a connection may experience a first delay in the frontend and a
9866 second delay in the backend if both have tcp-request rules.
9867
Willy Tarreau62644772008-07-16 18:36:06 +02009868 Note that when performing content inspection, haproxy will evaluate the whole
9869 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009870 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009871 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009872 contents are definitive. If no delay is set, haproxy will not wait at all
9873 and will immediately apply a verdict based on the available information.
9874 Obviously this is unlikely to be very useful and might even be racy, so such
9875 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009876
9877 As soon as a rule matches, the request is released and continues as usual. If
9878 the timeout is reached and no rule matches, the default policy will be to let
9879 it pass through unaffected.
9880
9881 For most protocols, it is enough to set it to a few seconds, as most clients
9882 send the full request immediately upon connection. Add 3 or more seconds to
9883 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009884 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009885 before the server (e.g. SMTP), or to wait for a client to talk before passing
9886 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009887 least the inspection delay, otherwise it will expire first. If the client
9888 closes the connection or if the buffer is full, the delay immediately expires
9889 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009890
Willy Tarreau55165fe2009-05-10 12:02:55 +02009891 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009892 "timeout client".
9893
9894
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009895tcp-response content <action> [{if | unless} <condition>]
9896 Perform an action on a session response depending on a layer 4-7 condition
9897 May be used in sections : defaults | frontend | listen | backend
9898 no | no | yes | yes
9899 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009900 <action> defines the action to perform if the condition applies. See
9901 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009902
9903 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9904
Davor Ocelice9ed2812017-12-25 17:49:28 +01009905 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009906 called "TCP content inspection". During this stage, ACL-based rules are
9907 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009908 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9909 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009910
9911 Most often, these decisions will consider a protocol recognition or validity.
9912
9913 Content-based rules are evaluated in their exact declaration order. If no
9914 rule matches or if there is no rule, the default action is to accept the
9915 contents. There is no specific limit to the number of rules which may be
9916 inserted.
9917
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009918 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009919 - accept :
9920 accepts the response if the condition is true (when used with "if")
9921 or false (when used with "unless"). The first such rule executed ends
9922 the rules evaluation.
9923
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009924 - close :
9925 immediately closes the connection with the server if the condition is
9926 true (when used with "if"), or false (when used with "unless"). The
9927 first such rule executed ends the rules evaluation. The main purpose of
9928 this action is to force a connection to be finished between a client
9929 and a server after an exchange when the application protocol expects
9930 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009931 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009932 protocols.
9933
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009934 - reject :
9935 rejects the response if the condition is true (when used with "if")
9936 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009937 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009938
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009939 - set-var(<var-name>) <expr>
9940 Sets a variable.
9941
Christopher Faulet85d79c92016-11-09 16:54:56 +01009942 - unset-var(<var-name>)
9943 Unsets a variable.
9944
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009945 - sc-inc-gpc0(<sc-id>):
9946 This action increments the GPC0 counter according to the sticky
9947 counter designated by <sc-id>. If an error occurs, this action fails
9948 silently and the actions evaluation continues.
9949
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009950 - sc-inc-gpc1(<sc-id>):
9951 This action increments the GPC1 counter according to the sticky
9952 counter designated by <sc-id>. If an error occurs, this action fails
9953 silently and the actions evaluation continues.
9954
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009955 - sc-set-gpt0(<sc-id>) <int> :
9956 This action sets the GPT0 tag according to the sticky counter designated
9957 by <sc-id> and the value of <int>. The expected result is a boolean. If
9958 an error occurs, this action silently fails and the actions evaluation
9959 continues.
9960
Willy Tarreau2d392c22015-08-24 01:43:45 +02009961 - "silent-drop" :
9962 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009963 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009964 to prevent the client from being notified. The effect it then that the
9965 client still sees an established connection while there's none on
9966 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9967 except that it doesn't use any local resource at all on the machine
9968 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009969 slow down stronger attackers. It is important to understand the impact
9970 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009971 client and HAProxy (firewalls, proxies, load balancers) will also keep
9972 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009973 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009974 TCP_REPAIR socket option is used to block the emission of a TCP
9975 reset. On other systems, the socket's TTL is reduced to 1 so that the
9976 TCP reset doesn't pass the first router, though it's still delivered to
9977 local networks. Do not use it unless you fully understand how it works.
9978
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009979 - send-spoe-group <engine-name> <group-name>
9980 Send a group of SPOE messages.
9981
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009982 Note that the "if/unless" condition is optional. If no condition is set on
9983 the action, it is simply performed unconditionally. That can be useful for
9984 for changing the default action to a reject.
9985
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009986 It is perfectly possible to match layer 7 contents with "tcp-response
9987 content" rules, but then it is important to ensure that a full response has
9988 been buffered, otherwise no contents will match. In order to achieve this,
9989 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009990 period.
9991
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009992 The "set-var" is used to set the content of a variable. The variable is
9993 declared inline.
9994
Daniel Schneller0b547052016-03-21 20:46:57 +01009995 <var-name> The name of the variable starts with an indication about
9996 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009997 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009998 "sess" : the variable is shared with the whole session
9999 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010000 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010001 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010002 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010003 "res" : the variable is shared only during response
10004 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010005 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010006 The name may only contain characters 'a-z', 'A-Z', '0-9',
10007 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010008
10009 <expr> Is a standard HAProxy expression formed by a sample-fetch
10010 followed by some converters.
10011
10012 Example:
10013
10014 tcp-request content set-var(sess.my_var) src
10015
Christopher Faulet85d79c92016-11-09 16:54:56 +010010016 The "unset-var" is used to unset a variable. See above for details about
10017 <var-name>.
10018
10019 Example:
10020
10021 tcp-request content unset-var(sess.my_var)
10022
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010023 The "send-spoe-group" is used to trigger sending of a group of SPOE
10024 messages. To do so, the SPOE engine used to send messages must be defined, as
10025 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10026 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10027 the SPOE agent name must be used.
10028
10029 <engine-name> The SPOE engine name.
10030
10031 <group-name> The SPOE group name as specified in the engine configuration.
10032
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010033 See section 7 about ACL usage.
10034
10035 See also : "tcp-request content", "tcp-response inspect-delay"
10036
10037
Willy Tarreau4f614292016-10-21 17:49:36 +020010038tcp-request session <action> [{if | unless} <condition>]
10039 Perform an action on a validated session depending on a layer 5 condition
10040 May be used in sections : defaults | frontend | listen | backend
10041 no | yes | yes | no
10042 Arguments :
10043 <action> defines the action to perform if the condition applies. See
10044 below.
10045
10046 <condition> is a standard layer5-only ACL-based condition (see section 7).
10047
Davor Ocelice9ed2812017-12-25 17:49:28 +010010048 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010049 it is possible to evaluate some conditions to decide whether this session
10050 must be accepted or dropped or have its counters tracked. Those conditions
10051 cannot make use of any data contents because no buffers are allocated yet and
10052 the processing cannot wait at this stage. The main use case it to copy some
10053 early information into variables (since variables are accessible in the
10054 session), or to keep track of some information collected after the handshake,
10055 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010056 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010057 extracted information can thus be copied to a variable or tracked using
10058 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10059 with other rulesets. Most operations performed here could also be performed
10060 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10061 for each new request, and that might not always be acceptable. For example a
10062 rule might increment a counter on each evaluation. It would also be possible
10063 that a country is resolved by geolocation from the source IP address,
10064 assigned to a session-wide variable, then the source address rewritten from
10065 an HTTP header for all requests. If some contents need to be inspected in
10066 order to take the decision, the "tcp-request content" statements must be used
10067 instead.
10068
10069 The "tcp-request session" rules are evaluated in their exact declaration
10070 order. If no rule matches or if there is no rule, the default action is to
10071 accept the incoming session. There is no specific limit to the number of
10072 rules which may be inserted.
10073
10074 Several types of actions are supported :
10075 - accept : the request is accepted
10076 - reject : the request is rejected and the connection is closed
10077 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10078 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010079 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010080 - sc-set-gpt0(<sc-id>) <int>
10081 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010082 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010083 - silent-drop
10084
10085 These actions have the same meaning as their respective counter-parts in
10086 "tcp-request connection" and "tcp-request content", so please refer to these
10087 sections for a complete description.
10088
10089 Note that the "if/unless" condition is optional. If no condition is set on
10090 the action, it is simply performed unconditionally. That can be useful for
10091 "track-sc*" actions as well as for changing the default action to a reject.
10092
10093 Example: track the original source address by default, or the one advertised
10094 in the PROXY protocol header for connection coming from the local
10095 proxies. The first connection-level rule enables receipt of the
10096 PROXY protocol for these ones, the second rule tracks whatever
10097 address we decide to keep after optional decoding.
10098
10099 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10100 tcp-request session track-sc0 src
10101
10102 Example: accept all sessions from white-listed hosts, reject too fast
10103 sessions without counting them, and track accepted sessions.
10104 This results in session rate being capped from abusive sources.
10105
10106 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10107 tcp-request session reject if { src_sess_rate gt 10 }
10108 tcp-request session track-sc0 src
10109
10110 Example: accept all sessions from white-listed hosts, count all other
10111 sessions and reject too fast ones. This results in abusive ones
10112 being blocked as long as they don't slow down.
10113
10114 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10115 tcp-request session track-sc0 src
10116 tcp-request session reject if { sc0_sess_rate gt 10 }
10117
10118 See section 7 about ACL usage.
10119
10120 See also : "tcp-request connection", "tcp-request content", "stick-table"
10121
10122
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010123tcp-response inspect-delay <timeout>
10124 Set the maximum allowed time to wait for a response during content inspection
10125 May be used in sections : defaults | frontend | listen | backend
10126 no | no | yes | yes
10127 Arguments :
10128 <timeout> is the timeout value specified in milliseconds by default, but
10129 can be in any other unit if the number is suffixed by the unit,
10130 as explained at the top of this document.
10131
10132 See also : "tcp-response content", "tcp-request inspect-delay".
10133
10134
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010135timeout check <timeout>
10136 Set additional check timeout, but only after a connection has been already
10137 established.
10138
10139 May be used in sections: defaults | frontend | listen | backend
10140 yes | no | yes | yes
10141 Arguments:
10142 <timeout> is the timeout value specified in milliseconds by default, but
10143 can be in any other unit if the number is suffixed by the unit,
10144 as explained at the top of this document.
10145
10146 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10147 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010148 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010149 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010150 (Please also note that there is no valid reason to have such long connect
10151 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10152 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010153
10154 If "timeout check" is not set haproxy uses "inter" for complete check
10155 timeout (connect + read) exactly like all <1.3.15 version.
10156
10157 In most cases check request is much simpler and faster to handle than normal
10158 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010159 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010160
10161 This parameter is specific to backends, but can be specified once for all in
10162 "defaults" sections. This is in fact one of the easiest solutions not to
10163 forget about it.
10164
Willy Tarreau41a340d2008-01-22 12:25:31 +010010165 See also: "timeout connect", "timeout queue", "timeout server",
10166 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010167
10168
Willy Tarreau0ba27502007-12-24 16:55:16 +010010169timeout client <timeout>
10170timeout clitimeout <timeout> (deprecated)
10171 Set the maximum inactivity time on the client side.
10172 May be used in sections : defaults | frontend | listen | backend
10173 yes | yes | yes | no
10174 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010175 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010176 can be in any other unit if the number is suffixed by the unit,
10177 as explained at the top of this document.
10178
10179 The inactivity timeout applies when the client is expected to acknowledge or
10180 send data. In HTTP mode, this timeout is particularly important to consider
10181 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010182 response while it is reading data sent by the server. That said, for the
10183 first phase, it is preferable to set the "timeout http-request" to better
10184 protect HAProxy from Slowloris like attacks. The value is specified in
10185 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010186 suffixed by the unit, as specified at the top of this document. In TCP mode
10187 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10188 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010189 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010190 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010191 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10192 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010193 which overrides "timeout client" and "timeout server" for tunnels, as well as
10194 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010195
10196 This parameter is specific to frontends, but can be specified once for all in
10197 "defaults" sections. This is in fact one of the easiest solutions not to
10198 forget about it. An unspecified timeout results in an infinite timeout, which
10199 is not recommended. Such a usage is accepted and works but reports a warning
10200 during startup because it may results in accumulation of expired sessions in
10201 the system if the system's timeouts are not configured either.
10202
Willy Tarreau95c4e142017-11-26 12:18:55 +010010203 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010204
Willy Tarreau0ba27502007-12-24 16:55:16 +010010205 This parameter replaces the old, deprecated "clitimeout". It is recommended
10206 to use it to write new configurations. The form "timeout clitimeout" is
10207 provided only by backwards compatibility but its use is strongly discouraged.
10208
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010209 See also : "clitimeout", "timeout server", "timeout tunnel",
10210 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010211
10212
Willy Tarreau05cdd962014-05-10 14:30:07 +020010213timeout client-fin <timeout>
10214 Set the inactivity timeout on the client side for half-closed connections.
10215 May be used in sections : defaults | frontend | listen | backend
10216 yes | yes | yes | no
10217 Arguments :
10218 <timeout> is the timeout value specified in milliseconds by default, but
10219 can be in any other unit if the number is suffixed by the unit,
10220 as explained at the top of this document.
10221
10222 The inactivity timeout applies when the client is expected to acknowledge or
10223 send data while one direction is already shut down. This timeout is different
10224 from "timeout client" in that it only applies to connections which are closed
10225 in one direction. This is particularly useful to avoid keeping connections in
10226 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10227 problem is particularly common long connections such as RDP or WebSocket.
10228 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010229 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10230 frame was sent, often indicating an expectation that the connection quickly
10231 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010232
10233 This parameter is specific to frontends, but can be specified once for all in
10234 "defaults" sections. By default it is not set, so half-closed connections
10235 will use the other timeouts (timeout.client or timeout.tunnel).
10236
10237 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10238
10239
Willy Tarreau0ba27502007-12-24 16:55:16 +010010240timeout connect <timeout>
10241timeout contimeout <timeout> (deprecated)
10242 Set the maximum time to wait for a connection attempt to a server to succeed.
10243 May be used in sections : defaults | frontend | listen | backend
10244 yes | no | yes | yes
10245 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010246 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010247 can be in any other unit if the number is suffixed by the unit,
10248 as explained at the top of this document.
10249
10250 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010251 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010252 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010253 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010254 connect timeout also presets both queue and tarpit timeouts to the same value
10255 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010256
10257 This parameter is specific to backends, but can be specified once for all in
10258 "defaults" sections. This is in fact one of the easiest solutions not to
10259 forget about it. An unspecified timeout results in an infinite timeout, which
10260 is not recommended. Such a usage is accepted and works but reports a warning
10261 during startup because it may results in accumulation of failed sessions in
10262 the system if the system's timeouts are not configured either.
10263
10264 This parameter replaces the old, deprecated "contimeout". It is recommended
10265 to use it to write new configurations. The form "timeout contimeout" is
10266 provided only by backwards compatibility but its use is strongly discouraged.
10267
Willy Tarreau41a340d2008-01-22 12:25:31 +010010268 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10269 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010270
10271
Willy Tarreaub16a5742010-01-10 14:46:16 +010010272timeout http-keep-alive <timeout>
10273 Set the maximum allowed time to wait for a new HTTP request to appear
10274 May be used in sections : defaults | frontend | listen | backend
10275 yes | yes | yes | yes
10276 Arguments :
10277 <timeout> is the timeout value specified in milliseconds by default, but
10278 can be in any other unit if the number is suffixed by the unit,
10279 as explained at the top of this document.
10280
10281 By default, the time to wait for a new request in case of keep-alive is set
10282 by "timeout http-request". However this is not always convenient because some
10283 people want very short keep-alive timeouts in order to release connections
10284 faster, and others prefer to have larger ones but still have short timeouts
10285 once the request has started to present itself.
10286
10287 The "http-keep-alive" timeout covers these needs. It will define how long to
10288 wait for a new HTTP request to start coming after a response was sent. Once
10289 the first byte of request has been seen, the "http-request" timeout is used
10290 to wait for the complete request to come. Note that empty lines prior to a
10291 new request do not refresh the timeout and are not counted as a new request.
10292
10293 There is also another difference between the two timeouts : when a connection
10294 expires during timeout http-keep-alive, no error is returned, the connection
10295 just closes. If the connection expires in "http-request" while waiting for a
10296 connection to complete, a HTTP 408 error is returned.
10297
10298 In general it is optimal to set this value to a few tens to hundreds of
10299 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010300 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010301 1 millisecond) it will probably only accept pipelined requests but not the
10302 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010303 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010304
10305 If this parameter is not set, the "http-request" timeout applies, and if both
10306 are not set, "timeout client" still applies at the lower level. It should be
10307 set in the frontend to take effect, unless the frontend is in TCP mode, in
10308 which case the HTTP backend's timeout will be used.
10309
Willy Tarreau95c4e142017-11-26 12:18:55 +010010310 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10311 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010312 (where we only have one connection per client and a connection setup).
10313
Willy Tarreaub16a5742010-01-10 14:46:16 +010010314 See also : "timeout http-request", "timeout client".
10315
10316
Willy Tarreau036fae02008-01-06 13:24:40 +010010317timeout http-request <timeout>
10318 Set the maximum allowed time to wait for a complete HTTP request
10319 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010320 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010321 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010322 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010323 can be in any other unit if the number is suffixed by the unit,
10324 as explained at the top of this document.
10325
10326 In order to offer DoS protection, it may be required to lower the maximum
10327 accepted time to receive a complete HTTP request without affecting the client
10328 timeout. This helps protecting against established connections on which
10329 nothing is sent. The client timeout cannot offer a good protection against
10330 this abuse because it is an inactivity timeout, which means that if the
10331 attacker sends one character every now and then, the timeout will not
10332 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010333 types, the request will be aborted if it does not complete in time. When the
10334 timeout expires, an HTTP 408 response is sent to the client to inform it
10335 about the problem, and the connection is closed. The logs will report
10336 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010337 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010338 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10339 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010340
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010341 By default, this timeout only applies to the header part of the request,
10342 and not to any data. As soon as the empty line is received, this timeout is
10343 not used anymore. When combined with "option http-buffer-request", this
10344 timeout also applies to the body of the request..
10345 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010346 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010347
10348 Generally it is enough to set it to a few seconds, as most clients send the
10349 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010350 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010351 generally work on local networks as long as there are no packet losses. This
10352 will prevent people from sending bare HTTP requests using telnet.
10353
10354 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010355 chunk of the incoming request. It should be set in the frontend to take
10356 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10357 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010358
Willy Tarreau0f228a02015-05-01 15:37:53 +020010359 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010360 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010361
Willy Tarreau844e3c52008-01-11 16:28:18 +010010362
10363timeout queue <timeout>
10364 Set the maximum time to wait in the queue for a connection slot to be free
10365 May be used in sections : defaults | frontend | listen | backend
10366 yes | no | yes | yes
10367 Arguments :
10368 <timeout> is the timeout value specified in milliseconds by default, but
10369 can be in any other unit if the number is suffixed by the unit,
10370 as explained at the top of this document.
10371
10372 When a server's maxconn is reached, connections are left pending in a queue
10373 which may be server-specific or global to the backend. In order not to wait
10374 indefinitely, a timeout is applied to requests pending in the queue. If the
10375 timeout is reached, it is considered that the request will almost never be
10376 served, so it is dropped and a 503 error is returned to the client.
10377
10378 The "timeout queue" statement allows to fix the maximum time for a request to
10379 be left pending in a queue. If unspecified, the same value as the backend's
10380 connection timeout ("timeout connect") is used, for backwards compatibility
10381 with older versions with no "timeout queue" parameter.
10382
10383 See also : "timeout connect", "contimeout".
10384
10385
10386timeout server <timeout>
10387timeout srvtimeout <timeout> (deprecated)
10388 Set the maximum inactivity time on the server side.
10389 May be used in sections : defaults | frontend | listen | backend
10390 yes | no | yes | yes
10391 Arguments :
10392 <timeout> is the timeout value specified in milliseconds by default, but
10393 can be in any other unit if the number is suffixed by the unit,
10394 as explained at the top of this document.
10395
10396 The inactivity timeout applies when the server is expected to acknowledge or
10397 send data. In HTTP mode, this timeout is particularly important to consider
10398 during the first phase of the server's response, when it has to send the
10399 headers, as it directly represents the server's processing time for the
10400 request. To find out what value to put there, it's often good to start with
10401 what would be considered as unacceptable response times, then check the logs
10402 to observe the response time distribution, and adjust the value accordingly.
10403
10404 The value is specified in milliseconds by default, but can be in any other
10405 unit if the number is suffixed by the unit, as specified at the top of this
10406 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10407 recommended that the client timeout remains equal to the server timeout in
10408 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010409 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010410 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010411 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10412 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010413 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10414 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010415
10416 This parameter is specific to backends, but can be specified once for all in
10417 "defaults" sections. This is in fact one of the easiest solutions not to
10418 forget about it. An unspecified timeout results in an infinite timeout, which
10419 is not recommended. Such a usage is accepted and works but reports a warning
10420 during startup because it may results in accumulation of expired sessions in
10421 the system if the system's timeouts are not configured either.
10422
10423 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10424 to use it to write new configurations. The form "timeout srvtimeout" is
10425 provided only by backwards compatibility but its use is strongly discouraged.
10426
Willy Tarreauce887fd2012-05-12 12:50:00 +020010427 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010428
Willy Tarreau05cdd962014-05-10 14:30:07 +020010429
10430timeout server-fin <timeout>
10431 Set the inactivity timeout on the server side for half-closed connections.
10432 May be used in sections : defaults | frontend | listen | backend
10433 yes | no | yes | yes
10434 Arguments :
10435 <timeout> is the timeout value specified in milliseconds by default, but
10436 can be in any other unit if the number is suffixed by the unit,
10437 as explained at the top of this document.
10438
10439 The inactivity timeout applies when the server is expected to acknowledge or
10440 send data while one direction is already shut down. This timeout is different
10441 from "timeout server" in that it only applies to connections which are closed
10442 in one direction. This is particularly useful to avoid keeping connections in
10443 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10444 This problem is particularly common long connections such as RDP or WebSocket.
10445 Note that this timeout can override "timeout tunnel" when a connection shuts
10446 down in one direction. This setting was provided for completeness, but in most
10447 situations, it should not be needed.
10448
10449 This parameter is specific to backends, but can be specified once for all in
10450 "defaults" sections. By default it is not set, so half-closed connections
10451 will use the other timeouts (timeout.server or timeout.tunnel).
10452
10453 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10454
Willy Tarreau844e3c52008-01-11 16:28:18 +010010455
10456timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010457 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010458 May be used in sections : defaults | frontend | listen | backend
10459 yes | yes | yes | yes
10460 Arguments :
10461 <timeout> is the tarpit duration specified in milliseconds by default, but
10462 can be in any other unit if the number is suffixed by the unit,
10463 as explained at the top of this document.
10464
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010465 When a connection is tarpitted using "http-request tarpit" or
10466 "reqtarpit", it is maintained open with no activity for a certain
10467 amount of time, then closed. "timeout tarpit" defines how long it will
10468 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010469
10470 The value is specified in milliseconds by default, but can be in any other
10471 unit if the number is suffixed by the unit, as specified at the top of this
10472 document. If unspecified, the same value as the backend's connection timeout
10473 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010474 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010475
10476 See also : "timeout connect", "contimeout".
10477
10478
Willy Tarreauce887fd2012-05-12 12:50:00 +020010479timeout tunnel <timeout>
10480 Set the maximum inactivity time on the client and server side for tunnels.
10481 May be used in sections : defaults | frontend | listen | backend
10482 yes | no | yes | yes
10483 Arguments :
10484 <timeout> is the timeout value specified in milliseconds by default, but
10485 can be in any other unit if the number is suffixed by the unit,
10486 as explained at the top of this document.
10487
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010488 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010489 between a client and a server, and the connection remains inactive in both
10490 directions. This timeout supersedes both the client and server timeouts once
10491 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010492 analyzer remains attached to either connection (e.g. tcp content rules are
10493 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010494 when switching to the WebSocket protocol, or forwarding a CONNECT request
10495 to a proxy), or after the first response when no keepalive/close option is
10496 specified.
10497
Willy Tarreau05cdd962014-05-10 14:30:07 +020010498 Since this timeout is usually used in conjunction with long-lived connections,
10499 it usually is a good idea to also set "timeout client-fin" to handle the
10500 situation where a client suddenly disappears from the net and does not
10501 acknowledge a close, or sends a shutdown and does not acknowledge pending
10502 data anymore. This can happen in lossy networks where firewalls are present,
10503 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10504 state.
10505
Willy Tarreauce887fd2012-05-12 12:50:00 +020010506 The value is specified in milliseconds by default, but can be in any other
10507 unit if the number is suffixed by the unit, as specified at the top of this
10508 document. Whatever the expected normal idle time, it is a good practice to
10509 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010510 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010511
10512 This parameter is specific to backends, but can be specified once for all in
10513 "defaults" sections. This is in fact one of the easiest solutions not to
10514 forget about it.
10515
10516 Example :
10517 defaults http
10518 option http-server-close
10519 timeout connect 5s
10520 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010521 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010522 timeout server 30s
10523 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10524
Willy Tarreau05cdd962014-05-10 14:30:07 +020010525 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010526
10527
Willy Tarreau844e3c52008-01-11 16:28:18 +010010528transparent (deprecated)
10529 Enable client-side transparent proxying
10530 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010531 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010532 Arguments : none
10533
10534 This keyword was introduced in order to provide layer 7 persistence to layer
10535 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10536 connection for a remote address to a local process (here HAProxy), and let
10537 this process know what address was initially requested. When this option is
10538 used, sessions without cookies will be forwarded to the original destination
10539 IP address of the incoming request (which should match that of another
10540 equipment), while requests with cookies will still be forwarded to the
10541 appropriate server.
10542
10543 The "transparent" keyword is deprecated, use "option transparent" instead.
10544
10545 Note that contrary to a common belief, this option does NOT make HAProxy
10546 present the client's IP to the server when establishing the connection.
10547
Willy Tarreau844e3c52008-01-11 16:28:18 +010010548 See also: "option transparent"
10549
William Lallemanda73203e2012-03-12 12:48:57 +010010550unique-id-format <string>
10551 Generate a unique ID for each request.
10552 May be used in sections : defaults | frontend | listen | backend
10553 yes | yes | yes | no
10554 Arguments :
10555 <string> is a log-format string.
10556
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010557 This keyword creates a ID for each request using the custom log format. A
10558 unique ID is useful to trace a request passing through many components of
10559 a complex infrastructure. The newly created ID may also be logged using the
10560 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010561
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010562 The format should be composed from elements that are guaranteed to be
10563 unique when combined together. For instance, if multiple haproxy instances
10564 are involved, it might be important to include the node name. It is often
10565 needed to log the incoming connection's source and destination addresses
10566 and ports. Note that since multiple requests may be performed over the same
10567 connection, including a request counter may help differentiate them.
10568 Similarly, a timestamp may protect against a rollover of the counter.
10569 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010570
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010571 It is recommended to use hexadecimal notation for many fields since it
10572 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010573
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010574 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010575
Julien Vehentf21be322014-03-07 08:27:34 -050010576 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010577
10578 will generate:
10579
10580 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10581
10582 See also: "unique-id-header"
10583
10584unique-id-header <name>
10585 Add a unique ID header in the HTTP request.
10586 May be used in sections : defaults | frontend | listen | backend
10587 yes | yes | yes | no
10588 Arguments :
10589 <name> is the name of the header.
10590
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010591 Add a unique-id header in the HTTP request sent to the server, using the
10592 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010593
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010594 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010595
Julien Vehentf21be322014-03-07 08:27:34 -050010596 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010597 unique-id-header X-Unique-ID
10598
10599 will generate:
10600
10601 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10602
10603 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010604
Willy Tarreauf51658d2014-04-23 01:21:56 +020010605use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010606 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010607 May be used in sections : defaults | frontend | listen | backend
10608 no | yes | yes | no
10609 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010610 <backend> is the name of a valid backend or "listen" section, or a
10611 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010612
Willy Tarreauf51658d2014-04-23 01:21:56 +020010613 <condition> is a condition composed of ACLs, as described in section 7. If
10614 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010615
10616 When doing content-switching, connections arrive on a frontend and are then
10617 dispatched to various backends depending on a number of conditions. The
10618 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010619 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010620 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010621 source address validation) or combined with a "tcp-request" rule to wait for
10622 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010623
10624 There may be as many "use_backend" rules as desired. All of these rules are
10625 evaluated in their declaration order, and the first one which matches will
10626 assign the backend.
10627
10628 In the first form, the backend will be used if the condition is met. In the
10629 second form, the backend will be used if the condition is not met. If no
10630 condition is valid, the backend defined with "default_backend" will be used.
10631 If no default backend is defined, either the servers in the same section are
10632 used (in case of a "listen" section) or, in case of a frontend, no server is
10633 used and a 503 service unavailable response is returned.
10634
Willy Tarreau51aecc72009-07-12 09:47:04 +020010635 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010636 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010637 and backend processing will immediately follow, or the backend will wait for
10638 a complete HTTP request to get in. This feature is useful when a frontend
10639 must decode several protocols on a unique port, one of them being HTTP.
10640
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010641 When <backend> is a simple name, it is resolved at configuration time, and an
10642 error is reported if the specified backend does not exist. If <backend> is
10643 a log-format string instead, no check may be done at configuration time, so
10644 the backend name is resolved dynamically at run time. If the resulting
10645 backend name does not correspond to any valid backend, no other rule is
10646 evaluated, and the default_backend directive is applied instead. Note that
10647 when using dynamic backend names, it is highly recommended to use a prefix
10648 that no other backend uses in order to ensure that an unauthorized backend
10649 cannot be forced from the request.
10650
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010651 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010652 used to detect the association between frontends and backends to compute the
10653 backend's "fullconn" setting. This cannot be done for dynamic names.
10654
10655 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10656 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010657
Willy Tarreau036fae02008-01-06 13:24:40 +010010658
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010659use-server <server> if <condition>
10660use-server <server> unless <condition>
10661 Only use a specific server if/unless an ACL-based condition is matched.
10662 May be used in sections : defaults | frontend | listen | backend
10663 no | no | yes | yes
10664 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010665 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010666
10667 <condition> is a condition composed of ACLs, as described in section 7.
10668
10669 By default, connections which arrive to a backend are load-balanced across
10670 the available servers according to the configured algorithm, unless a
10671 persistence mechanism such as a cookie is used and found in the request.
10672
10673 Sometimes it is desirable to forward a particular request to a specific
10674 server without having to declare a dedicated backend for this server. This
10675 can be achieved using the "use-server" rules. These rules are evaluated after
10676 the "redirect" rules and before evaluating cookies, and they have precedence
10677 on them. There may be as many "use-server" rules as desired. All of these
10678 rules are evaluated in their declaration order, and the first one which
10679 matches will assign the server.
10680
10681 If a rule designates a server which is down, and "option persist" is not used
10682 and no force-persist rule was validated, it is ignored and evaluation goes on
10683 with the next rules until one matches.
10684
10685 In the first form, the server will be used if the condition is met. In the
10686 second form, the server will be used if the condition is not met. If no
10687 condition is valid, the processing continues and the server will be assigned
10688 according to other persistence mechanisms.
10689
10690 Note that even if a rule is matched, cookie processing is still performed but
10691 does not assign the server. This allows prefixed cookies to have their prefix
10692 stripped.
10693
10694 The "use-server" statement works both in HTTP and TCP mode. This makes it
10695 suitable for use with content-based inspection. For instance, a server could
10696 be selected in a farm according to the TLS SNI field. And if these servers
10697 have their weight set to zero, they will not be used for other traffic.
10698
10699 Example :
10700 # intercept incoming TLS requests based on the SNI field
10701 use-server www if { req_ssl_sni -i www.example.com }
10702 server www 192.168.0.1:443 weight 0
10703 use-server mail if { req_ssl_sni -i mail.example.com }
10704 server mail 192.168.0.1:587 weight 0
10705 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010706 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010707 # all the rest is forwarded to this server
10708 server default 192.168.0.2:443 check
10709
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010710 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010711
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010712
Davor Ocelice9ed2812017-12-25 17:49:28 +0100107135. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010714--------------------------
10715
10716The "bind", "server" and "default-server" keywords support a number of settings
10717depending on some build options and on the system HAProxy was built on. These
10718settings generally each consist in one word sometimes followed by a value,
10719written on the same line as the "bind" or "server" line. All these options are
10720described in this section.
10721
10722
107235.1. Bind options
10724-----------------
10725
10726The "bind" keyword supports a certain number of settings which are all passed
10727as arguments on the same line. The order in which those arguments appear makes
10728no importance, provided that they appear after the bind address. All of these
10729parameters are optional. Some of them consist in a single words (booleans),
10730while other ones expect a value after them. In this case, the value must be
10731provided immediately after the setting name.
10732
10733The currently supported settings are the following ones.
10734
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010735accept-netscaler-cip <magic number>
10736 Enforces the use of the NetScaler Client IP insertion protocol over any
10737 connection accepted by any of the TCP sockets declared on the same line. The
10738 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10739 the incoming connection to be used everywhere an address is used, with the
10740 only exception of "tcp-request connection" rules which will only see the
10741 real connection address. Logs will reflect the addresses indicated in the
10742 protocol, unless it is violated, in which case the real address will still
10743 be used. This keyword combined with support from external components can be
10744 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010745 mechanism which is not always reliable and not even always usable. See also
10746 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10747 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010748
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010749accept-proxy
10750 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010751 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10752 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010753 3/4 addresses of the incoming connection to be used everywhere an address is
10754 used, with the only exception of "tcp-request connection" rules which will
10755 only see the real connection address. Logs will reflect the addresses
10756 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010757 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010758 components can be used as an efficient and reliable alternative to the
10759 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010760 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10761 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010762
Olivier Houchardc2aae742017-09-22 18:26:28 +020010763allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010764 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchardc2aae742017-09-22 18:26:28 +020010765 due to security considerations.
10766
Willy Tarreauab861d32013-04-02 02:30:41 +020010767alpn <protocols>
10768 This enables the TLS ALPN extension and advertises the specified protocol
10769 list as supported on top of ALPN. The protocol list consists in a comma-
10770 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10771 quotes). This requires that the SSL library is build with support for TLS
10772 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010773 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10774 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10775 now obsolete NPN extension. At the time of writing this, most browsers still
10776 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10777 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10778 are expected to be supported, both versions can be advertised, in order of
10779 preference, like below :
10780
10781 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010782
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010783backlog <backlog>
10784 Sets the socket's backlog to this value. If unspecified, the frontend's
10785 backlog is used instead, which generally defaults to the maxconn value.
10786
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010787curves <curves>
10788 This setting is only available when support for OpenSSL was built in. It sets
10789 the string describing the list of elliptic curves algorithms ("curve suite")
10790 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10791 string is a colon-delimited list of curve name.
10792 Example: "X25519:P-256" (without quote)
10793 When "curves" is set, "ecdhe" parameter is ignored.
10794
Emeric Brun7fb34422012-09-28 15:26:15 +020010795ecdhe <named curve>
10796 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010797 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10798 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010799
Emeric Brunfd33a262012-10-11 16:28:27 +020010800ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010801 This setting is only available when support for OpenSSL was built in. It
10802 designates a PEM file from which to load CA certificates used to verify
10803 client's certificate.
10804
Emeric Brunb6dc9342012-09-28 17:55:37 +020010805ca-ignore-err [all|<errorID>,...]
10806 This setting is only available when support for OpenSSL was built in.
10807 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10808 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10809 error is ignored.
10810
Christopher Faulet31af49d2015-06-09 17:29:50 +020010811ca-sign-file <cafile>
10812 This setting is only available when support for OpenSSL was built in. It
10813 designates a PEM file containing both the CA certificate and the CA private
10814 key used to create and sign server's certificates. This is a mandatory
10815 setting when the dynamic generation of certificates is enabled. See
10816 'generate-certificates' for details.
10817
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010818ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010819 This setting is only available when support for OpenSSL was built in. It is
10820 the CA private key passphrase. This setting is optional and used only when
10821 the dynamic generation of certificates is enabled. See
10822 'generate-certificates' for details.
10823
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010824ciphers <ciphers>
10825 This setting is only available when support for OpenSSL was built in. It sets
10826 the string describing the list of cipher algorithms ("cipher suite") that are
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010827 negotiated during the SSL/TLS handshake except for TLSv1.3. The format of the
10828 string is defined in "man 1 ciphers" from OpenSSL man pages, and can be for
10829 instance a string such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without
10830 quotes). Depending on the compatibility and security requirements, the list
10831 of suitable ciphers depends on a variety of variables. For background
10832 information and recommendations see e.g.
10833 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10834 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10835 cipher configuration, please check the "ciphersuites" keyword.
10836
10837ciphersuites <ciphersuites>
10838 This setting is only available when support for OpenSSL was built in and
10839 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10840 the list of cipher algorithms ("cipher suite") that are negotiated during the
10841 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
10842 OpenSSL man pages under the "ciphersuites" section, and can be for instance a
10843 string such as
10844 "TLS_AES_256_GCM_SHA384:TLS_CHACHA20_POLY1305_SHA256:TLS_AES_128_GCM_SHA256"
10845 (without quotes). For cipher configuration for TLSv1.2 and earlier, please check
10846 the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010847
Emeric Brunfd33a262012-10-11 16:28:27 +020010848crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010849 This setting is only available when support for OpenSSL was built in. It
10850 designates a PEM file from which to load certificate revocation list used
10851 to verify client's certificate.
10852
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010853crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010854 This setting is only available when support for OpenSSL was built in. It
10855 designates a PEM file containing both the required certificates and any
10856 associated private keys. This file can be built by concatenating multiple
10857 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10858 requires an intermediate certificate, this can also be concatenated into this
10859 file.
10860
10861 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10862 are loaded.
