blob: d6f8b8d3884fc7046b9892ee6067da4b0fdfa719 [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 Tarreau0b787922017-11-26 19:25:23 +01007 2017/11/26
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
Willy Tarreau70dffda2014-01-30 03:07:23 +0100184HAProxy supports 5 connection modes :
185 - 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.
188 - passive close : tunnel with "Connection: close" added in both directions.
189 - server close : the server-facing connection is closed after the response.
190 - forced close : the connection is actively closed after end of response.
191
Davor Ocelice9ed2812017-12-25 17:49:28 +0100192For HTTP/2, the connection mode resembles more the "server close" mode : given
193the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100194server connection after a response, so it is closed after the response. HTTP/2
195is only supported for incoming connections, not on connections going to
196servers.
197
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198
1991.2. HTTP request
200-----------------
201
202First, let's consider this HTTP request :
203
204 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100205 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200206 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
207 2 Host: www.mydomain.com
208 3 User-agent: my small browser
209 4 Accept: image/jpeg, image/gif
210 5 Accept: image/png
211
212
2131.2.1. The Request line
214-----------------------
215
216Line 1 is the "request line". It is always composed of 3 fields :
217
218 - a METHOD : GET
219 - a URI : /serv/login.php?lang=en&profile=2
220 - a version tag : HTTP/1.1
221
222All of them are delimited by what the standard calls LWS (linear white spaces),
223which are commonly spaces, but can also be tabs or line feeds/carriage returns
224followed by spaces/tabs. The method itself cannot contain any colon (':') and
225is limited to alphabetic letters. All those various combinations make it
226desirable that HAProxy performs the splitting itself rather than leaving it to
227the user to write a complex or inaccurate regular expression.
228
229The URI itself can have several forms :
230
231 - A "relative URI" :
232
233 /serv/login.php?lang=en&profile=2
234
235 It is a complete URL without the host part. This is generally what is
236 received by servers, reverse proxies and transparent proxies.
237
238 - An "absolute URI", also called a "URL" :
239
240 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
241
242 It is composed of a "scheme" (the protocol name followed by '://'), a host
243 name or address, optionally a colon (':') followed by a port number, then
244 a relative URI beginning at the first slash ('/') after the address part.
245 This is generally what proxies receive, but a server supporting HTTP/1.1
246 must accept this form too.
247
248 - a star ('*') : this form is only accepted in association with the OPTIONS
249 method and is not relayable. It is used to inquiry a next hop's
250 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100251
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200252 - an address:port combination : 192.168.0.12:80
253 This is used with the CONNECT method, which is used to establish TCP
254 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
255 other protocols too.
256
257In a relative URI, two sub-parts are identified. The part before the question
258mark is called the "path". It is typically the relative path to static objects
259on the server. The part after the question mark is called the "query string".
260It is mostly used with GET requests sent to dynamic scripts and is very
261specific to the language, framework or application in use.
262
Willy Tarreau95c4e142017-11-26 12:18:55 +0100263HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100264assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100265However, haproxy natively processes HTTP/1.x requests and headers, so requests
266received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
267processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
268as well as in server logs.
269
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200270
2711.2.2. The request headers
272--------------------------
273
274The headers start at the second line. They are composed of a name at the
275beginning of the line, immediately followed by a colon (':'). Traditionally,
276an LWS is added after the colon but that's not required. Then come the values.
277Multiple identical headers may be folded into one single line, delimiting the
278values with commas, provided that their order is respected. This is commonly
279encountered in the "Cookie:" field. A header may span over multiple lines if
280the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
281define a total of 3 values for the "Accept:" header.
282
Davor Ocelice9ed2812017-12-25 17:49:28 +0100283Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200284their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100285"Connection:" header). In HTTP/2, header names are always sent in lower case,
286as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200287
288The end of the headers is indicated by the first empty line. People often say
289that it's a double line feed, which is not exact, even if a double line feed
290is one valid form of empty line.
291
292Fortunately, HAProxy takes care of all these complex combinations when indexing
293headers, checking values and counting them, so there is no reason to worry
294about the way they could be written, but it is important not to accuse an
295application of being buggy if it does unusual, valid things.
296
297Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000298 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200299 in the middle of headers by LWS in order to join multi-line headers. This
300 is necessary for proper analysis and helps less capable HTTP parsers to work
301 correctly and not to be fooled by such complex constructs.
302
303
3041.3. HTTP response
305------------------
306
307An HTTP response looks very much like an HTTP request. Both are called HTTP
308messages. Let's consider this HTTP response :
309
310 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100311 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200312 1 HTTP/1.1 200 OK
313 2 Content-length: 350
314 3 Content-Type: text/html
315
Willy Tarreau816b9792009-09-15 21:25:21 +0200316As a special case, HTTP supports so called "Informational responses" as status
317codes 1xx. These messages are special in that they don't convey any part of the
318response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100319continue to post its request for instance. In the case of a status 100 response
320the requested information will be carried by the next non-100 response message
321following the informational one. This implies that multiple responses may be
322sent to a single request, and that this only works when keep-alive is enabled
323(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
324correctly forward and skip them, and only process the next non-100 response. As
325such, these messages are neither logged nor transformed, unless explicitly
326state otherwise. Status 101 messages indicate that the protocol is changing
327over the same connection and that haproxy must switch to tunnel mode, just as
328if a CONNECT had occurred. Then the Upgrade header would contain additional
329information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200330
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200331
Davor Ocelice9ed2812017-12-25 17:49:28 +01003321.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200333------------------------
334
335Line 1 is the "response line". It is always composed of 3 fields :
336
337 - a version tag : HTTP/1.1
338 - a status code : 200
339 - a reason : OK
340
341The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100342 - 1xx = informational message to be skipped (e.g. 100, 101)
343 - 2xx = OK, content is following (e.g. 200, 206)
344 - 3xx = OK, no content following (e.g. 302, 304)
345 - 4xx = error caused by the client (e.g. 401, 403, 404)
346 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200347
Lukas Tribus23953682017-04-28 13:24:30 +0000348Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100349"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200350found there, but it's a common practice to respect the well-established
351messages. It can be composed of one or multiple words, such as "OK", "Found",
352or "Authentication Required".
353
Davor Ocelice9ed2812017-12-25 17:49:28 +0100354HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200355
356 Code When / reason
357 200 access to stats page, and when replying to monitoring requests
358 301 when performing a redirection, depending on the configured code
359 302 when performing a redirection, depending on the configured code
360 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100361 307 when performing a redirection, depending on the configured code
362 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200363 400 for an invalid or too large request
364 401 when an authentication is required to perform the action (when
365 accessing the stats page)
366 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
367 408 when the request timeout strikes before the request is complete
368 500 when haproxy encounters an unrecoverable internal error, such as a
369 memory allocation failure, which should never happen
370 502 when the server returns an empty, invalid or incomplete response, or
371 when an "rspdeny" filter blocks the response.
372 503 when no server was available to handle the request, or in response to
373 monitoring requests which match the "monitor fail" condition
374 504 when the response timeout strikes before the server responds
375
376The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3774.2).
378
379
3801.3.2. The response headers
381---------------------------
382
383Response headers work exactly like request headers, and as such, HAProxy uses
384the same parsing function for both. Please refer to paragraph 1.2.2 for more
385details.
386
387
3882. Configuring HAProxy
389----------------------
390
3912.1. Configuration file format
392------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200393
394HAProxy's configuration process involves 3 major sources of parameters :
395
396 - the arguments from the command-line, which always take precedence
397 - the "global" section, which sets process-wide parameters
398 - the proxies sections which can take form of "defaults", "listen",
399 "frontend" and "backend".
400
Willy Tarreau0ba27502007-12-24 16:55:16 +0100401The configuration file syntax consists in lines beginning with a keyword
402referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200403delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100404
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200405
William Lallemandf9873ba2015-05-05 17:37:14 +02004062.2. Quoting and escaping
407-------------------------
408
409HAProxy's configuration introduces a quoting and escaping system similar to
410many programming languages. The configuration file supports 3 types: escaping
411with a backslash, weak quoting with double quotes, and strong quoting with
412single quotes.
413
414If spaces have to be entered in strings, then they must be escaped by preceding
415them by a backslash ('\') or by quoting them. Backslashes also have to be
416escaped by doubling or strong quoting them.
417
418Escaping is achieved by preceding a special character by a backslash ('\'):
419
420 \ to mark a space and differentiate it from a delimiter
421 \# to mark a hash and differentiate it from a comment
422 \\ to use a backslash
423 \' to use a single quote and differentiate it from strong quoting
424 \" to use a double quote and differentiate it from weak quoting
425
426Weak quoting is achieved by using double quotes (""). Weak quoting prevents
427the interpretation of:
428
429 space as a parameter separator
430 ' single quote as a strong quoting delimiter
431 # hash as a comment start
432
William Lallemandb2f07452015-05-12 14:27:13 +0200433Weak quoting permits the interpretation of variables, if you want to use a non
434-interpreted dollar within a double quoted string, you should escape it with a
435backslash ("\$"), it does not work outside weak quoting.
436
437Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200438quoting.
439
440Strong quoting is achieved by using single quotes (''). Inside single quotes,
441nothing is interpreted, it's the efficient way to quote regexes.
442
443Quoted and escaped strings are replaced in memory by their interpreted
444equivalent, it allows you to perform concatenation.
445
446 Example:
447 # those are equivalents:
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 log-format "%{+Q}o %t"' %s'\ %{-Q}r
453
454 # those are equivalents:
455 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
456 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
457 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
458 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
459
460
William Lallemandb2f07452015-05-12 14:27:13 +02004612.3. Environment variables
462--------------------------
463
464HAProxy's configuration supports environment variables. Those variables are
465interpreted only within double quotes. Variables are expanded during the
466configuration parsing. Variable names must be preceded by a dollar ("$") and
467optionally enclosed with braces ("{}") similarly to what is done in Bourne
468shell. Variable names can contain alphanumerical characters or the character
469underscore ("_") but should not start with a digit.
470
471 Example:
472
473 bind "fd@${FD_APP1}"
474
475 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
476
477 user "$HAPROXY_USER"
478
479
4802.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200481----------------
482
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100483Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100484values are generally expressed in milliseconds (unless explicitly stated
485otherwise) but may be expressed in any other unit by suffixing the unit to the
486numeric value. It is important to consider this because it will not be repeated
487for every keyword. Supported units are :
488
489 - us : microseconds. 1 microsecond = 1/1000000 second
490 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
491 - s : seconds. 1s = 1000ms
492 - m : minutes. 1m = 60s = 60000ms
493 - h : hours. 1h = 60m = 3600s = 3600000ms
494 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
495
496
Lukas Tribusaa83a312017-03-21 09:25:09 +00004972.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200498-------------
499
500 # Simple configuration for an HTTP proxy listening on port 80 on all
501 # interfaces and forwarding requests to a single backend "servers" with a
502 # single server "server1" listening on 127.0.0.1:8000
503 global
504 daemon
505 maxconn 256
506
507 defaults
508 mode http
509 timeout connect 5000ms
510 timeout client 50000ms
511 timeout server 50000ms
512
513 frontend http-in
514 bind *:80
515 default_backend servers
516
517 backend servers
518 server server1 127.0.0.1:8000 maxconn 32
519
520
521 # The same configuration defined with a single listen block. Shorter but
522 # less expressive, especially in HTTP mode.
523 global
524 daemon
525 maxconn 256
526
527 defaults
528 mode http
529 timeout connect 5000ms
530 timeout client 50000ms
531 timeout server 50000ms
532
533 listen http-in
534 bind *:80
535 server server1 127.0.0.1:8000 maxconn 32
536
537
538Assuming haproxy is in $PATH, test these configurations in a shell with:
539
Willy Tarreauccb289d2010-12-11 20:19:38 +0100540 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200541
542
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005433. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200544--------------------
545
546Parameters in the "global" section are process-wide and often OS-specific. They
547are generally set once for all and do not need being changed once correct. Some
548of them have command-line equivalents.
549
550The following keywords are supported in the "global" section :
551
552 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200553 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200554 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200555 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200556 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200557 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200558 - description
559 - deviceatlas-json-file
560 - deviceatlas-log-level
561 - deviceatlas-separator
562 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900563 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200564 - gid
565 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100566 - hard-stop-after
Willy Tarreau6a06a402007-07-15 20:15:28 +0200567 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200568 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100569 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200570 - lua-load
Willy Tarreau6a06a402007-07-15 20:15:28 +0200571 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200572 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200573 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200574 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100575 - presetenv
576 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577 - uid
578 - ulimit-n
579 - user
Willy Tarreau1d549722016-02-16 12:41:57 +0100580 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200581 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200582 - ssl-default-bind-ciphers
583 - ssl-default-bind-options
584 - ssl-default-server-ciphers
585 - ssl-default-server-options
586 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100587 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100588 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100589 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100590 - 51degrees-data-file
591 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200592 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200593 - 51degrees-cache-size
scientiamobiled0027ed2016-11-04 10:55:08 +0100594 - wurfl-data-file
595 - wurfl-information-list
596 - wurfl-information-list-separator
597 - wurfl-engine-mode
598 - wurfl-cache-size
599 - wurfl-useragent-priority
Willy Tarreaud72758d2010-01-12 10:42:19 +0100600
Willy Tarreau6a06a402007-07-15 20:15:28 +0200601 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200602 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200603 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200604 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100605 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100606 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100607 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200608 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200609 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200610 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200611 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200612 - noepoll
613 - nokqueue
614 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100615 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300616 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000617 - noreuseport
Willy Tarreaufe255b72007-10-14 23:09:26 +0200618 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200619 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200620 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000621 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000622 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200623 - tune.buffers.limit
624 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200625 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200626 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100627 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200628 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200629 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200630 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100631 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200632 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200633 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100634 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100635 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100636 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100637 - tune.lua.session-timeout
638 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200639 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100640 - tune.maxaccept
641 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200642 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200643 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200644 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100645 - tune.rcvbuf.client
646 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100647 - tune.recv_enough
Willy Tarreaue803de22010-01-21 17:43:04 +0100648 - tune.sndbuf.client
649 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100650 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100651 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200652 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100653 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200654 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200655 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100656 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200657 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100658 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200659 - tune.vars.reqres-max-size
660 - tune.vars.sess-max-size
661 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100662 - tune.zlib.memlevel
663 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100664
Willy Tarreau6a06a402007-07-15 20:15:28 +0200665 * Debugging
666 - debug
667 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200668
669
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006703.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200671------------------------------------
672
Emeric Brunc8e8d122012-10-02 18:42:10 +0200673ca-base <dir>
674 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200675 relative path is used with "ca-file" or "crl-file" directives. Absolute
676 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200677
Willy Tarreau6a06a402007-07-15 20:15:28 +0200678chroot <jail dir>
679 Changes current directory to <jail dir> and performs a chroot() there before
680 dropping privileges. This increases the security level in case an unknown
681 vulnerability would be exploited, since it would make it very hard for the
682 attacker to exploit the system. This only works when the process is started
683 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100684 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100685
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100686cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
687 On Linux 2.6 and above, it is possible to bind a process or a thread to a
688 specific CPU set. This means that the process or the thread will never run on
689 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
690 sets. The first argument is a process set, eventually followed by a thread
691 set. These sets have the format
692
693 all | odd | even | number[-[number]]
694
695 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100696 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100697 ignored. It is possible to specify a range with two such number delimited by
698 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100699 "all", only odd numbers using "odd" or even numbers using "even", just like
700 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100701 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 +0100702 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100703 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100704 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100705 specified. Each "cpu-map" directive will replace the previous ones when they
706 overlap. A thread will be bound on the intersection of its mapping and the
707 one of the process on which it is attached. If the intersection is null, no
708 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100709
Christopher Fauletff4121f2017-11-22 16:38:49 +0100710 Ranges can be partially defined. The higher bound can be omitted. In such
711 case, it is replaced by the corresponding maximum value, 32 or 64 depending
712 on the machine's word size.
713
Christopher Faulet26028f62017-11-22 15:01:51 +0100714 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100715 automatically bind a process or a thread to a CPU by incrementing
716 process/thread and CPU sets. To be valid, both sets must have the same
717 size. No matter the declaration order of the CPU sets, it will be bound from
718 the lowest to the highest bound. Having a process and a thread range with the
719 "auto:" prefix is not supported. Only one range is supported, the other one
720 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100721
722 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100723 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
724
725 cpu-map 1/all 0-3 # bind all threads of the first process on the
726 # first 4 CPUs
727
728 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
729 # or "cpu-map 1-32 0-31" depending on the machine's
730 # word size.
731
Christopher Faulet26028f62017-11-22 15:01:51 +0100732 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100733 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100734 cpu-map auto:1-4 0-3
735 cpu-map auto:1-4 0-1 2-3
736 cpu-map auto:1-4 3 2 1 0
737
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100738 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
739 # and so on.
740 cpu-map auto:1/1-4 0-3
741 cpu-map auto:1/1-4 0-1 2-3
742 cpu-map auto:1/1-4 3 2 1 0
743
Davor Ocelice9ed2812017-12-25 17:49:28 +0100744 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100745 cpu-map auto:all 0-63
746 cpu-map auto:even 0-31
747 cpu-map auto:odd 32-63
748
749 # invalid cpu-map because process and CPU sets have different sizes.
750 cpu-map auto:1-4 0 # invalid
751 cpu-map auto:1 0-3 # invalid
752
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100753 # invalid cpu-map because automatic binding is used with a process range
754 # and a thread range.
755 cpu-map auto:all/all 0 # invalid
756 cpu-map auto:all/1-4 0 # invalid
757 cpu-map auto:1-4/all 0 # invalid
758
Emeric Brunc8e8d122012-10-02 18:42:10 +0200759crt-base <dir>
760 Assigns a default directory to fetch SSL certificates from when a relative
761 path is used with "crtfile" directives. Absolute locations specified after
762 "crtfile" prevail and ignore "crt-base".
763
Willy Tarreau6a06a402007-07-15 20:15:28 +0200764daemon
765 Makes the process fork into background. This is the recommended mode of
766 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100767 disabled by the command line "-db" argument. This option is ignored in
768 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200769
David Carlier8167f302015-06-01 13:50:06 +0200770deviceatlas-json-file <path>
771 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100772 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200773
774deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100775 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200776 optional and set to 0 by default if not set.
777
778deviceatlas-separator <char>
779 Sets the character separator for the API properties results. This directive
780 is optional and set to | by default if not set.
781
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100782deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200783 Sets the client cookie's name used for the detection if the DeviceAtlas
784 Client-side component was used during the request. This directive is optional
785 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100786
Simon Horman98637e52014-06-20 12:30:16 +0900787external-check
788 Allows the use of an external agent to perform health checks.
789 This is disabled by default as a security precaution.
790 See "option external-check".
791
Willy Tarreau6a06a402007-07-15 20:15:28 +0200792gid <number>
793 Changes the process' group ID to <number>. It is recommended that the group
794 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
795 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100796 Note that if haproxy is started from a user having supplementary groups, it
797 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200798 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100799
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100800hard-stop-after <time>
801 Defines the maximum time allowed to perform a clean soft-stop.
802
803 Arguments :
804 <time> is the maximum time (by default in milliseconds) for which the
805 instance will remain alive when a soft-stop is received via the
806 SIGUSR1 signal.
807
808 This may be used to ensure that the instance will quit even if connections
809 remain opened during a soft-stop (for example with long timeouts for a proxy
810 in tcp mode). It applies both in TCP and HTTP mode.
811
812 Example:
813 global
814 hard-stop-after 30s
815
Willy Tarreau6a06a402007-07-15 20:15:28 +0200816group <group name>
817 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
818 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100819
Dragan Dosen7ad31542015-09-28 17:16:47 +0200820log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Willy Tarreau6a06a402007-07-15 20:15:28 +0200821 Adds a global syslog server. Up to two global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100822 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100823 configured with "log global".
824
825 <address> can be one of:
826
Willy Tarreau2769aa02007-12-27 18:26:09 +0100827 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100828 no port is specified, 514 is used by default (the standard syslog
829 port).
830
David du Colombier24bb5f52011-03-17 10:40:23 +0100831 - An IPv6 address followed by a colon and optionally a UDP port. If
832 no port is specified, 514 is used by default (the standard syslog
833 port).
834
Robert Tsai81ae1952007-12-05 10:47:29 +0100835 - A filesystem path to a UNIX domain socket, keeping in mind
836 considerations for chroot (be sure the path is accessible inside
837 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100838 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100839
William Lallemandb2f07452015-05-12 14:27:13 +0200840 You may want to reference some environment variables in the address
841 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100842
Willy Tarreau18324f52014-06-27 18:10:07 +0200843 <length> is an optional maximum line length. Log lines larger than this value
844 will be truncated before being sent. The reason is that syslog
845 servers act differently on log line length. All servers support the
846 default value of 1024, but some servers simply drop larger lines
847 while others do log them. If a server supports long lines, it may
848 make sense to set this value here in order to avoid truncating long
849 lines. Similarly, if a server drops long lines, it is preferable to
850 truncate them before sending them. Accepted values are 80 to 65535
851 inclusive. The default value of 1024 is generally fine for all
852 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100853 JSON-formatted logs may require larger values. You may also need to
854 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200855
Dragan Dosen7ad31542015-09-28 17:16:47 +0200856 <format> is the log format used when generating syslog messages. It may be
857 one of the following :
858
859 rfc3164 The RFC3164 syslog message format. This is the default.
860 (https://tools.ietf.org/html/rfc3164)
861
862 rfc5424 The RFC5424 syslog message format.
863 (https://tools.ietf.org/html/rfc5424)
864
Robert Tsai81ae1952007-12-05 10:47:29 +0100865 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200866
867 kern user mail daemon auth syslog lpr news
868 uucp cron auth2 ftp ntp audit alert cron2
869 local0 local1 local2 local3 local4 local5 local6 local7
870
871 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200872 all messages are sent. If a maximum level is specified, only messages with a
873 severity at least as important as this level will be sent. An optional minimum
874 level can be specified. If it is set, logs emitted with a more severe level
875 than this one will be capped to this level. This is used to avoid sending
876 "emerg" messages on all terminals on some default syslog configurations.
877 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200878
Cyril Bontédc4d9032012-04-08 21:57:39 +0200879 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200880
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100881log-send-hostname [<string>]
882 Sets the hostname field in the syslog header. If optional "string" parameter
883 is set the header is set to the string contents, otherwise uses the hostname
884 of the system. Generally used if one is not relaying logs through an
885 intermediate syslog server or for simply customizing the hostname printed in
886 the logs.
887
Kevinm48936af2010-12-22 16:08:21 +0000888log-tag <string>
889 Sets the tag field in the syslog header to this string. It defaults to the
890 program name as launched from the command line, which usually is "haproxy".
891 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100892 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000893
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100894lua-load <file>
895 This global directive loads and executes a Lua file. This directive can be
896 used multiple times.
897
William Lallemand4cfede82017-11-24 22:02:34 +0100898master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200899 Master-worker mode. It is equivalent to the command line "-W" argument.
900 This mode will launch a "master" which will monitor the "workers". Using
901 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100902 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200903 or daemon mode. It is recommended to use this mode with multiprocess and
904 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100905 By default, if a worker exits with a bad return code, in the case of a
906 segfault for example, all workers will be killed, and the master will leave.
907 It is convenient to combine this behavior with Restart=on-failure in a
908 systemd unit file in order to relaunch the whole process. If you don't want
909 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200910
William Lallemand4cfede82017-11-24 22:02:34 +0100911 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200912
Willy Tarreau6a06a402007-07-15 20:15:28 +0200913nbproc <number>
914 Creates <number> processes when going daemon. This requires the "daemon"
915 mode. By default, only one process is created, which is the recommended mode
916 of operation. For systems limited to small sets of file descriptors per
917 process, it may be needed to fork multiple daemons. USING MULTIPLE PROCESSES
918 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon".
919
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200920nbthread <number>
921 This setting is only available when support for threads was built in. It
922 creates <number> threads for each created processes. It means if HAProxy is
923 started in foreground, it only creates <number> threads for the first
924 process. FOR NOW, THREADS SUPPORT IN HAPROXY IS HIGHLY EXPERIMENTAL AND IT
925 MUST BE ENABLED WITH CAUTION AND AT YOUR OWN RISK. See also "nbproc".
926
Willy Tarreau6a06a402007-07-15 20:15:28 +0200927pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100928 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200929 the "-p" command line argument. The file must be accessible to the user
930 starting the process. See also "daemon".
931
Willy Tarreau1d549722016-02-16 12:41:57 +0100932presetenv <name> <value>
933 Sets environment variable <name> to value <value>. If the variable exists, it
934 is NOT overwritten. The changes immediately take effect so that the next line
935 in the configuration file sees the new value. See also "setenv", "resetenv",
936 and "unsetenv".
937
938resetenv [<name> ...]
939 Removes all environment variables except the ones specified in argument. It
940 allows to use a clean controlled environment before setting new values with
941 setenv or unsetenv. Please note that some internal functions may make use of
942 some environment variables, such as time manipulation functions, but also
943 OpenSSL or even external checks. This must be used with extreme care and only
944 after complete validation. The changes immediately take effect so that the
945 next line in the configuration file sees the new environment. See also
946 "setenv", "presetenv", and "unsetenv".
947
Christopher Fauletff4121f2017-11-22 16:38:49 +0100948stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200949 Limits the stats socket to a certain set of processes numbers. By default the
950 stats socket is bound to all processes, causing a warning to be emitted when
951 nbproc is greater than 1 because there is no way to select the target process
952 when connecting. However, by using this setting, it becomes possible to pin
953 the stats socket to a specific set of processes, typically the first one. The
954 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +0100955 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +0100956 word size (32 or 64). Ranges can be partially defined. The higher bound can
957 be omitted. In such case, it is replaced by the corresponding maximum
958 value. A better option consists in using the "process" setting of the "stats
959 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +0200960
Baptiste Assmann5626f482015-08-23 10:00:10 +0200961server-state-base <directory>
962 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200963 file names which do not start with a '/'. See also "server-state-file",
964 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200965
966server-state-file <file>
967 Specifies the path to the file containing state of servers. If the path starts
968 with a slash ('/'), it is considered absolute, otherwise it is considered
969 relative to the directory specified using "server-state-base" (if set) or to
970 the current directory. Before reloading HAProxy, it is possible to save the
971 servers' current state using the stats command "show servers state". The
972 output of this command must be written in the file pointed by <file>. When
973 starting up, before handling traffic, HAProxy will read, load and apply state
974 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +0200975 configuration. See also "server-state-base" and "show servers state",
976 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +0200977
Willy Tarreau1d549722016-02-16 12:41:57 +0100978setenv <name> <value>
979 Sets environment variable <name> to value <value>. If the variable exists, it
980 is overwritten. The changes immediately take effect so that the next line in
981 the configuration file sees the new value. See also "presetenv", "resetenv",
982 and "unsetenv".