10863
10864 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010865 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010866 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10867 specified multiple times in order to load certificates from multiple files or
10868 directories. The certificates will be presented to clients who provide a
10869 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010870 subjects. Wildcards are supported, where a wildcard character '*' is used
10871 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010872 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010873
10874 If no SNI is provided by the client or if the SSL library does not support
10875 TLS extensions, or if the client provides an SNI hostname which does not
10876 match any certificate, then the first loaded certificate will be presented.
10877 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010878 recommended to load the default one first as a file or to ensure that it will
10879 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010880
Emeric Brune032bfa2012-09-28 13:01:45 +020010881 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010882
Davor Ocelice9ed2812017-12-25 17:49:28 +010010883 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010884 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010885 choose a web server that the CA believes requires an intermediate CA (for
10886 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010887 others, e.g. nginx, result in a wrong bundle that will not work for some
10888 clients).
10889
Emeric Brun4147b2e2014-06-16 18:36:30 +020010890 For each PEM file, haproxy checks for the presence of file at the same path
10891 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10892 Status Request extension (also known as "OCSP stapling") is automatically
10893 enabled. The content of this file is optional. If not empty, it must contain
10894 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10895 must comply with the following rules: it has to indicate a good status,
10896 it has to be a single response for the certificate of the PEM file, and it
10897 has to be valid at the moment of addition. If these rules are not respected
10898 the OCSP Response is ignored and a warning is emitted. In order to identify
10899 which certificate an OCSP Response applies to, the issuer's certificate is
10900 necessary. If the issuer's certificate is not found in the PEM file, it will
10901 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10902 if it exists otherwise it will fail with an error.
10903
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010904 For each PEM file, haproxy also checks for the presence of file at the same
10905 path suffixed by ".sctl". If such file is found, support for Certificate
10906 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10907 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10908 to check basic syntax, but no signatures are verified.
10909
yanbzhu6c25e9e2016-01-05 12:52:02 -050010910 There are cases where it is desirable to support multiple key types, e.g. RSA
10911 and ECDSA in the cipher suites offered to the clients. This allows clients
10912 that support EC certificates to be able to use EC ciphers, while
10913 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010914
10915 In order to provide this functionality, multiple PEM files, each with a
10916 different key type, are required. To associate these PEM files into a
10917 "cert bundle" that is recognized by haproxy, they must be named in the
10918 following way: All PEM files that are to be bundled must have the same base
10919 name, with a suffix indicating the key type. Currently, three suffixes are
10920 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10921 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10922 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10923 suffix matters. To load this bundle into haproxy, specify the base name only:
10924
10925 Example : bind :8443 ssl crt example.pem
10926
yanbzhu6c25e9e2016-01-05 12:52:02 -050010927 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010928 a cert bundle.
10929
Davor Ocelice9ed2812017-12-25 17:49:28 +010010930 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010931 support multiple key types. PEM files are combined based on Common Name
10932 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10933 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10934 entries in the certificates in that bundle, haproxy will not be able to
10935 provide multi-cert support.
10936
10937 Assuming bundle in the example above contained the following:
10938
10939 Filename | CN | SAN
10940 -------------------+-----------------+-------------------
10941 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010942 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010943 example.pem.ecdsa | www.example.com | ecdsa.example.com
10944 -------------------+-----------------+-------------------
10945
10946 Users connecting with an SNI of "www.example.com" will be able
10947 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10948 "rsa.example.com" will only be able to use RSA cipher suites, and users
10949 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010950 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10951 no need to bundle certificates. ECDSA certificate will be preferred if client
10952 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010953
10954 If a directory name is given as the <cert> argument, haproxy will
10955 automatically search and load bundled files in that directory.
10956
10957 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10958 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10959 time, sctl is not supported in multi-certificate bundling.
10960
Emeric Brunb6dc9342012-09-28 17:55:37 +020010961crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010962 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010963 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010964 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010965 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010966
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010967crt-list <file>
10968 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010969 designates a list of PEM file with an optional ssl configuration and a SNI
10970 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010971
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010972 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10973
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010974 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10975 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010976 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010977 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010978
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010979 Wildcards are supported in the SNI filter. Negative filter are also supported,
10980 only useful in combination with a wildcard filter to exclude a particular SNI.
10981 The certificates will be presented to clients who provide a valid TLS Server
10982 Name Indication field matching one of the SNI filters. If no SNI filter is
10983 specified, the CN and alt subjects are used. This directive may be specified
10984 multiple times. See the "crt" option for more information. The default
10985 certificate is still needed to meet OpenSSL expectations. If it is not used,
10986 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010987
yanbzhu6c25e9e2016-01-05 12:52:02 -050010988 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010989 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010990 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10991 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10992 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010993
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010994 crt-list file example:
10995 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010996 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010997 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010998 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010999
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011000defer-accept
11001 Is an optional keyword which is supported only on certain Linux kernels. It
11002 states that a connection will only be accepted once some data arrive on it,
11003 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011004 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011005 performance by ensuring that most of the request is already available when
11006 the connection is accepted. On the other hand, it will not be able to detect
11007 connections which don't talk. It is important to note that this option is
11008 broken in all kernels up to 2.6.31, as the connection is never accepted until
11009 the client talks. This can cause issues with front firewalls which would see
11010 an established connection while the proxy will only see it in SYN_RECV. This
11011 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11012
William Lallemandf6975e92017-05-26 17:42:10 +020011013expose-fd listeners
11014 This option is only usable with the stats socket. It gives your stats socket
11015 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011016 During a reload with the master-worker mode, the process is automatically
11017 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011018 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011019
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011020force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011021 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011022 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011023 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011024 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011025
11026force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011027 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011028 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011029 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011030
11031force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011032 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011033 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011034 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011035
11036force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011037 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011038 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011039 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011040
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011041force-tlsv13
11042 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11043 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011044 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011045
Christopher Faulet31af49d2015-06-09 17:29:50 +020011046generate-certificates
11047 This setting is only available when support for OpenSSL was built in. It
11048 enables the dynamic SSL certificates generation. A CA certificate and its
11049 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11050 a transparent forward proxy, SSL requests generate errors because of a common
11051 name mismatch on the certificate presented to the client. With this option
11052 enabled, HAProxy will try to forge a certificate using the SNI hostname
11053 indicated by the client. This is done only if no certificate matches the SNI
11054 hostname (see 'crt-list'). If an error occurs, the default certificate is
11055 used, else the 'strict-sni' option is set.
11056 It can also be used when HAProxy is configured as a reverse proxy to ease the
11057 deployment of an architecture with many backends.
11058
11059 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11060 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011061 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011062 certificate is used many times.
11063
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011064gid <gid>
11065 Sets the group of the UNIX sockets to the designated system gid. It can also
11066 be set by default in the global section's "unix-bind" statement. Note that
11067 some platforms simply ignore this. This setting is equivalent to the "group"
11068 setting except that the group ID is used instead of its name. This setting is
11069 ignored by non UNIX sockets.
11070
11071group <group>
11072 Sets the group of the UNIX sockets to the designated system group. It can
11073 also be set by default in the global section's "unix-bind" statement. Note
11074 that some platforms simply ignore this. This setting is equivalent to the
11075 "gid" setting except that the group name is used instead of its gid. This
11076 setting is ignored by non UNIX sockets.
11077
11078id <id>
11079 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11080 sometimes it is more convenient to fix them to ease monitoring. This value
11081 must be strictly positive and unique within the listener/frontend. This
11082 option can only be used when defining only a single socket.
11083
11084interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011085 Restricts the socket to a specific interface. When specified, only packets
11086 received from that particular interface are processed by the socket. This is
11087 currently only supported on Linux. The interface must be a primary system
11088 interface, not an aliased interface. It is also possible to bind multiple
11089 frontends to the same address if they are bound to different interfaces. Note
11090 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011091 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11092 uses the same interface as inbound traffic, and its associated routing table,
11093 even if there are explicit routes through different interfaces configured.
11094 This can prove useful to address asymmetric routing issues when the same
11095 client IP addresses need to be able to reach frontends hosted on different
11096 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011097
Willy Tarreauabb175f2012-09-24 12:43:26 +020011098level <level>
11099 This setting is used with the stats sockets only to restrict the nature of
11100 the commands that can be issued on the socket. It is ignored by other
11101 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011102 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011103 read, and no change is allowed. It would make sense on systems where it
11104 is not easy to restrict access to the socket.
11105 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011106 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011107 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011108 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011109 all counters).
11110
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011111severity-output <format>
11112 This setting is used with the stats sockets only to configure severity
11113 level output prepended to informational feedback messages. Severity
11114 level of messages can range between 0 and 7, conforming to syslog
11115 rfc5424. Valid and successful socket commands requesting data
11116 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11117 prepended. It is ignored by other sockets. <format> can be one of :
11118 - "none" (default) no severity level is prepended to feedback messages.
11119 - "number" severity level is prepended as a number.
11120 - "string" severity level is prepended as a string following the
11121 rfc5424 convention.
11122
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011123maxconn <maxconn>
11124 Limits the sockets to this number of concurrent connections. Extraneous
11125 connections will remain in the system's backlog until a connection is
11126 released. If unspecified, the limit will be the same as the frontend's
11127 maxconn. Note that in case of port ranges or multiple addresses, the same
11128 value will be applied to each socket. This setting enables different
11129 limitations on expensive sockets, for instance SSL entries which may easily
11130 eat all memory.
11131
11132mode <mode>
11133 Sets the octal mode used to define access permissions on the UNIX socket. It
11134 can also be set by default in the global section's "unix-bind" statement.
11135 Note that some platforms simply ignore this. This setting is ignored by non
11136 UNIX sockets.
11137
11138mss <maxseg>
11139 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11140 connections. This can be used to force a lower MSS for certain specific
11141 ports, for instance for connections passing through a VPN. Note that this
11142 relies on a kernel feature which is theoretically supported under Linux but
11143 was buggy in all versions prior to 2.6.28. It may or may not work on other
11144 operating systems. It may also not change the advertised value but change the
11145 effective size of outgoing segments. The commonly advertised value for TCPv4
11146 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11147 positive, it will be used as the advertised MSS. If it is negative, it will
11148 indicate by how much to reduce the incoming connection's advertised MSS for
11149 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11150
11151name <name>
11152 Sets an optional name for these sockets, which will be reported on the stats
11153 page.
11154
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011155namespace <name>
11156 On Linux, it is possible to specify which network namespace a socket will
11157 belong to. This directive makes it possible to explicitly bind a listener to
11158 a namespace different from the default one. Please refer to your operating
11159 system's documentation to find more details about network namespaces.
11160
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011161nice <nice>
11162 Sets the 'niceness' of connections initiated from the socket. Value must be
11163 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11164 means that such connections are more friendly to others and easily offer
11165 their place in the scheduler. On the opposite, negative values mean that
11166 connections want to run with a higher priority than others. The difference
11167 only happens under high loads when the system is close to saturation.
11168 Negative values are appropriate for low-latency or administration services,
11169 and high values are generally recommended for CPU intensive tasks such as SSL
11170 processing or bulk transfers which are less sensible to latency. For example,
11171 it may make sense to use a positive value for an SMTP socket and a negative
11172 one for an RDP socket.
11173
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011174no-ca-names
11175 This setting is only available when support for OpenSSL was built in. It
11176 prevents from send CA names in server hello message when ca-file is used.
11177
Emeric Brun9b3009b2012-10-05 11:55:06 +020011178no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011179 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011180 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011181 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011182 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011183 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11184 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011185
Emeric Brun90ad8722012-10-02 14:00:59 +020011186no-tls-tickets
11187 This setting is only available when support for OpenSSL was built in. It
11188 disables the stateless session resumption (RFC 5077 TLS Ticket
11189 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011190 session resumption is more expensive in CPU usage. This option is also
11191 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011192
Emeric Brun9b3009b2012-10-05 11:55:06 +020011193no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011194 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011195 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011196 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011197 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011198 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11199 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011200
Emeric Brun9b3009b2012-10-05 11:55:06 +020011201no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011202 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011203 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011204 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011205 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011206 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11207 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011208
Emeric Brun9b3009b2012-10-05 11:55:06 +020011209no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011210 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011211 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011212 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011213 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011214 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11215 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011216
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011217no-tlsv13
11218 This setting is only available when support for OpenSSL was built in. It
11219 disables support for TLSv1.3 on any sockets instantiated from the listener
11220 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11221 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011222 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11223 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011224
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011225npn <protocols>
11226 This enables the NPN TLS extension and advertises the specified protocol list
11227 as supported on top of NPN. The protocol list consists in a comma-delimited
11228 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11229 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011230 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011231 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11232 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11233 version of OpenSSL, NPN might still be used as most clients still support it
11234 at the time of writing this. It is possible to enable both NPN and ALPN
11235 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011236
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011237prefer-client-ciphers
11238 Use the client's preference when selecting the cipher suite, by default
11239 the server's preference is enforced. This option is also available on
11240 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011241 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11242 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11243 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011244
Christopher Fauletc644fa92017-11-23 22:44:11 +010011245process <process-set>[/<thread-set>]
11246 This restricts the list of processes and/or threads on which this listener is
11247 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011248 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011249 between the two is applied. If in the end the listener is not allowed to run
11250 on any remaining process, a warning is emitted, and the listener will either
11251 run on the first process of the listener if a single process was specified,
11252 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011253 set is specified, it limits the threads allowed to process incoming
Christopher Fauletc644fa92017-11-23 22:44:11 +010011254 connections for this listener, for the corresponding process set. For the
11255 unlikely case where several ranges are needed, this directive may be
11256 repeated. <process-set> and <thread-set> must use the format
11257
11258 all | odd | even | number[-[number]]
11259
11260 Ranges can be partially defined. The higher bound can be omitted. In such
11261 case, it is replaced by the corresponding maximum value. The main purpose of
11262 this directive is to be used with the stats sockets and have one different
11263 socket per process. The second purpose is to have multiple bind lines sharing
11264 the same IP:port but not the same process in a listener, so that the system
11265 can distribute the incoming connections into multiple queues and allow a
11266 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11267 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011268
Christopher Fauleta717b992018-04-10 14:43:00 +020011269proto <name>
11270 Forces the multiplexer's protocol to use for the incoming connections. It
11271 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11272 be usable on the frontend side. The list of available protocols is reported
11273 in haproxy -vv.
11274 Idea behind this optipon is to bypass the selection of the best multiplexer's
11275 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011276 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011277 h2" on the bind line.
11278
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011279ssl
11280 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011281 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011282 certificate is necessary (see "crt" above). All contents in the buffers will
11283 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011284 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11285 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011286
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011287ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11288 This option enforces use of <version> or lower on SSL connections instantiated
11289 from this listener. This option is also available on global statement
11290 "ssl-default-bind-options". See also "ssl-min-ver".
11291
11292ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11293 This option enforces use of <version> or upper on SSL connections instantiated
11294 from this listener. This option is also available on global statement
11295 "ssl-default-bind-options". See also "ssl-max-ver".
11296
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011297strict-sni
11298 This setting is only available when support for OpenSSL was built in. The
11299 SSL/TLS negotiation is allow only if the client provided an SNI which match
11300 a certificate. The default certificate is not used.
11301 See the "crt" option for more information.
11302
Willy Tarreau2af207a2015-02-04 00:45:58 +010011303tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011304 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011305 listening socket. This option is available on Linux since version 2.6.37. It
11306 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011307 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011308 useful on long-lived connections experiencing long idle periods such as
11309 remote terminals or database connection pools, where the client and server
11310 timeouts must remain high to allow a long period of idle, but where it is
11311 important to detect that the client has disappeared in order to release all
11312 resources associated with its connection (and the server's session). The
11313 argument is a delay expressed in milliseconds by default. This only works
11314 for regular TCP connections, and is ignored for other protocols.
11315
Willy Tarreau1c862c52012-10-05 16:21:00 +020011316tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011317 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011318 enables TCP Fast Open on the listening socket, which means that clients which
11319 support this feature will be able to send a request and receive a response
11320 during the 3-way handshake starting from second connection, thus saving one
11321 round-trip after the first connection. This only makes sense with protocols
11322 that use high connection rates and where each round trip matters. This can
11323 possibly cause issues with many firewalls which do not accept data on SYN
11324 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011325 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11326 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11327 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011328
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011329tls-ticket-keys <keyfile>
11330 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
11331 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
11332 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
11333 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
11334 used for decryption and the penultimate one for encryption. This enables easy
11335 key rotation by just appending new key to the file and reloading the process.
11336 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
11337 is compromised. It is also a good idea to keep the keys off any permanent
11338 storage such as hard drives (hint: use tmpfs and don't swap those files).
11339 Lifetime hint can be changed using tune.ssl.timeout.
11340
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011341transparent
11342 Is an optional keyword which is supported only on certain Linux kernels. It
11343 indicates that the addresses will be bound even if they do not belong to the
11344 local machine, and that packets targeting any of these addresses will be
11345 intercepted just as if the addresses were locally configured. This normally
11346 requires that IP forwarding is enabled. Caution! do not use this with the
11347 default address '*', as it would redirect any traffic for the specified port.
11348 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11349 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11350 kernel version. Some distribution kernels include backports of the feature,
11351 so check for support with your vendor.
11352
Willy Tarreau77e3af92012-11-24 15:07:23 +010011353v4v6
11354 Is an optional keyword which is supported only on most recent systems
11355 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11356 and IPv6 when it uses the default address. Doing so is sometimes necessary
11357 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011358 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011359
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011360v6only
11361 Is an optional keyword which is supported only on most recent systems
11362 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11363 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011364 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11365 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011366
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011367uid <uid>
11368 Sets the owner of the UNIX sockets to the designated system uid. It can also
11369 be set by default in the global section's "unix-bind" statement. Note that
11370 some platforms simply ignore this. This setting is equivalent to the "user"
11371 setting except that the user numeric ID is used instead of its name. This
11372 setting is ignored by non UNIX sockets.
11373
11374user <user>
11375 Sets the owner of the UNIX sockets to the designated system user. It can also
11376 be set by default in the global section's "unix-bind" statement. Note that
11377 some platforms simply ignore this. This setting is equivalent to the "uid"
11378 setting except that the user name is used instead of its uid. This setting is
11379 ignored by non UNIX sockets.
11380
Emeric Brun1a073b42012-09-28 17:07:34 +020011381verify [none|optional|required]
11382 This setting is only available when support for OpenSSL was built in. If set
11383 to 'none', client certificate is not requested. This is the default. In other
11384 cases, a client certificate is requested. If the client does not provide a
11385 certificate after the request and if 'verify' is set to 'required', then the
11386 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011387 certificate provided by the client is always verified using CAs from
11388 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11389 is aborted, regardless of the 'verify' option, unless the error code exactly
11390 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011391
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200113925.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011393------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011394
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011395The "server" and "default-server" keywords support a certain number of settings
11396which are all passed as arguments on the server line. The order in which those
11397arguments appear does not count, and they are all optional. Some of those
11398settings are single words (booleans) while others expect one or several values
11399after them. In this case, the values must immediately follow the setting name.
11400Except default-server, all those settings must be specified after the server's
11401address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011402
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011403 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011404 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011405
Frédéric Lécailled2376272017-03-21 18:52:12 +010011406Note that all these settings are supported both by "server" and "default-server"
11407keywords, except "id" which is only supported by "server".
11408
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011409The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011410
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011411addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011412 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011413 to send health-checks or to probe the agent-check. On some servers, it may be
11414 desirable to dedicate an IP address to specific component able to perform
11415 complex tests which are more suitable to health-checks than the application.
11416 This parameter is ignored if the "check" parameter is not set. See also the
11417 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011418
Simon Hormand60d6912013-11-25 10:46:36 +090011419agent-check
11420 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011421 health check. An agent health check is performed by making a TCP connection
11422 to the port set by the "agent-port" parameter and reading an ASCII string.
11423 The string is made of a series of words delimited by spaces, tabs or commas
11424 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011425
Willy Tarreau81f5d942013-12-09 20:51:51 +010011426 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011427 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011428 weight of a server as configured when haproxy starts. Note that a zero
11429 weight is reported on the stats page as "DRAIN" since it has the same
11430 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011431
Davor Ocelice9ed2812017-12-25 17:49:28 +010011432 - The string "maxconn:" followed by an integer (no space between). Values
11433 in this format will set the maxconn of a server. The maximum number of
11434 connections advertised needs to be multiplied by the number of load
11435 balancers and different backends that use this health check to get the
11436 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011437
Willy Tarreau81f5d942013-12-09 20:51:51 +010011438 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011439 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011440
Willy Tarreau81f5d942013-12-09 20:51:51 +010011441 - The word "drain". This will turn the server's administrative state to the
11442 DRAIN mode, thus it will not accept any new connections other than those
11443 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011444
Willy Tarreau81f5d942013-12-09 20:51:51 +010011445 - The word "maint". This will turn the server's administrative state to the
11446 MAINT mode, thus it will not accept any new connections at all, and health
11447 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011448
Willy Tarreau81f5d942013-12-09 20:51:51 +010011449 - The words "down", "failed", or "stopped", optionally followed by a
11450 description string after a sharp ('#'). All of these mark the server's
11451 operating state as DOWN, but since the word itself is reported on the stats
11452 page, the difference allows an administrator to know if the situation was
11453 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011454 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011455 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011456
Willy Tarreau81f5d942013-12-09 20:51:51 +010011457 - The word "up" sets back the server's operating state as UP if health checks
11458 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011459
Willy Tarreau81f5d942013-12-09 20:51:51 +010011460 Parameters which are not advertised by the agent are not changed. For
11461 example, an agent might be designed to monitor CPU usage and only report a
11462 relative weight and never interact with the operating status. Similarly, an
11463 agent could be designed as an end-user interface with 3 radio buttons
11464 allowing an administrator to change only the administrative state. However,
11465 it is important to consider that only the agent may revert its own actions,
11466 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11467 agent must implement the other equivalent actions to bring the service into
11468 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011469
Simon Horman2f1f9552013-11-25 10:46:37 +090011470 Failure to connect to the agent is not considered an error as connectivity
11471 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011472 parameter. Warning though, it is not a good idea to stop an agent after it
11473 reports "down", since only an agent reporting "up" will be able to turn the
11474 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011475 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011476
Willy Tarreau81f5d942013-12-09 20:51:51 +010011477 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011478 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011479
James Brown55f9ff12015-10-21 18:19:05 -070011480agent-send <string>
11481 If this option is specified, haproxy will send the given string (verbatim)
11482 to the agent server upon connection. You could, for example, encode
11483 the backend name into this string, which would enable your agent to send
11484 different responses based on the backend. Make sure to include a '\n' if
11485 you want to terminate your request with a newline.
11486
Simon Hormand60d6912013-11-25 10:46:36 +090011487agent-inter <delay>
11488 The "agent-inter" parameter sets the interval between two agent checks
11489 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11490
11491 Just as with every other time-based parameter, it may be entered in any
11492 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11493 parameter also serves as a timeout for agent checks "timeout check" is
11494 not set. In order to reduce "resonance" effects when multiple servers are
11495 hosted on the same hardware, the agent and health checks of all servers
11496 are started with a small time offset between them. It is also possible to
11497 add some random noise in the agent and health checks interval using the
11498 global "spread-checks" keyword. This makes sense for instance when a lot
11499 of backends use the same servers.
11500
11501 See also the "agent-check" and "agent-port" parameters.
11502
Misiek768d8602017-01-09 09:52:43 +010011503agent-addr <addr>
11504 The "agent-addr" parameter sets address for agent check.
11505
11506 You can offload agent-check to another target, so you can make single place
11507 managing status and weights of servers defined in haproxy in case you can't
11508 make self-aware and self-managing services. You can specify both IP or
11509 hostname, it will be resolved.
11510
Simon Hormand60d6912013-11-25 10:46:36 +090011511agent-port <port>
11512 The "agent-port" parameter sets the TCP port used for agent checks.
11513
11514 See also the "agent-check" and "agent-inter" parameters.
11515
Olivier Houchardc7566002018-11-20 23:33:50 +010011516alpn <protocols>
11517 This enables the TLS ALPN extension and advertises the specified protocol
11518 list as supported on top of ALPN. The protocol list consists in a comma-
11519 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
11520 quotes). This requires that the SSL library is build with support for TLS
11521 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11522 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11523 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11524 now obsolete NPN extension.
11525 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11526 be advertised, in order of preference, like below :
11527
11528 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11529
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011530backup
11531 When "backup" is present on a server line, the server is only used in load
11532 balancing when all other non-backup servers are unavailable. Requests coming
11533 with a persistence cookie referencing the server will always be served
11534 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011535 the "allbackups" option is set in the backend. See also the "no-backup" and
11536 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011537
Emeric Brunef42d922012-10-11 16:11:36 +020011538ca-file <cafile>
11539 This setting is only available when support for OpenSSL was built in. It
11540 designates a PEM file from which to load CA certificates used to verify
11541 server's certificate.
11542
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011543check
11544 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011545 always considered available. If "check" is set, the server is available when
11546 accepting periodic TCP connections, to ensure that it is really able to serve
11547 requests. The default address and port to send the tests to are those of the
11548 server, and the default source is the same as the one defined in the
11549 backend. It is possible to change the address using the "addr" parameter, the
11550 port using the "port" parameter, the source address using the "source"
11551 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011552 parameters. The request method is define in the backend using the "httpchk",
11553 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011554 refer to those options and parameters for more information. See also
11555 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011556
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011557check-send-proxy
11558 This option forces emission of a PROXY protocol line with outgoing health
11559 checks, regardless of whether the server uses send-proxy or not for the
11560 normal traffic. By default, the PROXY protocol is enabled for health checks
11561 if it is already enabled for normal traffic and if no "port" nor "addr"
11562 directive is present. However, if such a directive is present, the
11563 "check-send-proxy" option needs to be used to force the use of the
11564 protocol. See also the "send-proxy" option for more information.
11565
Jérôme Magninae9bb762018-12-09 16:08:26 +010011566check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011567 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011568 over SSL. It is only possible to use a string to set <sni>. If you want to
11569 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011570
Willy Tarreau763a95b2012-10-04 23:15:39 +020011571check-ssl
11572 This option forces encryption of all health checks over SSL, regardless of
11573 whether the server uses SSL or not for the normal traffic. This is generally
11574 used when an explicit "port" or "addr" directive is specified and SSL health
11575 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011576 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011577 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11578 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011579 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011580 See the "ssl" option for more information and "no-check-ssl" to disable
11581 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011582
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011583ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011584 This setting is only available when support for OpenSSL was built in. This
11585 option sets the string describing the list of cipher algorithms that is
11586 negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011587 string is defined in "man 1 ciphers". When SSL is used to communicate with
11588 servers on the local network, it is common to see a weaker set of algorithms
11589 than what is used over the internet. Doing so reduces CPU usage on both the
11590 server and haproxy while still keeping it compatible with deployed software.
11591 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
11592 is needed and just connectivity, using DES can be appropriate.
11593
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011594ciphersuites <ciphersuites>
11595 This setting is only available when support for OpenSSL was built in and
11596 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11597 describing the list of cipher algorithms that is negotiated during the TLS
11598 1.3 handshake with the server. The format of the string is defined in
11599 "man 1 ciphers" under the "ciphersuites" section.
11600
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011601cookie <value>
11602 The "cookie" parameter sets the cookie value assigned to the server to
11603 <value>. This value will be checked in incoming requests, and the first
11604 operational server possessing the same value will be selected. In return, in
11605 cookie insertion or rewrite modes, this value will be assigned to the cookie
11606 sent to the client. There is nothing wrong in having several servers sharing
11607 the same cookie value, and it is in fact somewhat common between normal and
11608 backup servers. See also the "cookie" keyword in backend section.
11609
Emeric Brunef42d922012-10-11 16:11:36 +020011610crl-file <crlfile>
11611 This setting is only available when support for OpenSSL was built in. It
11612 designates a PEM file from which to load certificate revocation list used
11613 to verify server's certificate.
11614
Emeric Bruna7aa3092012-10-26 12:58:00 +020011615crt <cert>
11616 This setting is only available when support for OpenSSL was built in.
11617 It designates a PEM file from which to load both a certificate and the
11618 associated private key. This file can be built by concatenating both PEM
11619 files into one. This certificate will be sent if the server send a client
11620 certificate request.
11621
Willy Tarreau96839092010-03-29 10:02:24 +020011622disabled
11623 The "disabled" keyword starts the server in the "disabled" state. That means
11624 that it is marked down in maintenance mode, and no connection other than the
11625 ones allowed by persist mode will reach it. It is very well suited to setup
11626 new servers, because normal traffic will never reach them, while it is still
11627 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011628 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011629
Frédéric Lécailled2376272017-03-21 18:52:12 +010011630enabled
11631 This option may be used as 'server' setting to reset any 'disabled'
11632 setting which would have been inherited from 'default-server' directive as
11633 default value.
11634 It may also be used as 'default-server' setting to reset any previous
11635 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011636
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011637error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011638 If health observing is enabled, the "error-limit" parameter specifies the
11639 number of consecutive errors that triggers event selected by the "on-error"
11640 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011641
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011642 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011643
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011644fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011645 The "fall" parameter states that a server will be considered as dead after
11646 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11647 unspecified. See also the "check", "inter" and "rise" parameters.
11648
Emeric Brun8694b9a2012-10-05 14:39:07 +020011649force-sslv3
11650 This option enforces use of SSLv3 only when SSL is used to communicate with
11651 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011652 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011653 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011654
11655force-tlsv10
11656 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011657 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011658 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011659
11660force-tlsv11
11661 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011662 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011663 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011664
11665force-tlsv12
11666 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011667 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011668 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011669
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011670force-tlsv13
11671 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11672 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011673 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011674
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011675id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011676 Set a persistent ID for the server. This ID must be positive and unique for
11677 the proxy. An unused ID will automatically be assigned if unset. The first
11678 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011679
Olivier Houchard0c18a6f2018-12-02 14:11:41 +010011680idle-timeout <delay>
11681 Set the time to keep a connection alive before destroying it. By default
11682 connections are destroyed as soon as they are unused, if idle-timeout is
11683 non-zero, then connection are kept alive for up to <delay> before being
11684 destroyed, and can be reused if no other connection is available.
11685
Willy Tarreau6a031d12016-11-07 19:42:35 +010011686init-addr {last | libc | none | <ip>},[...]*
11687 Indicate in what order the server's address should be resolved upon startup
11688 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011689 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011690 method which succeeds is used. If the end of the list is reached without
11691 finding a working method, an error is thrown. Method "last" suggests to pick
11692 the address which appears in the state file (see "server-state-file"). Method
11693 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11694 depending on the operating system and build options). Method "none"
11695 specifically indicates that the server should start without any valid IP
11696 address in a down state. It can be useful to ignore some DNS issues upon
11697 startup, waiting for the situation to get fixed later. Finally, an IP address
11698 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011699 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011700 server used to catch old sessions and present them with a decent error
11701 message for example. When the "first" load balancing algorithm is used, this
11702 IP address could point to a fake server used to trigger the creation of new
11703 instances on the fly. This option defaults to "last,libc" indicating that the
11704 previous address found in the state file (if any) is used first, otherwise
11705 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011706 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011707
11708 Example:
11709 defaults
11710 # never fail on address resolution
11711 default-server init-addr last,libc,none
11712
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011713inter <delay>
11714fastinter <delay>
11715downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011716 The "inter" parameter sets the interval between two consecutive health checks
11717 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11718 It is also possible to use "fastinter" and "downinter" to optimize delays
11719 between checks depending on the server state :
11720
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011721 Server state | Interval used
11722 ----------------------------------------+----------------------------------
11723 UP 100% (non-transitional) | "inter"
11724 ----------------------------------------+----------------------------------
11725 Transitionally UP (going down "fall"), | "fastinter" if set,
11726 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11727 or yet unchecked. |
11728 ----------------------------------------+----------------------------------
11729 DOWN 100% (non-transitional) | "downinter" if set,
11730 | "inter" otherwise.
11731 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011732
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011733 Just as with every other time-based parameter, they can be entered in any
11734 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11735 serves as a timeout for health checks sent to servers if "timeout check" is
11736 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011737 hosted on the same hardware, the agent and health checks of all servers
11738 are started with a small time offset between them. It is also possible to
11739 add some random noise in the agent and health checks interval using the
11740 global "spread-checks" keyword. This makes sense for instance when a lot
11741 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011742
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011743maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011744 The "maxconn" parameter specifies the maximal number of concurrent
11745 connections that will be sent to this server. If the number of incoming
11746 concurrent requests goes higher than this value, they will be queued, waiting
11747 for a connection to be released. This parameter is very important as it can
11748 save fragile servers from going down under extreme loads. If a "minconn"
11749 parameter is specified, the limit becomes dynamic. The default value is "0"
11750 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11751 the backend's "fullconn" keyword.
11752
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011753maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011754 The "maxqueue" parameter specifies the maximal number of connections which
11755 will wait in the queue for this server. If this limit is reached, next
11756 requests will be redispatched to other servers instead of indefinitely
11757 waiting to be served. This will break persistence but may allow people to
11758 quickly re-log in when the server they try to connect to is dying. The
11759 default value is "0" which means the queue is unlimited. See also the
11760 "maxconn" and "minconn" parameters.
11761
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011762minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011763 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11764 limit following the backend's load. The server will always accept at least
11765 <minconn> connections, never more than <maxconn>, and the limit will be on
11766 the ramp between both values when the backend has less than <fullconn>
11767 concurrent connections. This makes it possible to limit the load on the
11768 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011769 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011770 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011771
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011772namespace <name>
11773 On Linux, it is possible to specify which network namespace a socket will
11774 belong to. This directive makes it possible to explicitly bind a server to
11775 a namespace different from the default one. Please refer to your operating
11776 system's documentation to find more details about network namespaces.
11777
Frédéric Lécailled2376272017-03-21 18:52:12 +010011778no-agent-check
11779 This option may be used as "server" setting to reset any "agent-check"
11780 setting which would have been inherited from "default-server" directive as
11781 default value.
11782 It may also be used as "default-server" setting to reset any previous
11783 "default-server" "agent-check" setting.
11784
11785no-backup
11786 This option may be used as "server" setting to reset any "backup"
11787 setting which would have been inherited from "default-server" directive as
11788 default value.
11789 It may also be used as "default-server" setting to reset any previous
11790 "default-server" "backup" setting.
11791
11792no-check
11793 This option may be used as "server" setting to reset any "check"
11794 setting which would have been inherited from "default-server" directive as
11795 default value.
11796 It may also be used as "default-server" setting to reset any previous
11797 "default-server" "check" setting.
11798
11799no-check-ssl
11800 This option may be used as "server" setting to reset any "check-ssl"
11801 setting which would have been inherited from "default-server" directive as
11802 default value.
11803 It may also be used as "default-server" setting to reset any previous
11804 "default-server" "check-ssl" setting.
11805
Frédéric Lécailled2376272017-03-21 18:52:12 +010011806no-send-proxy
11807 This option may be used as "server" setting to reset any "send-proxy"
11808 setting which would have been inherited from "default-server" directive as
11809 default value.
11810 It may also be used as "default-server" setting to reset any previous
11811 "default-server" "send-proxy" setting.
11812
11813no-send-proxy-v2
11814 This option may be used as "server" setting to reset any "send-proxy-v2"
11815 setting which would have been inherited from "default-server" directive as
11816 default value.
11817 It may also be used as "default-server" setting to reset any previous
11818 "default-server" "send-proxy-v2" setting.
11819
11820no-send-proxy-v2-ssl
11821 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11822 setting which would have been inherited from "default-server" directive as
11823 default value.
11824 It may also be used as "default-server" setting to reset any previous
11825 "default-server" "send-proxy-v2-ssl" setting.
11826
11827no-send-proxy-v2-ssl-cn
11828 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11829 setting which would have been inherited from "default-server" directive as
11830 default value.
11831 It may also be used as "default-server" setting to reset any previous
11832 "default-server" "send-proxy-v2-ssl-cn" setting.
11833
11834no-ssl
11835 This option may be used as "server" setting to reset any "ssl"
11836 setting which would have been inherited from "default-server" directive as
11837 default value.
11838 It may also be used as "default-server" setting to reset any previous
11839 "default-server" "ssl" setting.
11840
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011841no-ssl-reuse
11842 This option disables SSL session reuse when SSL is used to communicate with
11843 the server. It will force the server to perform a full handshake for every
11844 new connection. It's probably only useful for benchmarking, troubleshooting,
11845 and for paranoid users.
11846
Emeric Brun9b3009b2012-10-05 11:55:06 +020011847no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011848 This option disables support for SSLv3 when SSL is used to communicate with
11849 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011850 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011851
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011852 Supported in default-server: No
11853
Emeric Brunf9c5c472012-10-11 15:28:34 +020011854no-tls-tickets
11855 This setting is only available when support for OpenSSL was built in. It
11856 disables the stateless session resumption (RFC 5077 TLS Ticket
11857 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011858 session resumption is more expensive in CPU usage for servers. This option
11859 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011860 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011861
Emeric Brun9b3009b2012-10-05 11:55:06 +020011862no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011863 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011864 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11865 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011866 often makes sense to disable it when communicating with local servers. This
11867 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011868 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011869
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011870 Supported in default-server: No
11871
Emeric Brun9b3009b2012-10-05 11:55:06 +020011872no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011873 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011874 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11875 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011876 often makes sense to disable it when communicating with local servers. This
11877 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011878 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011879
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011880 Supported in default-server: No
11881
Emeric Brun9b3009b2012-10-05 11:55:06 +020011882no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011883 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011884 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11885 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011886 often makes sense to disable it when communicating with local servers. This
11887 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011888 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011889
11890 Supported in default-server: No
11891
11892no-tlsv13
11893 This option disables support for TLSv1.3 when SSL is used to communicate with
11894 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11895 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11896 often makes sense to disable it when communicating with local servers. This
11897 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011898 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011899
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011900 Supported in default-server: No
11901
Frédéric Lécailled2376272017-03-21 18:52:12 +010011902no-verifyhost
11903 This option may be used as "server" setting to reset any "verifyhost"
11904 setting which would have been inherited from "default-server" directive as
11905 default value.