983
Willy Tarreau610f04b2014-02-13 11:36:41 +0100984ssl-default-bind-ciphers <ciphers>
985 This setting is only available when support for OpenSSL was built in. It sets
986 the default string describing the list of cipher algorithms ("cipher suite")
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300987 that are negotiated during the SSL/TLS handshake for all "bind" lines which
Willy Tarreau610f04b2014-02-13 11:36:41 +0100988 do not explicitly define theirs. The format of the string is defined in
989 "man 1 ciphers" from OpenSSL man pages, and can be for instance a string such
990 as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes). Please check the
991 "bind" keyword for more information.
992
Emeric Brun2c86cbf2014-10-30 15:56:50 +0100993ssl-default-bind-options [<option>]...
994 This setting is only available when support for OpenSSL was built in. It sets
995 default ssl-options to force on all "bind" lines. Please check the "bind"
996 keyword to see available options.
997
998 Example:
999 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001000 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001001
Willy Tarreau610f04b2014-02-13 11:36:41 +01001002ssl-default-server-ciphers <ciphers>
1003 This setting is only available when support for OpenSSL was built in. It
1004 sets the default string describing the list of cipher algorithms that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001005 negotiated during the SSL/TLS handshake with the server, for all "server"
Willy Tarreau610f04b2014-02-13 11:36:41 +01001006 lines which do not explicitly define theirs. The format of the string is
1007 defined in "man 1 ciphers". Please check the "server" keyword for more
1008 information.
1009
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001010ssl-default-server-options [<option>]...
1011 This setting is only available when support for OpenSSL was built in. It sets
1012 default ssl-options to force on all "server" lines. Please check the "server"
1013 keyword to see available options.
1014
Remi Gacogne47783ef2015-05-29 15:53:22 +02001015ssl-dh-param-file <file>
1016 This setting is only available when support for OpenSSL was built in. It sets
1017 the default DH parameters that are used during the SSL/TLS handshake when
1018 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001019 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001020 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001021 are not specified either by using ssl-dh-param-file or by setting them
1022 directly in the certificate file, pre-generated DH parameters of the size
1023 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1024 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001025 Custom DH parameters may be generated by using the OpenSSL command
1026 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1027 parameters should not be considered secure anymore.
1028
Emeric Brun850efd52014-01-29 12:24:34 +01001029ssl-server-verify [none|required]
1030 The default behavior for SSL verify on servers side. If specified to 'none',
1031 servers certificates are not verified. The default is 'required' except if
1032 forced using cmdline option '-dV'.
1033
Willy Tarreauabb175f2012-09-24 12:43:26 +02001034stats socket [<address:port>|<path>] [param*]
1035 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1036 Connections to this socket will return various statistics outputs and even
1037 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001038 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001039 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001040
Willy Tarreauabb175f2012-09-24 12:43:26 +02001041 All parameters supported by "bind" lines are supported, for instance to
1042 restrict access to some users or their access rights. Please consult
1043 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001044
1045stats timeout <timeout, in milliseconds>
1046 The default timeout on the stats socket is set to 10 seconds. It is possible
1047 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001048 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001049
1050stats maxconn <connections>
1051 By default, the stats socket is limited to 10 concurrent connections. It is
1052 possible to change this value with "stats maxconn".
1053
Willy Tarreau6a06a402007-07-15 20:15:28 +02001054uid <number>
1055 Changes the process' user ID to <number>. It is recommended that the user ID
1056 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1057 be started with superuser privileges in order to be able to switch to another
1058 one. See also "gid" and "user".
1059
1060ulimit-n <number>
1061 Sets the maximum number of per-process file-descriptors to <number>. By
1062 default, it is automatically computed, so it is recommended not to use this
1063 option.
1064
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001065unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1066 [ group <group> ] [ gid <gid> ]
1067
1068 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1069 This is mainly used to simplify declaration of those UNIX sockets and reduce
1070 the risk of errors, since those settings are most commonly required but are
1071 also process-specific. The <prefix> setting can be used to force all socket
1072 path to be relative to that directory. This might be needed to access another
1073 component's chroot. Note that those paths are resolved before haproxy chroots
1074 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1075 all have the same meaning as their homonyms used by the "bind" statement. If
1076 both are specified, the "bind" statement has priority, meaning that the
1077 "unix-bind" settings may be seen as process-wide default settings.
1078
Willy Tarreau1d549722016-02-16 12:41:57 +01001079unsetenv [<name> ...]
1080 Removes environment variables specified in arguments. This can be useful to
1081 hide some sensitive information that are occasionally inherited from the
1082 user's environment during some operations. Variables which did not exist are
1083 silently ignored so that after the operation, it is certain that none of
1084 these variables remain. The changes immediately take effect so that the next
1085 line in the configuration file will not see these variables. See also
1086 "setenv", "presetenv", and "resetenv".
1087
Willy Tarreau6a06a402007-07-15 20:15:28 +02001088user <user name>
1089 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1090 See also "uid" and "group".
1091
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001092node <name>
1093 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1094
1095 This statement is useful in HA configurations where two or more processes or
1096 servers share the same IP address. By setting a different node-name on all
1097 nodes, it becomes easy to immediately spot what server is handling the
1098 traffic.
1099
1100description <text>
1101 Add a text that describes the instance.
1102
1103 Please note that it is required to escape certain characters (# for example)
1104 and this text is inserted into a html page so you should avoid using
1105 "<" and ">" characters.
1106
Thomas Holmesdb04f192015-05-18 13:21:39 +0100110751degrees-data-file <file path>
1108 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001109 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001110
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001111 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001112 compiled with USE_51DEGREES.
1113
Ben Shillitof25e8e52016-12-02 14:25:37 +0000111451degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001115 A list of 51Degrees property names to be load from the dataset. A full list
1116 of names is available on the 51Degrees website:
1117 https://51degrees.com/resources/property-dictionary
1118
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001119 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001120 compiled with USE_51DEGREES.
1121
Dragan Dosen93b38d92015-06-29 16:43:25 +0200112251degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001123 A char that will be appended to every property value in a response header
1124 containing 51Degrees results. If not set that will be set as ','.
1125
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001126 Please note that this option is only available when haproxy has been
1127 compiled with USE_51DEGREES.
1128
112951degrees-cache-size <number>
1130 Sets the size of the 51Degrees converter cache to <number> entries. This
1131 is an LRU cache which reminds previous device detections and their results.
1132 By default, this cache is disabled.
1133
1134 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001135 compiled with USE_51DEGREES.
1136
scientiamobiled0027ed2016-11-04 10:55:08 +01001137wurfl-data-file <file path>
1138 The path of the WURFL data file to provide device detection services. The
1139 file should be accessible by HAProxy with relevant permissions.
1140
1141 Please note that this option is only available when haproxy has been compiled
1142 with USE_WURFL=1.
1143
1144wurfl-information-list [<capability>]*
1145 A space-delimited list of WURFL capabilities, virtual capabilities, property
1146 names we plan to use in injected headers. A full list of capability and
1147 virtual capability names is available on the Scientiamobile website :
1148
1149 https://www.scientiamobile.com/wurflCapability
1150
1151 Valid WURFL properties are:
1152 - wurfl_id Contains the device ID of the matched device.
1153
1154 - wurfl_root_id Contains the device root ID of the matched
1155 device.
1156
1157 - wurfl_isdevroot Tells if the matched device is a root device.
1158 Possible values are "TRUE" or "FALSE".
1159
1160 - wurfl_useragent The original useragent coming with this
1161 particular web request.
1162
1163 - wurfl_api_version Contains a string representing the currently
1164 used Libwurfl API version.
1165
1166 - wurfl_engine_target Contains a string representing the currently
1167 set WURFL Engine Target. Possible values are
1168 "HIGH_ACCURACY", "HIGH_PERFORMANCE", "INVALID".
1169
1170 - wurfl_info A string containing information on the parsed
1171 wurfl.xml and its full path.
1172
1173 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1174 WURFL has been loaded successfully.
1175
1176 - wurfl_normalized_useragent The normalized useragent.
1177
1178 - wurfl_useragent_priority The user agent priority used by WURFL.
1179
1180 Please note that this option is only available when haproxy has been compiled
1181 with USE_WURFL=1.
1182
1183wurfl-information-list-separator <char>
1184 A char that will be used to separate values in a response header containing
1185 WURFL results. If not set that a comma (',') will be used by default.
1186
1187 Please note that this option is only available when haproxy has been compiled
1188 with USE_WURFL=1.
1189
1190wurfl-patch-file [<file path>]
1191 A list of WURFL patch file paths. Note that patches are loaded during startup
1192 thus before the chroot.
1193
1194 Please note that this option is only available when haproxy has been compiled
1195 with USE_WURFL=1.
1196
1197wurfl-engine-mode { accuracy | performance }
1198 Sets the WURFL engine target. You can choose between 'accuracy' or
1199 'performance' targets. In performance mode, desktop web browser detection is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001200 done programmatically without referencing the WURFL data. As a result, most
scientiamobiled0027ed2016-11-04 10:55:08 +01001201 desktop web browsers are returned as generic_web_browser WURFL ID for
Davor Ocelice9ed2812017-12-25 17:49:28 +01001202 performance. If either performance or accuracy are not defined, performance
scientiamobiled0027ed2016-11-04 10:55:08 +01001203 mode is enabled by default.
1204
1205 Please note that this option is only available when haproxy has been compiled
1206 with USE_WURFL=1.
1207
1208wurfl-cache-size <U>[,<D>]
1209 Sets the WURFL caching strategy. Here <U> is the Useragent cache size, and
1210 <D> is the internal device cache size. There are three possibilities here :
1211 - "0" : no cache is used.
1212 - <U> : the Single LRU cache is used, the size is expressed in elements.
1213 - <U>,<D> : the Double LRU cache is used, both sizes are in elements. This is
1214 the highest performing option.
1215
1216 Please note that this option is only available when haproxy has been compiled
1217 with USE_WURFL=1.
1218
1219wurfl-useragent-priority { plain | sideloaded_browser }
1220 Tells WURFL if it should prioritize use of the plain user agent ('plain')
1221 over the default sideloaded browser user agent ('sideloaded_browser').
1222
1223 Please note that this option is only available when haproxy has been compiled
1224 with USE_WURFL=1.
1225
Willy Tarreau6a06a402007-07-15 20:15:28 +02001226
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012273.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001228-----------------------
1229
Willy Tarreau1746eec2014-04-25 10:46:47 +02001230max-spread-checks <delay in milliseconds>
1231 By default, haproxy tries to spread the start of health checks across the
1232 smallest health check interval of all the servers in a farm. The principle is
1233 to avoid hammering services running on the same server. But when using large
1234 check intervals (10 seconds or more), the last servers in the farm take some
1235 time before starting to be tested, which can be a problem. This parameter is
1236 used to enforce an upper bound on delay between the first and the last check,
1237 even if the servers' check intervals are larger. When servers run with
1238 shorter intervals, their intervals will be respected though.
1239
Willy Tarreau6a06a402007-07-15 20:15:28 +02001240maxconn <number>
1241 Sets the maximum per-process number of concurrent connections to <number>. It
1242 is equivalent to the command-line argument "-n". Proxies will stop accepting
1243 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001244 automatically adjusted according to this value. See also "ulimit-n". Note:
1245 the "select" poller cannot reliably use more than 1024 file descriptors on
1246 some platforms. If your platform only supports select and reports "select
1247 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaud0256482015-01-15 21:45:22 +01001248 below 500 in general). If this value is not set, it will default to the value
1249 set in DEFAULT_MAXCONN at build time (reported in haproxy -vv) if no memory
1250 limit is enforced, or will be computed based on the memory limit, the buffer
1251 size, memory allocated to compression, SSL cache size, and use or not of SSL
1252 and the associated maxsslconn (which can also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001253
Willy Tarreau81c25d02011-09-07 15:17:21 +02001254maxconnrate <number>
1255 Sets the maximum per-process number of connections per second to <number>.
1256 Proxies will stop accepting connections when this limit is reached. It can be
1257 used to limit the global capacity regardless of each frontend capacity. It is
1258 important to note that this can only be used as a service protection measure,
1259 as there will not necessarily be a fair share between frontends when the
1260 limit is reached, so it's a good idea to also limit each frontend to some
1261 value close to its expected share. Also, lowering tune.maxaccept can improve
1262 fairness.
1263
William Lallemandd85f9172012-11-09 17:05:39 +01001264maxcomprate <number>
1265 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001266 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001267 level will be decreased during the session. If the maximum is reached at the
1268 beginning of a session, the session will not compress at all. If the maximum
1269 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001270 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001271 default value.
1272
William Lallemand072a2bf2012-11-20 17:01:01 +01001273maxcompcpuusage <number>
1274 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1275 for new requests or decreasing the compression level of current requests.
1276 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1277 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1278 case of multiple processes (nbproc > 1), each process manages its individual
1279 usage. A value of 100 disable the limit. The default value is 100. Setting
1280 a lower value will prevent the compression work from slowing the whole
1281 process down and from introducing high latencies.
1282
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001283maxpipes <number>
1284 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1285 are only used by kernel-based tcp splicing. Since a pipe contains two file
1286 descriptors, the "ulimit-n" value will be increased accordingly. The default
1287 value is maxconn/4, which seems to be more than enough for most heavy usages.
1288 The splice code dynamically allocates and releases pipes, and can fall back
1289 to standard copy, so setting this value too low may only impact performance.
1290
Willy Tarreau93e7c002013-10-07 18:51:07 +02001291maxsessrate <number>
1292 Sets the maximum per-process number of sessions per second to <number>.
1293 Proxies will stop accepting connections when this limit is reached. It can be
1294 used to limit the global capacity regardless of each frontend capacity. It is
1295 important to note that this can only be used as a service protection measure,
1296 as there will not necessarily be a fair share between frontends when the
1297 limit is reached, so it's a good idea to also limit each frontend to some
1298 value close to its expected share. Also, lowering tune.maxaccept can improve
1299 fairness.
1300
Willy Tarreau403edff2012-09-06 11:58:37 +02001301maxsslconn <number>
1302 Sets the maximum per-process number of concurrent SSL connections to
1303 <number>. By default there is no SSL-specific limit, which means that the
1304 global maxconn setting will apply to all connections. Setting this limit
1305 avoids having openssl use too much memory and crash when malloc returns NULL
1306 (since it unfortunately does not reliably check for such conditions). Note
1307 that the limit applies both to incoming and outgoing connections, so one
1308 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001309 If this value is not set, but a memory limit is enforced, this value will be
1310 automatically computed based on the memory limit, maxconn, the buffer size,
1311 memory allocated to compression, SSL cache size, and use of SSL in either
1312 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1313 when there is a memory limit, haproxy will automatically adjust these values
1314 so that 100% of the connections can be made over SSL with no risk, and will
1315 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001316
Willy Tarreaue43d5322013-10-07 20:01:52 +02001317maxsslrate <number>
1318 Sets the maximum per-process number of SSL sessions per second to <number>.
1319 SSL listeners will stop accepting connections when this limit is reached. It
1320 can be used to limit the global SSL CPU usage regardless of each frontend
1321 capacity. It is important to note that this can only be used as a service
1322 protection measure, as there will not necessarily be a fair share between
1323 frontends when the limit is reached, so it's a good idea to also limit each
1324 frontend to some value close to its expected share. It is also important to
1325 note that the sessions are accounted before they enter the SSL stack and not
1326 after, which also protects the stack against bad handshakes. Also, lowering
1327 tune.maxaccept can improve fairness.
1328
William Lallemand9d5f5482012-11-07 16:12:57 +01001329maxzlibmem <number>
1330 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1331 When the maximum amount is reached, future sessions will not compress as long
1332 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001333 The default value is 0. The value is available in bytes on the UNIX socket
1334 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1335 "ZlibMemUsage" in bytes.
1336
Willy Tarreau6a06a402007-07-15 20:15:28 +02001337noepoll
1338 Disables the use of the "epoll" event polling system on Linux. It is
1339 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001340 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001341
1342nokqueue
1343 Disables the use of the "kqueue" event polling system on BSD. It is
1344 equivalent to the command-line argument "-dk". The next polling system
1345 used will generally be "poll". See also "nopoll".
1346
1347nopoll
1348 Disables the use of the "poll" event polling system. It is equivalent to the
1349 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001350 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001351 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001352
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001353nosplice
1354 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001355 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001356 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001357 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001358 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1359 be used. This option makes it easier to globally disable kernel splicing in
1360 case of doubt. See also "option splice-auto", "option splice-request" and
1361 "option splice-response".
1362
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001363nogetaddrinfo
1364 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1365 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1366
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001367noreuseport
1368 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1369 command line argument "-dR".
1370
Willy Tarreaufe255b72007-10-14 23:09:26 +02001371spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001372 Sometimes it is desirable to avoid sending agent and health checks to
1373 servers at exact intervals, for instance when many logical servers are
1374 located on the same physical server. With the help of this parameter, it
1375 becomes possible to add some randomness in the check interval between 0
1376 and +/- 50%. A value between 2 and 5 seems to show good results. The
1377 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001378
Davor Ocelice9ed2812017-12-25 17:49:28 +01001379ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001380 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001381 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001382 multiple times, it will simply enable multiple crypto engines. Referencing an
1383 unsupported engine will prevent haproxy from starting. Note that many engines
1384 will lead to lower HTTPS performance than pure software with recent
1385 processors. The optional command "algo" sets the default algorithms an ENGINE
1386 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001387 of "ALL" uses the engine for all cryptographic operations. If no list of
1388 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001389 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1390 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1391 openssl configuration file uses:
1392 https://www.openssl.org/docs/man1.0.2/apps/config.html
1393
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001394ssl-mode-async
1395 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001396 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001397 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1398 doesn't support moving read/write buffers and is not compliant with
1399 haproxy's buffer management. So the asynchronous mode is disabled on
1400 read/write operations (it is only enabled during initial and reneg
1401 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001402
Willy Tarreau33cb0652014-12-23 22:52:37 +01001403tune.buffers.limit <number>
1404 Sets a hard limit on the number of buffers which may be allocated per process.
1405 The default value is zero which means unlimited. The minimum non-zero value
1406 will always be greater than "tune.buffers.reserve" and should ideally always
1407 be about twice as large. Forcing this value can be particularly useful to
1408 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001409 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001410 another one to be released by another session. Since buffers are dynamically
1411 allocated and released, the waiting time is very short and not perceptible
1412 provided that limits remain reasonable. In fact sometimes reducing the limit
1413 may even increase performance by increasing the CPU cache's efficiency. Tests
1414 have shown good results on average HTTP traffic with a limit to 1/10 of the
1415 expected global maxconn setting, which also significantly reduces memory
1416 usage. The memory savings come from the fact that a number of connections
1417 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1418 advised to do so by an haproxy core developer.
1419
Willy Tarreau1058ae72014-12-23 22:40:40 +01001420tune.buffers.reserve <number>
1421 Sets the number of buffers which are pre-allocated and reserved for use only
1422 during memory shortage conditions resulting in failed memory allocations. The
1423 minimum value is 2 and is also the default. There is no reason a user would
1424 want to change this value, it's mostly aimed at haproxy core developers.
1425
Willy Tarreau27a674e2009-08-17 07:23:33 +02001426tune.bufsize <number>
1427 Sets the buffer size to this size (in bytes). Lower values allow more
1428 sessions to coexist in the same amount of RAM, and higher values allow some
1429 applications with very large cookies to work. The default value is 16384 and
1430 can be changed at build time. It is strongly recommended not to change this
1431 from the default value, as very low values will break some services such as
1432 statistics, and values larger than default size will increase memory usage,
1433 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001434 parameter should be decreased by the same factor as this one is increased. In
1435 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1436 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001437 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
1438 than this size, haproxy will return HTTP 502 (Bad Gateway).
Willy Tarreau27a674e2009-08-17 07:23:33 +02001439
Willy Tarreau43961d52010-10-04 20:39:20 +02001440tune.chksize <number>
1441 Sets the check buffer size to this size (in bytes). Higher values may help
1442 find string or regex patterns in very large pages, though doing so may imply
1443 more memory and CPU usage. The default value is 16384 and can be changed at
1444 build time. It is not recommended to change this value, but to use better
1445 checks whenever possible.
1446
William Lallemandf3747832012-11-09 12:33:10 +01001447tune.comp.maxlevel <number>
1448 Sets the maximum compression level. The compression level affects CPU
1449 usage during compression. This value affects CPU usage during compression.
1450 Each session using compression initializes the compression algorithm with
1451 this value. The default value is 1.
1452
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001453tune.h2.header-table-size <number>
1454 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1455 cannot be larger than 65536 bytes. A larger value may help certain clients
1456 send more compact requests, depending on their capabilities. This amount of
1457 memory is consumed for each HTTP/2 connection. It is recommended not to
1458 change it.
1459
Willy Tarreaue6baec02017-07-27 11:45:11 +02001460tune.h2.initial-window-size <number>
1461 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001462 can upload before waiting for an acknowledgment from haproxy. This setting
1463 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001464 The default value is 65535, which roughly allows up to 5 Mbps of upload
1465 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1466 over a 1-ms local network. It can make sense to increase this value to allow
1467 faster uploads, or to reduce it to increase fairness when dealing with many
1468 clients. It doesn't affect resource usage.
1469
Willy Tarreau5242ef82017-07-27 11:47:28 +02001470tune.h2.max-concurrent-streams <number>
1471 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1472 number of outstanding requests on a single connection). The default value is
1473 100. A larger one may slightly improve page load time for complex sites when
1474 visited over high latency networks, but increases the amount of resources a
1475 single client may allocate. A value of zero disables the limit so a single
1476 client may create as many streams as allocatable by haproxy. It is highly
1477 recommended not to change this value.
1478
Willy Tarreau193b8c62012-11-22 00:17:38 +01001479tune.http.cookielen <number>
1480 Sets the maximum length of captured cookies. This is the maximum value that
1481 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1482 will automatically be truncated to this one. It is important not to set too
1483 high a value because all cookie captures still allocate this size whatever
1484 their configured value (they share a same pool). This value is per request
1485 per response, so the memory allocated is twice this value per connection.
1486 When not specified, the limit is set to 63 characters. It is recommended not
1487 to change this value.
1488
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001489tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001490 Sets the maximum length of request URI in logs. This prevents truncating long
1491 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001492 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001493 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001494 configuration directives too.
1495 The default value is 1024.
1496
Willy Tarreauac1932d2011-10-24 19:14:41 +02001497tune.http.maxhdr <number>
1498 Sets the maximum number of headers in a request. When a request comes with a
1499 number of headers greater than this value (including the first line), it is
1500 rejected with a "400 Bad Request" status code. Similarly, too large responses
1501 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1502 for all usages, considering that the widely deployed Apache server uses the
1503 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001504 a buggy application to work by the time it gets fixed. The accepted range is
1505 1..32767. Keep in mind that each new header consumes 32bits of memory for
1506 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001507
Willy Tarreau7e312732014-02-12 16:35:14 +01001508tune.idletimer <timeout>
1509 Sets the duration after which haproxy will consider that an empty buffer is
1510 probably associated with an idle stream. This is used to optimally adjust
1511 some packet sizes while forwarding large and small data alternatively. The
1512 decision to use splice() or to send large buffers in SSL is modulated by this
1513 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1514 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001515 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001516 clicking). There should be not reason for changing this value. Please check
1517 tune.ssl.maxrecord below.
1518
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001519tune.lua.forced-yield <number>
1520 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001521 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001522 HAProxy scheduler to process other tasks like accepting connections or
1523 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001524 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001525 lowered. If the Lua code is quite long and its result is absolutely required
1526 to process the data, the <number> can be increased.
1527
Willy Tarreau32f61e22015-03-18 17:54:59 +01001528tune.lua.maxmem
1529 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1530 default it is zero which means unlimited. It is important to set a limit to
1531 ensure that a bug in a script will not result in the system running out of
1532 memory.
1533
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001534tune.lua.session-timeout <timeout>
1535 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001536 preventing infinite loops or spending too much time in Lua. This timeout
1537 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001538 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001539
1540tune.lua.task-timeout <timeout>
1541 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1542 dedicated to the tasks. By default, this timeout isn't set because a task may
1543 remain alive during of the lifetime of HAProxy. For example, a task used to
1544 check servers.
1545
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001546tune.lua.service-timeout <timeout>
1547 This is the execution timeout for the Lua services. This is useful for
1548 preventing infinite loops or spending too much time in Lua. This timeout
1549 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001550 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001551
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001552tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001553 Sets the maximum number of consecutive connections a process may accept in a
1554 row before switching to other work. In single process mode, higher numbers
1555 give better performance at high connection rates. However in multi-process
1556 modes, keeping a bit of fairness between processes generally is better to
1557 increase performance. This value applies individually to each listener, so
1558 that the number of processes a listener is bound to is taken into account.
1559 This value defaults to 64. In multi-process mode, it is divided by twice
1560 the number of processes the listener is bound to. Setting this value to -1
1561 completely disables the limitation. It should normally not be needed to tweak
1562 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001563
1564tune.maxpollevents <number>
1565 Sets the maximum amount of events that can be processed at once in a call to
1566 the polling system. The default value is adapted to the operating system. It
1567 has been noticed that reducing it below 200 tends to slightly decrease
1568 latency at the expense of network bandwidth, and increasing it above 200
1569 tends to trade latency for slightly increased bandwidth.
1570
Willy Tarreau27a674e2009-08-17 07:23:33 +02001571tune.maxrewrite <number>
1572 Sets the reserved buffer space to this size in bytes. The reserved space is
1573 used for header rewriting or appending. The first reads on sockets will never
1574 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1575 bufsize, though that does not make much sense since there are rarely large
1576 numbers of headers to add. Setting it too high prevents processing of large
1577 requests or responses. Setting it too low prevents addition of new headers
1578 to already large requests or to POST requests. It is generally wise to set it
1579 to about 1024. It is automatically readjusted to half of bufsize if it is
1580 larger than that. This means you don't have to worry about it when changing
1581 bufsize.
1582
Willy Tarreauf3045d22015-04-29 16:24:50 +02001583tune.pattern.cache-size <number>
1584 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1585 cache which reminds previous lookups and their results. It is used by ACLs
1586 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1587 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1588 strings. It applies to pattern expressions which means that it will be able
1589 to memorize the result of a lookup among all the patterns specified on a
1590 configuration line (including all those loaded from files). It automatically
1591 invalidates entries which are updated using HTTP actions or on the CLI. The
1592 default cache size is set to 10000 entries, which limits its footprint to
1593 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1594 risk of collision in this cache, which is in the order of the size of the
1595 cache divided by 2^64. Typically, at 10000 requests per second with the
1596 default cache size of 10000 entries, there's 1% chance that a brute force
1597 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1598 This is considered much lower than the risk of a memory corruption caused by
1599 aging components. If this is not acceptable, the cache can be disabled by
1600 setting this parameter to 0.