11906 It may also be used as "default-server" setting to reset any previous
11907 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011908
Simon Hormanfa461682011-06-25 09:39:49 +090011909non-stick
11910 Never add connections allocated to this sever to a stick-table.
11911 This may be used in conjunction with backup to ensure that
11912 stick-table persistence is disabled for backup servers.
11913
Olivier Houchardc7566002018-11-20 23:33:50 +010011914npn <protocols>
11915 This enables the NPN TLS extension and advertises the specified protocol list
11916 as supported on top of NPN. The protocol list consists in a comma-delimited
11917 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11918 This requires that the SSL library is build with support for TLS extensions
11919 enabled (check with haproxy -vv). Note that the NPN extension has been
11920 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11921 only available starting with OpenSSL 1.0.2.
11922
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011923observe <mode>
11924 This option enables health adjusting based on observing communication with
11925 the server. By default this functionality is disabled and enabling it also
11926 requires to enable health checks. There are two supported modes: "layer4" and
11927 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11928 significant. In layer7, which is only allowed for http proxies, responses
11929 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011930 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011931
11932 See also the "check", "on-error" and "error-limit".
11933
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011934on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011935 Select what should happen when enough consecutive errors are detected.
11936 Currently, four modes are available:
11937 - fastinter: force fastinter
11938 - fail-check: simulate a failed check, also forces fastinter (default)
11939 - sudden-death: simulate a pre-fatal failed health check, one more failed
11940 check will mark a server down, forces fastinter
11941 - mark-down: mark the server immediately down and force fastinter
11942
11943 See also the "check", "observe" and "error-limit".
11944
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011945on-marked-down <action>
11946 Modify what occurs when a server is marked down.
11947 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011948 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11949 all connections to the server are immediately terminated when the server
11950 goes down. It might be used if the health check detects more complex cases
11951 than a simple connection status, and long timeouts would cause the service
11952 to remain unresponsive for too long a time. For instance, a health check
11953 might detect that a database is stuck and that there's no chance to reuse
11954 existing connections anymore. Connections killed this way are logged with
11955 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011956
11957 Actions are disabled by default
11958
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011959on-marked-up <action>
11960 Modify what occurs when a server is marked up.
11961 Currently one action is available:
11962 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11963 done only if the server is not in backup state and if it is not disabled
11964 (it must have an effective weight > 0). This can be used sometimes to force
11965 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011966 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11967 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011968 with extreme care. Sessions killed because a server comes up are logged
11969 with an 'U' termination code (for "Up").
11970
11971 Actions are disabled by default
11972
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011973port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011974 Using the "port" parameter, it becomes possible to use a different port to
11975 send health-checks. On some servers, it may be desirable to dedicate a port
11976 to a specific component able to perform complex tests which are more suitable
11977 to health-checks than the application. It is common to run a simple script in
11978 inetd for instance. This parameter is ignored if the "check" parameter is not
11979 set. See also the "addr" parameter.
11980
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020011981proto <name>
11982
11983 Forces the multiplexer's protocol to use for the outgoing connections to this
11984 server. It must be compatible with the mode of the backend (TCP or HTTP). It
11985 must also be usable on the backend side. The list of available protocols is
11986 reported in haproxy -vv.
11987 Idea behind this optipon is to bypass the selection of the best multiplexer's
11988 protocol for all connections established to this server.
11989
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011990redir <prefix>
11991 The "redir" parameter enables the redirection mode for all GET and HEAD
11992 requests addressing this server. This means that instead of having HAProxy
11993 forward the request to the server, it will send an "HTTP 302" response with
11994 the "Location" header composed of this prefix immediately followed by the
11995 requested URI beginning at the leading '/' of the path component. That means
11996 that no trailing slash should be used after <prefix>. All invalid requests
11997 will be rejected, and all non-GET or HEAD requests will be normally served by
11998 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011999 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012000 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012001 users to a remote location in case of local disaster. Main use consists in
12002 increasing bandwidth for static servers by having the clients directly
12003 connect to them. Note: never use a relative location here, it would cause a
12004 loop between the client and HAProxy!
12005
12006 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12007
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012008rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012009 The "rise" parameter states that a server will be considered as operational
12010 after <count> consecutive successful health checks. This value defaults to 2
12011 if unspecified. See also the "check", "inter" and "fall" parameters.
12012
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012013resolve-opts <option>,<option>,...
12014 Comma separated list of options to apply to DNS resolution linked to this
12015 server.
12016
12017 Available options:
12018
12019 * allow-dup-ip
12020 By default, HAProxy prevents IP address duplication in a backend when DNS
12021 resolution at runtime is in operation.
12022 That said, for some cases, it makes sense that two servers (in the same
12023 backend, being resolved by the same FQDN) have the same IP address.
12024 For such case, simply enable this option.
12025 This is the opposite of prevent-dup-ip.
12026
12027 * prevent-dup-ip
12028 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12029 an IP address already set to a server in the same backend and sharing the
12030 same fqdn.
12031 This is the opposite of allow-dup-ip.
12032
12033 Example:
12034 backend b_myapp
12035 default-server init-addr none resolvers dns
12036 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12037 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12038
12039 With the option allow-dup-ip set:
12040 * if the nameserver returns a single IP address, then both servers will use
12041 it
12042 * If the nameserver returns 2 IP addresses, then each server will pick up a
12043 different address
12044
12045 Default value: not set
12046
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012047resolve-prefer <family>
12048 When DNS resolution is enabled for a server and multiple IP addresses from
12049 different families are returned, HAProxy will prefer using an IP address
12050 from the family mentioned in the "resolve-prefer" parameter.
12051 Available families: "ipv4" and "ipv6"
12052
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012053 Default value: ipv6
12054
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012055 Example:
12056
12057 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012058
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012059resolve-net <network>[,<network[,...]]
12060 This options prioritize th choice of an ip address matching a network. This is
12061 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012062 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012063 different datacenters. The latency between datacenter is not negligible, so
12064 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012065 configured network, another address is selected.
12066
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012067 Example:
12068
12069 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012070
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012071resolvers <id>
12072 Points to an existing "resolvers" section to resolve current server's
12073 hostname.
12074
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012075 Example:
12076
12077 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012078
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012079 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012080
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012081send-proxy
12082 The "send-proxy" parameter enforces use of the PROXY protocol over any
12083 connection established to this server. The PROXY protocol informs the other
12084 end about the layer 3/4 addresses of the incoming connection, so that it can
12085 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012086 upper layer protocol. For connections accepted by an "accept-proxy" or
12087 "accept-netscaler-cip" listener, the advertised address will be used. Only
12088 TCPv4 and TCPv6 address families are supported. Other families such as
12089 Unix sockets, will report an UNKNOWN family. Servers using this option can
12090 fully be chained to another instance of haproxy listening with an
12091 "accept-proxy" setting. This setting must not be used if the server isn't
12092 aware of the protocol. When health checks are sent to the server, the PROXY
12093 protocol is automatically used when this option is set, unless there is an
12094 explicit "port" or "addr" directive, in which case an explicit
12095 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012096 See also the "no-send-proxy" option of this section and "accept-proxy" and
12097 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012098
David Safb76832014-05-08 23:42:08 -040012099send-proxy-v2
12100 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12101 over any connection established to this server. The PROXY protocol informs
12102 the other end about the layer 3/4 addresses of the incoming connection, so
12103 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012104 whatever the upper layer protocol. It also send ALPN information if an alpn
12105 have been negotiated. This setting must not be used if the server isn't aware
12106 of this version of the protocol. See also the "no-send-proxy-v2" option of
12107 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012108
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012109proxy-v2-options <option>[,<option>]*
12110 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12111 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012112 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12113 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012114 certificate, "cert-key": key algorithm of the used certificate), "authority":
12115 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012116 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012117
David Safb76832014-05-08 23:42:08 -040012118send-proxy-v2-ssl
12119 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12120 2 over any connection established to this server. The PROXY protocol informs
12121 the other end about the layer 3/4 addresses of the incoming connection, so
12122 that it can know the client's address or the public address it accessed to,
12123 whatever the upper layer protocol. In addition, the SSL information extension
12124 of the PROXY protocol is added to the PROXY protocol header. This setting
12125 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012126 See also the "no-send-proxy-v2-ssl" option of this section and the
12127 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012128
12129send-proxy-v2-ssl-cn
12130 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12131 2 over any connection established to this server. The PROXY protocol informs
12132 the other end about the layer 3/4 addresses of the incoming connection, so
12133 that it can know the client's address or the public address it accessed to,
12134 whatever the upper layer protocol. In addition, the SSL information extension
12135 of the PROXY protocol, along along with the Common Name from the subject of
12136 the client certificate (if any), is added to the PROXY protocol header. This
12137 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012138 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12139 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012140
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012141slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012142 The "slowstart" parameter for a server accepts a value in milliseconds which
12143 indicates after how long a server which has just come back up will run at
12144 full speed. Just as with every other time-based parameter, it can be entered
12145 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12146 linearly from 0 to 100% during this time. The limitation applies to two
12147 parameters :
12148
12149 - maxconn: the number of connections accepted by the server will grow from 1
12150 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12151
12152 - weight: when the backend uses a dynamic weighted algorithm, the weight
12153 grows linearly from 1 to 100%. In this case, the weight is updated at every
12154 health-check. For this reason, it is important that the "inter" parameter
12155 is smaller than the "slowstart", in order to maximize the number of steps.
12156
12157 The slowstart never applies when haproxy starts, otherwise it would cause
12158 trouble to running servers. It only applies when a server has been previously
12159 seen as failed.
12160
Willy Tarreau732eac42015-07-09 11:40:25 +020012161sni <expression>
12162 The "sni" parameter evaluates the sample fetch expression, converts it to a
12163 string and uses the result as the host name sent in the SNI TLS extension to
12164 the server. A typical use case is to send the SNI received from the client in
12165 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012166 expression, though alternatives such as req.hdr(host) can also make sense. If
12167 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012168 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012169 "verify" directive for more details. If you want to set a SNI for health
12170 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012171
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012172source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012173source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012174source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012175 The "source" parameter sets the source address which will be used when
12176 connecting to the server. It follows the exact same parameters and principle
12177 as the backend "source" keyword, except that it only applies to the server
12178 referencing it. Please consult the "source" keyword for details.
12179
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012180 Additionally, the "source" statement on a server line allows one to specify a
12181 source port range by indicating the lower and higher bounds delimited by a
12182 dash ('-'). Some operating systems might require a valid IP address when a
12183 source port range is specified. It is permitted to have the same IP/range for
12184 several servers. Doing so makes it possible to bypass the maximum of 64k
12185 total concurrent connections. The limit will then reach 64k connections per
12186 server.
12187
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012188 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12189 specifying the source address without port(s).
12190
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012191ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012192 This option enables SSL ciphering on outgoing connections to the server. It
12193 is critical to verify server certificates using "verify" when using SSL to
12194 connect to servers, otherwise the communication is prone to trivial man in
12195 the-middle attacks rendering SSL useless. When this option is used, health
12196 checks are automatically sent in SSL too unless there is a "port" or an
12197 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012198 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12199 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012200
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012201ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12202 This option enforces use of <version> or lower when SSL is used to communicate
12203 with the server. This option is also available on global statement
12204 "ssl-default-server-options". See also "ssl-min-ver".
12205
12206ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12207 This option enforces use of <version> or upper when SSL is used to communicate
12208 with the server. This option is also available on global statement
12209 "ssl-default-server-options". See also "ssl-max-ver".
12210
Frédéric Lécailled2376272017-03-21 18:52:12 +010012211ssl-reuse
12212 This option may be used as "server" setting to reset any "no-ssl-reuse"
12213 setting which would have been inherited from "default-server" directive as
12214 default value.
12215 It may also be used as "default-server" setting to reset any previous
12216 "default-server" "no-ssl-reuse" setting.
12217
12218stick
12219 This option may be used as "server" setting to reset any "non-stick"
12220 setting which would have been inherited from "default-server" directive as
12221 default value.
12222 It may also be used as "default-server" setting to reset any previous
12223 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012224
Willy Tarreau163d4622015-10-13 16:16:41 +020012225tcp-ut <delay>
12226 Sets the TCP User Timeout for all outgoing connections to this server. This
12227 option is available on Linux since version 2.6.37. It allows haproxy to
12228 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012229 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012230 long-lived connections experiencing long idle periods such as remote
12231 terminals or database connection pools, where the client and server timeouts
12232 must remain high to allow a long period of idle, but where it is important to
12233 detect that the server has disappeared in order to release all resources
12234 associated with its connection (and the client's session). One typical use
12235 case is also to force dead server connections to die when health checks are
12236 too slow or during a soft reload since health checks are then disabled. The
12237 argument is a delay expressed in milliseconds by default. This only works for
12238 regular TCP connections, and is ignored for other protocols.
12239
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012240track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012241 This option enables ability to set the current state of the server by tracking
12242 another one. It is possible to track a server which itself tracks another
12243 server, provided that at the end of the chain, a server has health checks
12244 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012245 used, it has to be enabled on both proxies.
12246
Frédéric Lécailled2376272017-03-21 18:52:12 +010012247tls-tickets
12248 This option may be used as "server" setting to reset any "no-tls-tickets"
12249 setting which would have been inherited from "default-server" directive as
12250 default value.
12251 It may also be used as "default-server" setting to reset any previous
12252 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012253
Emeric Brunef42d922012-10-11 16:11:36 +020012254verify [none|required]
12255 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012256 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012257 certificate provided by the server is verified using CAs from 'ca-file' and
12258 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012259 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012260 the name passed using the "sni" directive, or if not provided, the static
12261 host name passed using the "verifyhost" directive. When no name is found, the
12262 certificate's names are ignored. For this reason, without SNI it's important
12263 to use "verifyhost". On verification failure the handshake is aborted. It is
12264 critically important to verify server certificates when using SSL to connect
12265 to servers, otherwise the communication is prone to trivial man-in-the-middle
12266 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12267 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012268
Evan Broderbe554312013-06-27 00:05:25 -070012269verifyhost <hostname>
12270 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012271 only takes effect if 'verify required' is also specified. This directive sets
12272 a default static hostname to check the server's certificate against when no
12273 SNI was used to connect to the server. If SNI is not used, this is the only
12274 way to enable hostname verification. This static hostname, when set, will
12275 also be used for health checks (which cannot provide an SNI value). If none
12276 of the hostnames in the certificate match the specified hostname, the
12277 handshake is aborted. The hostnames in the server-provided certificate may
12278 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012279
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012280weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012281 The "weight" parameter is used to adjust the server's weight relative to
12282 other servers. All servers will receive a load proportional to their weight
12283 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012284 load. The default weight is 1, and the maximal value is 256. A value of 0
12285 means the server will not participate in load-balancing but will still accept
12286 persistent connections. If this parameter is used to distribute the load
12287 according to server's capacity, it is recommended to start with values which
12288 can both grow and shrink, for instance between 10 and 100 to leave enough
12289 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012290
12291
Cyril Bonté46175dd2015-07-02 22:45:32 +0200122925.3. Server IP address resolution using DNS
12293-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012294
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012295HAProxy allows using a host name on the server line to retrieve its IP address
12296using name servers. By default, HAProxy resolves the name when parsing the
12297configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012298This is not sufficient in some cases, such as in Amazon where a server's IP
12299can change after a reboot or an ELB Virtual IP can change based on current
12300workload.
12301This chapter describes how HAProxy can be configured to process server's name
12302resolution at run time.
12303Whether run time server name resolution has been enable or not, HAProxy will
12304carry on doing the first resolution when parsing the configuration.
12305
12306
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123075.3.1. Global overview
12308----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012309
12310As we've seen in introduction, name resolution in HAProxy occurs at two
12311different steps of the process life:
12312
12313 1. when starting up, HAProxy parses the server line definition and matches a
12314 host name. It uses libc functions to get the host name resolved. This
12315 resolution relies on /etc/resolv.conf file.
12316
Christopher Faulet67957bd2017-09-27 11:00:59 +020012317 2. at run time, HAProxy performs periodically name resolutions for servers
12318 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012319
12320A few other events can trigger a name resolution at run time:
12321 - when a server's health check ends up in a connection timeout: this may be
12322 because the server has a new IP address. So we need to trigger a name
12323 resolution to know this new IP.
12324
Christopher Faulet67957bd2017-09-27 11:00:59 +020012325When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012326HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012327SRV label is specified, then the corresponding SRV records will be retrieved
12328from the DNS server, and the provided hostnames will be used. The SRV label
12329will be checked periodically, and if any server are added or removed, haproxy
12330will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012331
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012332A few things important to notice:
12333 - all the name servers are queried in the mean time. HAProxy will process the
12334 first valid response.
12335
12336 - a resolution is considered as invalid (NX, timeout, refused), when all the
12337 servers return an error.
12338
12339
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123405.3.2. The resolvers section
12341----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012342
12343This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012344HAProxy. There can be as many as resolvers section as needed. Each section can
12345contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012346
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012347When multiple name servers are configured in a resolvers section, then HAProxy
12348uses the first valid response. In case of invalid responses, only the last one
12349is treated. Purpose is to give the chance to a slow server to deliver a valid
12350answer after a fast faulty or outdated server.
12351
12352When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012353used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012354
Christopher Faulet67957bd2017-09-27 11:00:59 +020012355 1. HAProxy retries the same DNS query with a new query type. The A queries are
12356 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12357 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012358
Christopher Faulet67957bd2017-09-27 11:00:59 +020012359 2. When the fallback on the query type was done (or not applicable), HAProxy
12360 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012361
Christopher Faulet67957bd2017-09-27 11:00:59 +020012362 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12363 response is received after that, it stops the DNS resolution and reports
12364 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012365
Christopher Faulet67957bd2017-09-27 11:00:59 +020012366For example, with 2 name servers configured in a resolvers section, the
12367following scenarios are possible:
12368
12369 - First response is valid and is applied directly, second response is
12370 ignored
12371
12372 - First response is invalid and second one is valid, then second response is
12373 applied
12374
12375 - First response is a NX domain and second one a truncated response, then
12376 HAProxy retries the query with a new type
12377
12378 - First response is a NX domain and second one is a timeout, then HAProxy
12379 retries the query with a new type
12380
12381 - Query timed out for both name servers, then HAProxy retries it with the
12382 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012383
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012384As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12385a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012386<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012387
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012388
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012389resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012390 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012391
12392A resolvers section accept the following parameters:
12393
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012394accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012395 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012396 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012397 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12398 by RFC 6891)
12399
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012400 Note: the maximum allowed value is 8192.
12401
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012402nameserver <id> <ip>:<port>
12403 DNS server description:
12404 <id> : label of the server, should be unique
12405 <ip> : IP address of the server
12406 <port> : port where the DNS service actually runs
12407
Ben Draut44e609b2018-05-29 15:40:08 -060012408parse-resolv-conf
12409 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12410 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12411 placed in the resolvers section in place of this directive.
12412
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012413hold <status> <period>
12414 Defines <period> during which the last name resolution should be kept based
12415 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012416 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012417 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012418 <period> : interval between two successive name resolution when the last
12419 answer was in <status>. It follows the HAProxy time format.
12420 <period> is in milliseconds by default.
12421
Baptiste Assmann686408b2017-08-18 10:15:42 +020012422 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012423
Christopher Faulet67957bd2017-09-27 11:00:59 +020012424resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012425 Defines the number of resolutions available in the pool for this resolvers.
12426 If not defines, it defaults to 64. If your configuration requires more than
12427 <nb>, then HAProxy will return an error when parsing the configuration.
12428
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012429resolve_retries <nb>
12430 Defines the number <nb> of queries to send to resolve a server name before
12431 giving up.
12432 Default value: 3
12433
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012434 A retry occurs on name server timeout or when the full sequence of DNS query
12435 type failover is over and we need to start up from the default ANY query
12436 type.
12437
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012438timeout <event> <time>
12439 Defines timeouts related to name resolution
12440 <event> : the event on which the <time> timeout period applies to.
12441 events available are:
Christopher Faulet67957bd2017-09-27 11:00:59 +020012442 - resolve : default time to trigger name resolutions when no
12443 other time applied.
12444 Default value: 1s
12445 - retry : time between two DNS queries, when no valid response
12446 have been received.
12447 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012448 <time> : time related to the event. It follows the HAProxy time format.
12449 <time> is expressed in milliseconds.
12450
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012451 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012452
12453 resolvers mydns
12454 nameserver dns1 10.0.0.1:53
12455 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012456 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012457 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012458 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012459 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012460 hold other 30s
12461 hold refused 30s
12462 hold nx 30s
12463 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012464 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012465 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012466
12467
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200124686. HTTP header manipulation
12469---------------------------
12470
12471In HTTP mode, it is possible to rewrite, add or delete some of the request and
12472response headers based on regular expressions. It is also possible to block a
12473request or a response if a particular header matches a regular expression,
12474which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012475against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012476
Willy Tarreau70dffda2014-01-30 03:07:23 +010012477If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12478to process all rsp* rules which can allow, deny, rewrite or delete a header,
12479but it will refuse to add a header to any such messages as this is not
12480HTTP-compliant. The reason for still processing headers in such responses is to
12481stop and/or fix any possible information leak which may happen, for instance
12482because another downstream equipment would unconditionally add a header, or if
12483a server name appears there. When such messages are seen, normal processing
12484still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012485
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012486This section covers common usage of the following keywords, described in detail
12487in section 4.2 :
12488
12489 - reqadd <string>
12490 - reqallow <search>
12491 - reqiallow <search>
12492 - reqdel <search>
12493 - reqidel <search>
12494 - reqdeny <search>
12495 - reqideny <search>
12496 - reqpass <search>
12497 - reqipass <search>
12498 - reqrep <search> <replace>
12499 - reqirep <search> <replace>
12500 - reqtarpit <search>
12501 - reqitarpit <search>
12502 - rspadd <string>
12503 - rspdel <search>
12504 - rspidel <search>
12505 - rspdeny <search>
12506 - rspideny <search>
12507 - rsprep <search> <replace>
12508 - rspirep <search> <replace>
12509
12510With all these keywords, the same conventions are used. The <search> parameter
12511is a POSIX extended regular expression (regex) which supports grouping through
12512parenthesis (without the backslash). Spaces and other delimiters must be
12513prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12514Other characters may be prefixed with a backslash to change their meaning :
12515
12516 \t for a tab
12517 \r for a carriage return (CR)
12518 \n for a new line (LF)
12519 \ to mark a space and differentiate it from a delimiter
12520 \# to mark a sharp and differentiate it from a comment
12521 \\ to use a backslash in a regex
12522 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12523 \xXX to write the ASCII hex code XX as in the C language
12524
12525The <replace> parameter contains the string to be used to replace the largest
12526portion of text matching the regex. It can make use of the special characters
12527above, and can reference a substring which is delimited by parenthesis in the
12528regex, by writing a backslash ('\') immediately followed by one digit from 0 to
125299 indicating the group position (0 designating the entire line). This practice
12530is very common to users of the "sed" program.
12531
12532The <string> parameter represents the string which will systematically be added
12533after the last header line. It can also use special character sequences above.
12534
12535Notes related to these keywords :
12536---------------------------------
12537 - these keywords are not always convenient to allow/deny based on header
12538 contents. It is strongly recommended to use ACLs with the "block" keyword
12539 instead, resulting in far more flexible and manageable rules.
12540
12541 - lines are always considered as a whole. It is not possible to reference
12542 a header name only or a value only. This is important because of the way
12543 headers are written (notably the number of spaces after the colon).
12544
12545 - the first line is always considered as a header, which makes it possible to
12546 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12547 it harder to distinguish between headers and request line. The regex prefix
12548 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12549 ^[^ \t:]*: matches any header name followed by a colon.
12550
12551 - for performances reasons, the number of characters added to a request or to
12552 a response is limited at build time to values between 1 and 4 kB. This
12553 should normally be far more than enough for most usages. If it is too short
12554 on occasional usages, it is possible to gain some space by removing some
12555 useless headers before adding new ones.
12556
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012557 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012558 without the 'i' letter except that they ignore case when matching patterns.
12559
12560 - when a request passes through a frontend then a backend, all req* rules
12561 from the frontend will be evaluated, then all req* rules from the backend
12562 will be evaluated. The reverse path is applied to responses.
12563
12564 - req* statements are applied after "block" statements, so that "block" is
12565 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012566 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012567
12568
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125697. Using ACLs and fetching samples
12570----------------------------------
12571
Davor Ocelice9ed2812017-12-25 17:49:28 +010012572HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012573client or server information, from tables, environmental information etc...
12574The action of extracting such data is called fetching a sample. Once retrieved,
12575these samples may be used for various purposes such as a key to a stick-table,
12576but most common usages consist in matching them against predefined constant
12577data called patterns.
12578
12579
125807.1. ACL basics
12581---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012582
12583The use of Access Control Lists (ACL) provides a flexible solution to perform
12584content switching and generally to take decisions based on content extracted
12585from the request, the response or any environmental status. The principle is
12586simple :
12587
Willy Tarreau74ca5042013-06-11 23:12:07 +020012588 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012589 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012590 - apply one or multiple pattern matching methods on this sample
12591 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012592
Willy Tarreau74ca5042013-06-11 23:12:07 +020012593The actions generally consist in blocking a request, selecting a backend, or
12594adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012595
12596In order to define a test, the "acl" keyword is used. The syntax is :
12597
Willy Tarreau74ca5042013-06-11 23:12:07 +020012598 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012599
12600This creates a new ACL <aclname> or completes an existing one with new tests.
12601Those tests apply to the portion of request/response specified in <criterion>
12602and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012603an operator which may be specified before the set of values. Optionally some
12604conversion operators may be applied to the sample, and they will be specified
12605as a comma-delimited list of keywords just after the first keyword. The values
12606are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012607
12608ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12609'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12610which means that "my_acl" and "My_Acl" are two different ACLs.
12611
12612There is no enforced limit to the number of ACLs. The unused ones do not affect
12613performance, they just consume a small amount of memory.
12614
Willy Tarreau74ca5042013-06-11 23:12:07 +020012615The criterion generally is the name of a sample fetch method, or one of its ACL
12616specific declinations. The default test method is implied by the output type of
12617this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012618methods of a same sample fetch method. The sample fetch methods are the only
12619ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012620
12621Sample fetch methods return data which can be of the following types :
12622 - boolean
12623 - integer (signed or unsigned)
12624 - IPv4 or IPv6 address
12625 - string
12626 - data block
12627
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012628Converters transform any of these data into any of these. For example, some
12629converters might convert a string to a lower-case string while other ones
12630would turn a string to an IPv4 address, or apply a netmask to an IP address.
12631The resulting sample is of the type of the last converter applied to the list,
12632which defaults to the type of the sample fetch method.
12633
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012634Each sample or converter returns data of a specific type, specified with its
12635keyword in this documentation. When an ACL is declared using a standard sample
12636fetch method, certain types automatically involved a default matching method
12637which are summarized in the table below :
12638
12639 +---------------------+-----------------+
12640 | Sample or converter | Default |
12641 | output type | matching method |
12642 +---------------------+-----------------+
12643 | boolean | bool |
12644 +---------------------+-----------------+
12645 | integer | int |
12646 +---------------------+-----------------+
12647 | ip | ip |
12648 +---------------------+-----------------+
12649 | string | str |
12650 +---------------------+-----------------+
12651 | binary | none, use "-m" |
12652 +---------------------+-----------------+
12653
12654Note that in order to match a binary samples, it is mandatory to specify a
12655matching method, see below.
12656
Willy Tarreau74ca5042013-06-11 23:12:07 +020012657The ACL engine can match these types against patterns of the following types :
12658 - boolean
12659 - integer or integer range
12660 - IP address / network
12661 - string (exact, substring, suffix, prefix, subdir, domain)
12662 - regular expression
12663 - hex block
12664
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012665The following ACL flags are currently supported :
12666
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012667 -i : ignore case during matching of all subsequent patterns.
12668 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012669 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012670 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012671 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012672 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012673 -- : force end of flags. Useful when a string looks like one of the flags.
12674
Willy Tarreau74ca5042013-06-11 23:12:07 +020012675The "-f" flag is followed by the name of a file from which all lines will be
12676read as individual values. It is even possible to pass multiple "-f" arguments
12677if the patterns are to be loaded from multiple files. Empty lines as well as
12678lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12679will be stripped. If it is absolutely necessary to insert a valid pattern
12680beginning with a sharp, just prefix it with a space so that it is not taken for
12681a comment. Depending on the data type and match method, haproxy may load the
12682lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12683exact string matching. In this case, duplicates will automatically be removed.
12684
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012685The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12686parsed as two column file. The first column contains the patterns used by the
12687ACL, and the second column contain the samples. The sample can be used later by
12688a map. This can be useful in some rare cases where an ACL would just be used to
12689check for the existence of a pattern in a map before a mapping is applied.
12690
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012691The "-u" flag forces the unique id of the ACL. This unique id is used with the
12692socket interface to identify ACL and dynamically change its values. Note that a
12693file is always identified by its name even if an id is set.
12694
Willy Tarreau74ca5042013-06-11 23:12:07 +020012695Also, note that the "-i" flag applies to subsequent entries and not to entries
12696loaded from files preceding it. For instance :
12697
12698 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12699
12700In this example, each line of "exact-ua.lst" will be exactly matched against
12701the "user-agent" header of the request. Then each line of "generic-ua" will be
12702case-insensitively matched. Then the word "test" will be insensitively matched
12703as well.
12704
12705The "-m" flag is used to select a specific pattern matching method on the input
12706sample. All ACL-specific criteria imply a pattern matching method and generally
12707do not need this flag. However, this flag is useful with generic sample fetch
12708methods to describe how they're going to be matched against the patterns. This
12709is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012710obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012711followed by a pattern matching method name, this method is used instead of the
12712default one for the criterion. This makes it possible to match contents in ways
12713that were not initially planned, or with sample fetch methods which return a
12714string. The matching method also affects the way the patterns are parsed.
12715
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012716The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12717By default, if the parser cannot parse ip address it considers that the parsed
12718string is maybe a domain name and try dns resolution. The flag "-n" disable this
12719resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12720server is not reachable, the haproxy configuration parsing may last many minutes
12721waiting fir the timeout. During this time no error messages are displayed. The
12722flag "-n" disable this behavior. Note also that during the runtime, this
12723function is disabled for the dynamic acl modifications.
12724
Willy Tarreau74ca5042013-06-11 23:12:07 +020012725There are some restrictions however. Not all methods can be used with all
12726sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12727be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012728
12729 - "found" : only check if the requested sample could be found in the stream,
12730 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012731 to pass any pattern to avoid confusion. This matching method is
12732 particularly useful to detect presence of certain contents such
12733 as headers, cookies, etc... even if they are empty and without
12734 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012735
12736 - "bool" : check the value as a boolean. It can only be applied to fetches
12737 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012738 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012739
12740 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012741 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012742
12743 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012744 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012745
Davor Ocelice9ed2812017-12-25 17:49:28 +010012746 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012747 binary sequence. This may be used with binary or string samples.
12748
12749 - "len" : match the sample's length as an integer. This may be used with
12750 binary or string samples.
12751
Willy Tarreau74ca5042013-06-11 23:12:07 +020012752 - "str" : exact match : match the contents against a string. This may be
12753 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012754
Willy Tarreau74ca5042013-06-11 23:12:07 +020012755 - "sub" : substring match : check that the contents contain at least one of
12756 the provided string patterns. This may be used with binary or
12757 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012758
Willy Tarreau74ca5042013-06-11 23:12:07 +020012759 - "reg" : regex match : match the contents against a list of regular
12760 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012761
Willy Tarreau74ca5042013-06-11 23:12:07 +020012762 - "beg" : prefix match : check that the contents begin like the provided
12763 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012764
Willy Tarreau74ca5042013-06-11 23:12:07 +020012765 - "end" : suffix match : check that the contents end like the provided
12766 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012767
Willy Tarreau74ca5042013-06-11 23:12:07 +020012768 - "dir" : subdir match : check that a slash-delimited portion of the
12769 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012770 This may be used with binary or string samples.
12771
Willy Tarreau74ca5042013-06-11 23:12:07 +020012772 - "dom" : domain match : check that a dot-delimited portion of the contents
12773 exactly match one of the provided string patterns. This may be
12774 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012775
12776For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12777request, it is possible to do :
12778
12779 acl jsess_present cook(JSESSIONID) -m found
12780
12781In order to apply a regular expression on the 500 first bytes of data in the
12782buffer, one would use the following acl :
12783
12784 acl script_tag payload(0,500) -m reg -i <script>
12785
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012786On systems where the regex library is much slower when using "-i", it is
12787possible to convert the sample to lowercase before matching, like this :
12788
12789 acl script_tag payload(0,500),lower -m reg <script>
12790
Willy Tarreau74ca5042013-06-11 23:12:07 +020012791All ACL-specific criteria imply a default matching method. Most often, these
12792criteria are composed by concatenating the name of the original sample fetch
12793method and the matching method. For example, "hdr_beg" applies the "beg" match
12794to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12795criteria rely on a sample fetch method, it is always possible instead to use
12796the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012797
Willy Tarreau74ca5042013-06-11 23:12:07 +020012798If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012799the matching method is simply applied to the underlying sample fetch method.
12800For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012801
Willy Tarreau74ca5042013-06-11 23:12:07 +020012802 acl short_form hdr_beg(host) www.
12803 acl alternate1 hdr_beg(host) -m beg www.
12804 acl alternate2 hdr_dom(host) -m beg www.
12805 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012806
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012807
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012808The table below summarizes the compatibility matrix between sample or converter
12809types and the pattern types to fetch against. It indicates for each compatible
12810combination the name of the matching method to be used, surrounded with angle
12811brackets ">" and "<" when the method is the default one and will work by
12812default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012813
Willy Tarreau74ca5042013-06-11 23:12:07 +020012814 +-------------------------------------------------+
12815 | Input sample type |
12816 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012817 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012818 +----------------------+---------+---------+---------+---------+---------+
12819 | none (presence only) | found | found | found | found | found |
12820 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012821 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012822 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012823 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012824 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012825 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012826 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012827 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012828 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012829 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012830 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012831 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012832 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012833 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012834 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012835 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012836 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012837 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012838 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012839 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012840 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012841 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012842 +----------------------+---------+---------+---------+---------+---------+
12843 | hex block | | | | bin | bin |
12844 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012845
12846
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128477.1.1. Matching booleans
12848------------------------
12849
12850In order to match a boolean, no value is needed and all values are ignored.
12851Boolean matching is used by default for all fetch methods of type "boolean".
12852When boolean matching is used, the fetched value is returned as-is, which means
12853that a boolean "true" will always match and a boolean "false" will never match.
12854
12855Boolean matching may also be enforced using "-m bool" on fetch methods which
12856return an integer value. Then, integer value 0 is converted to the boolean
12857"false" and all other values are converted to "true".
12858
Willy Tarreau6a06a402007-07-15 20:15:28 +020012859
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128607.1.2. Matching integers
12861------------------------
12862
12863Integer matching applies by default to integer fetch methods. It can also be
12864enforced on boolean fetches using "-m int". In this case, "false" is converted
12865to the integer 0, and "true" is converted to the integer 1.
12866
12867Integer matching also supports integer ranges and operators. Note that integer
12868matching only applies to positive values. A range is a value expressed with a
12869lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012870
12871For instance, "1024:65535" is a valid range to represent a range of
12872unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12873representation of privileged ports, and ":1023" would also work.
12874
Willy Tarreau62644772008-07-16 18:36:06 +020012875As a special case, some ACL functions support decimal numbers which are in fact
12876two integers separated by a dot. This is used with some version checks for
12877instance. All integer properties apply to those decimal numbers, including
12878ranges and operators.
12879
Willy Tarreau6a06a402007-07-15 20:15:28 +020012880For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012881operators with ranges does not make much sense and is strongly discouraged.
12882Similarly, it does not make much sense to perform order comparisons with a set
12883of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012884
Willy Tarreau0ba27502007-12-24 16:55:16 +010012885Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012886
12887 eq : true if the tested value equals at least one value
12888 ge : true if the tested value is greater than or equal to at least one value
12889 gt : true if the tested value is greater than at least one value
12890 le : true if the tested value is less than or equal to at least one value
12891 lt : true if the tested value is less than at least one value
12892
Willy Tarreau0ba27502007-12-24 16:55:16 +010012893For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012894
12895 acl negative-length hdr_val(content-length) lt 0
12896
Willy Tarreau62644772008-07-16 18:36:06 +020012897This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12898
12899 acl sslv3 req_ssl_ver 3:3.1
12900
Willy Tarreau6a06a402007-07-15 20:15:28 +020012901
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129027.1.3. Matching strings
12903-----------------------
12904
12905String matching applies to string or binary fetch methods, and exists in 6
12906different forms :
12907
12908 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012909 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012910
12911 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012912 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012913
12914 - prefix match (-m beg) : the patterns are compared with the beginning of
12915 the extracted string, and the ACL matches if any of them matches.
12916
12917 - suffix match (-m end) : the patterns are compared with the end of the
12918 extracted string, and the ACL matches if any of them matches.
12919
Baptiste Assmann33db6002016-03-06 23:32:10 +010012920 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012921 string, delimited with slashes ("/"), and the ACL matches if any of them
12922 matches.
12923
12924 - domain match (-m dom) : the patterns are looked up inside the extracted
12925 string, delimited with dots ("."), and the ACL matches if any of them
12926 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012927
12928String matching applies to verbatim strings as they are passed, with the
12929exception of the backslash ("\") which makes it possible to escape some
12930characters such as the space. If the "-i" flag is passed before the first
12931string, then the matching will be performed ignoring the case. In order
12932to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012933before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012934
12935
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129367.1.4. Matching regular expressions (regexes)
12937---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012938
12939Just like with string matching, regex matching applies to verbatim strings as
12940they are passed, with the exception of the backslash ("\") which makes it
12941possible to escape some characters such as the space. If the "-i" flag is
12942passed before the first regex, then the matching will be performed ignoring
12943the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012944the "--" flag before the first string. Same principle applies of course to
12945match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012946
12947
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129487.1.5. Matching arbitrary data blocks
12949-------------------------------------
12950
12951It is possible to match some extracted samples against a binary block which may
12952not safely be represented as a string. For this, the patterns must be passed as
12953a series of hexadecimal digits in an even number, when the match method is set
12954to binary. Each sequence of two digits will represent a byte. The hexadecimal
12955digits may be used upper or lower case.