1601
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001602tune.pipesize <number>
1603 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1604 are the default size for the system. But sometimes when using TCP splicing,
1605 it can improve performance to increase pipe sizes, especially if it is
1606 suspected that pipes are not filled and that many calls to splice() are
1607 performed. This has an impact on the kernel's memory footprint, so this must
1608 not be changed if impacts are not understood.
1609
Willy Tarreaue803de22010-01-21 17:43:04 +01001610tune.rcvbuf.client <number>
1611tune.rcvbuf.server <number>
1612 Forces the kernel socket receive buffer size on the client or the server side
1613 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1614 and backends. It should normally never be set, and the default size (0) lets
1615 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001616 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001617 order to save kernel memory by preventing it from buffering too large amounts
1618 of received data. Lower values will significantly increase CPU usage though.
1619
Willy Tarreaub22fc302015-12-14 12:04:35 +01001620tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001621 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001622 socket buffers. One of them is that a read returns more than <recv_enough>
1623 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1624 may be changed by this setting to better deal with workloads involving lots
1625 of short messages such as telnet or SSH sessions.
1626
Willy Tarreaue803de22010-01-21 17:43:04 +01001627tune.sndbuf.client <number>
1628tune.sndbuf.server <number>
1629 Forces the kernel socket send buffer size on the client or the server side to
1630 the specified value in bytes. This value applies to all TCP/HTTP frontends
1631 and backends. It should normally never be set, and the default size (0) lets
1632 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001633 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001634 order to save kernel memory by preventing it from buffering too large amounts
1635 of received data. Lower values will significantly increase CPU usage though.
1636 Another use case is to prevent write timeouts with extremely slow clients due
1637 to the kernel waiting for a large part of the buffer to be read before
1638 notifying haproxy again.
1639
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001640tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001641 Sets the size of the global SSL session cache, in a number of blocks. A block
1642 is large enough to contain an encoded session without peer certificate.
1643 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001644 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001645 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001646 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001647 and reassigned. Higher values reduce the occurrence of such a purge, hence
1648 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1649 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001650 and are shared between all processes if "nbproc" is greater than 1. Setting
1651 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001652
Emeric Brun8dc60392014-05-09 13:52:00 +02001653tune.ssl.force-private-cache
1654 This boolean disables SSL session cache sharing between all processes. It
1655 should normally not be used since it will force many renegotiations due to
1656 clients hitting a random process. But it may be required on some operating
1657 systems where none of the SSL cache synchronization method may be used. In
1658 this case, adding a first layer of hash-based load balancing before the SSL
1659 layer might limit the impact of the lack of session sharing.
1660
Emeric Brun4f65bff2012-11-16 15:11:00 +01001661tune.ssl.lifetime <timeout>
1662 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001663 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001664 does not guarantee that sessions will last that long, because if the cache is
1665 full, the longest idle sessions will be purged despite their configured
1666 lifetime. The real usefulness of this setting is to prevent sessions from
1667 being used for too long.
1668
Willy Tarreaubfd59462013-02-21 07:46:09 +01001669tune.ssl.maxrecord <number>
1670 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1671 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1672 data only once it has received a full record. With large records, it means
1673 that clients might have to download up to 16kB of data before starting to
1674 process them. Limiting the value can improve page load times on browsers
1675 located over high latency or low bandwidth networks. It is suggested to find
1676 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1677 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1678 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1679 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001680 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001681 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001682
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001683tune.ssl.default-dh-param <number>
1684 Sets the maximum size of the Diffie-Hellman parameters used for generating
1685 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1686 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1687 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1688 this maximum value. Default value if 1024. Only 1024 or higher values are
1689 allowed. Higher values will increase the CPU load, and values greater than
1690 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001691 used if static Diffie-Hellman parameters are supplied either directly
1692 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001693
Christopher Faulet31af49d2015-06-09 17:29:50 +02001694tune.ssl.ssl-ctx-cache-size <number>
1695 Sets the size of the cache used to store generated certificates to <number>
1696 entries. This is a LRU cache. Because generating a SSL certificate
1697 dynamically is expensive, they are cached. The default cache size is set to
1698 1000 entries.
1699
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001700tune.ssl.capture-cipherlist-size <number>
1701 Sets the maximum size of the buffer used for capturing client-hello cipher
1702 list. If the value is 0 (default value) the capture is disabled, otherwise
1703 a buffer is allocated for each SSL/TLS connection.
1704
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001705tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001706tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001707tune.vars.reqres-max-size <size>
1708tune.vars.sess-max-size <size>
1709tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001710 These five tunes help to manage the maximum amount of memory used by the
1711 variables system. "global" limits the overall amount of memory available for
1712 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1713 memory for the session scope, "txn" for the transaction scope, and "reqres"
1714 limits the memory for each request or response processing.
1715 Memory accounting is hierarchical, meaning more coarse grained limits include
1716 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1717 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001718
Daniel Schneller0b547052016-03-21 20:46:57 +01001719 For example, when "tune.vars.sess-max-size" is limited to 100,
1720 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1721 100 either. If we create a variable "txn.var" that contains 100 bytes,
1722 all available space is consumed.
1723 Notice that exceeding the limits at runtime will not result in an error
1724 message, but values might be cut off or corrupted. So make sure to accurately
1725 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001726
William Lallemanda509e4c2012-11-07 16:54:34 +01001727tune.zlib.memlevel <number>
1728 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001729 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001730 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001731 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001732 between 1 and 9. The default value is 8.
1733
1734tune.zlib.windowsize <number>
1735 Sets the window size (the size of the history buffer) as a parameter of the
1736 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001737 in better compression at the expense of memory usage. Can be a value between
1738 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001739
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017403.3. Debugging
1741--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001742
1743debug
1744 Enables debug mode which dumps to stdout all exchanges, and disables forking
1745 into background. It is the equivalent of the command-line argument "-d". It
1746 should never be used in a production configuration since it may prevent full
1747 system startup.
1748
1749quiet
1750 Do not display any message during startup. It is equivalent to the command-
1751 line argument "-q".
1752
Emeric Brunf099e792010-09-27 12:05:28 +02001753
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010017543.4. Userlists
1755--------------
1756It is possible to control access to frontend/backend/listen sections or to
1757http stats by allowing only authenticated and authorized users. To do this,
1758it is required to create at least one userlist and to define users.
1759
1760userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001761 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001762 used to store authentication & authorization data for independent customers.
1763
1764group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001765 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001766 attach users to this group by using a comma separated list of names
1767 proceeded by "users" keyword.
1768
Cyril Bontéf0c60612010-02-06 14:44:47 +01001769user <username> [password|insecure-password <password>]
1770 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001771 Adds user <username> to the current userlist. Both secure (encrypted) and
1772 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001773 evaluated using the crypt(3) function, so depending on the system's
1774 capabilities, different algorithms are supported. For example, modern Glibc
1775 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1776 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001777
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001778 Attention: Be aware that using encrypted passwords might cause significantly
1779 increased CPU usage, depending on the number of requests, and the algorithm
1780 used. For any of the hashed variants, the password for each request must
1781 be processed through the chosen algorithm, before it can be compared to the
1782 value specified in the config file. Most current algorithms are deliberately
1783 designed to be expensive to compute to achieve resistance against brute
1784 force attacks. They do not simply salt/hash the clear text password once,
1785 but thousands of times. This can quickly become a major factor in haproxy's
1786 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001787
1788 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001789 userlist L1
1790 group G1 users tiger,scott
1791 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001792
Cyril Bontéf0c60612010-02-06 14:44:47 +01001793 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1794 user scott insecure-password elgato
1795 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001796
Cyril Bontéf0c60612010-02-06 14:44:47 +01001797 userlist L2
1798 group G1
1799 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001800
Cyril Bontéf0c60612010-02-06 14:44:47 +01001801 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1802 user scott insecure-password elgato groups G1,G2
1803 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001804
1805 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001806
Emeric Brunf099e792010-09-27 12:05:28 +02001807
18083.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001809----------
Emeric Brun94900952015-06-11 18:25:54 +02001810It is possible to propagate entries of any data-types in stick-tables between
1811several haproxy instances over TCP connections in a multi-master fashion. Each
1812instance pushes its local updates and insertions to remote peers. The pushed
1813values overwrite remote ones without aggregation. Interrupted exchanges are
1814automatically detected and recovered from the last known point.
1815In addition, during a soft restart, the old process connects to the new one
1816using such a TCP connection to push all its entries before the new process
1817tries to connect to other peers. That ensures very fast replication during a
1818reload, it typically takes a fraction of a second even for large tables.
1819Note that Server IDs are used to identify servers remotely, so it is important
1820that configurations look similar or at least that the same IDs are forced on
1821each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001822
1823peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001824 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001825 which is referenced by one or more stick-tables.
1826
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001827disabled
1828 Disables a peers section. It disables both listening and any synchronization
1829 related to this section. This is provided to disable synchronization of stick
1830 tables without having to comment out all "peers" references.
1831
1832enable
1833 This re-enables a disabled peers section which was previously disabled.
1834
Emeric Brunf099e792010-09-27 12:05:28 +02001835peer <peername> <ip>:<port>
1836 Defines a peer inside a peers section.
1837 If <peername> is set to the local peer name (by default hostname, or forced
1838 using "-L" command line option), haproxy will listen for incoming remote peer
1839 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1840 to join the remote peer, and <peername> is used at the protocol level to
1841 identify and validate the remote peer on the server side.
1842
1843 During a soft restart, local peer <ip>:<port> is used by the old instance to
1844 connect the new one and initiate a complete replication (teaching process).
1845
1846 It is strongly recommended to have the exact same peers declaration on all
1847 peers and to only rely on the "-L" command line argument to change the local
1848 peer name. This makes it easier to maintain coherent configuration files
1849 across all peers.
1850
William Lallemandb2f07452015-05-12 14:27:13 +02001851 You may want to reference some environment variables in the address
1852 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001853
Cyril Bontédc4d9032012-04-08 21:57:39 +02001854 Example:
Emeric Brunf099e792010-09-27 12:05:28 +02001855 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001856 peer haproxy1 192.168.0.1:1024
1857 peer haproxy2 192.168.0.2:1024
1858 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001859
1860 backend mybackend
1861 mode tcp
1862 balance roundrobin
1863 stick-table type ip size 20k peers mypeers
1864 stick on src
1865
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001866 server srv1 192.168.0.30:80
1867 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001868
1869
Simon Horman51a1cf62015-02-03 13:00:44 +090018703.6. Mailers
1871------------
1872It is possible to send email alerts when the state of servers changes.
1873If configured email alerts are sent to each mailer that is configured
1874in a mailers section. Email is sent to mailers using SMTP.
1875
Pieter Baauw386a1272015-08-16 15:26:24 +02001876mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001877 Creates a new mailer list with the name <mailersect>. It is an
1878 independent section which is referenced by one or more proxies.
1879
1880mailer <mailername> <ip>:<port>
1881 Defines a mailer inside a mailers section.
1882
1883 Example:
1884 mailers mymailers
1885 mailer smtp1 192.168.0.1:587
1886 mailer smtp2 192.168.0.2:587
1887
1888 backend mybackend
1889 mode tcp
1890 balance roundrobin
1891
1892 email-alert mailers mymailers
1893 email-alert from test1@horms.org
1894 email-alert to test2@horms.org
1895
1896 server srv1 192.168.0.30:80
1897 server srv2 192.168.0.31:80
1898
Pieter Baauw235fcfc2016-02-13 15:33:40 +01001899timeout mail <time>
1900 Defines the time available for a mail/connection to be made and send to
1901 the mail-server. If not defined the default value is 10 seconds. To allow
1902 for at least two SYN-ACK packets to be send during initial TCP handshake it
1903 is advised to keep this value above 4 seconds.
1904
1905 Example:
1906 mailers mymailers
1907 timeout mail 20s
1908 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09001909
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019104. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02001911----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001912
Willy Tarreau6a06a402007-07-15 20:15:28 +02001913Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02001914 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001915 - frontend <name>
1916 - backend <name>
1917 - listen <name>
1918
1919A "defaults" section sets default parameters for all other sections following
1920its declaration. Those default parameters are reset by the next "defaults"
1921section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01001922section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001923
1924A "frontend" section describes a set of listening sockets accepting client
1925connections.
1926
1927A "backend" section describes a set of servers to which the proxy will connect
1928to forward incoming connections.
1929
1930A "listen" section defines a complete proxy with its frontend and backend
1931parts combined in one section. It is generally useful for TCP-only traffic.
1932
Willy Tarreau0ba27502007-12-24 16:55:16 +01001933All proxy names must be formed from upper and lower case letters, digits,
1934'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
1935case-sensitive, which means that "www" and "WWW" are two different proxies.
1936
1937Historically, all proxy names could overlap, it just caused troubles in the
1938logs. Since the introduction of content switching, it is mandatory that two
1939proxies with overlapping capabilities (frontend/backend) have different names.
1940However, it is still permitted that a frontend and a backend share the same
1941name, as this configuration seems to be commonly encountered.
1942
1943Right now, two major proxy modes are supported : "tcp", also known as layer 4,
1944and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001945bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01001946protocol, and can interact with it by allowing, blocking, switching, adding,
1947modifying, or removing arbitrary contents in requests or responses, based on
1948arbitrary criteria.
1949
Willy Tarreau70dffda2014-01-30 03:07:23 +01001950In HTTP mode, the processing applied to requests and responses flowing over
1951a connection depends in the combination of the frontend's HTTP options and
1952the backend's. HAProxy supports 5 connection modes :
1953
1954 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
1955 requests and responses are processed, and connections remain open but idle
1956 between responses and new requests.
1957
1958 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
1959 1.0 to 1.5-dev21 : only the first request and response are processed, and
1960 everything else is forwarded with no analysis at all. This mode should not
1961 be used as it creates lots of trouble with logging and HTTP processing.
1962
1963 - PCL: passive close ("option httpclose") : exactly the same as tunnel mode,
1964 but with "Connection: close" appended in both directions to try to make
1965 both ends close after the first request/response exchange.
1966
1967 - SCL: server close ("option http-server-close") : the server-facing
1968 connection is closed after the end of the response is received, but the
1969 client-facing connection remains open.
1970
1971 - FCL: forced close ("option forceclose") : the connection is actively closed
1972 after the end of the response.
1973
1974The effective mode that will be applied to a connection passing through a
1975frontend and a backend can be determined by both proxy modes according to the
1976following matrix, but in short, the modes are symmetric, keep-alive is the
1977weakest option and force close is the strongest.
1978
1979 Backend mode
1980
1981 | KAL | TUN | PCL | SCL | FCL
1982 ----+-----+-----+-----+-----+----
1983 KAL | KAL | TUN | PCL | SCL | FCL
1984 ----+-----+-----+-----+-----+----
1985 TUN | TUN | TUN | PCL | SCL | FCL
1986 Frontend ----+-----+-----+-----+-----+----
1987 mode PCL | PCL | PCL | PCL | FCL | FCL
1988 ----+-----+-----+-----+-----+----
1989 SCL | SCL | SCL | FCL | SCL | FCL
1990 ----+-----+-----+-----+-----+----
1991 FCL | FCL | FCL | FCL | FCL | FCL
1992
Willy Tarreau0ba27502007-12-24 16:55:16 +01001993
Willy Tarreau70dffda2014-01-30 03:07:23 +01001994
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019954.1. Proxy keywords matrix
1996--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01001997
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001998The following list of keywords is supported. Most of them may only be used in a
1999limited set of section types. Some of them are marked as "deprecated" because
2000they are inherited from an old syntax which may be confusing or functionally
2001limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002002marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002003option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002004and must be disabled for a specific instance. Such options may also be prefixed
2005with "default" in order to restore default settings regardless of what has been
2006specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002007
Willy Tarreau6a06a402007-07-15 20:15:28 +02002008
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002009 keyword defaults frontend listen backend
2010------------------------------------+----------+----------+---------+---------
2011acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002012appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002013backlog X X X -
2014balance X - X X
2015bind - X X -
2016bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002017block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002018capture cookie - X X -
2019capture request header - X X -
2020capture response header - X X -
2021clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002022compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002023contimeout (deprecated) X - X X
2024cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002025declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002026default-server X - X X
2027default_backend X X X -
2028description - X X X
2029disabled X X X X
2030dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002031email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002032email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002033email-alert mailers X X X X
2034email-alert myhostname X X X X
2035email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002036enabled X X X X
2037errorfile X X X X
2038errorloc X X X X
2039errorloc302 X X X X
2040-- keyword -------------------------- defaults - frontend - listen -- backend -
2041errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002042force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002043filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002044fullconn X - X X
2045grace X X X X
2046hash-type X - X X
2047http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002048http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002049http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002050http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002051http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002052http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002053http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002054id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002055ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002056load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002057log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002058log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002059log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002060log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002061max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002062maxconn X X X -
2063mode X X X X
2064monitor fail - X X -
2065monitor-net X X X -
2066monitor-uri X X X -
2067option abortonclose (*) X - X X
2068option accept-invalid-http-request (*) X X X -
2069option accept-invalid-http-response (*) X - X X
2070option allbackups (*) X - X X
2071option checkcache (*) X - X X
2072option clitcpka (*) X X X -
2073option contstats (*) X X X -
2074option dontlog-normal (*) X X X -
2075option dontlognull (*) X X X -
2076option forceclose (*) X X X X
2077-- keyword -------------------------- defaults - frontend - listen -- backend -
2078option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002079option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002080option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002081option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002082option http-no-delay (*) X X X X
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02002083option http-pretend-keepalive (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002084option http-server-close (*) X X X X
Willy Tarreau02bce8b2014-01-30 00:15:28 +01002085option http-tunnel (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002086option http-use-proxy-header (*) X X X -
2087option httpchk X - X X
2088option httpclose (*) X X X X
2089option httplog X X X X
2090option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002091option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002092option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002093option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002094option log-health-checks (*) X - X X
2095option log-separate-errors (*) X X X -
2096option logasap (*) X X X -
2097option mysql-check X - X X
2098option nolinger (*) X X X X
2099option originalto X X X X
2100option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002101option pgsql-check X - X X
2102option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002103option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002104option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002105option smtpchk X - X X
2106option socket-stats (*) X X X -
2107option splice-auto (*) X X X X
2108option splice-request (*) X X X X
2109option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002110option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002111option srvtcpka (*) X - X X
2112option ssl-hello-chk X - X X
2113-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002114option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002115option tcp-smart-accept (*) X X X -
2116option tcp-smart-connect (*) X - X X
2117option tcpka X X X X
2118option tcplog X X X X
2119option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002120external-check command X - X X
2121external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002122persist rdp-cookie X - X X
2123rate-limit sessions X X X -
2124redirect - X X X
2125redisp (deprecated) X - X X
2126redispatch (deprecated) X - X X
2127reqadd - X X X
2128reqallow - X X X
2129reqdel - X X X
2130reqdeny - X X X
2131reqiallow - X X X
2132reqidel - X X X
2133reqideny - X X X
2134reqipass - X X X
2135reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002136reqitarpit - X X X
2137reqpass - X X X
2138reqrep - X X X
2139-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002140reqtarpit - X X X
2141retries X - X X
2142rspadd - X X X
2143rspdel - X X X
2144rspdeny - X X X
2145rspidel - X X X
2146rspideny - X X X
2147rspirep - X X X
2148rsprep - X X X
2149server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002150server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002151server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002152source X - X X
2153srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002154stats admin - X X X
2155stats auth X X X X
2156stats enable X X X X
2157stats hide-version X X X X
2158stats http-request - X X X
2159stats realm X X X X
2160stats refresh X X X X
2161stats scope X X X X
2162stats show-desc X X X X
2163stats show-legends X X X X
2164stats show-node X X X X
2165stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002166-- keyword -------------------------- defaults - frontend - listen -- backend -
2167stick match - - X X
2168stick on - - X X
2169stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002170stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002171stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002172tcp-check connect - - X X
2173tcp-check expect - - X X
2174tcp-check send - - X X
2175tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002176tcp-request connection - X X -
2177tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002178tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002179tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002180tcp-response content - - X X
2181tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002182timeout check X - X X
2183timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002184timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002185timeout clitimeout (deprecated) X X X -
2186timeout connect X - X X
2187timeout contimeout (deprecated) X - X X
2188timeout http-keep-alive X X X X
2189timeout http-request X X X X
2190timeout queue X - X X
2191timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002192timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002193timeout srvtimeout (deprecated) X - X X
2194timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002195timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002196transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002197unique-id-format X X X -
2198unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002199use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002200use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002201------------------------------------+----------+----------+---------+---------
2202 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002203
Willy Tarreau0ba27502007-12-24 16:55:16 +01002204
Willy Tarreauc57f0e22009-05-10 13:12:33 +020022054.2. Alphabetically sorted keywords reference
2206---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002207
2208This section provides a description of each keyword and its usage.
2209
2210
2211acl <aclname> <criterion> [flags] [operator] <value> ...
2212 Declare or complete an access list.
2213 May be used in sections : defaults | frontend | listen | backend
2214 no | yes | yes | yes
2215 Example:
2216 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2217 acl invalid_src src_port 0:1023
2218 acl local_dst hdr(host) -i localhost
2219
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002220 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002221
2222
Cyril Bontéb21570a2009-11-29 20:04:48 +01002223appsession <cookie> len <length> timeout <holdtime>
2224 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002225 Define session stickiness on an existing application cookie.
2226 May be used in sections : defaults | frontend | listen | backend
2227 no | no | yes | yes
2228 Arguments :
2229 <cookie> this is the name of the cookie used by the application and which
2230 HAProxy will have to learn for each new session.
2231
Cyril Bontéb21570a2009-11-29 20:04:48 +01002232 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002233 checked in each cookie value.
2234
2235 <holdtime> this is the time after which the cookie will be removed from
2236 memory if unused. If no unit is specified, this time is in
2237 milliseconds.
2238
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002239 request-learn
2240 If this option is specified, then haproxy will be able to learn
2241 the cookie found in the request in case the server does not
2242 specify any in response. This is typically what happens with
2243 PHPSESSID cookies, or when haproxy's session expires before
2244 the application's session and the correct server is selected.
2245 It is recommended to specify this option to improve reliability.
2246
Cyril Bontéb21570a2009-11-29 20:04:48 +01002247 prefix When this option is specified, haproxy will match on the cookie
2248 prefix (or URL parameter prefix). The appsession value is the
2249 data following this prefix.
2250
2251 Example :
2252 appsession ASPSESSIONID len 64 timeout 3h prefix
2253
Davor Ocelice9ed2812017-12-25 17:49:28 +01002254 This will match the cookie ASPSESSIONIDXXX=XXXX,
2255 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002256
2257 mode This option allows to change the URL parser mode.
2258 2 modes are currently supported :
2259 - path-parameters :
2260 The parser looks for the appsession in the path parameters
2261 part (each parameter is separated by a semi-colon), which is
2262 convenient for JSESSIONID for example.
2263 This is the default mode if the option is not set.
2264 - query-string :
2265 In this mode, the parser will look for the appsession in the
2266 query string.
2267
Willy Tarreau294d0f02015-08-10 19:40:12 +02002268 As of version 1.6, appsessions was removed. It is more flexible and more
2269 convenient to use stick-tables instead, and stick-tables support multi-master
2270 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002271
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002272 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2273 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002274
2275
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002276backlog <conns>
2277 Give hints to the system about the approximate listen backlog desired size
2278 May be used in sections : defaults | frontend | listen | backend
2279 yes | yes | yes | no
2280 Arguments :
2281 <conns> is the number of pending connections. Depending on the operating
2282 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002283 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002284
2285 In order to protect against SYN flood attacks, one solution is to increase
2286 the system's SYN backlog size. Depending on the system, sometimes it is just
2287 tunable via a system parameter, sometimes it is not adjustable at all, and
2288 sometimes the system relies on hints given by the application at the time of
2289 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2290 to the listen() syscall. On systems which can make use of this value, it can
2291 sometimes be useful to be able to specify a different value, hence this
2292 backlog parameter.
2293
2294 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2295 used as a hint and the system accepts up to the smallest greater power of
2296 two, and never more than some limits (usually 32768).
2297
2298 See also : "maxconn" and the target operating system's tuning guide.
2299
2300
Willy Tarreau0ba27502007-12-24 16:55:16 +01002301balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002302balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002303 Define the load balancing algorithm to be used in a backend.
2304 May be used in sections : defaults | frontend | listen | backend
2305 yes | no | yes | yes
2306 Arguments :
2307 <algorithm> is the algorithm used to select a server when doing load
2308 balancing. This only applies when no persistence information
2309 is available, or when a connection is redispatched to another
2310 server. <algorithm> may be one of the following :
2311
2312 roundrobin Each server is used in turns, according to their weights.
2313 This is the smoothest and fairest algorithm when the server's
2314 processing time remains equally distributed. This algorithm
2315 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002316 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002317 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002318 large farms, when a server becomes up after having been down
2319 for a very short time, it may sometimes take a few hundreds
2320 requests for it to be re-integrated into the farm and start
2321 receiving traffic. This is normal, though very rare. It is
2322 indicated here in case you would have the chance to observe
2323 it, so that you don't worry.
2324
2325 static-rr Each server is used in turns, according to their weights.
2326 This algorithm is as similar to roundrobin except that it is
2327 static, which means that changing a server's weight on the
2328 fly will have no effect. On the other hand, it has no design
2329 limitation on the number of servers, and when a server goes
2330 up, it is always immediately reintroduced into the farm, once
2331 the full map is recomputed. It also uses slightly less CPU to
2332 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002333
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002334 leastconn The server with the lowest number of connections receives the
2335 connection. Round-robin is performed within groups of servers
2336 of the same load to ensure that all servers will be used. Use
2337 of this algorithm is recommended where very long sessions are
2338 expected, such as LDAP, SQL, TSE, etc... but is not very well
2339 suited for protocols using short sessions such as HTTP. This
2340 algorithm is dynamic, which means that server weights may be
2341 adjusted on the fly for slow starts for instance.
2342
Willy Tarreauf09c6602012-02-13 17:12:08 +01002343 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002344 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002345 identifier to the highest (see server parameter "id"), which
2346 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002347 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002348 not make sense to use this algorithm without setting maxconn.