12956
12957Example :
12958 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12959 acl hello payload(0,6) -m bin 48656c6c6f0a
12960
12961
129627.1.6. Matching IPv4 and IPv6 addresses
12963---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012964
12965IPv4 addresses values can be specified either as plain addresses or with a
12966netmask appended, in which case the IPv4 address matches whenever it is
12967within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012968host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012969difficult to read and debug configurations. If hostnames are used, you should
12970at least ensure that they are present in /etc/hosts so that the configuration
12971does not depend on any random DNS match at the moment the configuration is
12972parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012973
Daniel Schnellereba56342016-04-13 00:26:52 +020012974The dotted IPv4 address notation is supported in both regular as well as the
12975abbreviated form with all-0-octets omitted:
12976
12977 +------------------+------------------+------------------+
12978 | Example 1 | Example 2 | Example 3 |
12979 +------------------+------------------+------------------+
12980 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12981 | 192.168.1 | 10.12 | 127.1 |
12982 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12983 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12984 +------------------+------------------+------------------+
12985
12986Notice that this is different from RFC 4632 CIDR address notation in which
12987192.168.42/24 would be equivalent to 192.168.42.0/24.
12988
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012989IPv6 may be entered in their usual form, with or without a netmask appended.
12990Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12991trouble with randomly resolved IP addresses, host names are never allowed in
12992IPv6 patterns.
12993
12994HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12995following situations :
12996 - tested address is IPv4, pattern address is IPv4, the match applies
12997 in IPv4 using the supplied mask if any.
12998 - tested address is IPv6, pattern address is IPv6, the match applies
12999 in IPv6 using the supplied mask if any.
13000 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13001 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13002 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13003 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13004 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13005 applied in IPv6 using the supplied IPv6 mask.
13006
Willy Tarreau74ca5042013-06-11 23:12:07 +020013007
130087.2. Using ACLs to form conditions
13009----------------------------------
13010
13011Some actions are only performed upon a valid condition. A condition is a
13012combination of ACLs with operators. 3 operators are supported :
13013
13014 - AND (implicit)
13015 - OR (explicit with the "or" keyword or the "||" operator)
13016 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013017
Willy Tarreau74ca5042013-06-11 23:12:07 +020013018A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013019
Willy Tarreau74ca5042013-06-11 23:12:07 +020013020 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013021
Willy Tarreau74ca5042013-06-11 23:12:07 +020013022Such conditions are generally used after an "if" or "unless" statement,
13023indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013024
Willy Tarreau74ca5042013-06-11 23:12:07 +020013025For instance, to block HTTP requests to the "*" URL with methods other than
13026"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13027requests with a content-length greater than 0, and finally every request which
13028is not either GET/HEAD/POST/OPTIONS !
13029
13030 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013031 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13032 http-request deny if METH_GET HTTP_CONTENT
13033 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013034
13035To select a different backend for requests to static contents on the "www" site
13036and to every request on the "img", "video", "download" and "ftp" hosts :
13037
13038 acl url_static path_beg /static /images /img /css
13039 acl url_static path_end .gif .png .jpg .css .js
13040 acl host_www hdr_beg(host) -i www
13041 acl host_static hdr_beg(host) -i img. video. download. ftp.
13042
Davor Ocelice9ed2812017-12-25 17:49:28 +010013043 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013044 # of host "www". Use backend "www" for the rest.
13045 use_backend static if host_static or host_www url_static
13046 use_backend www if host_www
13047
13048It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13049expressions that are built on the fly without needing to be declared. They must
13050be enclosed between braces, with a space before and after each brace (because
13051the braces must be seen as independent words). Example :
13052
13053 The following rule :
13054
13055 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013056 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013057
13058 Can also be written that way :
13059
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013060 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013061
13062It is generally not recommended to use this construct because it's a lot easier
13063to leave errors in the configuration when written that way. However, for very
13064simple rules matching only one source IP address for instance, it can make more
13065sense to use them than to declare ACLs with random names. Another example of
13066good use is the following :
13067
13068 With named ACLs :
13069
13070 acl site_dead nbsrv(dynamic) lt 2
13071 acl site_dead nbsrv(static) lt 2
13072 monitor fail if site_dead
13073
13074 With anonymous ACLs :
13075
13076 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13077
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013078See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13079keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013080
13081
130827.3. Fetching samples
13083---------------------
13084
13085Historically, sample fetch methods were only used to retrieve data to match
13086against patterns using ACLs. With the arrival of stick-tables, a new class of
13087sample fetch methods was created, most often sharing the same syntax as their
13088ACL counterpart. These sample fetch methods are also known as "fetches". As
13089of now, ACLs and fetches have converged. All ACL fetch methods have been made
13090available as fetch methods, and ACLs may use any sample fetch method as well.
13091
13092This section details all available sample fetch methods and their output type.
13093Some sample fetch methods have deprecated aliases that are used to maintain
13094compatibility with existing configurations. They are then explicitly marked as
13095deprecated and should not be used in new setups.
13096
13097The ACL derivatives are also indicated when available, with their respective
13098matching methods. These ones all have a well defined default pattern matching
13099method, so it is never necessary (though allowed) to pass the "-m" option to
13100indicate how the sample will be matched using ACLs.
13101
13102As indicated in the sample type versus matching compatibility matrix above,
13103when using a generic sample fetch method in an ACL, the "-m" option is
13104mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13105the same keyword exists as an ACL keyword and as a standard fetch method, the
13106ACL engine will automatically pick the ACL-only one by default.
13107
13108Some of these keywords support one or multiple mandatory arguments, and one or
13109multiple optional arguments. These arguments are strongly typed and are checked
13110when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013111incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13112are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013113is optional, it will be indicated below between square brackets ('[ ]'). When
13114all arguments are optional, the parenthesis may be omitted.
13115
13116Thus, the syntax of a standard sample fetch method is one of the following :
13117 - name
13118 - name(arg1)
13119 - name(arg1,arg2)
13120
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013121
131227.3.1. Converters
13123-----------------
13124
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013125Sample fetch methods may be combined with transformations to be applied on top
13126of the fetched sample (also called "converters"). These combinations form what
13127is called "sample expressions" and the result is a "sample". Initially this
13128was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013129has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013130unique-id-format, add-header, ...).
13131
13132These transformations are enumerated as a series of specific keywords after the
13133sample fetch method. These keywords may equally be appended immediately after
13134the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013135support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013136
Willy Tarreau97707872015-01-27 15:12:13 +010013137A certain category of converters are bitwise and arithmetic operators which
13138support performing basic operations on integers. Some bitwise operations are
13139supported (and, or, xor, cpl) and some arithmetic operations are supported
13140(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13141bool) which make it possible to report a match without having to write an ACL.
13142
Willy Tarreau74ca5042013-06-11 23:12:07 +020013143The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013144
Ben Shillitof25e8e52016-12-02 14:25:37 +00001314551d.single(<prop>[,<prop>*])
13146 Returns values for the properties requested as a string, where values are
13147 separated by the delimiter specified with "51degrees-property-separator".
13148 The device is identified using the User-Agent header passed to the
13149 converter. The function can be passed up to five property names, and if a
13150 property name can't be found, the value "NoData" is returned.
13151
13152 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013153 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13154 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013155 # User-Agent passed to the converter.
13156 frontend http-in
13157 bind *:8081
13158 default_backend servers
13159 http-request set-header X-51D-DeviceTypeMobileTablet \
13160 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13161
Willy Tarreau97707872015-01-27 15:12:13 +010013162add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013163 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013164 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013165 name. The name of the variable starts with an indication about its scope. The
13166 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013167 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013168 "sess" : the variable is shared with the whole session
13169 "txn" : the variable is shared with the transaction (request and response)
13170 "req" : the variable is shared only during request processing
13171 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013172 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013173 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013174
13175and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013176 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013177 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013178 numeric value or a variable name. The name of the variable starts with an
13179 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013180 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013181 "sess" : the variable is shared with the whole session
13182 "txn" : the variable is shared with the transaction (request and response)
13183 "req" : the variable is shared only during request processing
13184 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013185 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013186 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013187
Holger Just1bfc24b2017-05-06 00:56:53 +020013188b64dec
13189 Converts (decodes) a base64 encoded input string to its binary
13190 representation. It performs the inverse operation of base64().
13191
Emeric Brun53d1a982014-04-30 18:21:37 +020013192base64
13193 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013194 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013195 an SSL ID can be copied in a header).
13196
Willy Tarreau97707872015-01-27 15:12:13 +010013197bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013198 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013199 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013200 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013201 presence of a flag).
13202
Emeric Brun54c4ac82014-11-03 15:32:43 +010013203bytes(<offset>[,<length>])
13204 Extracts some bytes from an input binary sample. The result is a binary
13205 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013206 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013207
Willy Tarreau280f42b2018-02-19 15:34:12 +010013208concat([<start>],[<var>],[<end>])
13209 Concatenates up to 3 fields after the current sample which is then turned to
13210 a string. The first one, <start>, is a constant string, that will be appended
13211 immediately after the existing sample. It may be omitted if not used. The
13212 second one, <var>, is a variable name. The variable will be looked up, its
13213 contents converted to a string, and it will be appended immediately after the
13214 <first> part. If the variable is not found, nothing is appended. It may be
13215 omitted as well. The third field, <end> is a constant string that will be
13216 appended after the variable. It may also be omitted. Together, these elements
13217 allow to concatenate variables with delimiters to an existing set of
13218 variables. This can be used to build new variables made of a succession of
13219 other variables, such as colon-delimited varlues. Note that due to the config
13220 parser, it is not possible to use a comma nor a closing parenthesis as
13221 delimitors.
13222
13223 Example:
13224 tcp-request session set-var(sess.src) src
13225 tcp-request session set-var(sess.dn) ssl_c_s_dn
13226 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13227 http-request set-header x-hap-sig %[var(txn.sig)]
13228
Willy Tarreau97707872015-01-27 15:12:13 +010013229cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013230 Takes the input value of type signed integer, applies a ones-complement
13231 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013232
Willy Tarreau80599772015-01-20 19:35:24 +010013233crc32([<avalanche>])
13234 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13235 hash function. Optionally, it is possible to apply a full avalanche hash
13236 function to the output if the optional <avalanche> argument equals 1. This
13237 converter uses the same functions as used by the various hash-based load
13238 balancing algorithms, so it will provide exactly the same results. It is
13239 provided for compatibility with other software which want a CRC32 to be
13240 computed on some input keys, so it follows the most common implementation as
13241 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13242 but may provide a better or at least less predictable distribution. It must
13243 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013244 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13245
13246crc32c([<avalanche>])
13247 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13248 hash function. Optionally, it is possible to apply a full avalanche hash
13249 function to the output if the optional <avalanche> argument equals 1. This
13250 converter uses the same functions as described in RFC4960, Appendix B [8].
13251 It is provided for compatibility with other software which want a CRC32C to be
13252 computed on some input keys. It is slower than the other algorithms and it must
13253 not be used for security purposes as a 32-bit hash is trivial to break. See
13254 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013255
David Carlier29b3ca32015-09-25 14:09:21 +010013256da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013257 Asks the DeviceAtlas converter to identify the User Agent string passed on
13258 input, and to emit a string made of the concatenation of the properties
13259 enumerated in argument, delimited by the separator defined by the global
13260 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013261 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013262 configuration language.
13263
13264 Example:
13265 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013266 bind *:8881
13267 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013268 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 +020013269
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013270debug
13271 This converter is used as debug tool. It dumps on screen the content and the
13272 type of the input sample. The sample is returned as is on its output. This
13273 converter only exists when haproxy was built with debugging enabled.
13274
Willy Tarreau97707872015-01-27 15:12:13 +010013275div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013276 Divides the input value of type signed integer by <value>, and returns the
13277 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013278 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013279 variable name. The name of the variable starts with an indication about its
13280 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013281 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013282 "sess" : the variable is shared with the whole session
13283 "txn" : the variable is shared with the transaction (request and response)
13284 "req" : the variable is shared only during request processing
13285 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013286 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013287 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013288
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013289djb2([<avalanche>])
13290 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13291 hash function. Optionally, it is possible to apply a full avalanche hash
13292 function to the output if the optional <avalanche> argument equals 1. This
13293 converter uses the same functions as used by the various hash-based load
13294 balancing algorithms, so it will provide exactly the same results. It is
13295 mostly intended for debugging, but can be used as a stick-table entry to
13296 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013297 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13298 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013299
Willy Tarreau97707872015-01-27 15:12:13 +010013300even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013301 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013302 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13303
Marcin Deranek9631a282018-04-16 14:30:46 +020013304field(<index>,<delimiters>[,<count>])
13305 Extracts the substring at the given index counting from the beginning
13306 (positive index) or from the end (negative index) considering given delimiters
13307 from an input string. Indexes start at 1 or -1 and delimiters are a string
13308 formatted list of chars. Optionally you can specify <count> of fields to
13309 extract (default: 1). Value of 0 indicates extraction of all remaining
13310 fields.
13311
13312 Example :
13313 str(f1_f2_f3__f5),field(5,_) # f5
13314 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13315 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13316 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13317 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013318
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013319hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013320 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013321 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013322 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 +020013323 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013324
Dragan Dosen3f957b22017-10-24 09:27:34 +020013325hex2i
13326 Converts a hex string containing two hex digits per input byte to an
13327 integer. If the input value can not be converted, then zero is returned.
13328
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013329http_date([<offset>])
13330 Converts an integer supposed to contain a date since epoch to a string
13331 representing this date in a format suitable for use in HTTP header fields. If
13332 an offset value is specified, then it is a number of seconds that is added to
13333 the date before the conversion is operated. This is particularly useful to
13334 emit Date header fields, Expires values in responses when combined with a
13335 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013336
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013337in_table(<table>)
13338 Uses the string representation of the input sample to perform a look up in
13339 the specified table. If the key is not found in the table, a boolean false
13340 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013341 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013342 or not a source IP address or an Authorization header was already seen).
13343
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013344ipmask(<mask4>, [<mask6>])
13345 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013346 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013347 table entries and as such use the same server. The mask4 can be passed in
13348 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13349 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13350 If no mask6 is given IPv6 addresses will fail to convert for backwards
13351 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013352
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013353json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013354 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013355 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013356 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013357 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13358 of errors:
13359 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13360 bytes, ...)
13361 - invalid range (the decoded value is within a UTF-8 prohibited range),
13362 - code overlong (the value is encoded with more bytes than necessary).
13363
13364 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13365 character is greater than 0xffff because the JSON string escape specification
13366 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13367 in 4 variants designated by a combination of two suffix letters : "p" for
13368 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13369 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013370 - "ascii" : never fails;
13371 - "utf8" : fails on any detected errors;
13372 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013373 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013374 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013375 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13376 characters corresponding to the other errors.
13377
13378 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013379 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013380
13381 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013382 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013383 capture request header user-agent len 150
13384 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013385
13386 Input request from client 127.0.0.1:
13387 GET / HTTP/1.0
13388 User-Agent: Very "Ugly" UA 1/2
13389
13390 Output log:
13391 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13392
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013393language(<value>[,<default>])
13394 Returns the value with the highest q-factor from a list as extracted from the
13395 "accept-language" header using "req.fhdr". Values with no q-factor have a
13396 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13397 belong to the list of semi-colon delimited <values> will be considered. The
13398 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13399 given list and a default value is provided, it is returned. Note that language
13400 names may have a variant after a dash ('-'). If this variant is present in the
13401 list, it will be matched, but if it is not, only the base language is checked.
13402 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013403 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013404 ordering of the values in the request counts, as the first value among
13405 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013406
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013407 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013408
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013409 # this configuration switches to the backend matching a
13410 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013411
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013412 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13413 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13414 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13415 use_backend spanish if es
13416 use_backend french if fr
13417 use_backend english if en
13418 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013419
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013420length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013421 Get the length of the string. This can only be placed after a string
13422 sample fetch function or after a transformation keyword returning a string
13423 type. The result is of type integer.
13424
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013425lower
13426 Convert a string sample to lower case. This can only be placed after a string
13427 sample fetch function or after a transformation keyword returning a string
13428 type. The result is of type string.
13429
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013430ltime(<format>[,<offset>])
13431 Converts an integer supposed to contain a date since epoch to a string
13432 representing this date in local time using a format defined by the <format>
13433 string using strftime(3). The purpose is to allow any date format to be used
13434 in logs. An optional <offset> in seconds may be applied to the input date
13435 (positive or negative). See the strftime() man page for the format supported
13436 by your operating system. See also the utime converter.
13437
13438 Example :
13439
13440 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013441 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013442 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13443
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013444map(<map_file>[,<default_value>])
13445map_<match_type>(<map_file>[,<default_value>])
13446map_<match_type>_<output_type>(<map_file>[,<default_value>])
13447 Search the input value from <map_file> using the <match_type> matching method,
13448 and return the associated value converted to the type <output_type>. If the
13449 input value cannot be found in the <map_file>, the converter returns the
13450 <default_value>. If the <default_value> is not set, the converter fails and
13451 acts as if no input value could be fetched. If the <match_type> is not set, it
13452 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13453 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13454 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013455
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013456 It is important to avoid overlapping between the keys : IP addresses and
13457 strings are stored in trees, so the first of the finest match will be used.
13458 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013459
Tim Düsterhus4896c442016-11-29 02:15:19 +010013460 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013461 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013462
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013463 input type | match method | output type str | output type int | output type ip
13464 -----------+--------------+-----------------+-----------------+---------------
13465 str | str | map_str | map_str_int | map_str_ip
13466 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013467 str | beg | map_beg | map_beg_int | map_end_ip
13468 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013469 str | sub | map_sub | map_sub_int | map_sub_ip
13470 -----------+--------------+-----------------+-----------------+---------------
13471 str | dir | map_dir | map_dir_int | map_dir_ip
13472 -----------+--------------+-----------------+-----------------+---------------
13473 str | dom | map_dom | map_dom_int | map_dom_ip
13474 -----------+--------------+-----------------+-----------------+---------------
13475 str | end | map_end | map_end_int | map_end_ip
13476 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013477 str | reg | map_reg | map_reg_int | map_reg_ip
13478 -----------+--------------+-----------------+-----------------+---------------
13479 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013480 -----------+--------------+-----------------+-----------------+---------------
13481 int | int | map_int | map_int_int | map_int_ip
13482 -----------+--------------+-----------------+-----------------+---------------
13483 ip | ip | map_ip | map_ip_int | map_ip_ip
13484 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013485
Thierry Fournier8feaa662016-02-10 22:55:20 +010013486 The special map called "map_regm" expect matching zone in the regular
13487 expression and modify the output replacing back reference (like "\1") by
13488 the corresponding match text.
13489
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013490 The file contains one key + value per line. Lines which start with '#' are
13491 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13492 is then the first "word" (series of non-space/tabs characters), and the value
13493 is what follows this series of space/tab till the end of the line excluding
13494 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013495
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013496 Example :
13497
13498 # this is a comment and is ignored
13499 2.22.246.0/23 United Kingdom \n
13500 <-><-----------><--><------------><---->
13501 | | | | `- trailing spaces ignored
13502 | | | `---------- value
13503 | | `-------------------- middle spaces ignored
13504 | `---------------------------- key
13505 `------------------------------------ leading spaces ignored
13506
Willy Tarreau97707872015-01-27 15:12:13 +010013507mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013508 Divides the input value of type signed integer by <value>, and returns the
13509 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013510 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013511 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013512 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013513 "sess" : the variable is shared with the whole session
13514 "txn" : the variable is shared with the transaction (request and response)
13515 "req" : the variable is shared only during request processing
13516 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013517 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013518 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013519
13520mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013521 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013522 the product as an signed integer. In case of overflow, the largest possible
13523 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013524 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013525 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013526 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013527 "sess" : the variable is shared with the whole session
13528 "txn" : the variable is shared with the transaction (request and response)
13529 "req" : the variable is shared only during request processing
13530 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013531 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013532 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013533
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013534nbsrv
13535 Takes an input value of type string, interprets it as a backend name and
13536 returns the number of usable servers in that backend. Can be used in places
13537 where we want to look up a backend from a dynamic name, like a result of a
13538 map lookup.
13539
Willy Tarreau97707872015-01-27 15:12:13 +010013540neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013541 Takes the input value of type signed integer, computes the opposite value,
13542 and returns the remainder as an signed integer. 0 is identity. This operator
13543 is provided for reversed subtracts : in order to subtract the input from a
13544 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013545
13546not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013547 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013548 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013549 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013550 absence of a flag).
13551
13552odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013553 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013554 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13555
13556or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013557 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013558 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013559 numeric value or a variable name. The name of the variable starts with an
13560 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013561 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013562 "sess" : the variable is shared with the whole session
13563 "txn" : the variable is shared with the transaction (request and response)
13564 "req" : the variable is shared only during request processing
13565 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013566 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013567 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013568
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013569regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013570 Applies a regex-based substitution to the input string. It does the same
13571 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13572 default it will replace in the input string the first occurrence of the
13573 largest part matching the regular expression <regex> with the substitution
13574 string <subst>. It is possible to replace all occurrences instead by adding
13575 the flag "g" in the third argument <flags>. It is also possible to make the
13576 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13577 string, it is made up from the concatenation of all desired flags. Thus if
13578 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13579 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013580 configuration parser, some characters such as closing parenthesis, closing
13581 square brackets or comma are not possible to use in the arguments. The first
13582 use of this converter is to replace certain characters or sequence of
13583 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013584
13585 Example :
13586
13587 # de-duplicate "/" in header "x-path".
13588 # input: x-path: /////a///b/c/xzxyz/
13589 # output: x-path: /a/b/c/xzxyz/
13590 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13591
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013592capture-req(<id>)
13593 Capture the string entry in the request slot <id> and returns the entry as
13594 is. If the slot doesn't exist, the capture fails silently.
13595
13596 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013597 "http-response capture", "capture.req.hdr" and
13598 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013599
13600capture-res(<id>)
13601 Capture the string entry in the response slot <id> and returns the entry as
13602 is. If the slot doesn't exist, the capture fails silently.
13603
13604 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013605 "http-response capture", "capture.req.hdr" and
13606 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013607
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013608sdbm([<avalanche>])
13609 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13610 hash function. Optionally, it is possible to apply a full avalanche hash
13611 function to the output if the optional <avalanche> argument equals 1. This
13612 converter uses the same functions as used by the various hash-based load
13613 balancing algorithms, so it will provide exactly the same results. It is
13614 mostly intended for debugging, but can be used as a stick-table entry to
13615 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013616 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13617 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013618
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013619set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013620 Sets a variable with the input content and returns the content on the output
13621 as-is. The variable keeps the value and the associated input type. The name of
13622 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013623 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013624 "sess" : the variable is shared with the whole session
13625 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013626 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013627 "req" : the variable is shared only during request processing,
13628 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013629 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013630 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013631
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013632sha1
13633 Converts a binary input sample to a SHA1 digest. The result is a binary
13634 sample with length of 20 bytes.
13635
Tim Duesterhusca097c12018-04-27 21:18:45 +020013636strcmp(<var>)
13637 Compares the contents of <var> with the input value of type string. Returns
13638 the result as a signed integer compatible with strcmp(3): 0 if both strings
13639 are identical. A value less than 0 if the left string is lexicographically
13640 smaller than the right string or if the left string is shorter. A value greater
13641 than 0 otherwise (right string greater than left string or the right string is
13642 shorter).
13643
13644 Example :
13645
13646 http-request set-var(txn.host) hdr(host)
13647 # Check whether the client is attempting domain fronting.
13648 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13649
13650
Willy Tarreau97707872015-01-27 15:12:13 +010013651sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013652 Subtracts <value> from the input value of type signed integer, and returns
13653 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013654 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013655 or a variable name. The name of the variable starts with an indication about
13656 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013657 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013658 "sess" : the variable is shared with the whole session
13659 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013660 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013661 "req" : the variable is shared only during request processing,
13662 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013663 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013664 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013665
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013666table_bytes_in_rate(<table>)
13667 Uses the string representation of the input sample to perform a look up in
13668 the specified table. If the key is not found in the table, integer value zero
13669 is returned. Otherwise the converter returns the average client-to-server
13670 bytes rate associated with the input sample in the designated table, measured
13671 in amount of bytes over the period configured in the table. See also the
13672 sc_bytes_in_rate sample fetch keyword.
13673
13674
13675table_bytes_out_rate(<table>)
13676 Uses the string representation of the input sample to perform a look up in
13677 the specified table. If the key is not found in the table, integer value zero
13678 is returned. Otherwise the converter returns the average server-to-client
13679 bytes rate associated with the input sample in the designated table, measured
13680 in amount of bytes over the period configured in the table. See also the
13681 sc_bytes_out_rate sample fetch keyword.
13682
13683table_conn_cnt(<table>)
13684 Uses the string representation of the input sample to perform a look up in
13685 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013686 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013687 connections associated with the input sample in the designated table. See
13688 also the sc_conn_cnt sample fetch keyword.
13689
13690table_conn_cur(<table>)
13691 Uses the string representation of the input sample to perform a look up in
13692 the specified table. If the key is not found in the table, integer value zero
13693 is returned. Otherwise the converter returns the current amount of concurrent
13694 tracked connections associated with the input sample in the designated table.
13695 See also the sc_conn_cur sample fetch keyword.
13696
13697table_conn_rate(<table>)
13698 Uses the string representation of the input sample to perform a look up in
13699 the specified table. If the key is not found in the table, integer value zero
13700 is returned. Otherwise the converter returns the average incoming connection
13701 rate associated with the input sample in the designated table. See also the
13702 sc_conn_rate sample fetch keyword.
13703
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013704table_gpt0(<table>)
13705 Uses the string representation of the input sample to perform a look up in
13706 the specified table. If the key is not found in the table, boolean value zero
13707 is returned. Otherwise the converter returns the current value of the first
13708 general purpose tag associated with the input sample in the designated table.
13709 See also the sc_get_gpt0 sample fetch keyword.
13710
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013711table_gpc0(<table>)
13712 Uses the string representation of the input sample to perform a look up in
13713 the specified table. If the key is not found in the table, integer value zero
13714 is returned. Otherwise the converter returns the current value of the first
13715 general purpose counter associated with the input sample in the designated
13716 table. See also the sc_get_gpc0 sample fetch keyword.
13717
13718table_gpc0_rate(<table>)
13719 Uses the string representation of the input sample to perform a look up in
13720 the specified table. If the key is not found in the table, integer value zero
13721 is returned. Otherwise the converter returns the frequency which the gpc0
13722 counter was incremented over the configured period in the table, associated
13723 with the input sample in the designated table. See also the sc_get_gpc0_rate
13724 sample fetch keyword.
13725
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013726table_gpc1(<table>)
13727 Uses the string representation of the input sample to perform a look up in
13728 the specified table. If the key is not found in the table, integer value zero
13729 is returned. Otherwise the converter returns the current value of the second
13730 general purpose counter associated with the input sample in the designated
13731 table. See also the sc_get_gpc1 sample fetch keyword.
13732
13733table_gpc1_rate(<table>)
13734 Uses the string representation of the input sample to perform a look up in
13735 the specified table. If the key is not found in the table, integer value zero
13736 is returned. Otherwise the converter returns the frequency which the gpc1
13737 counter was incremented over the configured period in the table, associated
13738 with the input sample in the designated table. See also the sc_get_gpc1_rate
13739 sample fetch keyword.
13740
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013741table_http_err_cnt(<table>)
13742 Uses the string representation of the input sample to perform a look up in
13743 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013744 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013745 errors associated with the input sample in the designated table. See also the
13746 sc_http_err_cnt sample fetch keyword.
13747
13748table_http_err_rate(<table>)
13749 Uses the string representation of the input sample to perform a look up in
13750 the specified table. If the key is not found in the table, integer value zero
13751 is returned. Otherwise the average rate of HTTP errors associated with the
13752 input sample in the designated table, measured in amount of errors over the
13753 period configured in the table. See also the sc_http_err_rate sample fetch
13754 keyword.
13755
13756table_http_req_cnt(<table>)
13757 Uses the string representation of the input sample to perform a look up in
13758 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013759 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013760 requests associated with the input sample in the designated table. See also
13761 the sc_http_req_cnt sample fetch keyword.
13762
13763table_http_req_rate(<table>)
13764 Uses the string representation of the input sample to perform a look up in
13765 the specified table. If the key is not found in the table, integer value zero
13766 is returned. Otherwise the average rate of HTTP requests associated with the
13767 input sample in the designated table, measured in amount of requests over the
13768 period configured in the table. See also the sc_http_req_rate sample fetch
13769 keyword.
13770
13771table_kbytes_in(<table>)
13772 Uses the string representation of the input sample to perform a look up in
13773 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013774 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013775 to-server data associated with the input sample in the designated table,
13776 measured in kilobytes. The test is currently performed on 32-bit integers,
13777 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13778 keyword.
13779
13780table_kbytes_out(<table>)
13781 Uses the string representation of the input sample to perform a look up in
13782 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013783 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013784 to-client data associated with the input sample in the designated table,
13785 measured in kilobytes. The test is currently performed on 32-bit integers,
13786 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13787 keyword.
13788
13789table_server_id(<table>)
13790 Uses the string representation of the input sample to perform a look up in
13791 the specified table. If the key is not found in the table, integer value zero
13792 is returned. Otherwise the converter returns the server ID associated with
13793 the input sample in the designated table. A server ID is associated to a
13794 sample by a "stick" rule when a connection to a server succeeds. A server ID
13795 zero means that no server is associated with this key.
13796
13797table_sess_cnt(<table>)
13798 Uses the string representation of the input sample to perform a look up in
13799 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013800 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013801 sessions associated with the input sample in the designated table. Note that
13802 a session here refers to an incoming connection being accepted by the
13803 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13804 keyword.
13805
13806table_sess_rate(<table>)
13807 Uses the string representation of the input sample to perform a look up in
13808 the specified table. If the key is not found in the table, integer value zero
13809 is returned. Otherwise the converter returns the average incoming session
13810 rate associated with the input sample in the designated table. Note that a
13811 session here refers to an incoming connection being accepted by the
13812 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13813 keyword.
13814
13815table_trackers(<table>)
13816 Uses the string representation of the input sample to perform a look up in
13817 the specified table. If the key is not found in the table, integer value zero
13818 is returned. Otherwise the converter returns the current amount of concurrent
13819 connections tracking the same key as the input sample in the designated
13820 table. It differs from table_conn_cur in that it does not rely on any stored
13821 information but on the table's reference count (the "use" value which is
13822 returned by "show table" on the CLI). This may sometimes be more suited for
13823 layer7 tracking. It can be used to tell a server how many concurrent
13824 connections there are from a given address for example. See also the
13825 sc_trackers sample fetch keyword.
13826
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013827upper
13828 Convert a string sample to upper case. This can only be placed after a string
13829 sample fetch function or after a transformation keyword returning a string
13830 type. The result is of type string.
13831
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013832url_dec
13833 Takes an url-encoded string provided as input and returns the decoded
13834 version as output. The input and the output are of type string.
13835
Christopher Faulet85d79c92016-11-09 16:54:56 +010013836unset-var(<var name>)
13837 Unsets a variable if the input content is defined. The name of the variable
13838 starts with an indication about its scope. The scopes allowed are:
13839 "proc" : the variable is shared with the whole process
13840 "sess" : the variable is shared with the whole session
13841 "txn" : the variable is shared with the transaction (request and
13842 response),
13843 "req" : the variable is shared only during request processing,
13844 "res" : the variable is shared only during response processing.
13845 This prefix is followed by a name. The separator is a '.'. The name may only
13846 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13847
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013848utime(<format>[,<offset>])
13849 Converts an integer supposed to contain a date since epoch to a string
13850 representing this date in UTC time using a format defined by the <format>
13851 string using strftime(3). The purpose is to allow any date format to be used
13852 in logs. An optional <offset> in seconds may be applied to the input date
13853 (positive or negative). See the strftime() man page for the format supported
13854 by your operating system. See also the ltime converter.
13855
13856 Example :
13857
13858 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013859 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013860 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13861
Marcin Deranek9631a282018-04-16 14:30:46 +020013862word(<index>,<delimiters>[,<count>])
13863 Extracts the nth word counting from the beginning (positive index) or from
13864 the end (negative index) considering given delimiters from an input string.
13865 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13866 Optionally you can specify <count> of words to extract (default: 1).
13867 Value of 0 indicates extraction of all remaining words.
13868
13869 Example :
13870 str(f1_f2_f3__f5),word(4,_) # f5
13871 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13872 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13873 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13874 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013875
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013876wt6([<avalanche>])
13877 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13878 hash function. Optionally, it is possible to apply a full avalanche hash
13879 function to the output if the optional <avalanche> argument equals 1. This
13880 converter uses the same functions as used by the various hash-based load
13881 balancing algorithms, so it will provide exactly the same results. It is
13882 mostly intended for debugging, but can be used as a stick-table entry to
13883 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013884 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13885 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013886
Willy Tarreau97707872015-01-27 15:12:13 +010013887xor(<value>)
13888 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013889 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013890 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013891 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013892 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013893 "sess" : the variable is shared with the whole session
13894 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013895 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013896 "req" : the variable is shared only during request processing,
13897 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013898 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013899 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013900
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013901xxh32([<seed>])
13902 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13903 variant of the XXHash hash function. This hash supports a seed which defaults
13904 to zero but a different value maybe passed as the <seed> argument. This hash
13905 is known to be very good and very fast so it can be used to hash URLs and/or
13906 URL parameters for use as stick-table keys to collect statistics with a low
13907 collision rate, though care must be taken as the algorithm is not considered
13908 as cryptographically secure.
13909
13910xxh64([<seed>])
13911 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13912 variant of the XXHash hash function. This hash supports a seed which defaults
13913 to zero but a different value maybe passed as the <seed> argument. This hash
13914 is known to be very good and very fast so it can be used to hash URLs and/or
13915 URL parameters for use as stick-table keys to collect statistics with a low
13916 collision rate, though care must be taken as the algorithm is not considered
13917 as cryptographically secure.
13918
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013919
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200139207.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013921--------------------------------------------
13922
13923A first set of sample fetch methods applies to internal information which does
13924not even relate to any client information. These ones are sometimes used with
13925"monitor-fail" directives to report an internal status to external watchers.
13926The sample fetch methods described in this section are usable anywhere.
13927
13928always_false : boolean
13929 Always returns the boolean "false" value. It may be used with ACLs as a
13930 temporary replacement for another one when adjusting configurations.
13931
13932always_true : boolean
13933 Always returns the boolean "true" value. It may be used with ACLs as a
13934 temporary replacement for another one when adjusting configurations.
13935
13936avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013937 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013938 divided by the number of active servers. The current backend is used if no
13939 backend is specified. This is very similar to "queue" except that the size of
13940 the farm is considered, in order to give a more accurate measurement of the
13941 time it may take for a new connection to be processed. The main usage is with
13942 ACL to return a sorry page to new users when it becomes certain they will get
13943 a degraded service, or to pass to the backend servers in a header so that
13944 they decide to work in degraded mode or to disable some functions to speed up
13945 the processing a bit. Note that in the event there would not be any active
13946 server anymore, twice the number of queued connections would be considered as
13947 the measured value. This is a fair estimate, as we expect one server to get
13948 back soon anyway, but we still prefer to send new traffic to another backend
13949 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13950 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013951
Willy Tarreau74ca5042013-06-11 23:12:07 +020013952be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013953 Applies to the number of currently established connections on the backend,
13954 possibly including the connection being evaluated. If no backend name is
13955 specified, the current one is used. But it is also possible to check another
13956 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013957 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
13958
13959be_conn_free([<backend>]) : integer
13960 Returns an integer value corresponding to the number of available connections
13961 across available servers in the backend. Queue slots are not included. Backup
13962 servers are also not included, unless all other servers are down. If no
13963 backend name is specified, the current one is used. But it is also possible
13964 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040013965 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
13966 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040013967
13968 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
13969 (meaning unlimited), then this fetch clearly does not make sense, in which
13970 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013971
Willy Tarreau74ca5042013-06-11 23:12:07 +020013972be_sess_rate([<backend>]) : integer
13973 Returns an integer value corresponding to the sessions creation rate on the
13974 backend, in number of new sessions per second. This is used with ACLs to
13975 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013976 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013977 online dictionary). It can also be useful to add this element to logs using a
13978 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013979
13980 Example :
13981 # Redirect to an error page if the dictionary is requested too often
13982 backend dynamic
13983 mode http
13984 acl being_scanned be_sess_rate gt 100
13985 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013986
Davor Ocelice9ed2812017-12-25 17:49:28 +010013987bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013988 Returns a binary chain. The input is the hexadecimal representation
13989 of the string.
13990
13991bool(<bool>) : bool
13992 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
13993 'false' and '0' are the same. 'true' and '1' are the same.
13994
Willy Tarreau74ca5042013-06-11 23:12:07 +020013995connslots([<backend>]) : integer
13996 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013997 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020013998 connections on all servers and the maximum queue size. This is probably only
13999 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014000
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014001 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014002 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014003 usage; see "use_backend" keyword) can be redirected to a different backend.
14004
Willy Tarreau55165fe2009-05-10 12:02:55 +020014005 'connslots' = number of available server connection slots, + number of
14006 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014007
Willy Tarreaua36af912009-10-10 12:02:45 +020014008 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014009 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014010 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014011 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014012 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014013 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014014 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014015
Willy Tarreau55165fe2009-05-10 12:02:55 +020014016 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14017 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014018 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014019 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014020
Willy Tarreau70fe9442018-11-22 16:07:39 +010014021cpu_calls : integer
14022 Returns the number of calls to the task processing the stream or current
14023 request since it was allocated. This number is reset for each new request on
14024 the same connections in case of HTTP keep-alive. This value should usually be
14025 low and stable (around 2 calls for a typically simple request) but may become
14026 high if some processing (compression, caching or analysis) is performed. This
14027 is purely for performance monitoring purposes.
14028
14029cpu_ns_avg : integer
14030 Returns the average number of nanoseconds spent in each call to the task
14031 processing the stream or current request. This number is reset for each new
14032 request on the same connections in case of HTTP keep-alive. This value
14033 indicates the overall cost of processing the request or the connection for
14034 each call. There is no good nor bad value but the time spent in a call
14035 automatically causes latency for other processing (see lat_ns_avg below),
14036 and may affect other connection's apparent response time. Certain operations
14037 like compression, complex regex matching or heavy Lua operations may directly
14038 affect this value, and having it in the logs will make it easier to spot the
14039 faulty processing that needs to be fixed to recover decent performance.
14040 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14041
14042cpu_ns_tot : integer
14043 Returns the total number of nanoseconds spent in each call to the task
14044 processing the stream or current request. This number is reset for each new
14045 request on the same connections in case of HTTP keep-alive. This value
14046 indicates the overall cost of processing the request or the connection for
14047 each call. There is no good nor bad value but the time spent in a call
14048 automatically causes latency for other processing (see lat_ns_avg below),
14049 induces CPU costs on the machine, and may affect other connection's apparent
14050 response time. Certain operations like compression, complex regex matching or
14051 heavy Lua operations may directly affect this value, and having it in the
14052 logs will make it easier to spot the faulty processing that needs to be fixed
14053 to recover decent performance. The value may be artificially high due to a
14054 high cpu_calls count, for example when processing many HTTP chunks, and for
14055 this reason it is often preferred to log cpu_ns_avg instead.
14056
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014057date([<offset>]) : integer
14058 Returns the current date as the epoch (number of seconds since 01/01/1970).
14059 If an offset value is specified, then it is a number of seconds that is added
14060 to the current date before returning the value. This is particularly useful
14061 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014062 It is useful combined with the http_date converter.
14063
14064 Example :
14065
14066 # set an expires header to now+1 hour in every response
14067 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014068
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014069date_us : integer
14070 Return the microseconds part of the date (the "second" part is returned by
14071 date sample). This sample is coherent with the date sample as it is comes
14072 from the same timeval structure.
14073
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014074distcc_body(<token>[,<occ>]) : binary
14075 Parses a distcc message and returns the body associated to occurrence #<occ>
14076 of the token <token>. Occurrences start at 1, and when unspecified, any may
14077 match though in practice only the first one is checked for now. This can be
14078 used to extract file names or arguments in files built using distcc through
14079 haproxy. Please refer to distcc's protocol documentation for the complete
14080 list of supported tokens.
14081
14082distcc_param(<token>[,<occ>]) : integer
14083 Parses a distcc message and returns the parameter associated to occurrence
14084 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14085 any may match though in practice only the first one is checked for now. This
14086 can be used to extract certain information such as the protocol version, the
14087 file size or the argument in files built using distcc through haproxy.
14088 Another use case consists in waiting for the start of the preprocessed file
14089 contents before connecting to the server to avoid keeping idle connections.
14090 Please refer to distcc's protocol documentation for the complete list of
14091 supported tokens.
14092
14093 Example :
14094 # wait up to 20s for the pre-processed file to be uploaded
14095 tcp-request inspect-delay 20s
14096 tcp-request content accept if { distcc_param(DOTI) -m found }
14097 # send large files to the big farm
14098 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14099
Willy Tarreau595ec542013-06-12 21:34:28 +020014100env(<name>) : string
14101 Returns a string containing the value of environment variable <name>. As a
14102 reminder, environment variables are per-process and are sampled when the
14103 process starts. This can be useful to pass some information to a next hop
14104 server, or with ACLs to take specific action when the process is started a
14105 certain way.
14106
14107 Examples :
14108 # Pass the Via header to next hop with the local hostname in it
14109 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14110
14111 # reject cookie-less requests when the STOP environment variable is set
14112 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14113
Willy Tarreau74ca5042013-06-11 23:12:07 +020014114fe_conn([<frontend>]) : integer
14115 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014116 possibly including the connection being evaluated. If no frontend name is
14117 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014118 frontend. It can be used to return a sorry page before hard-blocking, or to
14119 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014120 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014121 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14122 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014123
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014124fe_req_rate([<frontend>]) : integer
14125 Returns an integer value corresponding to the number of HTTP requests per
14126 second sent to a frontend. This number can differ from "fe_sess_rate" in
14127 situations where client-side keep-alive is enabled.
14128
Willy Tarreau74ca5042013-06-11 23:12:07 +020014129fe_sess_rate([<frontend>]) : integer
14130 Returns an integer value corresponding to the sessions creation rate on the
14131 frontend, in number of new sessions per second. This is used with ACLs to
14132 limit the incoming session rate to an acceptable range in order to prevent
14133 abuse of service at the earliest moment, for example when combined with other
14134 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14135 down below the limit. It can also be useful to add this element to logs using
14136 a log-format directive. See also the "rate-limit sessions" directive for use
14137 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014138
14139 Example :
14140 # This frontend limits incoming mails to 10/s with a max of 100
14141 # concurrent connections. We accept any connection below 10/s, and
14142 # force excess clients to wait for 100 ms. Since clients are limited to
14143 # 100 max, there cannot be more than 10 incoming mails per second.
14144 frontend mail
14145 bind :25
14146 mode tcp
14147 maxconn 100
14148 acl too_fast fe_sess_rate ge 10
14149 tcp-request inspect-delay 100ms
14150 tcp-request content accept if ! too_fast
14151 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014152
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014153hostname : string
14154 Returns the system hostname.
14155
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014156int(<integer>) : signed integer
14157 Returns a signed integer.
14158
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014159ipv4(<ipv4>) : ipv4
14160 Returns an ipv4.
14161
14162ipv6(<ipv6>) : ipv6
14163 Returns an ipv6.
14164
Willy Tarreau70fe9442018-11-22 16:07:39 +010014165lat_ns_avg : integer
14166 Returns the average number of nanoseconds spent between the moment the task
14167 handling the stream is woken up and the moment it is effectively called. This
14168 number is reset for each new request on the same connections in case of HTTP
14169 keep-alive. This value indicates the overall latency inflicted to the current
14170 request by all other requests being processed in parallel, and is a direct
14171 indicator of perceived performance due to noisy neighbours. In order to keep
14172 the value low, it is possible to reduce the scheduler's run queue depth using
14173 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14174 once using "tune.maxpollevents", to decrease the stream's nice value using
14175 the "nice" option on the "bind" lines or in the frontend, or to look for
14176 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14177 whose processing needs to be adjusted or fixed. Compression of large buffers
14178 could be a culprit, like heavy regex or long lists of regex.
14179 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14180
14181lat_ns_tot : integer
14182 Returns the total number of nanoseconds spent between the moment the task
14183 handling the stream is woken up and the moment it is effectively called. This
14184 number is reset for each new request on the same connections in case of HTTP
14185 keep-alive. This value indicates the overall latency inflicted to the current
14186 request by all other requests being processed in parallel, and is a direct
14187 indicator of perceived performance due to noisy neighbours. In order to keep
14188 the value low, it is possible to reduce the scheduler's run queue depth using
14189 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14190 once using "tune.maxpollevents", to decrease the stream's nice value using
14191 the "nice" option on the "bind" lines or in the frontend, or to look for
14192 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14193 whose processing needs to be adjusted or fixed. Compression of large buffers
14194 could be a culprit, like heavy regex or long lists of regex. Note: while it
14195 may intuitively seem that the total latency adds to a transfer time, it is
14196 almost never true because while a task waits for the CPU, network buffers
14197 continue to fill up and the next call will process more at once. The value
14198 may be artificially high due to a high cpu_calls count, for example when
14199 processing many HTTP chunks, and for this reason it is often preferred to log
14200 lat_ns_avg instead, which is a more relevant performance indicator.
14201
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014202meth(<method>) : method
14203 Returns a method.
14204
Willy Tarreau0f30d262014-11-24 16:02:05 +010014205nbproc : integer
14206 Returns an integer value corresponding to the number of processes that were
14207 started (it equals the global "nbproc" setting). This is useful for logging
14208 and debugging purposes.
14209
Willy Tarreau74ca5042013-06-11 23:12:07 +020014210nbsrv([<backend>]) : integer
14211 Returns an integer value corresponding to the number of usable servers of
14212 either the current backend or the named backend. This is mostly used with
14213 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014214 switch to an alternate backend when the number of servers is too low to
14215 to handle some load. It is useful to report a failure when combined with
14216 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014217
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014218prio_class : integer
14219 Returns the priority class of the current session for http mode or connection
14220 for tcp mode. The value will be that set by the last call to "http-request
14221 set-priority-class" or "tcp-request content set-priority-class".
14222
14223prio_offset : integer
14224 Returns the priority offset of the current session for http mode or
14225 connection for tcp mode. The value will be that set by the last call to
14226 "http-request set-priority-offset" or "tcp-request content
14227 set-priority-offset".
14228
Willy Tarreau0f30d262014-11-24 16:02:05 +010014229proc : integer
14230 Returns an integer value corresponding to the position of the process calling
14231 the function, between 1 and global.nbproc. This is useful for logging and
14232 debugging purposes.
14233
Willy Tarreau74ca5042013-06-11 23:12:07 +020014234queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014235 Returns the total number of queued connections of the designated backend,
14236 including all the connections in server queues. If no backend name is
14237 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014238 one. This is useful with ACLs or to pass statistics to backend servers. This
14239 can be used to take actions when queuing goes above a known level, generally
14240 indicating a surge of traffic or a massive slowdown on the servers. One
14241 possible action could be to reject new users but still accept old ones. See
14242 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14243
Willy Tarreau84310e22014-02-14 11:59:04 +010014244rand([<range>]) : integer
14245 Returns a random integer value within a range of <range> possible values,
14246 starting at zero. If the range is not specified, it defaults to 2^32, which
14247 gives numbers between 0 and 4294967295. It can be useful to pass some values
14248 needed to take some routing decisions for example, or just for debugging
14249 purposes. This random must not be used for security purposes.
14250
Willy Tarreau74ca5042013-06-11 23:12:07 +020014251srv_conn([<backend>/]<server>) : integer
14252 Returns an integer value corresponding to the number of currently established
14253 connections on the designated server, possibly including the connection being
14254 evaluated. If <backend> is omitted, then the server is looked up in the
14255 current backend. It can be used to use a specific farm when one server is
14256 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014257 connections with it. See also the "fe_conn", "be_conn", "queue", and
14258 "srv_conn_free" fetch methods.
14259
14260srv_conn_free([<backend>/]<server>) : integer
14261 Returns an integer value corresponding to the number of available connections
14262 on the designated server, possibly including the connection being evaluated.
14263 The value does not include queue slots. If <backend> is omitted, then the
14264 server is looked up in the current backend. It can be used to use a specific
14265 farm when one server is full, or to inform the server about our view of the
14266 number of active connections with it. See also the "be_conn_free" and
14267 "srv_conn" fetch methods.
14268
14269 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14270 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014271
14272srv_is_up([<backend>/]<server>) : boolean
14273 Returns true when the designated server is UP, and false when it is either
14274 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14275 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014276 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014277 availability). Another possible use which is more of a hack consists in
14278 using dummy servers as boolean variables that can be enabled or disabled from
14279 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14280
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014281srv_queue([<backend>/]<server>) : integer
14282 Returns an integer value corresponding to the number of connections currently
14283 pending in the designated server's queue. If <backend> is omitted, then the
14284 server is looked up in the current backend. It can sometimes be used together
14285 with the "use-server" directive to force to use a known faster server when it
14286 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14287 fetch methods.
14288
Willy Tarreau74ca5042013-06-11 23:12:07 +020014289srv_sess_rate([<backend>/]<server>) : integer
14290 Returns an integer corresponding to the sessions creation rate on the
14291 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014292 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014293 used with ACLs but can make sense with logs too. This is used to switch to an
14294 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014295 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014296 overloading servers).
14297
14298 Example :
14299 # Redirect to a separate back
14300 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14301 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14302 use_backend be2 if srv1_full or srv2_full
14303
Willy Tarreau0f30d262014-11-24 16:02:05 +010014304stopping : boolean
14305 Returns TRUE if the process calling the function is currently stopping. This
14306 can be useful for logging, or for relaxing certain checks or helping close
14307 certain connections upon graceful shutdown.
14308
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014309str(<string>) : string
14310 Returns a string.
14311
Willy Tarreau74ca5042013-06-11 23:12:07 +020014312table_avl([<table>]) : integer
14313 Returns the total number of available entries in the current proxy's
14314 stick-table or in the designated stick-table. See also table_cnt.
14315
14316table_cnt([<table>]) : integer
14317 Returns the total number of entries currently in use in the current proxy's
14318 stick-table or in the designated stick-table. See also src_conn_cnt and
14319 table_avl for other entry counting methods.
14320
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014321thread : integer
14322 Returns an integer value corresponding to the position of the thread calling
14323 the function, between 0 and (global.nbthread-1). This is useful for logging
14324 and debugging purposes.
14325
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014326var(<var-name>) : undefined
14327 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014328 sample fetch fails. The name of the variable starts with an indication
14329 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014330 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014331 "sess" : the variable is shared with the whole session
14332 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014333 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014334 "req" : the variable is shared only during request processing,
14335 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014336 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014337 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014338
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200143397.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014340----------------------------------
14341
14342The layer 4 usually describes just the transport layer which in haproxy is
14343closest to the connection, where no content is yet made available. The fetch
14344methods described here are usable as low as the "tcp-request connection" rule
14345sets unless they require some future information. Those generally include
14346TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014347the incoming connection. For retrieving a value from a sticky counters, the
14348counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014349"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14350used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14351can be specified as the first integer argument when using the "sc_" prefix.
14352Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14353table may be specified with the "sc*" form, in which case the currently
14354tracked key will be looked up into this alternate table instead of the table
14355currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014356
Jérôme Magnin86577422018-12-07 09:03:11 +010014357bc_http_major: integer
14358 Returns the backend connection's HTTP major version encoding, which may be 1
14359 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14360 encoding and not the version present in the request header.
14361
Willy Tarreau74ca5042013-06-11 23:12:07 +020014362be_id : integer
14363 Returns an integer containing the current backend's id. It can be used in
14364 frontends with responses to check which backend processed the request.
14365
Marcin Deranekd2471c22016-12-12 14:08:05 +010014366be_name : string
14367 Returns a string containing the current backend's name. It can be used in
14368 frontends with responses to check which backend processed the request.
14369
Willy Tarreau74ca5042013-06-11 23:12:07 +020014370dst : ip
14371 This is the destination IPv4 address of the connection on the client side,
14372 which is the address the client connected to. It can be useful when running
14373 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14374 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
14375 RFC 4291.
14376
14377dst_conn : integer
14378 Returns an integer value corresponding to the number of currently established
14379 connections on the same socket including the one being evaluated. It is
14380 normally used with ACLs but can as well be used to pass the information to
14381 servers in an HTTP header or in logs. It can be used to either return a sorry
14382 page before hard-blocking, or to use a specific backend to drain new requests
14383 when the socket is considered saturated. This offers the ability to assign
14384 different limits to different listening ports or addresses. See also the
14385 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014386
Willy Tarreau16e01562016-08-09 16:46:18 +020014387dst_is_local : boolean
14388 Returns true if the destination address of the incoming connection is local
14389 to the system, or false if the address doesn't exist on the system, meaning
14390 that it was intercepted in transparent mode. It can be useful to apply
14391 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014392 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014393 be delivered only on this address, or SSH access could be locally redirected.
14394 Please note that the check involves a few system calls, so it's better to do
14395 it only once per connection.
14396
Willy Tarreau74ca5042013-06-11 23:12:07 +020014397dst_port : integer
14398 Returns an integer value corresponding to the destination TCP port of the
14399 connection on the client side, which is the port the client connected to.
14400 This might be used when running in transparent mode, when assigning dynamic
14401 ports to some clients for a whole application session, to stick all users to
14402 a same server, or to pass the destination port information to a server using
14403 an HTTP header.
14404
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014405fc_http_major : integer
14406 Reports the front connection's HTTP major version encoding, which may be 1
14407 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14408 encoding and not on the version present in the request header.
14409
Emeric Brun4f603012017-01-05 15:11:44 +010014410fc_rcvd_proxy : boolean
14411 Returns true if the client initiated the connection with a PROXY protocol
14412 header.
14413
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014414fc_rtt(<unit>) : integer
14415 Returns the Round Trip Time (RTT) measured by the kernel for the client
14416 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14417 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14418 connection is not established, if the connection is not TCP or if the
14419 operating system does not support TCP_INFO, for example Linux kernels before
14420 2.4, the sample fetch fails.
14421
14422fc_rttvar(<unit>) : integer
14423 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14424 client connection. <unit> is facultative, by default the unit is milliseconds.
14425 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14426 server connection is not established, if the connection is not TCP or if the
14427 operating system does not support TCP_INFO, for example Linux kernels before
14428 2.4, the sample fetch fails.
14429
Joe Williams30fcd392016-08-10 07:06:44 -070014430fc_unacked(<unit>) : integer
14431 Returns the unacked counter measured by the kernel for the client connection.
14432 If the server connection is not established, if the connection is not TCP or
14433 if the operating system does not support TCP_INFO, for example Linux kernels
14434 before 2.4, the sample fetch fails.
14435
14436fc_sacked(<unit>) : integer
14437 Returns the sacked counter measured by the kernel for the client connection.
14438 If the server connection is not established, if the connection is not TCP or
14439 if the operating system does not support TCP_INFO, for example Linux kernels
14440 before 2.4, the sample fetch fails.
14441
14442fc_retrans(<unit>) : integer
14443 Returns the retransmits counter measured by the kernel for the client
14444 connection. If the server connection is not established, if the connection is
14445 not TCP or if the operating system does not support TCP_INFO, for example
14446 Linux kernels before 2.4, the sample fetch fails.
14447
14448fc_fackets(<unit>) : integer
14449 Returns the fack counter measured by the kernel for the client
14450 connection. If the server connection is not established, if the connection is
14451 not TCP or if the operating system does not support TCP_INFO, for example
14452 Linux kernels before 2.4, the sample fetch fails.
14453
14454fc_lost(<unit>) : integer
14455 Returns the lost counter measured by the kernel for the client
14456 connection. If the server connection is not established, if the connection is
14457 not TCP or if the operating system does not support TCP_INFO, for example
14458 Linux kernels before 2.4, the sample fetch fails.
14459
14460fc_reordering(<unit>) : integer
14461 Returns the reordering counter measured by the kernel for the client
14462 connection. If the server connection is not established, if the connection is
14463 not TCP or if the operating system does not support TCP_INFO, for example
14464 Linux kernels before 2.4, the sample fetch fails.
14465
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014466fe_defbe : string
14467 Returns a string containing the frontend's default backend name. It can be
14468 used in frontends to check which backend will handle requests by default.
14469
Willy Tarreau74ca5042013-06-11 23:12:07 +020014470fe_id : integer
14471 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014472 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014473 coming via a same frontend to the same server.
14474
Marcin Deranekd2471c22016-12-12 14:08:05 +010014475fe_name : string
14476 Returns a string containing the current frontend's name. It can be used in
14477 backends to check from which frontend it was called, or to stick all users
14478 coming via a same frontend to the same server.
14479
Cyril Bonté62ba8702014-04-22 23:52:25 +020014480sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014481sc0_bytes_in_rate([<table>]) : integer
14482sc1_bytes_in_rate([<table>]) : integer
14483sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014484 Returns the average client-to-server bytes rate from the currently tracked
14485 counters, measured in amount of bytes over the period configured in the
14486 table. See also src_bytes_in_rate.
14487
Cyril Bonté62ba8702014-04-22 23:52:25 +020014488sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014489sc0_bytes_out_rate([<table>]) : integer
14490sc1_bytes_out_rate([<table>]) : integer
14491sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014492 Returns the average server-to-client bytes rate from the currently tracked
14493 counters, measured in amount of bytes over the period configured in the
14494 table. See also src_bytes_out_rate.
14495
Cyril Bonté62ba8702014-04-22 23:52:25 +020014496sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014497sc0_clr_gpc0([<table>]) : integer
14498sc1_clr_gpc0([<table>]) : integer
14499sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014500 Clears the first General Purpose Counter associated to the currently tracked
14501 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014502 stored value is zero, so first invocation will always return zero. This is
14503 typically used as a second ACL in an expression in order to mark a connection
14504 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014505
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014506 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014507 # block if 5 consecutive requests continue to come faster than 10 sess
14508 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014509 acl abuse sc0_http_req_rate gt 10
14510 acl kill sc0_inc_gpc0 gt 5
14511 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014512 tcp-request connection accept if !abuse save
14513 tcp-request connection reject if abuse kill
14514
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014515sc_clr_gpc1(<ctr>[,<table>]) : integer
14516sc0_clr_gpc1([<table>]) : integer
14517sc1_clr_gpc1([<table>]) : integer
14518sc2_clr_gpc1([<table>]) : integer
14519 Clears the second General Purpose Counter associated to the currently tracked
14520 counters, and returns its previous value. Before the first invocation, the
14521 stored value is zero, so first invocation will always return zero. This is
14522 typically used as a second ACL in an expression in order to mark a connection
14523 when a first ACL was verified.
14524
Cyril Bonté62ba8702014-04-22 23:52:25 +020014525sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014526sc0_conn_cnt([<table>]) : integer
14527sc1_conn_cnt([<table>]) : integer
14528sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014529 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014530 counters. See also src_conn_cnt.
14531
Cyril Bonté62ba8702014-04-22 23:52:25 +020014532sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014533sc0_conn_cur([<table>]) : integer
14534sc1_conn_cur([<table>]) : integer
14535sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014536 Returns the current amount of concurrent connections tracking the same
14537 tracked counters. This number is automatically incremented when tracking
14538 begins and decremented when tracking stops. See also src_conn_cur.
14539
Cyril Bonté62ba8702014-04-22 23:52:25 +020014540sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014541sc0_conn_rate([<table>]) : integer
14542sc1_conn_rate([<table>]) : integer
14543sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014544 Returns the average connection rate from the currently tracked counters,
14545 measured in amount of connections over the period configured in the table.
14546 See also src_conn_rate.
14547
Cyril Bonté62ba8702014-04-22 23:52:25 +020014548sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014549sc0_get_gpc0([<table>]) : integer
14550sc1_get_gpc0([<table>]) : integer
14551sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014552 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014553 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014554
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014555sc_get_gpc1(<ctr>[,<table>]) : integer
14556sc0_get_gpc1([<table>]) : integer
14557sc1_get_gpc1([<table>]) : integer
14558sc2_get_gpc1([<table>]) : integer
14559 Returns the value of the second General Purpose Counter associated to the
14560 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14561
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014562sc_get_gpt0(<ctr>[,<table>]) : integer
14563sc0_get_gpt0([<table>]) : integer
14564sc1_get_gpt0([<table>]) : integer
14565sc2_get_gpt0([<table>]) : integer
14566 Returns the value of the first General Purpose Tag associated to the
14567 currently tracked counters. See also src_get_gpt0.
14568
Cyril Bonté62ba8702014-04-22 23:52:25 +020014569sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014570sc0_gpc0_rate([<table>]) : integer
14571sc1_gpc0_rate([<table>]) : integer
14572sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014573 Returns the average increment rate of the first General Purpose Counter
14574 associated to the currently tracked counters. It reports the frequency
14575 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014576 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14577 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14578 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014579
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014580sc_gpc1_rate(<ctr>[,<table>]) : integer
14581sc0_gpc1_rate([<table>]) : integer
14582sc1_gpc1_rate([<table>]) : integer
14583sc2_gpc1_rate([<table>]) : integer
14584 Returns the average increment rate of the second General Purpose Counter
14585 associated to the currently tracked counters. It reports the frequency
14586 which the gpc1 counter was incremented over the configured period. See also
14587 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14588 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14589 be returned, as "gpc1" only holds the event count.
14590
Cyril Bonté62ba8702014-04-22 23:52:25 +020014591sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014592sc0_http_err_cnt([<table>]) : integer
14593sc1_http_err_cnt([<table>]) : integer
14594sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014595 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014596 counters. This includes the both request errors and 4xx error responses.
14597 See also src_http_err_cnt.
14598
Cyril Bonté62ba8702014-04-22 23:52:25 +020014599sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014600sc0_http_err_rate([<table>]) : integer
14601sc1_http_err_rate([<table>]) : integer
14602sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014603 Returns the average rate of HTTP errors from the currently tracked counters,
14604 measured in amount of errors over the period configured in the table. This
14605 includes the both request errors and 4xx error responses. See also
14606 src_http_err_rate.
14607
Cyril Bonté62ba8702014-04-22 23:52:25 +020014608sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014609sc0_http_req_cnt([<table>]) : integer
14610sc1_http_req_cnt([<table>]) : integer
14611sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014612 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014613 counters. This includes every started request, valid or not. See also
14614 src_http_req_cnt.
14615
Cyril Bonté62ba8702014-04-22 23:52:25 +020014616sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014617sc0_http_req_rate([<table>]) : integer
14618sc1_http_req_rate([<table>]) : integer
14619sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014620 Returns the average rate of HTTP requests from the currently tracked
14621 counters, measured in amount of requests over the period configured in
14622 the table. This includes every started request, valid or not. See also
14623 src_http_req_rate.
14624
Cyril Bonté62ba8702014-04-22 23:52:25 +020014625sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014626sc0_inc_gpc0([<table>]) : integer
14627sc1_inc_gpc0([<table>]) : integer
14628sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014629 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014630 tracked counters, and returns its new value. Before the first invocation,
14631 the stored value is zero, so first invocation will increase it to 1 and will
14632 return 1. This is typically used as a second ACL in an expression in order
14633 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014634
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014635 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014636 acl abuse sc0_http_req_rate gt 10
14637 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014638 tcp-request connection reject if abuse kill
14639
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014640sc_inc_gpc1(<ctr>[,<table>]) : integer
14641sc0_inc_gpc1([<table>]) : integer
14642sc1_inc_gpc1([<table>]) : integer
14643sc2_inc_gpc1([<table>]) : integer
14644 Increments the second General Purpose Counter associated to the currently
14645 tracked counters, and returns its new value. Before the first invocation,
14646 the stored value is zero, so first invocation will increase it to 1 and will
14647 return 1. This is typically used as a second ACL in an expression in order
14648 to mark a connection when a first ACL was verified.
14649
Cyril Bonté62ba8702014-04-22 23:52:25 +020014650sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014651sc0_kbytes_in([<table>]) : integer
14652sc1_kbytes_in([<table>]) : integer
14653sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014654 Returns the total amount of client-to-server data from the currently tracked
14655 counters, measured in kilobytes. The test is currently performed on 32-bit
14656 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014657
Cyril Bonté62ba8702014-04-22 23:52:25 +020014658sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014659sc0_kbytes_out([<table>]) : integer
14660sc1_kbytes_out([<table>]) : integer
14661sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014662 Returns the total amount of server-to-client data from the currently tracked
14663 counters, measured in kilobytes. The test is currently performed on 32-bit
14664 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014665
Cyril Bonté62ba8702014-04-22 23:52:25 +020014666sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014667sc0_sess_cnt([<table>]) : integer
14668sc1_sess_cnt([<table>]) : integer
14669sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014670 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014671 into sessions, which means that they were accepted by a "tcp-request
14672 connection" rule, from the currently tracked counters. A backend may count
14673 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014674 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014675 with the client. See also src_sess_cnt.
14676
Cyril Bonté62ba8702014-04-22 23:52:25 +020014677sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014678sc0_sess_rate([<table>]) : integer
14679sc1_sess_rate([<table>]) : integer
14680sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014681 Returns the average session rate from the currently tracked counters,
14682 measured in amount of sessions over the period configured in the table. A
14683 session is a connection that got past the early "tcp-request connection"
14684 rules. A backend may count more sessions than connections because each
14685 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014686 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014687
Cyril Bonté62ba8702014-04-22 23:52:25 +020014688sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014689sc0_tracked([<table>]) : boolean
14690sc1_tracked([<table>]) : boolean
14691sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014692 Returns true if the designated session counter is currently being tracked by
14693 the current session. This can be useful when deciding whether or not we want
14694 to set some values in a header passed to the server.
14695
Cyril Bonté62ba8702014-04-22 23:52:25 +020014696sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014697sc0_trackers([<table>]) : integer
14698sc1_trackers([<table>]) : integer
14699sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014700 Returns the current amount of concurrent connections tracking the same
14701 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014702 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014703 that it does not rely on any stored information but on the table's reference
14704 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014705 may sometimes be more suited for layer7 tracking. It can be used to tell a
14706 server how many concurrent connections there are from a given address for
14707 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014708
Willy Tarreau74ca5042013-06-11 23:12:07 +020014709so_id : integer
14710 Returns an integer containing the current listening socket's id. It is useful
14711 in frontends involving many "bind" lines, or to stick all users coming via a
14712 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014713
Willy Tarreau74ca5042013-06-11 23:12:07 +020014714src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014715 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014716 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14717 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14718 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014719 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14720 directive is used, it can be the address of a client behind another
14721 PROXY-protocol compatible component for all rule sets except
14722 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014723
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014724 Example:
14725 # add an HTTP header in requests with the originating address' country
14726 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14727
Willy Tarreau74ca5042013-06-11 23:12:07 +020014728src_bytes_in_rate([<table>]) : integer
14729 Returns the average bytes rate from the incoming connection's source address
14730 in the current proxy's stick-table or in the designated stick-table, measured
14731 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014732 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014733
Willy Tarreau74ca5042013-06-11 23:12:07 +020014734src_bytes_out_rate([<table>]) : integer
14735 Returns the average bytes rate to the incoming connection's source address in
14736 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014737 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014738 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014739
Willy Tarreau74ca5042013-06-11 23:12:07 +020014740src_clr_gpc0([<table>]) : integer
14741 Clears the first General Purpose Counter associated to the incoming
14742 connection's source address in the current proxy's stick-table or in the
14743 designated stick-table, and returns its previous value. If the address is not
14744 found, an entry is created and 0 is returned. This is typically used as a
14745 second ACL in an expression in order to mark a connection when a first ACL
14746 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014747
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014748 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014749 # block if 5 consecutive requests continue to come faster than 10 sess
14750 # per second, and reset the counter as soon as the traffic slows down.
14751 acl abuse src_http_req_rate gt 10
14752 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014753 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014754 tcp-request connection accept if !abuse save
14755 tcp-request connection reject if abuse kill
14756
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014757src_clr_gpc1([<table>]) : integer
14758 Clears the second General Purpose Counter associated to the incoming
14759 connection's source address in the current proxy's stick-table or in the
14760 designated stick-table, and returns its previous value. If the address is not
14761 found, an entry is created and 0 is returned. This is typically used as a
14762 second ACL in an expression in order to mark a connection when a first ACL
14763 was verified.
14764
Willy Tarreau74ca5042013-06-11 23:12:07 +020014765src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014766 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014767 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014768 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014769 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014770
Willy Tarreau74ca5042013-06-11 23:12:07 +020014771src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014772 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014773 current incoming connection's source address in the current proxy's
14774 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014775 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014776
Willy Tarreau74ca5042013-06-11 23:12:07 +020014777src_conn_rate([<table>]) : integer
14778 Returns the average connection rate from the incoming connection's source
14779 address in the current proxy's stick-table or in the designated stick-table,
14780 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014781 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014782
Willy Tarreau74ca5042013-06-11 23:12:07 +020014783src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014784 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014785 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014786 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014787 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014788
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014789src_get_gpc1([<table>]) : integer
14790 Returns the value of the second General Purpose Counter associated to the
14791 incoming connection's source address in the current proxy's stick-table or in
14792 the designated stick-table. If the address is not found, zero is returned.
14793 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14794
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014795src_get_gpt0([<table>]) : integer
14796 Returns the value of the first General Purpose Tag associated to the
14797 incoming connection's source address in the current proxy's stick-table or in
14798 the designated stick-table. If the address is not found, zero is returned.
14799 See also sc/sc0/sc1/sc2_get_gpt0.
14800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014802 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014803 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014804 stick-table or in the designated stick-table. It reports the frequency
14805 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014806 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14807 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14808 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014809
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014810src_gpc1_rate([<table>]) : integer
14811 Returns the average increment rate of the second General Purpose Counter
14812 associated to the incoming connection's source address in the current proxy's
14813 stick-table or in the designated stick-table. It reports the frequency
14814 which the gpc1 counter was incremented over the configured period. See also
14815 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14816 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14817 be returned, as "gpc1" only holds the event count.
14818
Willy Tarreau74ca5042013-06-11 23:12:07 +020014819src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014820 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014821 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014822 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014823 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014825
Willy Tarreau74ca5042013-06-11 23:12:07 +020014826src_http_err_rate([<table>]) : integer
14827 Returns the average rate of HTTP errors from the incoming connection's source
14828 address in the current proxy's stick-table or in the designated stick-table,
14829 measured in amount of errors over the period configured in the table. This
14830 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014831 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014832
Willy Tarreau74ca5042013-06-11 23:12:07 +020014833src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014834 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014835 source address in the current proxy's stick-table or in the designated stick-
14836 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014837 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014838
Willy Tarreau74ca5042013-06-11 23:12:07 +020014839src_http_req_rate([<table>]) : integer
14840 Returns the average rate of HTTP requests from the incoming connection's
14841 source address in the current proxy's stick-table or in the designated stick-
14842 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014843 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014844 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014845
Willy Tarreau74ca5042013-06-11 23:12:07 +020014846src_inc_gpc0([<table>]) : integer
14847 Increments the first General Purpose Counter associated to the incoming
14848 connection's source address in the current proxy's stick-table or in the
14849 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014850 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014851 This is typically used as a second ACL in an expression in order to mark a
14852 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014853
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014854 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014855 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014856 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014857 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014858
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014859src_inc_gpc1([<table>]) : integer
14860 Increments the second General Purpose Counter associated to the incoming
14861 connection's source address in the current proxy's stick-table or in the
14862 designated stick-table, and returns its new value. If the address is not
14863 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14864 This is typically used as a second ACL in an expression in order to mark a
14865 connection when a first ACL was verified.
14866
Willy Tarreau16e01562016-08-09 16:46:18 +020014867src_is_local : boolean
14868 Returns true if the source address of the incoming connection is local to the
14869 system, or false if the address doesn't exist on the system, meaning that it
14870 comes from a remote machine. Note that UNIX addresses are considered local.
14871 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014872 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014873 note that the check involves a few system calls, so it's better to do it only
14874 once per connection.
14875
Willy Tarreau74ca5042013-06-11 23:12:07 +020014876src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014877 Returns the total amount of data received from the incoming connection's
14878 source address in the current proxy's stick-table or in the designated
14879 stick-table, measured in kilobytes. If the address is not found, zero is
14880 returned. The test is currently performed on 32-bit integers, which limits
14881 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014882
Willy Tarreau74ca5042013-06-11 23:12:07 +020014883src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014884 Returns the total amount of data sent to the incoming connection's source
14885 address in the current proxy's stick-table or in the designated stick-table,
14886 measured in kilobytes. If the address is not found, zero is returned. The
14887 test is currently performed on 32-bit integers, which limits values to 4
14888 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014889
Willy Tarreau74ca5042013-06-11 23:12:07 +020014890src_port : integer
14891 Returns an integer value corresponding to the TCP source port of the
14892 connection on the client side, which is the port the client connected from.
14893 Usage of this function is very limited as modern protocols do not care much
14894 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014895
Willy Tarreau74ca5042013-06-11 23:12:07 +020014896src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014897 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014898 connection's source IPv4 address in the current proxy's stick-table or in the
14899 designated stick-table, that were transformed into sessions, which means that
14900 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014901 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014902
Willy Tarreau74ca5042013-06-11 23:12:07 +020014903src_sess_rate([<table>]) : integer
14904 Returns the average session rate from the incoming connection's source
14905 address in the current proxy's stick-table or in the designated stick-table,
14906 measured in amount of sessions over the period configured in the table. A
14907 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014908 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014909
Willy Tarreau74ca5042013-06-11 23:12:07 +020014910src_updt_conn_cnt([<table>]) : integer
14911 Creates or updates the entry associated to the incoming connection's source
14912 address in the current proxy's stick-table or in the designated stick-table.
14913 This table must be configured to store the "conn_cnt" data type, otherwise
14914 the match will be ignored. The current count is incremented by one, and the
14915 expiration timer refreshed. The updated count is returned, so this match
14916 can't return zero. This was used to reject service abusers based on their
14917 source address. Note: it is recommended to use the more complete "track-sc*"
14918 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014919
14920 Example :
14921 # This frontend limits incoming SSH connections to 3 per 10 second for
14922 # each source address, and rejects excess connections until a 10 second
14923 # silence is observed. At most 20 addresses are tracked.
14924 listen ssh
14925 bind :22
14926 mode tcp
14927 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014928 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014929 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014930 server local 127.0.0.1:22
14931
Willy Tarreau74ca5042013-06-11 23:12:07 +020014932srv_id : integer
14933 Returns an integer containing the server's id when processing the response.
14934 While it's almost only used with ACLs, it may be used for logging or
14935 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014936
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200149377.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014938----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014939
Willy Tarreau74ca5042013-06-11 23:12:07 +020014940The layer 5 usually describes just the session layer which in haproxy is
14941closest to the session once all the connection handshakes are finished, but
14942when no content is yet made available. The fetch methods described here are
14943usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014944future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014945
Ben Shillitof25e8e52016-12-02 14:25:37 +00001494651d.all(<prop>[,<prop>*]) : string
14947 Returns values for the properties requested as a string, where values are
14948 separated by the delimiter specified with "51degrees-property-separator".