2349 The purpose of this algorithm is to always use the smallest
2350 number of servers so that extra servers can be powered off
2351 during non-intensive hours. This algorithm ignores the server
2352 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002353 or IMAP than HTTP, though it can be useful there too. In
2354 order to use this algorithm efficiently, it is recommended
2355 that a cloud controller regularly checks server usage to turn
2356 them off when unused, and regularly checks backend queue to
2357 turn new servers on when the queue inflates. Alternatively,
2358 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002359
Willy Tarreau0ba27502007-12-24 16:55:16 +01002360 source The source IP address is hashed and divided by the total
2361 weight of the running servers to designate which server will
2362 receive the request. This ensures that the same client IP
2363 address will always reach the same server as long as no
2364 server goes down or up. If the hash result changes due to the
2365 number of running servers changing, many clients will be
2366 directed to a different server. This algorithm is generally
2367 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002368 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002369 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002370 static by default, which means that changing a server's
2371 weight on the fly will have no effect, but this can be
2372 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002373
Oskar Stolc8dc41842012-05-19 10:19:54 +01002374 uri This algorithm hashes either the left part of the URI (before
2375 the question mark) or the whole URI (if the "whole" parameter
2376 is present) and divides the hash value by the total weight of
2377 the running servers. The result designates which server will
2378 receive the request. This ensures that the same URI will
2379 always be directed to the same server as long as no server
2380 goes up or down. This is used with proxy caches and
2381 anti-virus proxies in order to maximize the cache hit rate.
2382 Note that this algorithm may only be used in an HTTP backend.
2383 This algorithm is static by default, which means that
2384 changing a server's weight on the fly will have no effect,
2385 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002386
Oskar Stolc8dc41842012-05-19 10:19:54 +01002387 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002388 "depth", both followed by a positive integer number. These
2389 options may be helpful when it is needed to balance servers
2390 based on the beginning of the URI only. The "len" parameter
2391 indicates that the algorithm should only consider that many
2392 characters at the beginning of the URI to compute the hash.
2393 Note that having "len" set to 1 rarely makes sense since most
2394 URIs start with a leading "/".
2395
2396 The "depth" parameter indicates the maximum directory depth
2397 to be used to compute the hash. One level is counted for each
2398 slash in the request. If both parameters are specified, the
2399 evaluation stops when either is reached.
2400
Willy Tarreau0ba27502007-12-24 16:55:16 +01002401 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002402 the query string of each HTTP GET request.
2403
2404 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002405 request entity will be searched for the parameter argument,
2406 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002407 ('?') in the URL. The message body will only start to be
2408 analyzed once either the advertised amount of data has been
2409 received or the request buffer is full. In the unlikely event
2410 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002411 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002412 be randomly balanced if at all. This keyword used to support
2413 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002414
2415 If the parameter is found followed by an equal sign ('=') and
2416 a value, then the value is hashed and divided by the total
2417 weight of the running servers. The result designates which
2418 server will receive the request.
2419
2420 This is used to track user identifiers in requests and ensure
2421 that a same user ID will always be sent to the same server as
2422 long as no server goes up or down. If no value is found or if
2423 the parameter is not found, then a round robin algorithm is
2424 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002425 backend. This algorithm is static by default, which means
2426 that changing a server's weight on the fly will have no
2427 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002428
Cyril Bontédc4d9032012-04-08 21:57:39 +02002429 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2430 request. Just as with the equivalent ACL 'hdr()' function,
2431 the header name in parenthesis is not case sensitive. If the
2432 header is absent or if it does not contain any value, the
2433 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002434
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002435 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002436 reducing the hash algorithm to the main domain part with some
2437 specific headers such as 'Host'. For instance, in the Host
2438 value "haproxy.1wt.eu", only "1wt" will be considered.
2439
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002440 This algorithm is static by default, which means that
2441 changing a server's weight on the fly will have no effect,
2442 but this can be changed using "hash-type".
2443
Emeric Brun736aa232009-06-30 17:56:00 +02002444 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002445 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002446 The RDP cookie <name> (or "mstshash" if omitted) will be
2447 looked up and hashed for each incoming TCP request. Just as
2448 with the equivalent ACL 'req_rdp_cookie()' function, the name
2449 is not case-sensitive. This mechanism is useful as a degraded
2450 persistence mode, as it makes it possible to always send the
2451 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002452 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002453 used instead.
2454
2455 Note that for this to work, the frontend must ensure that an
2456 RDP cookie is already present in the request buffer. For this
2457 you must use 'tcp-request content accept' rule combined with
2458 a 'req_rdp_cookie_cnt' ACL.
2459
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002460 This algorithm is static by default, which means that
2461 changing a server's weight on the fly will have no effect,
2462 but this can be changed using "hash-type".
2463
Cyril Bontédc4d9032012-04-08 21:57:39 +02002464 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002465
Willy Tarreau0ba27502007-12-24 16:55:16 +01002466 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002467 algorithms. Right now, only "url_param" and "uri" support an
2468 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002469
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002470 The load balancing algorithm of a backend is set to roundrobin when no other
2471 algorithm, mode nor option have been set. The algorithm may only be set once
2472 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002473
2474 Examples :
2475 balance roundrobin
2476 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002477 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002478 balance hdr(User-Agent)
2479 balance hdr(host)
2480 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002481
2482 Note: the following caveats and limitations on using the "check_post"
2483 extension with "url_param" must be considered :
2484
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002485 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002486 to determine if the parameters will be found in the body or entity which
2487 may contain binary data. Therefore another method may be required to
2488 restrict consideration of POST requests that have no URL parameters in
2489 the body. (see acl reqideny http_end)
2490
2491 - using a <max_wait> value larger than the request buffer size does not
2492 make sense and is useless. The buffer size is set at build time, and
2493 defaults to 16 kB.
2494
2495 - Content-Encoding is not supported, the parameter search will probably
2496 fail; and load balancing will fall back to Round Robin.
2497
2498 - Expect: 100-continue is not supported, load balancing will fall back to
2499 Round Robin.
2500
Lukas Tribus23953682017-04-28 13:24:30 +00002501 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002502 If the entire parameter value is not present in the first chunk, the
2503 selection of server is undefined (actually, defined by how little
2504 actually appeared in the first chunk).
2505
2506 - This feature does not support generation of a 100, 411 or 501 response.
2507
2508 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002509 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002510 white space or control characters are found, indicating the end of what
2511 might be a URL parameter list. This is probably not a concern with SGML
2512 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002513
Willy Tarreau294d0f02015-08-10 19:40:12 +02002514 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002515
2516
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002517bind [<address>]:<port_range> [, ...] [param*]
2518bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002519 Define one or several listening addresses and/or ports in a frontend.
2520 May be used in sections : defaults | frontend | listen | backend
2521 no | yes | yes | no
2522 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002523 <address> is optional and can be a host name, an IPv4 address, an IPv6
2524 address, or '*'. It designates the address the frontend will
2525 listen on. If unset, all IPv4 addresses of the system will be
2526 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002527 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002528 Optionally, an address family prefix may be used before the
2529 address to force the family regardless of the address format,
2530 which can be useful to specify a path to a unix socket with
2531 no slash ('/'). Currently supported prefixes are :
2532 - 'ipv4@' -> address is always IPv4
2533 - 'ipv6@' -> address is always IPv6
2534 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002535 - 'abns@' -> address is in abstract namespace (Linux only).
2536 Note: since abstract sockets are not "rebindable", they
2537 do not cope well with multi-process mode during
2538 soft-restart, so it is better to avoid them if
2539 nbproc is greater than 1. The effect is that if the
2540 new process fails to start, only one of the old ones
2541 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002542 - 'fd@<n>' -> use file descriptor <n> inherited from the
2543 parent. The fd must be bound and may or may not already
2544 be listening.
William Lallemandb2f07452015-05-12 14:27:13 +02002545 You may want to reference some environment variables in the
2546 address parameter, see section 2.3 about environment
2547 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002548
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002549 <port_range> is either a unique TCP port, or a port range for which the
2550 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002551 above. The port is mandatory for TCP listeners. Note that in
2552 the case of an IPv6 address, the port is always the number
2553 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002554 - a numerical port (ex: '80')
2555 - a dash-delimited ports range explicitly stating the lower
2556 and upper bounds (ex: '2000-2100') which are included in
2557 the range.
2558
2559 Particular care must be taken against port ranges, because
2560 every <address:port> couple consumes one socket (= a file
2561 descriptor), so it's easy to consume lots of descriptors
2562 with a simple range, and to run out of sockets. Also, each
2563 <address:port> couple must be used only once among all
2564 instances running on a same system. Please note that binding
2565 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002566 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002567 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002568
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002569 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002570 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002571 receive UNIX connections on the socket located at this place.
2572 The path must begin with a slash and by default is absolute.
2573 It can be relative to the prefix defined by "unix-bind" in
2574 the global section. Note that the total length of the prefix
2575 followed by the socket path cannot exceed some system limits
2576 for UNIX sockets, which commonly are set to 107 characters.
2577
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002578 <param*> is a list of parameters common to all sockets declared on the
2579 same line. These numerous parameters depend on OS and build
2580 options and have a complete section dedicated to them. Please
2581 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002582
Willy Tarreau0ba27502007-12-24 16:55:16 +01002583 It is possible to specify a list of address:port combinations delimited by
2584 commas. The frontend will then listen on all of these addresses. There is no
2585 fixed limit to the number of addresses and ports which can be listened on in
2586 a frontend, as well as there is no limit to the number of "bind" statements
2587 in a frontend.
2588
2589 Example :
2590 listen http_proxy
2591 bind :80,:443
2592 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002593 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002594
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002595 listen http_https_proxy
2596 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002597 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002598
Willy Tarreau24709282013-03-10 21:32:12 +01002599 listen http_https_proxy_explicit
2600 bind ipv6@:80
2601 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2602 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2603
Willy Tarreaudad36a32013-03-11 01:20:04 +01002604 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002605 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002606
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002607 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2608 sun_path length is used for the address length. Some other programs
2609 such as socat use the string length only by default. Pass the option
2610 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2611 make it compatible with HAProxy's.
2612
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002613 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002614 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002615
2616
Christopher Fauletff4121f2017-11-22 16:38:49 +01002617bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002618 Limit visibility of an instance to a certain set of processes numbers.
2619 May be used in sections : defaults | frontend | listen | backend
2620 yes | yes | yes | yes
2621 Arguments :
2622 all All process will see this instance. This is the default. It
2623 may be used to override a default value.
2624
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002625 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002626 option may be combined with other numbers.
2627
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002628 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002629 option may be combined with other numbers. Do not use it
2630 with less than 2 processes otherwise some instances might be
2631 missing from all processes.
2632
Christopher Fauletff4121f2017-11-22 16:38:49 +01002633 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002634 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002635 the machine's word size. Ranges can be partially defined. The
2636 higher bound can be omitted. In such case, it is replaced by
2637 the corresponding maximum value. If a proxy is bound to
2638 process numbers greater than the configured global.nbproc, it
2639 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002640 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002641
2642 This keyword limits binding of certain instances to certain processes. This
2643 is useful in order not to have too many processes listening to the same
2644 ports. For instance, on a dual-core machine, it might make sense to set
2645 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2646 and 'even' instances.
2647
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002648 At the moment, it is not possible to reference more than 32 or 64 processes
2649 using this keyword, but this should be more than enough for most setups.
2650 Please note that 'all' really means all processes regardless of the machine's
2651 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002652
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002653 Each "bind" line may further be limited to a subset of the proxy's processes,
2654 please consult the "process" bind keyword in section 5.1.
2655
Willy Tarreaub369a042014-09-16 13:21:03 +02002656 When a frontend has no explicit "bind-process" line, it tries to bind to all
2657 the processes referenced by its "bind" lines. That means that frontends can
2658 easily adapt to their listeners' processes.
2659
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002660 If some backends are referenced by frontends bound to other processes, the
2661 backend automatically inherits the frontend's processes.
2662
2663 Example :
2664 listen app_ip1
2665 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002666 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002667
2668 listen app_ip2
2669 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002670 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002671
2672 listen management
2673 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002674 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002675
Willy Tarreau110ecc12012-11-15 17:50:01 +01002676 listen management
2677 bind 10.0.0.4:80
2678 bind-process 1-4
2679
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002680 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002681
2682
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002683block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002684 Block a layer 7 request if/unless a condition is matched
2685 May be used in sections : defaults | frontend | listen | backend
2686 no | yes | yes | yes
2687
2688 The HTTP request will be blocked very early in the layer 7 processing
2689 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002690 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002691 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002692 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002693 "block" statements per instance. To block connections at layer 4 (without
2694 sending a 403 error) see "tcp-request connection reject" and
2695 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002696
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002697 This form is deprecated, do not use it in any new configuration, use the new
2698 "http-request deny" instead.
2699
Willy Tarreau0ba27502007-12-24 16:55:16 +01002700 Example:
2701 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2702 acl invalid_src src_port 0:1023
2703 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002704 # block is deprecated. Use http-request deny instead:
2705 #block if invalid_src || local_dst
2706 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002707
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002708 See also : section 7 about ACL usage, "http-request deny",
2709 "http-response deny", "tcp-request connection reject" and
2710 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002711
2712capture cookie <name> len <length>
2713 Capture and log a cookie in the request and in the response.
2714 May be used in sections : defaults | frontend | listen | backend
2715 no | yes | yes | no
2716 Arguments :
2717 <name> is the beginning of the name of the cookie to capture. In order
2718 to match the exact name, simply suffix the name with an equal
2719 sign ('='). The full name will appear in the logs, which is
2720 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002721 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002722
2723 <length> is the maximum number of characters to report in the logs, which
2724 include the cookie name, the equal sign and the value, all in the
2725 standard "name=value" form. The string will be truncated on the
2726 right if it exceeds <length>.