14949 The device is identified using all the important HTTP headers from the
14950 request. The function can be passed up to five property names, and if a
14951 property name can't be found, the value "NoData" is returned.
14952
14953 Example :
14954 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14955 # containing the three properties requested using all relevant headers from
14956 # the request.
14957 frontend http-in
14958 bind *:8081
14959 default_backend servers
14960 http-request set-header X-51D-DeviceTypeMobileTablet \
14961 %[51d.all(DeviceType,IsMobile,IsTablet)]
14962
Emeric Brun645ae792014-04-30 14:21:06 +020014963ssl_bc : boolean
14964 Returns true when the back connection was made via an SSL/TLS transport
14965 layer and is locally deciphered. This means the outgoing connection was made
14966 other a server with the "ssl" option.
14967
14968ssl_bc_alg_keysize : integer
14969 Returns the symmetric cipher key size supported in bits when the outgoing
14970 connection was made over an SSL/TLS transport layer.
14971
Olivier Houchard6b77f492018-11-22 18:18:29 +010014972ssl_bc_alpn : string
14973 This extracts the Application Layer Protocol Negotiation field from an
14974 outgoing connection made via a TLS transport layer.
14975 The result is a string containing the protocol name negociated with the
14976 server. The SSL library must have been built with support for TLS
14977 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
14978 not advertised unless the "alpn" keyword on the "server" line specifies a
14979 protocol list. Also, nothing forces the server to pick a protocol from this
14980 list, any other one may be requested. The TLS ALPN extension is meant to
14981 replace the TLS NPN extension. See also "ssl_bc_npn".
14982
Emeric Brun645ae792014-04-30 14:21:06 +020014983ssl_bc_cipher : string
14984 Returns the name of the used cipher when the outgoing connection was made
14985 over an SSL/TLS transport layer.
14986
Emeric Brun74f7ffa2018-02-19 16:14:12 +010014987ssl_bc_is_resumed : boolean
14988 Returns true when the back connection was made over an SSL/TLS transport
14989 layer and the newly created SSL session was resumed using a cached
14990 session or a TLS ticket.
14991
Olivier Houchard6b77f492018-11-22 18:18:29 +010014992ssl_bc_npn : string
14993 This extracts the Next Protocol Negotiation field from an outgoing connection
14994 made via a TLS transport layer. The result is a string containing the
14995 protocol name negociated with the server . The SSL library must have been
14996 built with support for TLS extensions enabled (check haproxy -vv). Note that
14997 the TLS NPN extension is not advertised unless the "npn" keyword on the
14998 "server" line specifies a protocol list. Also, nothing forces the server to
14999 pick a protocol from this list, any other one may be used. Please note that
15000 the TLS NPN extension was replaced with ALPN.
15001
Emeric Brun645ae792014-04-30 14:21:06 +020015002ssl_bc_protocol : string
15003 Returns the name of the used protocol when the outgoing connection was made
15004 over an SSL/TLS transport layer.
15005
Emeric Brunb73a9b02014-04-30 18:49:19 +020015006ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015007 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015008 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15009 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015010
15011ssl_bc_session_id : binary
15012 Returns the SSL ID of the back connection when the outgoing connection was
15013 made over an SSL/TLS transport layer. It is useful to log if we want to know
15014 if session was reused or not.
15015
Patrick Hemmere0275472018-04-28 19:15:51 -040015016ssl_bc_session_key : binary
15017 Returns the SSL session master key of the back connection when the outgoing
15018 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15019 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15020 BoringSSL.
15021
Emeric Brun645ae792014-04-30 14:21:06 +020015022ssl_bc_use_keysize : integer
15023 Returns the symmetric cipher key size used in bits when the outgoing
15024 connection was made over an SSL/TLS transport layer.
15025
Willy Tarreau74ca5042013-06-11 23:12:07 +020015026ssl_c_ca_err : integer
15027 When the incoming connection was made over an SSL/TLS transport layer,
15028 returns the ID of the first error detected during verification of the client
15029 certificate at depth > 0, or 0 if no error was encountered during this
15030 verification process. Please refer to your SSL library's documentation to
15031 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015032
Willy Tarreau74ca5042013-06-11 23:12:07 +020015033ssl_c_ca_err_depth : integer
15034 When the incoming connection was made over an SSL/TLS transport layer,
15035 returns the depth in the CA chain of the first error detected during the
15036 verification of the client certificate. If no error is encountered, 0 is
15037 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015038
Emeric Brun43e79582014-10-29 19:03:26 +010015039ssl_c_der : binary
15040 Returns the DER formatted certificate presented by the client when the
15041 incoming connection was made over an SSL/TLS transport layer. When used for
15042 an ACL, the value(s) to match against can be passed in hexadecimal form.
15043
Willy Tarreau74ca5042013-06-11 23:12:07 +020015044ssl_c_err : integer
15045 When the incoming connection was made over an SSL/TLS transport layer,
15046 returns the ID of the first error detected during verification at depth 0, or
15047 0 if no error was encountered during this verification process. Please refer
15048 to your SSL library's documentation to find the exhaustive list of error
15049 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015050
Willy Tarreau74ca5042013-06-11 23:12:07 +020015051ssl_c_i_dn([<entry>[,<occ>]]) : string
15052 When the incoming connection was made over an SSL/TLS transport layer,
15053 returns the full distinguished name of the issuer of the certificate
15054 presented by the client when no <entry> is specified, or the value of the
15055 first given entry found from the beginning of the DN. If a positive/negative
15056 occurrence number is specified as the optional second argument, it returns
15057 the value of the nth given entry value from the beginning/end of the DN.
15058 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15059 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015060
Willy Tarreau74ca5042013-06-11 23:12:07 +020015061ssl_c_key_alg : string
15062 Returns the name of the algorithm used to generate the key of the certificate
15063 presented by the client when the incoming connection was made over an SSL/TLS
15064 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015065
Willy Tarreau74ca5042013-06-11 23:12:07 +020015066ssl_c_notafter : string
15067 Returns the end date presented by the client as a formatted string
15068 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15069 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015070
Willy Tarreau74ca5042013-06-11 23:12:07 +020015071ssl_c_notbefore : string
15072 Returns the start date presented by the client as a formatted string
15073 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15074 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015075
Willy Tarreau74ca5042013-06-11 23:12:07 +020015076ssl_c_s_dn([<entry>[,<occ>]]) : string
15077 When the incoming connection was made over an SSL/TLS transport layer,
15078 returns the full distinguished name of the subject of the certificate
15079 presented by the client when no <entry> is specified, or the value of the
15080 first given entry found from the beginning of the DN. If a positive/negative
15081 occurrence number is specified as the optional second argument, it returns
15082 the value of the nth given entry value from the beginning/end of the DN.
15083 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15084 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015085
Willy Tarreau74ca5042013-06-11 23:12:07 +020015086ssl_c_serial : binary
15087 Returns the serial of the certificate presented by the client when the
15088 incoming connection was made over an SSL/TLS transport layer. When used for
15089 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015090
Willy Tarreau74ca5042013-06-11 23:12:07 +020015091ssl_c_sha1 : binary
15092 Returns the SHA-1 fingerprint of the certificate presented by the client when
15093 the incoming connection was made over an SSL/TLS transport layer. This can be
15094 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015095 Note that the output is binary, so if you want to pass that signature to the
15096 server, you need to encode it in hex or base64, such as in the example below:
15097
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015098 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015099 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015100
Willy Tarreau74ca5042013-06-11 23:12:07 +020015101ssl_c_sig_alg : string
15102 Returns the name of the algorithm used to sign the certificate presented by
15103 the client when the incoming connection was made over an SSL/TLS transport
15104 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015105
Willy Tarreau74ca5042013-06-11 23:12:07 +020015106ssl_c_used : boolean
15107 Returns true if current SSL session uses a client certificate even if current
15108 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015109
Willy Tarreau74ca5042013-06-11 23:12:07 +020015110ssl_c_verify : integer
15111 Returns the verify result error ID when the incoming connection was made over
15112 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15113 refer to your SSL library's documentation for an exhaustive list of error
15114 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015115
Willy Tarreau74ca5042013-06-11 23:12:07 +020015116ssl_c_version : integer
15117 Returns the version of the certificate presented by the client when the
15118 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015119
Emeric Brun43e79582014-10-29 19:03:26 +010015120ssl_f_der : binary
15121 Returns the DER formatted certificate presented by the frontend when the
15122 incoming connection was made over an SSL/TLS transport layer. When used for
15123 an ACL, the value(s) to match against can be passed in hexadecimal form.
15124
Willy Tarreau74ca5042013-06-11 23:12:07 +020015125ssl_f_i_dn([<entry>[,<occ>]]) : string
15126 When the incoming connection was made over an SSL/TLS transport layer,
15127 returns the full distinguished name of the issuer of the certificate
15128 presented by the frontend when no <entry> is specified, or the value of the
15129 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015130 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015131 the value of the nth given entry value from the beginning/end of the DN.
15132 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15133 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015134
Willy Tarreau74ca5042013-06-11 23:12:07 +020015135ssl_f_key_alg : string
15136 Returns the name of the algorithm used to generate the key of the certificate
15137 presented by the frontend when the incoming connection was made over an
15138 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015139
Willy Tarreau74ca5042013-06-11 23:12:07 +020015140ssl_f_notafter : string
15141 Returns the end date presented by the frontend as a formatted string
15142 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15143 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015144
Willy Tarreau74ca5042013-06-11 23:12:07 +020015145ssl_f_notbefore : string
15146 Returns the start date presented by the frontend as a formatted string
15147 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15148 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015149
Willy Tarreau74ca5042013-06-11 23:12:07 +020015150ssl_f_s_dn([<entry>[,<occ>]]) : string
15151 When the incoming connection was made over an SSL/TLS transport layer,
15152 returns the full distinguished name of the subject of the certificate
15153 presented by the frontend when no <entry> is specified, or the value of the
15154 first given entry found from the beginning of the DN. If a positive/negative
15155 occurrence number is specified as the optional second argument, it returns
15156 the value of the nth given entry value from the beginning/end of the DN.
15157 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15158 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015159
Willy Tarreau74ca5042013-06-11 23:12:07 +020015160ssl_f_serial : binary
15161 Returns the serial of the certificate presented by the frontend when the
15162 incoming connection was made over an SSL/TLS transport layer. When used for
15163 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015164
Emeric Brun55f4fa82014-04-30 17:11:25 +020015165ssl_f_sha1 : binary
15166 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15167 when the incoming connection was made over an SSL/TLS transport layer. This
15168 can be used to know which certificate was chosen using SNI.
15169
Willy Tarreau74ca5042013-06-11 23:12:07 +020015170ssl_f_sig_alg : string
15171 Returns the name of the algorithm used to sign the certificate presented by
15172 the frontend when the incoming connection was made over an SSL/TLS transport
15173 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015174
Willy Tarreau74ca5042013-06-11 23:12:07 +020015175ssl_f_version : integer
15176 Returns the version of the certificate presented by the frontend when the
15177 incoming connection was made over an SSL/TLS transport layer.
15178
15179ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015180 Returns true when the front connection was made via an SSL/TLS transport
15181 layer and is locally deciphered. This means it has matched a socket declared
15182 with a "bind" line having the "ssl" option.
15183
Willy Tarreau74ca5042013-06-11 23:12:07 +020015184 Example :
15185 # This passes "X-Proto: https" to servers when client connects over SSL
15186 listen http-https
15187 bind :80
15188 bind :443 ssl crt /etc/haproxy.pem
15189 http-request add-header X-Proto https if { ssl_fc }
15190
15191ssl_fc_alg_keysize : integer
15192 Returns the symmetric cipher key size supported in bits when the incoming
15193 connection was made over an SSL/TLS transport layer.
15194
15195ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015196 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015197 incoming connection made via a TLS transport layer and locally deciphered by
15198 haproxy. The result is a string containing the protocol name advertised by
15199 the client. The SSL library must have been built with support for TLS
15200 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15201 not advertised unless the "alpn" keyword on the "bind" line specifies a
15202 protocol list. Also, nothing forces the client to pick a protocol from this
15203 list, any other one may be requested. The TLS ALPN extension is meant to
15204 replace the TLS NPN extension. See also "ssl_fc_npn".
15205
Willy Tarreau74ca5042013-06-11 23:12:07 +020015206ssl_fc_cipher : string
15207 Returns the name of the used cipher when the incoming connection was made
15208 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015209
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015210ssl_fc_cipherlist_bin : binary
15211 Returns the binary form of the client hello cipher list. The maximum returned
15212 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015213 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015214
15215ssl_fc_cipherlist_hex : string
15216 Returns the binary form of the client hello cipher list encoded as
15217 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015218 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015219
15220ssl_fc_cipherlist_str : string
15221 Returns the decoded text form of the client hello cipher list. The maximum
15222 number of ciphers returned is according with the value of
15223 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015224 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015225 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015226
15227ssl_fc_cipherlist_xxh : integer
15228 Returns a xxh64 of the cipher list. This hash can be return only is the value
15229 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015230 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015231
Willy Tarreau74ca5042013-06-11 23:12:07 +020015232ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015233 Returns true if a client certificate is present in an incoming connection over
15234 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015235 Note: on SSL session resumption with Session ID or TLS ticket, client
15236 certificate is not present in the current connection but may be retrieved
15237 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15238 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015239
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015240ssl_fc_has_early : boolean
15241 Returns true if early data were sent, and the handshake didn't happen yet. As
15242 it has security implications, it is useful to be able to refuse those, or
15243 wait until the handshake happened.
15244
Willy Tarreau74ca5042013-06-11 23:12:07 +020015245ssl_fc_has_sni : boolean
15246 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015247 in an incoming connection was made over an SSL/TLS transport layer. Returns
15248 true when the incoming connection presents a TLS SNI field. This requires
15249 that the SSL library is build with support for TLS extensions enabled (check
15250 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015251
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015252ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015253 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015254 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15255 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015256
Willy Tarreau74ca5042013-06-11 23:12:07 +020015257ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015258 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015259 made via a TLS transport layer and locally deciphered by haproxy. The result
15260 is a string containing the protocol name advertised by the client. The SSL
15261 library must have been built with support for TLS extensions enabled (check
15262 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15263 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15264 forces the client to pick a protocol from this list, any other one may be
15265 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015266
Willy Tarreau74ca5042013-06-11 23:12:07 +020015267ssl_fc_protocol : string
15268 Returns the name of the used protocol when the incoming connection was made
15269 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015270
Emeric Brunb73a9b02014-04-30 18:49:19 +020015271ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015272 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015273 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15274 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015275
Willy Tarreau74ca5042013-06-11 23:12:07 +020015276ssl_fc_session_id : binary
15277 Returns the SSL ID of the front connection when the incoming connection was
15278 made over an SSL/TLS transport layer. It is useful to stick a given client to
15279 a server. It is important to note that some browsers refresh their session ID
15280 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015281
Patrick Hemmere0275472018-04-28 19:15:51 -040015282ssl_fc_session_key : binary
15283 Returns the SSL session master key of the front connection when the incoming
15284 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15285 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15286 BoringSSL.
15287
15288
Willy Tarreau74ca5042013-06-11 23:12:07 +020015289ssl_fc_sni : string
15290 This extracts the Server Name Indication TLS extension (SNI) field from an
15291 incoming connection made via an SSL/TLS transport layer and locally
15292 deciphered by haproxy. The result (when present) typically is a string
15293 matching the HTTPS host name (253 chars or less). The SSL library must have
15294 been built with support for TLS extensions enabled (check haproxy -vv).
15295
15296 This fetch is different from "req_ssl_sni" above in that it applies to the
15297 connection being deciphered by haproxy and not to SSL contents being blindly
15298 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015299 requires that the SSL library is build with support for TLS extensions
15300 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015301
Willy Tarreau74ca5042013-06-11 23:12:07 +020015302 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015303 ssl_fc_sni_end : suffix match
15304 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015305
Willy Tarreau74ca5042013-06-11 23:12:07 +020015306ssl_fc_use_keysize : integer
15307 Returns the symmetric cipher key size used in bits when the incoming
15308 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015309
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015310
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200153117.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015312------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015313
Willy Tarreau74ca5042013-06-11 23:12:07 +020015314Fetching samples from buffer contents is a bit different from the previous
15315sample fetches above because the sampled data are ephemeral. These data can
15316only be used when they're available and will be lost when they're forwarded.
15317For this reason, samples fetched from buffer contents during a request cannot
15318be used in a response for example. Even while the data are being fetched, they
15319can change. Sometimes it is necessary to set some delays or combine multiple
15320sample fetch methods to ensure that the expected data are complete and usable,
15321for example through TCP request content inspection. Please see the "tcp-request
15322content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015323
Willy Tarreau74ca5042013-06-11 23:12:07 +020015324payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015325 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015326 "stick on", "stick match"), and for "res.payload" when used in the context of
15327 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015328
Willy Tarreau74ca5042013-06-11 23:12:07 +020015329payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15330 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015331 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015332 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015333
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015334req.hdrs : string
15335 Returns the current request headers as string including the last empty line
15336 separating headers from the request body. The last empty line can be used to
15337 detect a truncated header block. This sample fetch is useful for some SPOE
15338 headers analyzers and for advanced logging.
15339
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015340req.hdrs_bin : binary
15341 Returns the current request headers contained in preparsed binary form. This
15342 is useful for offloading some processing with SPOE. Each string is described
15343 by a length followed by the number of bytes indicated in the length. The
15344 length is represented using the variable integer encoding detailed in the
15345 SPOE documentation. The end of the list is marked by a couple of empty header
15346 names and values (length of 0 for both).
15347
15348 *(<str:header-name><str:header-value>)<empty string><empty string>
15349
15350 int: refer to the SPOE documentation for the encoding
15351 str: <int:length><bytes>
15352
Willy Tarreau74ca5042013-06-11 23:12:07 +020015353req.len : integer
15354req_len : integer (deprecated)
15355 Returns an integer value corresponding to the number of bytes present in the
15356 request buffer. This is mostly used in ACL. It is important to understand
15357 that this test does not return false as long as the buffer is changing. This
15358 means that a check with equality to zero will almost always immediately match
15359 at the beginning of the session, while a test for more data will wait for
15360 that data to come in and return false only when haproxy is certain that no
15361 more data will come in. This test was designed to be used with TCP request
15362 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015363
Willy Tarreau74ca5042013-06-11 23:12:07 +020015364req.payload(<offset>,<length>) : binary
15365 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015366 in the request buffer. As a special case, if the <length> argument is zero,
15367 the the whole buffer from <offset> to the end is extracted. This can be used
15368 with ACLs in order to check for the presence of some content in a buffer at
15369 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015370
Willy Tarreau74ca5042013-06-11 23:12:07 +020015371 ACL alternatives :
15372 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015373
Willy Tarreau74ca5042013-06-11 23:12:07 +020015374req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15375 This extracts a binary block whose size is specified at <offset1> for <length>
15376 bytes, and which starts at <offset2> if specified or just after the length in
15377 the request buffer. The <offset2> parameter also supports relative offsets if
15378 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015379
Willy Tarreau74ca5042013-06-11 23:12:07 +020015380 ACL alternatives :
15381 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015382
Willy Tarreau74ca5042013-06-11 23:12:07 +020015383 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015384
Willy Tarreau74ca5042013-06-11 23:12:07 +020015385req.proto_http : boolean
15386req_proto_http : boolean (deprecated)
15387 Returns true when data in the request buffer look like HTTP and correctly
15388 parses as such. It is the same parser as the common HTTP request parser which
15389 is used so there should be no surprises. The test does not match until the
15390 request is complete, failed or timed out. This test may be used to report the
15391 protocol in TCP logs, but the biggest use is to block TCP request analysis
15392 until a complete HTTP request is present in the buffer, for example to track
15393 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015394
Willy Tarreau74ca5042013-06-11 23:12:07 +020015395 Example:
15396 # track request counts per "base" (concatenation of Host+URL)
15397 tcp-request inspect-delay 10s
15398 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015399 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015400
Willy Tarreau74ca5042013-06-11 23:12:07 +020015401req.rdp_cookie([<name>]) : string
15402rdp_cookie([<name>]) : string (deprecated)
15403 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15404 <name>, or any cookie if unspecified. The parser only checks for the first
15405 cookie, as illustrated in the RDP protocol specification. The cookie name is
15406 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15407 contain the user name of the client connecting to the server if properly
15408 configured on the client. The "MSTSHASH" cookie is often used as well for
15409 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015410
Willy Tarreau74ca5042013-06-11 23:12:07 +020015411 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15412 used and thus the distribution of clients to backend servers is not linked to
15413 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15414 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15415 distribution of clients to backend servers than the hash used by "balance
15416 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015417
Willy Tarreau74ca5042013-06-11 23:12:07 +020015418 ACL derivatives :
15419 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015420
Willy Tarreau74ca5042013-06-11 23:12:07 +020015421 Example :
15422 listen tse-farm
15423 bind 0.0.0.0:3389
15424 # wait up to 5s for an RDP cookie in the request
15425 tcp-request inspect-delay 5s
15426 tcp-request content accept if RDP_COOKIE
15427 # apply RDP cookie persistence
15428 persist rdp-cookie
15429 # Persist based on the mstshash cookie
15430 # This is only useful makes sense if
15431 # balance rdp-cookie is not used
15432 stick-table type string size 204800
15433 stick on req.rdp_cookie(mstshash)
15434 server srv1 1.1.1.1:3389
15435 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015436
Willy Tarreau74ca5042013-06-11 23:12:07 +020015437 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15438 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015439
Willy Tarreau74ca5042013-06-11 23:12:07 +020015440req.rdp_cookie_cnt([name]) : integer
15441rdp_cookie_cnt([name]) : integer (deprecated)
15442 Tries to parse the request buffer as RDP protocol, then returns an integer
15443 corresponding to the number of RDP cookies found. If an optional cookie name
15444 is passed, only cookies matching this name are considered. This is mostly
15445 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015446
Willy Tarreau74ca5042013-06-11 23:12:07 +020015447 ACL derivatives :
15448 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015449
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015450req.ssl_ec_ext : boolean
15451 Returns a boolean identifying if client sent the Supported Elliptic Curves
15452 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015453 message. This can be used to present ECC compatible clients with EC
15454 certificate and to use RSA for all others, on the same IP address. Note that
15455 this only applies to raw contents found in the request buffer and not to
15456 contents deciphered via an SSL data layer, so this will not work with "bind"
15457 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015458
Willy Tarreau74ca5042013-06-11 23:12:07 +020015459req.ssl_hello_type : integer
15460req_ssl_hello_type : integer (deprecated)
15461 Returns an integer value containing the type of the SSL hello message found
15462 in the request buffer if the buffer contains data that parse as a complete
15463 SSL (v3 or superior) client hello message. Note that this only applies to raw
15464 contents found in the request buffer and not to contents deciphered via an
15465 SSL data layer, so this will not work with "bind" lines having the "ssl"
15466 option. This is mostly used in ACL to detect presence of an SSL hello message
15467 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015468
Willy Tarreau74ca5042013-06-11 23:12:07 +020015469req.ssl_sni : string
15470req_ssl_sni : string (deprecated)
15471 Returns a string containing the value of the Server Name TLS extension sent
15472 by a client in a TLS stream passing through the request buffer if the buffer
15473 contains data that parse as a complete SSL (v3 or superior) client hello
15474 message. Note that this only applies to raw contents found in the request
15475 buffer and not to contents deciphered via an SSL data layer, so this will not
15476 work with "bind" lines having the "ssl" option. SNI normally contains the
15477 name of the host the client tries to connect to (for recent browsers). SNI is
15478 useful for allowing or denying access to certain hosts when SSL/TLS is used
15479 by the client. This test was designed to be used with TCP request content
15480 inspection. If content switching is needed, it is recommended to first wait
15481 for a complete client hello (type 1), like in the example below. See also
15482 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015483
Willy Tarreau74ca5042013-06-11 23:12:07 +020015484 ACL derivatives :
15485 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015486
Willy Tarreau74ca5042013-06-11 23:12:07 +020015487 Examples :
15488 # Wait for a client hello for at most 5 seconds
15489 tcp-request inspect-delay 5s
15490 tcp-request content accept if { req_ssl_hello_type 1 }
15491 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15492 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015493
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015494req.ssl_st_ext : integer
15495 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15496 Returns 1 if the client sent SessionTicket TLS Extension
15497 Returns 2 if the client also sent non-zero length TLS SessionTicket
15498 Note that this only applies to raw contents found in the request buffer and
15499 not to contents deciphered via an SSL data layer, so this will not work with
15500 "bind" lines having the "ssl" option. This can for example be used to detect
15501 whether the client sent a SessionTicket or not and stick it accordingly, if
15502 no SessionTicket then stick on SessionID or don't stick as there's no server
15503 side state is there when SessionTickets are in use.
15504
Willy Tarreau74ca5042013-06-11 23:12:07 +020015505req.ssl_ver : integer
15506req_ssl_ver : integer (deprecated)
15507 Returns an integer value containing the version of the SSL/TLS protocol of a
15508 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15509 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15510 composed of the major version multiplied by 65536, added to the minor
15511 version. Note that this only applies to raw contents found in the request
15512 buffer and not to contents deciphered via an SSL data layer, so this will not
15513 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015514 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015515 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015516
Willy Tarreau74ca5042013-06-11 23:12:07 +020015517 ACL derivatives :
15518 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015519
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015520res.len : integer
15521 Returns an integer value corresponding to the number of bytes present in the
15522 response buffer. This is mostly used in ACL. It is important to understand
15523 that this test does not return false as long as the buffer is changing. This
15524 means that a check with equality to zero will almost always immediately match
15525 at the beginning of the session, while a test for more data will wait for
15526 that data to come in and return false only when haproxy is certain that no
15527 more data will come in. This test was designed to be used with TCP response
15528 content inspection.
15529
Willy Tarreau74ca5042013-06-11 23:12:07 +020015530res.payload(<offset>,<length>) : binary
15531 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015532 in the response buffer. As a special case, if the <length> argument is zero,
15533 the the whole buffer from <offset> to the end is extracted. This can be used
15534 with ACLs in order to check for the presence of some content in a buffer at
15535 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015536
Willy Tarreau74ca5042013-06-11 23:12:07 +020015537res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15538 This extracts a binary block whose size is specified at <offset1> for <length>
15539 bytes, and which starts at <offset2> if specified or just after the length in
15540 the response buffer. The <offset2> parameter also supports relative offsets
15541 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015542
Willy Tarreau74ca5042013-06-11 23:12:07 +020015543 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015544
Willy Tarreau971f7b62015-09-29 14:06:59 +020015545res.ssl_hello_type : integer
15546rep_ssl_hello_type : integer (deprecated)
15547 Returns an integer value containing the type of the SSL hello message found
15548 in the response buffer if the buffer contains data that parses as a complete
15549 SSL (v3 or superior) hello message. Note that this only applies to raw
15550 contents found in the response buffer and not to contents deciphered via an
15551 SSL data layer, so this will not work with "server" lines having the "ssl"
15552 option. This is mostly used in ACL to detect presence of an SSL hello message
15553 that is supposed to contain an SSL session ID usable for stickiness.
15554
Willy Tarreau74ca5042013-06-11 23:12:07 +020015555wait_end : boolean
15556 This fetch either returns true when the inspection period is over, or does
15557 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015558 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015559 actions, such as a delayed reject for some special addresses. Since it either
15560 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015561 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015562 "WAIT_END" is always usable without prior declaration. This test was designed
15563 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015564
Willy Tarreau74ca5042013-06-11 23:12:07 +020015565 Examples :
15566 # delay every incoming request by 2 seconds
15567 tcp-request inspect-delay 2s
15568 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015569
Willy Tarreau74ca5042013-06-11 23:12:07 +020015570 # don't immediately tell bad guys they are rejected
15571 tcp-request inspect-delay 10s
15572 acl goodguys src 10.0.0.0/24
15573 acl badguys src 10.0.1.0/24
15574 tcp-request content accept if goodguys
15575 tcp-request content reject if badguys WAIT_END
15576 tcp-request content reject
15577
15578
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200155797.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015580--------------------------------------
15581
15582It is possible to fetch samples from HTTP contents, requests and responses.
15583This application layer is also called layer 7. It is only possible to fetch the
15584data in this section when a full HTTP request or response has been parsed from
15585its respective request or response buffer. This is always the case with all
15586HTTP specific rules and for sections running with "mode http". When using TCP
15587content inspection, it may be necessary to support an inspection delay in order
15588to let the request or response come in first. These fetches may require a bit
15589more CPU resources than the layer 4 ones, but not much since the request and
15590response are indexed.
15591
15592base : string
15593 This returns the concatenation of the first Host header and the path part of
15594 the request, which starts at the first slash and ends before the question
15595 mark. It can be useful in virtual hosted environments to detect URL abuses as
15596 well as to improve shared caches efficiency. Using this with a limited size
15597 stick table also allows one to collect statistics about most commonly
15598 requested objects by host/path. With ACLs it can allow simple content
15599 switching rules involving the host and the path at the same time, such as
15600 "www.example.com/favicon.ico". See also "path" and "uri".
15601
15602 ACL derivatives :
15603 base : exact string match
15604 base_beg : prefix match
15605 base_dir : subdir match
15606 base_dom : domain match
15607 base_end : suffix match
15608 base_len : length match
15609 base_reg : regex match
15610 base_sub : substring match
15611
15612base32 : integer
15613 This returns a 32-bit hash of the value returned by the "base" fetch method
15614 above. This is useful to track per-URL activity on high traffic sites without
15615 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015616 memory. The output type is an unsigned integer. The hash function used is
15617 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15618 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015619
15620base32+src : binary
15621 This returns the concatenation of the base32 fetch above and the src fetch
15622 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15623 depending on the source address family. This can be used to track per-IP,
15624 per-URL counters.
15625
William Lallemand65ad6e12014-01-31 15:08:02 +010015626capture.req.hdr(<idx>) : string
15627 This extracts the content of the header captured by the "capture request
15628 header", idx is the position of the capture keyword in the configuration.
15629 The first entry is an index of 0. See also: "capture request header".
15630
15631capture.req.method : string
15632 This extracts the METHOD of an HTTP request. It can be used in both request
15633 and response. Unlike "method", it can be used in both request and response
15634 because it's allocated.
15635
15636capture.req.uri : string
15637 This extracts the request's URI, which starts at the first slash and ends
15638 before the first space in the request (without the host part). Unlike "path"
15639 and "url", it can be used in both request and response because it's
15640 allocated.
15641
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015642capture.req.ver : string
15643 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15644 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15645 logs because it relies on a persistent flag.
15646
William Lallemand65ad6e12014-01-31 15:08:02 +010015647capture.res.hdr(<idx>) : string
15648 This extracts the content of the header captured by the "capture response
15649 header", idx is the position of the capture keyword in the configuration.
15650 The first entry is an index of 0.
15651 See also: "capture response header"
15652
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015653capture.res.ver : string
15654 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15655 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15656 persistent flag.
15657
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015658req.body : binary
15659 This returns the HTTP request's available body as a block of data. It
15660 requires that the request body has been buffered made available using
15661 "option http-buffer-request". In case of chunked-encoded body, currently only
15662 the first chunk is analyzed.
15663
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015664req.body_param([<name>) : string
15665 This fetch assumes that the body of the POST request is url-encoded. The user
15666 can check if the "content-type" contains the value
15667 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15668 parameter <name> in the body, which ends before '&'. The parameter name is
15669 case-sensitive. If no name is given, any parameter will match, and the first
15670 one will be returned. The result is a string corresponding to the value of the
15671 parameter <name> as presented in the request body (no URL decoding is
15672 performed). Note that the ACL version of this fetch iterates over multiple
15673 parameters and will iteratively report all parameters values if no name is
15674 given.
15675
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015676req.body_len : integer
15677 This returns the length of the HTTP request's available body in bytes. It may
15678 be lower than the advertised length if the body is larger than the buffer. It
15679 requires that the request body has been buffered made available using
15680 "option http-buffer-request".
15681
15682req.body_size : integer
15683 This returns the advertised length of the HTTP request's body in bytes. It
15684 will represent the advertised Content-Length header, or the size of the first
15685 chunk in case of chunked encoding. In order to parse the chunks, it requires
15686 that the request body has been buffered made available using
15687 "option http-buffer-request".
15688
Willy Tarreau74ca5042013-06-11 23:12:07 +020015689req.cook([<name>]) : string
15690cook([<name>]) : string (deprecated)
15691 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15692 header line from the request, and returns its value as string. If no name is
15693 specified, the first cookie value is returned. When used with ACLs, all
15694 matching cookies are evaluated. Spaces around the name and the value are
15695 ignored as requested by the Cookie header specification (RFC6265). The cookie
15696 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15697 well return an empty value if it is present. Use the "found" match to detect
15698 presence. Use the res.cook() variant for response cookies sent by the server.
15699
15700 ACL derivatives :
15701 cook([<name>]) : exact string match
15702 cook_beg([<name>]) : prefix match
15703 cook_dir([<name>]) : subdir match
15704 cook_dom([<name>]) : domain match
15705 cook_end([<name>]) : suffix match
15706 cook_len([<name>]) : length match
15707 cook_reg([<name>]) : regex match
15708 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015709
Willy Tarreau74ca5042013-06-11 23:12:07 +020015710req.cook_cnt([<name>]) : integer
15711cook_cnt([<name>]) : integer (deprecated)
15712 Returns an integer value representing the number of occurrences of the cookie
15713 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015714
Willy Tarreau74ca5042013-06-11 23:12:07 +020015715req.cook_val([<name>]) : integer
15716cook_val([<name>]) : integer (deprecated)
15717 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15718 header line from the request, and converts its value to an integer which is
15719 returned. If no name is specified, the first cookie value is returned. When
15720 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015721
Willy Tarreau74ca5042013-06-11 23:12:07 +020015722cookie([<name>]) : string (deprecated)
15723 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15724 header line from the request, or a "Set-Cookie" header from the response, and
15725 returns its value as a string. A typical use is to get multiple clients
15726 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015727 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015728 multi-peer synchronization and state keeping across restarts. If no name is
15729 specified, the first cookie value is returned. This fetch should not be used
15730 anymore and should be replaced by req.cook() or res.cook() instead as it
15731 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015732
Willy Tarreau74ca5042013-06-11 23:12:07 +020015733hdr([<name>[,<occ>]]) : string
15734 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15735 used on responses. Please refer to these respective fetches for more details.
15736 In case of doubt about the fetch direction, please use the explicit ones.
15737 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015738 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015739
Willy Tarreau74ca5042013-06-11 23:12:07 +020015740req.fhdr(<name>[,<occ>]) : string
15741 This extracts the last occurrence of header <name> in an HTTP request. When
15742 used from an ACL, all occurrences are iterated over until a match is found.
15743 Optionally, a specific occurrence might be specified as a position number.
15744 Positive values indicate a position from the first occurrence, with 1 being
15745 the first one. Negative values indicate positions relative to the last one,
15746 with -1 being the last one. It differs from req.hdr() in that any commas
15747 present in the value are returned and are not used as delimiters. This is
15748 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015749
Willy Tarreau74ca5042013-06-11 23:12:07 +020015750req.fhdr_cnt([<name>]) : integer
15751 Returns an integer value representing the number of occurrences of request
15752 header field name <name>, or the total number of header fields if <name> is
15753 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15754 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015755
Willy Tarreau74ca5042013-06-11 23:12:07 +020015756req.hdr([<name>[,<occ>]]) : string
15757 This extracts the last occurrence of header <name> in an HTTP request. When
15758 used from an ACL, all occurrences are iterated over until a match is found.
15759 Optionally, a specific occurrence might be specified as a position number.
15760 Positive values indicate a position from the first occurrence, with 1 being
15761 the first one. Negative values indicate positions relative to the last one,
15762 with -1 being the last one. A typical use is with the X-Forwarded-For header
15763 once converted to IP, associated with an IP stick-table. The function
15764 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015765 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015766 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015767 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015768
Willy Tarreau74ca5042013-06-11 23:12:07 +020015769 ACL derivatives :
15770 hdr([<name>[,<occ>]]) : exact string match
15771 hdr_beg([<name>[,<occ>]]) : prefix match
15772 hdr_dir([<name>[,<occ>]]) : subdir match
15773 hdr_dom([<name>[,<occ>]]) : domain match
15774 hdr_end([<name>[,<occ>]]) : suffix match
15775 hdr_len([<name>[,<occ>]]) : length match
15776 hdr_reg([<name>[,<occ>]]) : regex match
15777 hdr_sub([<name>[,<occ>]]) : substring match
15778
15779req.hdr_cnt([<name>]) : integer
15780hdr_cnt([<header>]) : integer (deprecated)
15781 Returns an integer value representing the number of occurrences of request
15782 header field name <name>, or the total number of header field values if
15783 <name> is not specified. It is important to remember that one header line may
15784 count as several headers if it has several values. The function considers any
15785 comma as a delimiter for distinct values. If full-line headers are desired
15786 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15787 detect presence, absence or abuse of a specific header, as well as to block
15788 request smuggling attacks by rejecting requests which contain more than one
15789 of certain headers. See "req.hdr" for more information on header matching.
15790
15791req.hdr_ip([<name>[,<occ>]]) : ip
15792hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15793 This extracts the last occurrence of header <name> in an HTTP request,
15794 converts it to an IPv4 or IPv6 address and returns this address. When used
15795 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15796 of every header is checked. Optionally, a specific occurrence might be
15797 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015798 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015799 positions relative to the last one, with -1 being the last one. A typical use
15800 is with the X-Forwarded-For and X-Client-IP headers.
15801
15802req.hdr_val([<name>[,<occ>]]) : integer
15803hdr_val([<name>[,<occ>]]) : integer (deprecated)
15804 This extracts the last occurrence of header <name> in an HTTP request, and
15805 converts it to an integer value. When used with ACLs, all occurrences are
15806 checked, and if <name> is omitted, every value of every header is checked.
15807 Optionally, a specific occurrence might be specified as a position number.
15808 Positive values indicate a position from the first occurrence, with 1 being
15809 the first one. Negative values indicate positions relative to the last one,
15810 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15811
15812http_auth(<userlist>) : boolean
15813 Returns a boolean indicating whether the authentication data received from
15814 the client match a username & password stored in the specified userlist. This
15815 fetch function is not really useful outside of ACLs. Currently only http
15816 basic auth is supported.
15817
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015818http_auth_group(<userlist>) : string
15819 Returns a string corresponding to the user name found in the authentication
15820 data received from the client if both the user name and password are valid
15821 according to the specified userlist. The main purpose is to use it in ACLs
15822 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015823 This fetch function is not really useful outside of ACLs. Currently only http
15824 basic auth is supported.
15825
15826 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015827 http_auth_group(<userlist>) : group ...
15828 Returns true when the user extracted from the request and whose password is
15829 valid according to the specified userlist belongs to at least one of the
15830 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015831
15832http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015833 Returns true when the request being processed is the first one of the
15834 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015835 from some requests when a request is not the first one, or to help grouping
15836 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015837
Willy Tarreau74ca5042013-06-11 23:12:07 +020015838method : integer + string
15839 Returns an integer value corresponding to the method in the HTTP request. For
15840 example, "GET" equals 1 (check sources to establish the matching). Value 9
15841 means "other method" and may be converted to a string extracted from the
15842 stream. This should not be used directly as a sample, this is only meant to
15843 be used from ACLs, which transparently convert methods from patterns to these
15844 integer + string values. Some predefined ACL already check for most common
15845 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015846
Willy Tarreau74ca5042013-06-11 23:12:07 +020015847 ACL derivatives :
15848 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015849
Willy Tarreau74ca5042013-06-11 23:12:07 +020015850 Example :
15851 # only accept GET and HEAD requests
15852 acl valid_method method GET HEAD
15853 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015854
Willy Tarreau74ca5042013-06-11 23:12:07 +020015855path : string
15856 This extracts the request's URL path, which starts at the first slash and
15857 ends before the question mark (without the host part). A typical use is with
15858 prefetch-capable caches, and with portals which need to aggregate multiple
15859 information from databases and keep them in caches. Note that with outgoing
15860 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015861 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015862 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015863
Willy Tarreau74ca5042013-06-11 23:12:07 +020015864 ACL derivatives :
15865 path : exact string match
15866 path_beg : prefix match
15867 path_dir : subdir match
15868 path_dom : domain match
15869 path_end : suffix match
15870 path_len : length match
15871 path_reg : regex match
15872 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015873
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015874query : string
15875 This extracts the request's query string, which starts after the first
15876 question mark. If no question mark is present, this fetch returns nothing. If
15877 a question mark is present but nothing follows, it returns an empty string.
15878 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015879 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015880 which stops before the question mark.
15881
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015882req.hdr_names([<delim>]) : string
15883 This builds a string made from the concatenation of all header names as they
15884 appear in the request when the rule is evaluated. The default delimiter is
15885 the comma (',') but it may be overridden as an optional argument <delim>. In
15886 this case, only the first character of <delim> is considered.
15887
Willy Tarreau74ca5042013-06-11 23:12:07 +020015888req.ver : string
15889req_ver : string (deprecated)
15890 Returns the version string from the HTTP request, for example "1.1". This can
15891 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15892 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015893
Willy Tarreau74ca5042013-06-11 23:12:07 +020015894 ACL derivatives :
15895 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015896
Willy Tarreau74ca5042013-06-11 23:12:07 +020015897res.comp : boolean
15898 Returns the boolean "true" value if the response has been compressed by
15899 HAProxy, otherwise returns boolean "false". This may be used to add
15900 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015901
Willy Tarreau74ca5042013-06-11 23:12:07 +020015902res.comp_algo : string
15903 Returns a string containing the name of the algorithm used if the response
15904 was compressed by HAProxy, for example : "deflate". This may be used to add
15905 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015906
Willy Tarreau74ca5042013-06-11 23:12:07 +020015907res.cook([<name>]) : string
15908scook([<name>]) : string (deprecated)
15909 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15910 header line from the response, and returns its value as string. If no name is
15911 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015912
Willy Tarreau74ca5042013-06-11 23:12:07 +020015913 ACL derivatives :
15914 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015915
Willy Tarreau74ca5042013-06-11 23:12:07 +020015916res.cook_cnt([<name>]) : integer
15917scook_cnt([<name>]) : integer (deprecated)
15918 Returns an integer value representing the number of occurrences of the cookie
15919 <name> in the response, or all cookies if <name> is not specified. This is
15920 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015921
Willy Tarreau74ca5042013-06-11 23:12:07 +020015922res.cook_val([<name>]) : integer
15923scook_val([<name>]) : integer (deprecated)
15924 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15925 header line from the response, and converts its value to an integer which is
15926 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015927
Willy Tarreau74ca5042013-06-11 23:12:07 +020015928res.fhdr([<name>[,<occ>]]) : string
15929 This extracts the last occurrence of header <name> in an HTTP response, or of
15930 the last header if no <name> is specified. Optionally, a specific occurrence
15931 might be specified as a position number. Positive values indicate a position
15932 from the first occurrence, with 1 being the first one. Negative values
15933 indicate positions relative to the last one, with -1 being the last one. It
15934 differs from res.hdr() in that any commas present in the value are returned
15935 and are not used as delimiters. If this is not desired, the res.hdr() fetch
15936 should be used instead. This is sometimes useful with headers such as Date or
15937 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015938
Willy Tarreau74ca5042013-06-11 23:12:07 +020015939res.fhdr_cnt([<name>]) : integer
15940 Returns an integer value representing the number of occurrences of response
15941 header field name <name>, or the total number of header fields if <name> is
15942 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
15943 the number of full line headers and does not stop on commas. If this is not
15944 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015945
Willy Tarreau74ca5042013-06-11 23:12:07 +020015946res.hdr([<name>[,<occ>]]) : string
15947shdr([<name>[,<occ>]]) : string (deprecated)
15948 This extracts the last occurrence of header <name> in an HTTP response, or of
15949 the last header if no <name> is specified. Optionally, a specific occurrence
15950 might be specified as a position number. Positive values indicate a position
15951 from the first occurrence, with 1 being the first one. Negative values
15952 indicate positions relative to the last one, with -1 being the last one. This
15953 can be useful to learn some data into a stick-table. The function considers
15954 any comma as a delimiter for distinct values. If this is not desired, the
15955 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015956
Willy Tarreau74ca5042013-06-11 23:12:07 +020015957 ACL derivatives :
15958 shdr([<name>[,<occ>]]) : exact string match
15959 shdr_beg([<name>[,<occ>]]) : prefix match
15960 shdr_dir([<name>[,<occ>]]) : subdir match
15961 shdr_dom([<name>[,<occ>]]) : domain match
15962 shdr_end([<name>[,<occ>]]) : suffix match
15963 shdr_len([<name>[,<occ>]]) : length match
15964 shdr_reg([<name>[,<occ>]]) : regex match
15965 shdr_sub([<name>[,<occ>]]) : substring match
15966
15967res.hdr_cnt([<name>]) : integer
15968shdr_cnt([<name>]) : integer (deprecated)
15969 Returns an integer value representing the number of occurrences of response
15970 header field name <name>, or the total number of header fields if <name> is
15971 not specified. The function considers any comma as a delimiter for distinct
15972 values. If this is not desired, the res.fhdr_cnt() fetch should be used
15973 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015974
Willy Tarreau74ca5042013-06-11 23:12:07 +020015975res.hdr_ip([<name>[,<occ>]]) : ip
15976shdr_ip([<name>[,<occ>]]) : ip (deprecated)
15977 This extracts the last occurrence of header <name> in an HTTP response,
15978 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
15979 specific occurrence might be specified as a position number. Positive values
15980 indicate a position from the first occurrence, with 1 being the first one.
15981 Negative values indicate positions relative to the last one, with -1 being
15982 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015983
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015984res.hdr_names([<delim>]) : string
15985 This builds a string made from the concatenation of all header names as they
15986 appear in the response when the rule is evaluated. The default delimiter is
15987 the comma (',') but it may be overridden as an optional argument <delim>. In
15988 this case, only the first character of <delim> is considered.
15989
Willy Tarreau74ca5042013-06-11 23:12:07 +020015990res.hdr_val([<name>[,<occ>]]) : integer
15991shdr_val([<name>[,<occ>]]) : integer (deprecated)
15992 This extracts the last occurrence of header <name> in an HTTP response, and
15993 converts it to an integer value. Optionally, a specific occurrence might be
15994 specified as a position number. Positive values indicate a position from the
15995 first occurrence, with 1 being the first one. Negative values indicate
15996 positions relative to the last one, with -1 being the last one. This can be
15997 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015998
Willy Tarreau74ca5042013-06-11 23:12:07 +020015999res.ver : string
16000resp_ver : string (deprecated)
16001 Returns the version string from the HTTP response, for example "1.1". This
16002 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016003
Willy Tarreau74ca5042013-06-11 23:12:07 +020016004 ACL derivatives :
16005 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016006
Willy Tarreau74ca5042013-06-11 23:12:07 +020016007set-cookie([<name>]) : string (deprecated)
16008 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16009 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016010 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016011 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016012
Willy Tarreau74ca5042013-06-11 23:12:07 +020016013 This fetch function is deprecated and has been superseded by the "res.cook"
16014 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016015
Willy Tarreau74ca5042013-06-11 23:12:07 +020016016status : integer
16017 Returns an integer containing the HTTP status code in the HTTP response, for
16018 example, 302. It is mostly used within ACLs and integer ranges, for example,
16019 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016020
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016021unique-id : string
16022 Returns the unique-id attached to the request. The directive
16023 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16024 fails. Note that the unique-id is usually used with HTTP requests, however this
16025 sample fetch can be used with other protocols. Obviously, if it is used with
16026 other protocols than HTTP, the unique-id-format directive must not contain
16027 HTTP parts. See: unique-id-format and unique-id-header
16028
Willy Tarreau74ca5042013-06-11 23:12:07 +020016029url : string
16030 This extracts the request's URL as presented in the request. A typical use is
16031 with prefetch-capable caches, and with portals which need to aggregate
16032 multiple information from databases and keep them in caches. With ACLs, using
16033 "path" is preferred over using "url", because clients may send a full URL as
16034 is normally done with proxies. The only real use is to match "*" which does
16035 not match in "path", and for which there is already a predefined ACL. See
16036 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016037
Willy Tarreau74ca5042013-06-11 23:12:07 +020016038 ACL derivatives :
16039 url : exact string match
16040 url_beg : prefix match
16041 url_dir : subdir match
16042 url_dom : domain match
16043 url_end : suffix match
16044 url_len : length match
16045 url_reg : regex match
16046 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016047
Willy Tarreau74ca5042013-06-11 23:12:07 +020016048url_ip : ip
16049 This extracts the IP address from the request's URL when the host part is
16050 presented as an IP address. Its use is very limited. For instance, a
16051 monitoring system might use this field as an alternative for the source IP in
16052 order to test what path a given source address would follow, or to force an
16053 entry in a table for a given source address. With ACLs it can be used to
16054 restrict access to certain systems through a proxy, for example when combined
16055 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016056
Willy Tarreau74ca5042013-06-11 23:12:07 +020016057url_port : integer
16058 This extracts the port part from the request's URL. Note that if the port is
16059 not specified in the request, port 80 is assumed. With ACLs it can be used to
16060 restrict access to certain systems through a proxy, for example when combined
16061 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016062
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016063urlp([<name>[,<delim>]]) : string
16064url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016065 This extracts the first occurrence of the parameter <name> in the query
16066 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016067 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16068 any parameter will match, and the first one will be returned. The result is
16069 a string corresponding to the value of the parameter <name> as presented in
16070 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016071 stickiness based on a client ID, to extract an application cookie passed as a
16072 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016073 this fetch iterates over multiple parameters and will iteratively report all
16074 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016075
Willy Tarreau74ca5042013-06-11 23:12:07 +020016076 ACL derivatives :
16077 urlp(<name>[,<delim>]) : exact string match
16078 urlp_beg(<name>[,<delim>]) : prefix match
16079 urlp_dir(<name>[,<delim>]) : subdir match
16080 urlp_dom(<name>[,<delim>]) : domain match
16081 urlp_end(<name>[,<delim>]) : suffix match
16082 urlp_len(<name>[,<delim>]) : length match
16083 urlp_reg(<name>[,<delim>]) : regex match
16084 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016085
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016086
Willy Tarreau74ca5042013-06-11 23:12:07 +020016087 Example :
16088 # match http://example.com/foo?PHPSESSIONID=some_id
16089 stick on urlp(PHPSESSIONID)
16090 # match http://example.com/foo;JSESSIONID=some_id
16091 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016092
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016093urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016094 See "urlp" above. This one extracts the URL parameter <name> in the request
16095 and converts it to an integer value. This can be used for session stickiness
16096 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016097
Dragan Dosen0070cd52016-06-16 12:19:49 +020016098url32 : integer
16099 This returns a 32-bit hash of the value obtained by concatenating the first
16100 Host header and the whole URL including parameters (not only the path part of
16101 the request, as in the "base32" fetch above). This is useful to track per-URL
16102 activity. A shorter hash is stored, saving a lot of memory. The output type
16103 is an unsigned integer.
16104
16105url32+src : binary
16106 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16107 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16108 the source address family. This can be used to track per-IP, per-URL counters.
16109
Willy Tarreau198a7442008-01-17 12:05:32 +010016110
Willy Tarreau74ca5042013-06-11 23:12:07 +0200161117.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016112---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016113
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016114Some predefined ACLs are hard-coded so that they do not have to be declared in
16115every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016116order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016117
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016118ACL name Equivalent to Usage
16119---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016120FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016121HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016122HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16123HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016124HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16125HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16126HTTP_URL_SLASH url_beg / match URL beginning with "/"
16127HTTP_URL_STAR url * match URL equal to "*"
16128LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016129METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016130METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016131METH_GET method GET HEAD match HTTP GET or HEAD method
16132METH_HEAD method HEAD match HTTP HEAD method
16133METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16134METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016135METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016136METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016137RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016138REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016139TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016140WAIT_END wait_end wait for end of content analysis
16141---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016142
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016143
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161448. Logging
16145----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016146
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016147One of HAProxy's strong points certainly lies is its precise logs. It probably
16148provides the finest level of information available for such a product, which is
16149very important for troubleshooting complex environments. Standard information
16150provided in logs include client ports, TCP/HTTP state timers, precise session
16151state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016152to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016153headers.
16154
16155In order to improve administrators reactivity, it offers a great transparency
16156about encountered problems, both internal and external, and it is possible to
16157send logs to different sources at the same time with different level filters :
16158
16159 - global process-level logs (system errors, start/stop, etc..)
16160 - per-instance system and internal errors (lack of resource, bugs, ...)
16161 - per-instance external troubles (servers up/down, max connections)
16162 - per-instance activity (client connections), either at the establishment or
16163 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016164 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016165 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016166
16167The ability to distribute different levels of logs to different log servers
16168allow several production teams to interact and to fix their problems as soon
16169as possible. For example, the system team might monitor system-wide errors,
16170while the application team might be monitoring the up/down for their servers in
16171real time, and the security team might analyze the activity logs with one hour
16172delay.
16173
16174
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161758.1. Log levels
16176---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016177
Simon Hormandf791f52011-05-29 15:01:10 +090016178TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016179source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016180HTTP request, HTTP return code, number of bytes transmitted, conditions
16181in which the session ended, and even exchanged cookies values. For example
16182track a particular user's problems. All messages may be sent to up to two
16183syslog servers. Check the "log" keyword in section 4.2 for more information
16184about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016185
16186
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161878.2. Log formats
16188----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016189
William Lallemand48940402012-01-30 16:47:22 +010016190HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016191and will be detailed in the following sections. A few of them may vary
16192slightly with the configuration, due to indicators specific to certain
16193options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016194
16195 - the default format, which is very basic and very rarely used. It only
16196 provides very basic information about the incoming connection at the moment
16197 it is accepted : source IP:port, destination IP:port, and frontend-name.
16198 This mode will eventually disappear so it will not be described to great
16199 extents.
16200
16201 - the TCP format, which is more advanced. This format is enabled when "option
16202 tcplog" is set on the frontend. HAProxy will then usually wait for the
16203 connection to terminate before logging. This format provides much richer
16204 information, such as timers, connection counts, queue size, etc... This
16205 format is recommended for pure TCP proxies.
16206
16207 - the HTTP format, which is the most advanced for HTTP proxying. This format
16208 is enabled when "option httplog" is set on the frontend. It provides the
16209 same information as the TCP format with some HTTP-specific fields such as
16210 the request, the status code, and captures of headers and cookies. This
16211 format is recommended for HTTP proxies.
16212
Emeric Brun3a058f32009-06-30 18:26:00 +020016213 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16214 fields arranged in the same order as the CLF format. In this mode, all
16215 timers, captures, flags, etc... appear one per field after the end of the
16216 common fields, in the same order they appear in the standard HTTP format.
16217
William Lallemand48940402012-01-30 16:47:22 +010016218 - the custom log format, allows you to make your own log line.
16219
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016220Next sections will go deeper into details for each of these formats. Format
16221specification will be performed on a "field" basis. Unless stated otherwise, a
16222field is a portion of text delimited by any number of spaces. Since syslog
16223servers are susceptible of inserting fields at the beginning of a line, it is
16224always assumed that the first field is the one containing the process name and
16225identifier.
16226
16227Note : Since log lines may be quite long, the log examples in sections below
16228 might be broken into multiple lines. The example log lines will be
16229 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16230 broken into multiple lines, each non-final line will end with a
16231 backslash ('\') and the next line will start indented by two characters.
16232
16233
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162348.2.1. Default log format
16235-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016236
16237This format is used when no specific option is set. The log is emitted as soon
16238as the connection is accepted. One should note that this currently is the only
16239format which logs the request's destination IP and ports.
16240
16241 Example :
16242 listen www
16243 mode http
16244 log global
16245 server srv1 127.0.0.1:8000
16246
16247 >>> Feb 6 12:12:09 localhost \
16248 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16249 (www/HTTP)
16250
16251 Field Format Extract from the example above
16252 1 process_name '[' pid ']:' haproxy[14385]:
16253 2 'Connect from' Connect from
16254 3 source_ip ':' source_port 10.0.1.2:33312
16255 4 'to' to
16256 5 destination_ip ':' destination_port 10.0.3.31:8012
16257 6 '(' frontend_name '/' mode ')' (www/HTTP)
16258
16259Detailed fields description :
16260 - "source_ip" is the IP address of the client which initiated the connection.
16261 - "source_port" is the TCP port of the client which initiated the connection.
16262 - "destination_ip" is the IP address the client connected to.
16263 - "destination_port" is the TCP port the client connected to.
16264 - "frontend_name" is the name of the frontend (or listener) which received
16265 and processed the connection.
16266 - "mode is the mode the frontend is operating (TCP or HTTP).
16267
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016268In case of a UNIX socket, the source and destination addresses are marked as
16269"unix:" and the ports reflect the internal ID of the socket which accepted the
16270connection (the same ID as reported in the stats).
16271
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016272It is advised not to use this deprecated format for newer installations as it
16273will eventually disappear.
16274
16275
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162768.2.2. TCP log format
16277---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016278
16279The TCP format is used when "option tcplog" is specified in the frontend, and
16280is the recommended format for pure TCP proxies. It provides a lot of precious
16281information for troubleshooting. Since this format includes timers and byte
16282counts, the log is normally emitted at the end of the session. It can be
16283emitted earlier if "option logasap" is specified, which makes sense in most
16284environments with long sessions such as remote terminals. Sessions which match
16285the "monitor" rules are never logged. It is also possible not to emit logs for
16286sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016287specifying "option dontlognull" in the frontend. Successful connections will
16288not be logged if "option dontlog-normal" is specified in the frontend. A few
16289fields may slightly vary depending on some configuration options, those are
16290marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016291
16292 Example :
16293 frontend fnt
16294 mode tcp
16295 option tcplog
16296 log global
16297 default_backend bck
16298
16299 backend bck
16300 server srv1 127.0.0.1:8000
16301
16302 >>> Feb 6 12:12:56 localhost \
16303 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16304 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16305
16306 Field Format Extract from the example above
16307 1 process_name '[' pid ']:' haproxy[14387]:
16308 2 client_ip ':' client_port 10.0.1.2:33313
16309 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16310 4 frontend_name fnt
16311 5 backend_name '/' server_name bck/srv1
16312 6 Tw '/' Tc '/' Tt* 0/0/5007
16313 7 bytes_read* 212
16314 8 termination_state --
16315 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16316 10 srv_queue '/' backend_queue 0/0
16317
16318Detailed fields description :
16319 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016320 connection to haproxy. If the connection was accepted on a UNIX socket
16321 instead, the IP address would be replaced with the word "unix". Note that
16322 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016323 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016324 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016325 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016326
16327 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016328 If the connection was accepted on a UNIX socket instead, the port would be
16329 replaced with the ID of the accepting socket, which is also reported in the
16330 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016331
16332 - "accept_date" is the exact date when the connection was received by haproxy
16333 (which might be very slightly different from the date observed on the
16334 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016335 the same date which may appear in any upstream firewall's log. When used in
16336 HTTP mode, the accept_date field will be reset to the first moment the
16337 connection is ready to receive a new request (end of previous response for
16338 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016339
16340 - "frontend_name" is the name of the frontend (or listener) which received
16341 and processed the connection.
16342
16343 - "backend_name" is the name of the backend (or listener) which was selected
16344 to manage the connection to the server. This will be the same as the
16345 frontend if no switching rule has been applied, which is common for TCP
16346 applications.
16347
16348 - "server_name" is the name of the last server to which the connection was
16349 sent, which might differ from the first one if there were connection errors
16350 and a redispatch occurred. Note that this server belongs to the backend
16351 which processed the request. If the connection was aborted before reaching
16352 a server, "<NOSRV>" is indicated instead of a server name.
16353
16354 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16355 It can be "-1" if the connection was aborted before reaching the queue.
16356 See "Timers" below for more details.
16357
16358 - "Tc" is the total time in milliseconds spent waiting for the connection to
16359 establish to the final server, including retries. It can be "-1" if the
16360 connection was aborted before a connection could be established. See
16361 "Timers" below for more details.
16362
16363 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016364 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016365 "option logasap" was specified, then the time counting stops at the moment
16366 the log is emitted. In this case, a '+' sign is prepended before the value,
16367 indicating that the final one will be larger. See "Timers" below for more
16368 details.
16369
16370 - "bytes_read" is the total number of bytes transmitted from the server to
16371 the client when the log is emitted. If "option logasap" is specified, the
16372 this value will be prefixed with a '+' sign indicating that the final one
16373 may be larger. Please note that this value is a 64-bit counter, so log
16374 analysis tools must be able to handle it without overflowing.
16375
16376 - "termination_state" is the condition the session was in when the session
16377 ended. This indicates the session state, which side caused the end of
16378 session to happen, and for what reason (timeout, error, ...). The normal
16379 flags should be "--", indicating the session was closed by either end with
16380 no data remaining in buffers. See below "Session state at disconnection"
16381 for more details.
16382
16383 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016384 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016385 limits have been reached. For instance, if actconn is close to 512 when
16386 multiple connection errors occur, chances are high that the system limits
16387 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016388 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016389
16390 - "feconn" is the total number of concurrent connections on the frontend when
16391 the session was logged. It is useful to estimate the amount of resource
16392 required to sustain high loads, and to detect when the frontend's "maxconn"
16393 has been reached. Most often when this value increases by huge jumps, it is
16394 because there is congestion on the backend servers, but sometimes it can be
16395 caused by a denial of service attack.
16396
16397 - "beconn" is the total number of concurrent connections handled by the
16398 backend when the session was logged. It includes the total number of
16399 concurrent connections active on servers as well as the number of
16400 connections pending in queues. It is useful to estimate the amount of
16401 additional servers needed to support high loads for a given application.
16402 Most often when this value increases by huge jumps, it is because there is
16403 congestion on the backend servers, but sometimes it can be caused by a
16404 denial of service attack.
16405
16406 - "srv_conn" is the total number of concurrent connections still active on
16407 the server when the session was logged. It can never exceed the server's
16408 configured "maxconn" parameter. If this value is very often close or equal
16409 to the server's "maxconn", it means that traffic regulation is involved a
16410 lot, meaning that either the server's maxconn value is too low, or that
16411 there aren't enough servers to process the load with an optimal response
16412 time. When only one of the server's "srv_conn" is high, it usually means
16413 that this server has some trouble causing the connections to take longer to
16414 be processed than on other servers.
16415
16416 - "retries" is the number of connection retries experienced by this session
16417 when trying to connect to the server. It must normally be zero, unless a
16418 server is being stopped at the same moment the connection was attempted.
16419 Frequent retries generally indicate either a network problem between
16420 haproxy and the server, or a misconfigured system backlog on the server
16421 preventing new connections from being queued. This field may optionally be
16422 prefixed with a '+' sign, indicating that the session has experienced a
16423 redispatch after the maximal retry count has been reached on the initial
16424 server. In this case, the server name appearing in the log is the one the
16425 connection was redispatched to, and not the first one, though both may
16426 sometimes be the same in case of hashing for instance. So as a general rule
16427 of thumb, when a '+' is present in front of the retry count, this count
16428 should not be attributed to the logged server.
16429
16430 - "srv_queue" is the total number of requests which were processed before
16431 this one in the server queue. It is zero when the request has not gone
16432 through the server queue. It makes it possible to estimate the approximate
16433 server's response time by dividing the time spent in queue by the number of
16434 requests in the queue. It is worth noting that if a session experiences a
16435 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016436 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016437 backend queue unless a redispatch occurs.
16438
16439 - "backend_queue" is the total number of requests which were processed before
16440 this one in the backend's global queue. It is zero when the request has not
16441 gone through the global queue. It makes it possible to estimate the average
16442 queue length, which easily translates into a number of missing servers when
16443 divided by a server's "maxconn" parameter. It is worth noting that if a
16444 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016445 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016446 through both the server queue and the backend queue unless a redispatch
16447 occurs.
16448
16449
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164508.2.3. HTTP log format
16451----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016452
16453The HTTP format is the most complete and the best suited for HTTP proxies. It
16454is enabled by when "option httplog" is specified in the frontend. It provides
16455the same level of information as the TCP format with additional features which
16456are specific to the HTTP protocol. Just like the TCP format, the log is usually
16457emitted at the end of the session, unless "option logasap" is specified, which
16458generally only makes sense for download sites. A session which matches the
16459"monitor" rules will never logged. It is also possible not to log sessions for
16460which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016461frontend. Successful connections will not be logged if "option dontlog-normal"
16462is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016463
16464Most fields are shared with the TCP log, some being different. A few fields may
16465slightly vary depending on some configuration options. Those ones are marked
16466with a star ('*') after the field name below.
16467
16468 Example :
16469 frontend http-in
16470 mode http
16471 option httplog
16472 log global
16473 default_backend bck
16474
16475 backend static
16476 server srv1 127.0.0.1:8000
16477
16478 >>> Feb 6 12:14:14 localhost \
16479 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16480 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 +010016481 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016482
16483 Field Format Extract from the example above
16484 1 process_name '[' pid ']:' haproxy[14389]:
16485 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016486 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016487 4 frontend_name http-in
16488 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016489 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016490 7 status_code 200
16491 8 bytes_read* 2750
16492 9 captured_request_cookie -
16493 10 captured_response_cookie -
16494 11 termination_state ----
16495 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16496 13 srv_queue '/' backend_queue 0/0
16497 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16498 15 '{' captured_response_headers* '}' {}
16499 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016500
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016501Detailed fields description :
16502 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016503 connection to haproxy. If the connection was accepted on a UNIX socket
16504 instead, the IP address would be replaced with the word "unix". Note that
16505 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016506 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016507 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016508 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016509
16510 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016511 If the connection was accepted on a UNIX socket instead, the port would be
16512 replaced with the ID of the accepting socket, which is also reported in the
16513 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016514
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016515 - "request_date" is the exact date when the first byte of the HTTP request
16516 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016517
16518 - "frontend_name" is the name of the frontend (or listener) which received
16519 and processed the connection.
16520
16521 - "backend_name" is the name of the backend (or listener) which was selected
16522 to manage the connection to the server. This will be the same as the
16523 frontend if no switching rule has been applied.
16524
16525 - "server_name" is the name of the last server to which the connection was
16526 sent, which might differ from the first one if there were connection errors
16527 and a redispatch occurred. Note that this server belongs to the backend
16528 which processed the request. If the request was aborted before reaching a
16529 server, "<NOSRV>" is indicated instead of a server name. If the request was
16530 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16531
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016532 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16533 request from the client (not counting body) after the first byte was
16534 received. It can be "-1" if the connection was aborted before a complete
16535 request could be received or the a bad request was received. It should
16536 always be very small because a request generally fits in one single packet.
16537 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016538 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16539 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016540
16541 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16542 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016543 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016544
16545 - "Tc" is the total time in milliseconds spent waiting for the connection to
16546 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016547 request was aborted before a connection could be established. See section
16548 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016549
16550 - "Tr" is the total time in milliseconds spent waiting for the server to send
16551 a full HTTP response, not counting data. It can be "-1" if the request was
16552 aborted before a complete response could be received. It generally matches
16553 the server's processing time for the request, though it may be altered by
16554 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016555 "GET" requests generally indicate an overloaded server. See section 8.4
16556 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016557
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016558 - "Ta" is the time the request remained active in haproxy, which is the total
16559 time in milliseconds elapsed between the first byte of the request was
16560 received and the last byte of response was sent. It covers all possible
16561 processing except the handshake (see Th) and idle time (see Ti). There is
16562 one exception, if "option logasap" was specified, then the time counting
16563 stops at the moment the log is emitted. In this case, a '+' sign is
16564 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016565 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016566
16567 - "status_code" is the HTTP status code returned to the client. This status
16568 is generally set by the server, but it might also be set by haproxy when
16569 the server cannot be reached or when its response is blocked by haproxy.
16570
16571 - "bytes_read" is the total number of bytes transmitted to the client when
16572 the log is emitted. This does include HTTP headers. If "option logasap" is
16573 specified, the this value will be prefixed with a '+' sign indicating that
16574 the final one may be larger. Please note that this value is a 64-bit
16575 counter, so log analysis tools must be able to handle it without
16576 overflowing.
16577
16578 - "captured_request_cookie" is an optional "name=value" entry indicating that
16579 the client had this cookie in the request. The cookie name and its maximum
16580 length are defined by the "capture cookie" statement in the frontend
16581 configuration. The field is a single dash ('-') when the option is not
16582 set. Only one cookie may be captured, it is generally used to track session
16583 ID exchanges between a client and a server to detect session crossing
16584 between clients due to application bugs. For more details, please consult
16585 the section "Capturing HTTP headers and cookies" below.
16586
16587 - "captured_response_cookie" is an optional "name=value" entry indicating
16588 that the server has returned a cookie with its response. The cookie name
16589 and its maximum length are defined by the "capture cookie" statement in the
16590 frontend configuration. The field is a single dash ('-') when the option is
16591 not set. Only one cookie may be captured, it is generally used to track
16592 session ID exchanges between a client and a server to detect session
16593 crossing between clients due to application bugs. For more details, please
16594 consult the section "Capturing HTTP headers and cookies" below.
16595
16596 - "termination_state" is the condition the session was in when the session
16597 ended. This indicates the session state, which side caused the end of
16598 session to happen, for what reason (timeout, error, ...), just like in TCP
16599 logs, and information about persistence operations on cookies in the last
16600 two characters. The normal flags should begin with "--", indicating the
16601 session was closed by either end with no data remaining in buffers. See
16602 below "Session state at disconnection" for more details.
16603
16604 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016605 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016606 limits have been reached. For instance, if actconn is close to 512 or 1024
16607 when multiple connection errors occur, chances are high that the system
16608 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016609 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016610 system.
16611
16612 - "feconn" is the total number of concurrent connections on the frontend when
16613 the session was logged. It is useful to estimate the amount of resource
16614 required to sustain high loads, and to detect when the frontend's "maxconn"
16615 has been reached. Most often when this value increases by huge jumps, it is
16616 because there is congestion on the backend servers, but sometimes it can be
16617 caused by a denial of service attack.
16618
16619 - "beconn" is the total number of concurrent connections handled by the
16620 backend when the session was logged. It includes the total number of
16621 concurrent connections active on servers as well as the number of
16622 connections pending in queues. It is useful to estimate the amount of
16623 additional servers needed to support high loads for a given application.
16624 Most often when this value increases by huge jumps, it is because there is
16625 congestion on the backend servers, but sometimes it can be caused by a
16626 denial of service attack.
16627
16628 - "srv_conn" is the total number of concurrent connections still active on
16629 the server when the session was logged. It can never exceed the server's
16630 configured "maxconn" parameter. If this value is very often close or equal
16631 to the server's "maxconn", it means that traffic regulation is involved a
16632 lot, meaning that either the server's maxconn value is too low, or that
16633 there aren't enough servers to process the load with an optimal response
16634 time. When only one of the server's "srv_conn" is high, it usually means
16635 that this server has some trouble causing the requests to take longer to be
16636 processed than on other servers.
16637
16638 - "retries" is the number of connection retries experienced by this session
16639 when trying to connect to the server. It must normally be zero, unless a
16640 server is being stopped at the same moment the connection was attempted.
16641 Frequent retries generally indicate either a network problem between
16642 haproxy and the server, or a misconfigured system backlog on the server
16643 preventing new connections from being queued. This field may optionally be
16644 prefixed with a '+' sign, indicating that the session has experienced a
16645 redispatch after the maximal retry count has been reached on the initial
16646 server. In this case, the server name appearing in the log is the one the
16647 connection was redispatched to, and not the first one, though both may
16648 sometimes be the same in case of hashing for instance. So as a general rule
16649 of thumb, when a '+' is present in front of the retry count, this count
16650 should not be attributed to the logged server.
16651
16652 - "srv_queue" is the total number of requests which were processed before
16653 this one in the server queue. It is zero when the request has not gone
16654 through the server queue. It makes it possible to estimate the approximate
16655 server's response time by dividing the time spent in queue by the number of
16656 requests in the queue. It is worth noting that if a session experiences a
16657 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016658 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016659 backend queue unless a redispatch occurs.
16660
16661 - "backend_queue" is the total number of requests which were processed before
16662 this one in the backend's global queue. It is zero when the request has not
16663 gone through the global queue. It makes it possible to estimate the average
16664 queue length, which easily translates into a number of missing servers when
16665 divided by a server's "maxconn" parameter. It is worth noting that if a
16666 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016667 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016668 through both the server queue and the backend queue unless a redispatch
16669 occurs.
16670
16671 - "captured_request_headers" is a list of headers captured in the request due
16672 to the presence of the "capture request header" statement in the frontend.
16673 Multiple headers can be captured, they will be delimited by a vertical bar
16674 ('|'). When no capture is enabled, the braces do not appear, causing a
16675 shift of remaining fields. It is important to note that this field may
16676 contain spaces, and that using it requires a smarter log parser than when
16677 it's not used. Please consult the section "Capturing HTTP headers and
16678 cookies" below for more details.
16679
16680 - "captured_response_headers" is a list of headers captured in the response
16681 due to the presence of the "capture response header" statement in the
16682 frontend. Multiple headers can be captured, they will be delimited by a
16683 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16684 causing a shift of remaining fields. It is important to note that this
16685 field may contain spaces, and that using it requires a smarter log parser
16686 than when it's not used. Please consult the section "Capturing HTTP headers
16687 and cookies" below for more details.
16688
16689 - "http_request" is the complete HTTP request line, including the method,
16690 request and HTTP version string. Non-printable characters are encoded (see
16691 below the section "Non-printable characters"). This is always the last
16692 field, and it is always delimited by quotes and is the only one which can
16693 contain quotes. If new fields are added to the log format, they will be
16694 added before this field. This field might be truncated if the request is
16695 huge and does not fit in the standard syslog buffer (1024 characters). This
16696 is the reason why this field must always remain the last one.
16697
16698
Cyril Bontédc4d9032012-04-08 21:57:39 +0200166998.2.4. Custom log format
16700------------------------
William Lallemand48940402012-01-30 16:47:22 +010016701
Willy Tarreau2beef582012-12-20 17:22:52 +010016702The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016703mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016704
Davor Ocelice9ed2812017-12-25 17:49:28 +010016705HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016706Variables can take arguments using braces ('{}'), and multiple arguments are
16707separated by commas within the braces. Flags may be added or removed by
16708prefixing them with a '+' or '-' sign.
16709
16710Special variable "%o" may be used to propagate its flags to all other
16711variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016712("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016713
Willy Tarreauc8368452012-12-21 00:09:23 +010016714If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016715as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016716less common information such as the client's SSL certificate's DN, or to log
16717the key that would be used to store an entry into a stick table.
16718
William Lallemand48940402012-01-30 16:47:22 +010016719Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016720In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016721in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016722
Dragan Dosen835b9212016-02-12 13:23:03 +010016723Note: when using the RFC5424 syslog message format, the characters '"',
16724'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16725https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16726such cases, the use of the flag "E" should be considered.
16727
William Lallemand48940402012-01-30 16:47:22 +010016728Flags are :
16729 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016730 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016731 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16732 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016733
16734 Example:
16735
16736 log-format %T\ %t\ Some\ Text
16737 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16738
Dragan Dosen835b9212016-02-12 13:23:03 +010016739 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16740
William Lallemand48940402012-01-30 16:47:22 +010016741At the moment, the default HTTP format is defined this way :
16742
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016743 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16744 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016745
William Lallemandbddd4fd2012-02-27 11:23:10 +010016746the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016747
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016748 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16749 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16750 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016751
William Lallemandbddd4fd2012-02-27 11:23:10 +010016752and the default TCP format is defined this way :
16753
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016754 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16755 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016756
William Lallemand48940402012-01-30 16:47:22 +010016757Please refer to the table below for currently defined variables :
16758
William Lallemandbddd4fd2012-02-27 11:23:10 +010016759 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016760 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016761 +---+------+-----------------------------------------------+-------------+
16762 | | %o | special variable, apply flags on all next var | |
16763 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016764 | | %B | bytes_read (from server to client) | numeric |
16765 | H | %CC | captured_request_cookie | string |
16766 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016767 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016768 | H | %HM | HTTP method (ex: POST) | string |
16769 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016770 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016771 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16772 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016773 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016774 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016775 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016776 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016777 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016778 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016779 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016780 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16781 | H | %Ti | idle time before the HTTP request | numeric |
16782 | H | %Tq | Th + Ti + TR | numeric |
16783 | H | %TR | time to receive the full request from 1st byte| numeric |
16784 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016785 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016786 | | %Tt | Tt | numeric |
16787 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016788 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016789 | | %ac | actconn | numeric |
16790 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016791 | | %bc | beconn (backend concurrent connections) | numeric |
16792 | | %bi | backend_source_ip (connecting address) | IP |
16793 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016794 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016795 | | %ci | client_ip (accepted address) | IP |
16796 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016797 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016798 | | %fc | feconn (frontend concurrent connections) | numeric |
16799 | | %fi | frontend_ip (accepting address) | IP |
16800 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016801 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016802 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016803 | | %hr | captured_request_headers default style | string |
16804 | | %hrl | captured_request_headers CLF style | string list |
16805 | | %hs | captured_response_headers default style | string |
16806 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016807 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016808 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016809 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016810 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016811 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016812 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016813 | | %sc | srv_conn (server concurrent connections) | numeric |
16814 | | %si | server_IP (target address) | IP |
16815 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016816 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016817 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16818 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016819 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016820 | H | %tr | date_time of HTTP request | date |
16821 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016822 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016823 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016824 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016825 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016826
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016827 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016828
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016829
168308.2.5. Error log format
16831-----------------------
16832
16833When an incoming connection fails due to an SSL handshake or an invalid PROXY
16834protocol header, haproxy will log the event using a shorter, fixed line format.
16835By default, logs are emitted at the LOG_INFO level, unless the option
16836"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016837will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016838logged if the "dontlognull" option is set.
16839
16840The format looks like this :
16841
16842 >>> Dec 3 18:27:14 localhost \
16843 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16844 Connection error during SSL handshake
16845
16846 Field Format Extract from the example above
16847 1 process_name '[' pid ']:' haproxy[6103]:
16848 2 client_ip ':' client_port 127.0.0.1:56059
16849 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16850 4 frontend_name "/" bind_name ":" frt/f1:
16851 5 message Connection error during SSL handshake
16852
16853These fields just provide minimal information to help debugging connection
16854failures.
16855
16856
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168578.3. Advanced logging options
16858-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016859
16860Some advanced logging options are often looked for but are not easy to find out
16861just by looking at the various options. Here is an entry point for the few
16862options which can enable better logging. Please refer to the keywords reference
16863for more information about their usage.
16864
16865
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168668.3.1. Disabling logging of external tests
16867------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016868
16869It is quite common to have some monitoring tools perform health checks on
16870haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16871commercial load-balancer, and sometimes it will simply be a more complete
16872monitoring system such as Nagios. When the tests are very frequent, users often
16873ask how to disable logging for those checks. There are three possibilities :
16874
16875 - if connections come from everywhere and are just TCP probes, it is often
16876 desired to simply disable logging of connections without data exchange, by
16877 setting "option dontlognull" in the frontend. It also disables logging of
16878 port scans, which may or may not be desired.
16879
16880 - if the connection come from a known source network, use "monitor-net" to
16881 declare this network as monitoring only. Any host in this network will then
16882 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016883 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016884 such as other load-balancers.
16885
16886 - if the tests are performed on a known URI, use "monitor-uri" to declare
16887 this URI as dedicated to monitoring. Any host sending this request will
16888 only get the result of a health-check, and the request will not be logged.
16889
16890
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168918.3.2. Logging before waiting for the session to terminate
16892----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016893
16894The problem with logging at end of connection is that you have no clue about
16895what is happening during very long sessions, such as remote terminal sessions
16896or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016897"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016898just before data transfer begins. This means that in case of TCP, it will still
16899log the connection status to the server, and in case of HTTP, it will log just
16900after processing the server headers. In this case, the number of bytes reported
16901is the number of header bytes sent to the client. In order to avoid confusion
16902with normal logs, the total time field and the number of bytes are prefixed
16903with a '+' sign which means that real numbers are certainly larger.
16904
16905
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169068.3.3. Raising log level upon errors
16907------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016908
16909Sometimes it is more convenient to separate normal traffic from errors logs,
16910for instance in order to ease error monitoring from log files. When the option
16911"log-separate-errors" is used, connections which experience errors, timeouts,
16912retries, redispatches or HTTP status codes 5xx will see their syslog level
16913raised from "info" to "err". This will help a syslog daemon store the log in
16914a separate file. It is very important to keep the errors in the normal traffic
16915file too, so that log ordering is not altered. You should also be careful if
16916you already have configured your syslog daemon to store all logs higher than
16917"notice" in an "admin" file, because the "err" level is higher than "notice".
16918
16919
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169208.3.4. Disabling logging of successful connections
16921--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016922
16923Although this may sound strange at first, some large sites have to deal with
16924multiple thousands of logs per second and are experiencing difficulties keeping
16925them intact for a long time or detecting errors within them. If the option
16926"dontlog-normal" is set on the frontend, all normal connections will not be
16927logged. In this regard, a normal connection is defined as one without any
16928error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
16929and a response with a status 5xx is not considered normal and will be logged
16930too. Of course, doing is is really discouraged as it will remove most of the
16931useful information from the logs. Do this only if you have no other
16932alternative.
16933
16934
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169358.4. Timing events
16936------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016937
16938Timers provide a great help in troubleshooting network problems. All values are
16939reported in milliseconds (ms). These timers should be used in conjunction with
16940the session termination flags. In TCP mode with "option tcplog" set on the
16941frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016942mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
16943addition, three other measures are provided, "Th", "Ti", and "Tq".
16944
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010016945Timings events in HTTP mode:
16946
16947 first request 2nd request
16948 |<-------------------------------->|<-------------- ...
16949 t tr t tr ...
16950 ---|----|----|----|----|----|----|----|----|--
16951 : Th Ti TR Tw Tc Tr Td : Ti ...
16952 :<---- Tq ---->: :
16953 :<-------------- Tt -------------->:
16954 :<--------- Ta --------->:
16955
16956Timings events in TCP mode:
16957
16958 TCP session
16959 |<----------------->|
16960 t t
16961 ---|----|----|----|----|---
16962 | Th Tw Tc Td |
16963 |<------ Tt ------->|
16964
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016965 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016966 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016967 only happen once during the whole connection's lifetime. A large time here
16968 may indicate that the client only pre-established the connection without
16969 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010016970 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020016971 SSL handshake was very expensive to compute. Please note that this time is
16972 reported only before the first request, so it is safe to average it over
16973 all request to calculate the amortized value. The second and subsequent
16974 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016975
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016976 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
16977 counts between the end of the handshakes and the first byte of the HTTP
16978 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020016979 to count after the end of the transmission the previous response. When a
16980 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
16981 after the previous request. Some browsers pre-establish connections to a
16982 server in order to reduce the latency of a future request, and keep them
16983 pending until they need it. This delay will be reported as the idle time. A
16984 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016985
16986 - TR: total time to get the client request (HTTP mode only). It's the time
16987 elapsed between the first bytes received and the moment the proxy received
16988 the empty line marking the end of the HTTP headers. The value "-1"
16989 indicates that the end of headers has never been seen. This happens when
16990 the client closes prematurely or times out. This time is usually very short
16991 since most requests fit in a single packet. A large time may indicate a
16992 request typed by hand during a test.
16993
16994 - Tq: total time to get the client request from the accept date or since the
16995 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010016996 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 +020016997 returns -1 as well. This timer used to be very useful before the arrival of
16998 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
16999 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17000 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017001
17002 - Tw: total time spent in the queues waiting for a connection slot. It
17003 accounts for backend queue as well as the server queues, and depends on the
17004 queue size, and the time needed for the server to complete previous
17005 requests. The value "-1" means that the request was killed before reaching
17006 the queue, which is generally what happens with invalid or denied requests.
17007
17008 - Tc: total time to establish the TCP connection to the server. It's the time
17009 elapsed between the moment the proxy sent the connection request, and the
17010 moment it was acknowledged by the server, or between the TCP SYN packet and
17011 the matching SYN/ACK packet in return. The value "-1" means that the
17012 connection never established.
17013
17014 - Tr: server response time (HTTP mode only). It's the time elapsed between
17015 the moment the TCP connection was established to the server and the moment
17016 the server sent its complete response headers. It purely shows its request
17017 processing time, without the network overhead due to the data transmission.
17018 It is worth noting that when the client has data to send to the server, for
17019 instance during a POST request, the time already runs, and this can distort
17020 apparent response time. For this reason, it's generally wise not to trust
17021 too much this field for POST requests initiated from clients behind an
17022 untrusted network. A value of "-1" here means that the last the response
17023 header (empty line) was never seen, most likely because the server timeout
17024 stroke before the server managed to process the request.
17025
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017026 - Ta: total active time for the HTTP request, between the moment the proxy
17027 received the first byte of the request header and the emission of the last
17028 byte of the response body. The exception is when the "logasap" option is
17029 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17030 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17031 by subtracting other timers when valid :
17032
17033 Td = Ta - (TR + Tw + Tc + Tr)
17034
17035 Timers with "-1" values have to be excluded from this equation. Note that
17036 "Ta" can never be negative.
17037
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017038 - Tt: total session duration time, between the moment the proxy accepted it
17039 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017040 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17041 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017042 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017043
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017044 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017045
17046 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017047 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17048 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017049
17050These timers provide precious indications on trouble causes. Since the TCP
17051protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17052that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017053due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17054"Tt" is close to a timeout value specified in the configuration, it often means
17055that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017056
17057Most common cases :
17058
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017059 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17060 the client and the proxy. This is very rare on local networks but might
17061 happen when clients are on far remote networks and send large requests. It
17062 may happen that values larger than usual appear here without any network
17063 cause. Sometimes, during an attack or just after a resource starvation has
17064 ended, haproxy may accept thousands of connections in a few milliseconds.
17065 The time spent accepting these connections will inevitably slightly delay
17066 processing of other connections, and it can happen that request times in the
17067 order of a few tens of milliseconds are measured after a few thousands of
17068 new connections have been accepted at once. Using one of the keep-alive
17069 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017070 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017071
17072 - If "Tc" is close to 3000, a packet has probably been lost between the
17073 server and the proxy during the server connection phase. This value should
17074 always be very low, such as 1 ms on local networks and less than a few tens
17075 of ms on remote networks.
17076
Willy Tarreau55165fe2009-05-10 12:02:55 +020017077 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17078 to be the average majored by 3000, there are probably some packets lost
17079 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017080
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017081 - If "Ta" is large even for small byte counts, it generally is because
17082 neither the client nor the server decides to close the connection while
17083 haproxy is running in tunnel mode and both have agreed on a keep-alive
17084 connection mode. In order to solve this issue, it will be needed to specify
17085 one of the HTTP options to manipulate keep-alive or close options on either
17086 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17087 important when connection regulation is used with the "maxconn" option on
17088 the servers, since no new connection will be sent to the server until
17089 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017090
17091Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17092
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017093 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017094 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017095 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017096
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017097 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017098 or it aborted too early. Check the session termination flags
17099 then "timeout http-request" and "timeout client" settings.
17100
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017101 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017102 servers were out of order, because the request was invalid
17103 or forbidden by ACL rules. Check the session termination
17104 flags.
17105
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017106 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17107 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017108 Check the session termination flags, then check the
17109 "timeout connect" setting. Note that the tarpit action might
17110 return similar-looking patterns, with "Tw" equal to the time
17111 the client connection was maintained open.
17112
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017113 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017114 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017115 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017116 termination flags, then check the "timeout server" setting.
17117
17118
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171198.5. Session state at disconnection
17120-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017121
17122TCP and HTTP logs provide a session termination indicator in the
17123"termination_state" field, just before the number of active connections. It is
171242-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17125each of which has a special meaning :
17126
17127 - On the first character, a code reporting the first event which caused the
17128 session to terminate :
17129
17130 C : the TCP session was unexpectedly aborted by the client.
17131
17132 S : the TCP session was unexpectedly aborted by the server, or the
17133 server explicitly refused it.
17134
17135 P : the session was prematurely aborted by the proxy, because of a
17136 connection limit enforcement, because a DENY filter was matched,
17137 because of a security check which detected and blocked a dangerous
17138 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017139 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017140
17141 L : the session was locally processed by haproxy and was not passed to
17142 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017143
17144 R : a resource on the proxy has been exhausted (memory, sockets, source
17145 ports, ...). Usually, this appears during the connection phase, and
17146 system logs should contain a copy of the precise error. If this
17147 happens, it must be considered as a very serious anomaly which
17148 should be fixed as soon as possible by any means.
17149
17150 I : an internal error was identified by the proxy during a self-check.
17151 This should NEVER happen, and you are encouraged to report any log
17152 containing this, because this would almost certainly be a bug. It
17153 would be wise to preventively restart the process after such an
17154 event too, in case it would be caused by memory corruption.
17155
Simon Horman752dc4a2011-06-21 14:34:59 +090017156 D : the session was killed by haproxy because the server was detected
17157 as down and was configured to kill all connections when going down.
17158
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017159 U : the session was killed by haproxy on this backup server because an
17160 active server was detected as up and was configured to kill all
17161 backup connections when going up.
17162
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017163 K : the session was actively killed by an admin operating on haproxy.
17164
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017165 c : the client-side timeout expired while waiting for the client to
17166 send or receive data.
17167
17168 s : the server-side timeout expired while waiting for the server to
17169 send or receive data.
17170
17171 - : normal session completion, both the client and the server closed
17172 with nothing left in the buffers.
17173
17174 - on the second character, the TCP or HTTP session state when it was closed :
17175
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017176 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017177 (HTTP mode only). Nothing was sent to any server.
17178
17179 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17180 only happen when servers have a 'maxconn' parameter set. It can
17181 also happen in the global queue after a redispatch consecutive to
17182 a failed attempt to connect to a dying server. If no redispatch is
17183 reported, then no connection attempt was made to any server.
17184
17185 C : the proxy was waiting for the CONNECTION to establish on the
17186 server. The server might at most have noticed a connection attempt.
17187
17188 H : the proxy was waiting for complete, valid response HEADERS from the
17189 server (HTTP only).
17190
17191 D : the session was in the DATA phase.
17192
17193 L : the proxy was still transmitting LAST data to the client while the
17194 server had already finished. This one is very rare as it can only
17195 happen when the client dies while receiving the last packets.
17196
17197 T : the request was tarpitted. It has been held open with the client
17198 during the whole "timeout tarpit" duration or until the client
17199 closed, both of which will be reported in the "Tw" timer.
17200
17201 - : normal session completion after end of data transfer.
17202
17203 - the third character tells whether the persistence cookie was provided by
17204 the client (only in HTTP mode) :
17205
17206 N : the client provided NO cookie. This is usually the case for new
17207 visitors, so counting the number of occurrences of this flag in the
17208 logs generally indicate a valid trend for the site frequentation.
17209
17210 I : the client provided an INVALID cookie matching no known server.
17211 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017212 cookies between HTTP/HTTPS sites, persistence conditionally
17213 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017214
17215 D : the client provided a cookie designating a server which was DOWN,
17216 so either "option persist" was used and the client was sent to
17217 this server, or it was not set and the client was redispatched to
17218 another server.
17219
Willy Tarreau996a92c2010-10-13 19:30:47 +020017220 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017221 server.
17222
Willy Tarreau996a92c2010-10-13 19:30:47 +020017223 E : the client provided a valid cookie, but with a last date which was
17224 older than what is allowed by the "maxidle" cookie parameter, so
17225 the cookie is consider EXPIRED and is ignored. The request will be
17226 redispatched just as if there was no cookie.
17227
17228 O : the client provided a valid cookie, but with a first date which was
17229 older than what is allowed by the "maxlife" cookie parameter, so
17230 the cookie is consider too OLD and is ignored. The request will be
17231 redispatched just as if there was no cookie.
17232
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017233 U : a cookie was present but was not used to select the server because
17234 some other server selection mechanism was used instead (typically a
17235 "use-server" rule).
17236
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017237 - : does not apply (no cookie set in configuration).
17238
17239 - the last character reports what operations were performed on the persistence
17240 cookie returned by the server (only in HTTP mode) :
17241
17242 N : NO cookie was provided by the server, and none was inserted either.
17243
17244 I : no cookie was provided by the server, and the proxy INSERTED one.
17245 Note that in "cookie insert" mode, if the server provides a cookie,
17246 it will still be overwritten and reported as "I" here.
17247
Willy Tarreau996a92c2010-10-13 19:30:47 +020017248 U : the proxy UPDATED the last date in the cookie that was presented by
17249 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017250 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017251 date indicated in the cookie. If any other change happens, such as
17252 a redispatch, then the cookie will be marked as inserted instead.
17253
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017254 P : a cookie was PROVIDED by the server and transmitted as-is.
17255
17256 R : the cookie provided by the server was REWRITTEN by the proxy, which
17257 happens in "cookie rewrite" or "cookie prefix" modes.
17258
17259 D : the cookie provided by the server was DELETED by the proxy.
17260
17261 - : does not apply (no cookie set in configuration).
17262
Willy Tarreau996a92c2010-10-13 19:30:47 +020017263The combination of the two first flags gives a lot of information about what
17264was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017265helpful to detect server saturation, network troubles, local system resource
17266starvation, attacks, etc...
17267
17268The most common termination flags combinations are indicated below. They are
17269alphabetically sorted, with the lowercase set just after the upper case for
17270easier finding and understanding.
17271
17272 Flags Reason
17273
17274 -- Normal termination.
17275
17276 CC The client aborted before the connection could be established to the
17277 server. This can happen when haproxy tries to connect to a recently
17278 dead (or unchecked) server, and the client aborts while haproxy is
17279 waiting for the server to respond or for "timeout connect" to expire.
17280
17281 CD The client unexpectedly aborted during data transfer. This can be
17282 caused by a browser crash, by an intermediate equipment between the
17283 client and haproxy which decided to actively break the connection,
17284 by network routing issues between the client and haproxy, or by a
17285 keep-alive session between the server and the client terminated first
17286 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017287
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017288 cD The client did not send nor acknowledge any data for as long as the
17289 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017290 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017291
17292 CH The client aborted while waiting for the server to start responding.
17293 It might be the server taking too long to respond or the client
17294 clicking the 'Stop' button too fast.
17295
17296 cH The "timeout client" stroke while waiting for client data during a
17297 POST request. This is sometimes caused by too large TCP MSS values
17298 for PPPoE networks which cannot transport full-sized packets. It can
17299 also happen when client timeout is smaller than server timeout and
17300 the server takes too long to respond.
17301
17302 CQ The client aborted while its session was queued, waiting for a server
17303 with enough empty slots to accept it. It might be that either all the
17304 servers were saturated or that the assigned server was taking too
17305 long a time to respond.
17306
17307 CR The client aborted before sending a full HTTP request. Most likely
17308 the request was typed by hand using a telnet client, and aborted
17309 too early. The HTTP status code is likely a 400 here. Sometimes this
17310 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017311 and the client. "option http-ignore-probes" can be used to ignore
17312 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017313
17314 cR The "timeout http-request" stroke before the client sent a full HTTP
17315 request. This is sometimes caused by too large TCP MSS values on the
17316 client side for PPPoE networks which cannot transport full-sized
17317 packets, or by clients sending requests by hand and not typing fast
17318 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017319 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017320 some browsers started to implement a "pre-connect" feature consisting
17321 in speculatively connecting to some recently visited web sites just
17322 in case the user would like to visit them. This results in many
17323 connections being established to web sites, which end up in 408
17324 Request Timeout if the timeout strikes first, or 400 Bad Request when
17325 the browser decides to close them first. These ones pollute the log
17326 and feed the error counters. Some versions of some browsers have even
17327 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017328 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017329 http-ignore-probes" in the frontend, resulting in connections with
17330 zero data transfer to be totally ignored. This will definitely hide
17331 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017332
17333 CT The client aborted while its session was tarpitted. It is important to
17334 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017335 wrong tarpit rules have been written. If a lot of them happen, it
17336 might make sense to lower the "timeout tarpit" value to something
17337 closer to the average reported "Tw" timer, in order not to consume
17338 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017339
Willy Tarreau570f2212013-06-10 16:42:09 +020017340 LR The request was intercepted and locally handled by haproxy. Generally
17341 it means that this was a redirect or a stats request.
17342
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017343 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017344 the TCP connection (the proxy received a TCP RST or an ICMP message
17345 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017346 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017347 or no ARP response on local network). When this happens in HTTP mode,
17348 the status code is likely a 502 or 503 here.
17349
17350 sC The "timeout connect" stroke before a connection to the server could
17351 complete. When this happens in HTTP mode, the status code is likely a
17352 503 or 504 here.
17353
17354 SD The connection to the server died with an error during the data
17355 transfer. This usually means that haproxy has received an RST from
17356 the server or an ICMP message from an intermediate equipment while
17357 exchanging data with the server. This can be caused by a server crash
17358 or by a network issue on an intermediate equipment.
17359
17360 sD The server did not send nor acknowledge any data for as long as the
17361 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017362 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017363 load-balancers, ...), as well as keep-alive sessions maintained
17364 between the client and the server expiring first on haproxy.
17365
17366 SH The server aborted before sending its full HTTP response headers, or
17367 it crashed while processing the request. Since a server aborting at
17368 this moment is very rare, it would be wise to inspect its logs to
17369 control whether it crashed and why. The logged request may indicate a
17370 small set of faulty requests, demonstrating bugs in the application.
17371 Sometimes this might also be caused by an IDS killing the connection
17372 between haproxy and the server.
17373
17374 sH The "timeout server" stroke before the server could return its
17375 response headers. This is the most common anomaly, indicating too
17376 long transactions, probably caused by server or database saturation.
17377 The immediate workaround consists in increasing the "timeout server"
17378 setting, but it is important to keep in mind that the user experience
17379 will suffer from these long response times. The only long term
17380 solution is to fix the application.
17381
17382 sQ The session spent too much time in queue and has been expired. See
17383 the "timeout queue" and "timeout connect" settings to find out how to
17384 fix this if it happens too often. If it often happens massively in
17385 short periods, it may indicate general problems on the affected
17386 servers due to I/O or database congestion, or saturation caused by
17387 external attacks.
17388
17389 PC The proxy refused to establish a connection to the server because the
17390 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017391 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017392 so that it does not happen anymore. This status is very rare and
17393 might happen when the global "ulimit-n" parameter is forced by hand.
17394
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017395 PD The proxy blocked an incorrectly formatted chunked encoded message in
17396 a request or a response, after the server has emitted its headers. In
17397 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017398 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017399 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017400
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017401 PH The proxy blocked the server's response, because it was invalid,
17402 incomplete, dangerous (cache control), or matched a security filter.
17403 In any case, an HTTP 502 error is sent to the client. One possible
17404 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017405 containing unauthorized characters. It is also possible but quite
17406 rare, that the proxy blocked a chunked-encoding request from the
17407 client due to an invalid syntax, before the server responded. In this
17408 case, an HTTP 400 error is sent to the client and reported in the
17409 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017410
17411 PR The proxy blocked the client's HTTP request, either because of an
17412 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17413 the client, or because a deny filter matched, in which case it
17414 returned an HTTP 403 error.
17415
17416 PT The proxy blocked the client's request and has tarpitted its
17417 connection before returning it a 500 server error. Nothing was sent
17418 to the server. The connection was maintained open for as long as
17419 reported by the "Tw" timer field.
17420
17421 RC A local resource has been exhausted (memory, sockets, source ports)
17422 preventing the connection to the server from establishing. The error
17423 logs will tell precisely what was missing. This is very rare and can
17424 only be solved by proper system tuning.
17425
Willy Tarreau996a92c2010-10-13 19:30:47 +020017426The combination of the two last flags gives a lot of information about how
17427persistence was handled by the client, the server and by haproxy. This is very
17428important to troubleshoot disconnections, when users complain they have to
17429re-authenticate. The commonly encountered flags are :
17430
17431 -- Persistence cookie is not enabled.
17432
17433 NN No cookie was provided by the client, none was inserted in the
17434 response. For instance, this can be in insert mode with "postonly"
17435 set on a GET request.
17436
17437 II A cookie designating an invalid server was provided by the client,
17438 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017439 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017440 value can be presented by a client when no other server knows it.
17441
17442 NI No cookie was provided by the client, one was inserted in the
17443 response. This typically happens for first requests from every user
17444 in "insert" mode, which makes it an easy way to count real users.
17445
17446 VN A cookie was provided by the client, none was inserted in the
17447 response. This happens for most responses for which the client has
17448 already got a cookie.
17449
17450 VU A cookie was provided by the client, with a last visit date which is
17451 not completely up-to-date, so an updated cookie was provided in
17452 response. This can also happen if there was no date at all, or if
17453 there was a date but the "maxidle" parameter was not set, so that the
17454 cookie can be switched to unlimited time.
17455
17456 EI A cookie was provided by the client, with a last visit date which is
17457 too old for the "maxidle" parameter, so the cookie was ignored and a
17458 new cookie was inserted in the response.
17459
17460 OI A cookie was provided by the client, with a first visit date which is
17461 too old for the "maxlife" parameter, so the cookie was ignored and a
17462 new cookie was inserted in the response.
17463
17464 DI The server designated by the cookie was down, a new server was
17465 selected and a new cookie was emitted in the response.
17466
17467 VI The server designated by the cookie was not marked dead but could not
17468 be reached. A redispatch happened and selected another one, which was
17469 then advertised in the response.
17470
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017471
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174728.6. Non-printable characters
17473-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017474
17475In order not to cause trouble to log analysis tools or terminals during log
17476consulting, non-printable characters are not sent as-is into log files, but are
17477converted to the two-digits hexadecimal representation of their ASCII code,
17478prefixed by the character '#'. The only characters that can be logged without
17479being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17480escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17481is the same for the character '"' which becomes "#22", as well as '{', '|' and
17482'}' when logging headers.
17483
17484Note that the space character (' ') is not encoded in headers, which can cause
17485issues for tools relying on space count to locate fields. A typical header
17486containing spaces is "User-Agent".
17487
17488Last, it has been observed that some syslog daemons such as syslog-ng escape
17489the quote ('"') with a backslash ('\'). The reverse operation can safely be
17490performed since no quote may appear anywhere else in the logs.
17491
17492
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174938.7. Capturing HTTP cookies
17494---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017495
17496Cookie capture simplifies the tracking a complete user session. This can be
17497achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017498section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017499cookie will simultaneously be checked in the request ("Cookie:" header) and in
17500the response ("Set-Cookie:" header). The respective values will be reported in
17501the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017502locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017503not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17504user switches to a new session for example, because the server will reassign it
17505a new cookie. It is also possible to detect if a server unexpectedly sets a
17506wrong cookie to a client, leading to session crossing.
17507
17508 Examples :
17509 # capture the first cookie whose name starts with "ASPSESSION"
17510 capture cookie ASPSESSION len 32
17511
17512 # capture the first cookie whose name is exactly "vgnvisitor"
17513 capture cookie vgnvisitor= len 32
17514
17515
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175168.8. Capturing HTTP headers
17517---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017518
17519Header captures are useful to track unique request identifiers set by an upper
17520proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17521the response, one can search for information about the response length, how the
17522server asked the cache to behave, or an object location during a redirection.
17523
17524Header captures are performed using the "capture request header" and "capture
17525response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017526section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017527
17528It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017529time. Non-existent headers are logged as empty strings, and if one header
17530appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017531are grouped within braces '{' and '}' in the same order as they were declared,
17532and delimited with a vertical bar '|' without any space. Response headers
17533follow the same representation, but are displayed after a space following the
17534request headers block. These blocks are displayed just before the HTTP request
17535in the logs.
17536
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017537As a special case, it is possible to specify an HTTP header capture in a TCP
17538frontend. The purpose is to enable logging of headers which will be parsed in
17539an HTTP backend if the request is then switched to this HTTP backend.
17540
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017541 Example :
17542 # This instance chains to the outgoing proxy
17543 listen proxy-out
17544 mode http
17545 option httplog
17546 option logasap
17547 log global
17548 server cache1 192.168.1.1:3128
17549
17550 # log the name of the virtual server
17551 capture request header Host len 20
17552
17553 # log the amount of data uploaded during a POST
17554 capture request header Content-Length len 10
17555
17556 # log the beginning of the referrer
17557 capture request header Referer len 20
17558
17559 # server name (useful for outgoing proxies only)
17560 capture response header Server len 20
17561
17562 # logging the content-length is useful with "option logasap"
17563 capture response header Content-Length len 10
17564
Davor Ocelice9ed2812017-12-25 17:49:28 +010017565 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017566 capture response header Cache-Control len 8
17567
17568 # the Via header will report the next proxy's name
17569 capture response header Via len 20
17570
17571 # log the URL location during a redirection
17572 capture response header Location len 20
17573
17574 >>> Aug 9 20:26:09 localhost \
17575 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17576 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17577 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17578 "GET http://fr.adserver.yahoo.com/"
17579
17580 >>> Aug 9 20:30:46 localhost \
17581 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17582 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17583 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017584 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017585
17586 >>> Aug 9 20:30:46 localhost \
17587 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17588 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17589 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17590 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017591 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017592
17593
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175948.9. Examples of logs
17595---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017596
17597These are real-world examples of logs accompanied with an explanation. Some of
17598them have been made up by hand. The syslog part has been removed for better
17599reading. Their sole purpose is to explain how to decipher them.
17600
17601 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17602 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17603 "HEAD / HTTP/1.0"
17604
17605 => long request (6.5s) entered by hand through 'telnet'. The server replied
17606 in 147 ms, and the session ended normally ('----')
17607
17608 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17609 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17610 0/9 "HEAD / HTTP/1.0"
17611
17612 => Idem, but the request was queued in the global queue behind 9 other
17613 requests, and waited there for 1230 ms.
17614
17615 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17616 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17617 "GET /image.iso HTTP/1.0"
17618
17619 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017620 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017621 14 ms, 243 bytes of headers were sent to the client, and total time from
17622 accept to first data byte is 30 ms.
17623
17624 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17625 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17626 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17627
17628 => the proxy blocked a server response either because of an "rspdeny" or
17629 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017630 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017631 risked being cached. In this case, the response is replaced with a "502
17632 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17633 to return the 502 and not the server.
17634
17635 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017636 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 +010017637
17638 => the client never completed its request and aborted itself ("C---") after
17639 8.5s, while the proxy was waiting for the request headers ("-R--").
17640 Nothing was sent to any server.
17641
17642 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17643 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17644
17645 => The client never completed its request, which was aborted by the
17646 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017647 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017648 send a 408 return code to the client.
17649
17650 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17651 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17652
17653 => This log was produced with "option tcplog". The client timed out after
17654 5 seconds ("c----").
17655
17656 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17657 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017658 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017659
17660 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017661 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017662 (config says 'retries 3'), and no redispatch (otherwise we would have
17663 seen "/+3"). Status code 503 was returned to the client. There were 115
17664 connections on this server, 202 connections on this proxy, and 205 on
17665 the global process. It is possible that the server refused the
17666 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017667
Willy Tarreau52b2d222011-09-07 23:48:48 +020017668
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200176699. Supported filters
17670--------------------
17671
17672Here are listed officially supported filters with the list of parameters they
17673accept. Depending on compile options, some of these filters might be
17674unavailable. The list of available filters is reported in haproxy -vv.
17675
17676See also : "filter"
17677
176789.1. Trace
17679----------
17680
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017681filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017682
17683 Arguments:
17684 <name> is an arbitrary name that will be reported in
17685 messages. If no name is provided, "TRACE" is used.
17686
17687 <random-parsing> enables the random parsing of data exchanged between
17688 the client and the server. By default, this filter
17689 parses all available data. With this parameter, it
17690 only parses a random amount of the available data.
17691
Davor Ocelice9ed2812017-12-25 17:49:28 +010017692 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017693 default, this filter forwards all previously parsed
17694 data. With this parameter, it only forwards a random
17695 amount of the parsed data.
17696
Davor Ocelice9ed2812017-12-25 17:49:28 +010017697 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017698
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017699This filter can be used as a base to develop new filters. It defines all
17700callbacks and print a message on the standard error stream (stderr) with useful
17701information for all of them. It may be useful to debug the activity of other
17702filters or, quite simply, HAProxy's activity.
17703
17704Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17705tests the behavior of a filter that parses data exchanged between a client and
17706a server by adding some latencies in the processing.
17707
17708
177099.2. HTTP compression
17710---------------------
17711
17712filter compression
17713
17714The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17715keyword must still be used to enable and configure the HTTP compression. And
17716when no other filter is used, it is enough. But it is mandatory to explicitly
17717use a filter line to enable the HTTP compression when two or more filters are
17718used for the same listener/frontend/backend. This is important to know the
17719filters evaluation order.
17720
17721See also : "compression"
17722
17723
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200177249.3. Stream Processing Offload Engine (SPOE)
17725--------------------------------------------
17726
17727filter spoe [engine <name>] config <file>
17728
17729 Arguments :
17730
17731 <name> is the engine name that will be used to find the right scope in
17732 the configuration file. If not provided, all the file will be
17733 parsed.
17734
17735 <file> is the path of the engine configuration file. This file can
17736 contain configuration of several engines. In this case, each
17737 part must be placed in its own scope.
17738
17739The Stream Processing Offload Engine (SPOE) is a filter communicating with
17740external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017741streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017742exchanged with them are configured in dedicated files, for the main part. It
17743also requires dedicated backends, defined in HAProxy configuration.
17744
17745SPOE communicates with external components using an in-house binary protocol,
17746the Stream Processing Offload Protocol (SPOP).
17747
Tim Düsterhus4896c442016-11-29 02:15:19 +010017748For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017749"doc/SPOE.txt".
17750
17751Important note:
17752 The SPOE filter is highly experimental for now and was not heavily
17753 tested. It is really not production ready. So use it carefully.
17754
William Lallemand86d0df02017-11-24 21:36:45 +01001775510. Cache
17756---------
17757
17758HAProxy provides a cache, which was designed to perform cache on small objects
17759(favicon, css...). This is a minimalist low-maintenance cache which runs in
17760RAM.
17761
17762The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017763this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017764
17765If an object is not used anymore, it can be deleted to store a new object
17766independently of its expiration date. The oldest objects are deleted first
17767when we try to allocate a new one.
17768
Cyril Bonté7b888f12017-11-26 22:24:31 +010017769The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017770
17771It's possible to view the status of a cache using the Unix socket command
17772"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17773for more details.
17774
17775When an object is delivered from the cache, the server name in the log is
17776replaced by "<CACHE>".
17777
Cyril Bonté7b888f12017-11-26 22:24:31 +01001777810.1. Limitation
17779----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017780
17781The cache won't store and won't deliver objects in these cases:
17782
17783- If the response is not a 200
17784- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017785- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010017786- If the response is not cacheable
17787
17788- If the request is not a GET
17789- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020017790- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010017791
17792Caution!: Due to the current limitation of the filters, it is not recommended
17793to use the cache with other filters. Using them can cause undefined behavior
Cyril Bonté7b888f12017-11-26 22:24:31 +010017794if they modify the response (compression for example).
William Lallemand86d0df02017-11-24 21:36:45 +010017795
Cyril Bonté7b888f12017-11-26 22:24:31 +01001779610.2. Setup
17797-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017798
17799To setup a cache, you must define a cache section and use it in a proxy with
17800the corresponding http-request and response actions.
17801
Cyril Bonté7b888f12017-11-26 22:24:31 +01001780210.2.1. Cache section
17803---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017804
17805cache <name>
17806 Declare a cache section, allocate a shared cache memory named <name>, the
17807 size of cache is mandatory.
17808
17809total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017810 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017811 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010017812
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017813max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017814 Define the maximum size of the objects to be cached. Must not be greater than
17815 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
17816 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017817
William Lallemand86d0df02017-11-24 21:36:45 +010017818max-age <seconds>
17819 Define the maximum expiration duration. The expiration is set has the lowest
17820 value between the s-maxage or max-age (in this order) directive in the
17821 Cache-Control response header and this value. The default value is 60
17822 seconds, which means that you can't cache an object more than 60 seconds by
17823 default.
17824
Cyril Bonté7b888f12017-11-26 22:24:31 +01001782510.2.2. Proxy section
17826---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017827
17828http-request cache-use <name>
17829 Try to deliver a cached object from the cache <name>. This directive is also
17830 mandatory to store the cache as it calculates the cache hash. If you want to
17831 use a condition for both storage and delivering that's a good idea to put it
17832 after this one.
17833
17834http-response cache-store <name>
17835 Store an http-response within the cache. The storage of the response headers
17836 is done at this step, which means you can use others http-response actions
17837 to modify headers before or after the storage of the response. This action
17838 is responsible for the setup of the cache storage filter.
17839
17840
17841Example:
17842
17843 backend bck1
17844 mode http
17845
17846 http-request cache-use foobar
17847 http-response cache-store foobar
17848 server srv1 127.0.0.1:80
17849
17850 cache foobar
17851 total-max-size 4
17852 max-age 240
17853
Willy Tarreau0ba27502007-12-24 16:55:16 +010017854/*
17855 * Local variables:
17856 * fill-column: 79
17857 * End:
17858 */