2727
2728 Only the first cookie is captured. Both the "cookie" request headers and the
2729 "set-cookie" response headers are monitored. This is particularly useful to
2730 check for application bugs causing session crossing or stealing between
2731 users, because generally the user's cookies can only change on a login page.
2732
2733 When the cookie was not presented by the client, the associated log column
2734 will report "-". When a request does not cause a cookie to be assigned by the
2735 server, a "-" is reported in the response column.
2736
2737 The capture is performed in the frontend only because it is necessary that
2738 the log format does not change for a given frontend depending on the
2739 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002740 "capture cookie" statement in a frontend. The maximum capture length is set
2741 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2742 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002743
2744 Example:
2745 capture cookie ASPSESSION len 32
2746
2747 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002748 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002749
2750
2751capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002752 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002753 May be used in sections : defaults | frontend | listen | backend
2754 no | yes | yes | no
2755 Arguments :
2756 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002757 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002758 appear in the requests, with the first letter of each word in
2759 upper case. The header name will not appear in the logs, only the
2760 value is reported, but the position in the logs is respected.
2761
2762 <length> is the maximum number of characters to extract from the value and
2763 report in the logs. The string will be truncated on the right if
2764 it exceeds <length>.
2765
Willy Tarreau4460d032012-11-21 23:37:37 +01002766 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002767 value will be added to the logs between braces ('{}'). If multiple headers
2768 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002769 in the same order they were declared in the configuration. Non-existent
2770 headers will be logged just as an empty string. Common uses for request
2771 header captures include the "Host" field in virtual hosting environments, the
2772 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002773 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002774 environments to find where the request came from.
2775
2776 Note that when capturing headers such as "User-agent", some spaces may be
2777 logged, making the log analysis more difficult. Thus be careful about what
2778 you log if you know your log parser is not smart enough to rely on the
2779 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002780
Willy Tarreau0900abb2012-11-22 00:21:46 +01002781 There is no limit to the number of captured request headers nor to their
2782 length, though it is wise to keep them low to limit memory usage per session.
2783 In order to keep log format consistent for a same frontend, header captures
2784 can only be declared in a frontend. It is not possible to specify a capture
2785 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002786
2787 Example:
2788 capture request header Host len 15
2789 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002790 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002791
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002792 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002793 about logging.
2794
2795
2796capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002797 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002798 May be used in sections : defaults | frontend | listen | backend
2799 no | yes | yes | no
2800 Arguments :
2801 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002802 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002803 appear in the response, with the first letter of each word in
2804 upper case. The header name will not appear in the logs, only the
2805 value is reported, but the position in the logs is respected.
2806
2807 <length> is the maximum number of characters to extract from the value and
2808 report in the logs. The string will be truncated on the right if
2809 it exceeds <length>.
2810
Willy Tarreau4460d032012-11-21 23:37:37 +01002811 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002812 result will be added to the logs between braces ('{}') after the captured
2813 request headers. If multiple headers are captured, they will be delimited by
2814 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002815 the configuration. Non-existent headers will be logged just as an empty
2816 string. Common uses for response header captures include the "Content-length"
2817 header which indicates how many bytes are expected to be returned, the
2818 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002819
Willy Tarreau0900abb2012-11-22 00:21:46 +01002820 There is no limit to the number of captured response headers nor to their
2821 length, though it is wise to keep them low to limit memory usage per session.
2822 In order to keep log format consistent for a same frontend, header captures
2823 can only be declared in a frontend. It is not possible to specify a capture
2824 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002825
2826 Example:
2827 capture response header Content-length len 9
2828 capture response header Location len 15
2829
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002830 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002831 about logging.
2832
2833
Cyril Bontéf0c60612010-02-06 14:44:47 +01002834clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002835 Set the maximum inactivity time on the client side.
2836 May be used in sections : defaults | frontend | listen | backend
2837 yes | yes | yes | no
2838 Arguments :
2839 <timeout> is the timeout value is specified in milliseconds by default, but
2840 can be in any other unit if the number is suffixed by the unit,
2841 as explained at the top of this document.
2842
2843 The inactivity timeout applies when the client is expected to acknowledge or
2844 send data. In HTTP mode, this timeout is particularly important to consider
2845 during the first phase, when the client sends the request, and during the
2846 response while it is reading data sent by the server. The value is specified
2847 in milliseconds by default, but can be in any other unit if the number is
2848 suffixed by the unit, as specified at the top of this document. In TCP mode
2849 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2850 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002851 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002852 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002853 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002854
2855 This parameter is specific to frontends, but can be specified once for all in
2856 "defaults" sections. This is in fact one of the easiest solutions not to
2857 forget about it. An unspecified timeout results in an infinite timeout, which
2858 is not recommended. Such a usage is accepted and works but reports a warning
2859 during startup because it may results in accumulation of expired sessions in
2860 the system if the system's timeouts are not configured either.
2861
2862 This parameter is provided for compatibility but is currently deprecated.
2863 Please use "timeout client" instead.
2864
Willy Tarreau036fae02008-01-06 13:24:40 +01002865 See also : "timeout client", "timeout http-request", "timeout server", and
2866 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002867
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002868compression algo <algorithm> ...
2869compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002870compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002871 Enable HTTP compression.
2872 May be used in sections : defaults | frontend | listen | backend
2873 yes | yes | yes | yes
2874 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002875 algo is followed by the list of supported compression algorithms.
2876 type is followed by the list of MIME types that will be compressed.
2877 offload makes haproxy work as a compression offloader only (see notes).
2878
2879 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01002880 identity this is mostly for debugging, and it was useful for developing
2881 the compression feature. Identity does not apply any change on
2882 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002883
Willy Tarreauc91840a2015-03-28 17:00:39 +01002884 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01002885 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01002886
2887 deflate same as "gzip", but with deflate algorithm and zlib format.
2888 Note that this algorithm has ambiguous support on many
2889 browsers and no support at all from recent ones. It is
2890 strongly recommended not to use it for anything else than
2891 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01002892 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002893
Willy Tarreauc91840a2015-03-28 17:00:39 +01002894 raw-deflate same as "deflate" without the zlib wrapper, and used as an
2895 alternative when the browser wants "deflate". All major
2896 browsers understand it and despite violating the standards,
2897 it is known to work better than "deflate", at least on MSIE
2898 and some versions of Safari. Do not use it in conjunction
2899 with "deflate", use either one or the other since both react
2900 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01002901 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002902
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04002903 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002904 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002905 If backend servers support HTTP compression, these directives
2906 will be no-op: haproxy will see the compressed response and will not
2907 compress again. If backend servers do not support HTTP compression and
2908 there is Accept-Encoding header in request, haproxy will compress the
2909 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02002910
2911 The "offload" setting makes haproxy remove the Accept-Encoding header to
2912 prevent backend servers from compressing responses. It is strongly
2913 recommended not to do this because this means that all the compression work
2914 will be done on the single point where haproxy is located. However in some
2915 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04002916 with broken HTTP compression implementation which can't be turned off.
2917 In that case haproxy can be used to prevent that gateway from emitting
2918 invalid payloads. In this case, simply removing the header in the
2919 configuration does not work because it applies before the header is parsed,
2920 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02002921 then be used for such scenarios. Note: for now, the "offload" setting is
2922 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02002923
William Lallemand05097442012-11-20 12:14:28 +01002924 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002925 * the request does not advertise a supported compression algorithm in the
2926 "Accept-Encoding" header
2927 * the response message is not HTTP/1.1
William Lallemandd3002612012-11-26 14:34:47 +01002928 * HTTP status code is not 200
William Lallemand8bb4e342013-12-10 17:28:48 +01002929 * response header "Transfer-Encoding" contains "chunked" (Temporary
2930 Workaround)
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002931 * response contain neither a "Content-Length" header nor a
2932 "Transfer-Encoding" whose last value is "chunked"
2933 * response contains a "Content-Type" header whose first value starts with
2934 "multipart"
2935 * the response contains the "no-transform" value in the "Cache-control"
2936 header
2937 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
2938 and later
2939 * The response contains a "Content-Encoding" header, indicating that the
2940 response is already compressed (see compression offload)
William Lallemand05097442012-11-20 12:14:28 +01002941
Baptiste Assmann650d53d2013-01-05 15:44:44 +01002942 Note: The compression does not rewrite Etag headers, and does not emit the
2943 Warning header.
William Lallemand05097442012-11-20 12:14:28 +01002944
William Lallemand82fe75c2012-10-23 10:25:10 +02002945 Examples :
2946 compression algo gzip
2947 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01002948
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002949
Cyril Bontéf0c60612010-02-06 14:44:47 +01002950contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002951 Set the maximum time to wait for a connection attempt to a server to succeed.
2952 May be used in sections : defaults | frontend | listen | backend
2953 yes | no | yes | yes
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 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002960 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01002961 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01002962 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002963 connect timeout also presets the queue timeout to the same value if this one
2964 has not been specified. Historically, the contimeout was also used to set the
2965 tarpit timeout in a listen section, which is not possible in a pure frontend.
2966
2967 This parameter is specific to backends, but can be specified once for all in
2968 "defaults" sections. This is in fact one of the easiest solutions not to
2969 forget about it. An unspecified timeout results in an infinite timeout, which
2970 is not recommended. Such a usage is accepted and works but reports a warning
2971 during startup because it may results in accumulation of failed sessions in
2972 the system if the system's timeouts are not configured either.
2973
2974 This parameter is provided for backwards compatibility but is currently
2975 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
2976 instead.
2977
2978 See also : "timeout connect", "timeout queue", "timeout tarpit",
2979 "timeout server", "contimeout".
2980
2981
Willy Tarreau55165fe2009-05-10 12:02:55 +02002982cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02002983 [ postonly ] [ preserve ] [ httponly ] [ secure ]
2984 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01002985 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002986 Enable cookie-based persistence in a backend.
2987 May be used in sections : defaults | frontend | listen | backend
2988 yes | no | yes | yes
2989 Arguments :
2990 <name> is the name of the cookie which will be monitored, modified or
2991 inserted in order to bring persistence. This cookie is sent to
2992 the client via a "Set-Cookie" header in the response, and is
2993 brought back by the client in a "Cookie" header in all requests.
2994 Special care should be taken to choose a name which does not
2995 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01002996 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002997 HTTP/HTTPS), care should be taken to use different cookie names
2998 between all backends if persistence between them is not desired.
2999
3000 rewrite This keyword indicates that the cookie will be provided by the
3001 server and that haproxy will have to modify its value to set the
3002 server's identifier in it. This mode is handy when the management
3003 of complex combinations of "Set-cookie" and "Cache-control"
3004 headers is left to the application. The application can then
3005 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003006 cookie. Since all responses should be monitored, this mode
3007 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003008 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003009 start with this mode for new deployments. This keyword is
3010 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003011
3012 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003013 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003014
Willy Tarreaua79094d2010-08-31 22:54:15 +02003015 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003016 server. When used without the "preserve" option, if the server
3017 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003018 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003019 existing configurations running in the "rewrite" mode. The cookie
3020 will only be a session cookie and will not be stored on the
3021 client's disk. By default, unless the "indirect" option is added,
3022 the server will see the cookies emitted by the client. Due to
3023 caching effects, it is generally wise to add the "nocache" or
3024 "postonly" keywords (see below). The "insert" keyword is not
3025 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003026
3027 prefix This keyword indicates that instead of relying on a dedicated
3028 cookie for the persistence, an existing one will be completed.
3029 This may be needed in some specific environments where the client
3030 does not support more than one single cookie and the application
3031 already needs it. In this case, whenever the server sets a cookie
3032 named <name>, it will be prefixed with the server's identifier
3033 and a delimiter. The prefix will be removed from all client
3034 requests so that the server still finds the cookie it emitted.
3035 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003036 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003037 not compatible with "rewrite" and "insert". Note: it is highly
3038 recommended not to use "indirect" with "prefix", otherwise server
3039 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003040
Willy Tarreaua79094d2010-08-31 22:54:15 +02003041 indirect When this option is specified, no cookie will be emitted to a
3042 client which already has a valid one for the server which has
3043 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003044 it will be removed, unless the "preserve" option is also set. In
3045 "insert" mode, this will additionally remove cookies from the
3046 requests transmitted to the server, making the persistence
3047 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003048 Note: it is highly recommended not to use "indirect" with
3049 "prefix", otherwise server cookie updates would not be sent to
3050 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003051
3052 nocache This option is recommended in conjunction with the insert mode
3053 when there is a cache between the client and HAProxy, as it
3054 ensures that a cacheable response will be tagged non-cacheable if
3055 a cookie needs to be inserted. This is important because if all
3056 persistence cookies are added on a cacheable home page for
3057 instance, then all customers will then fetch the page from an
3058 outer cache and will all share the same persistence cookie,
3059 leading to one server receiving much more traffic than others.
3060 See also the "insert" and "postonly" options.
3061
3062 postonly This option ensures that cookie insertion will only be performed
3063 on responses to POST requests. It is an alternative to the
3064 "nocache" option, because POST responses are not cacheable, so
3065 this ensures that the persistence cookie will never get cached.
3066 Since most sites do not need any sort of persistence before the
3067 first POST which generally is a login request, this is a very
3068 efficient method to optimize caching without risking to find a
3069 persistence cookie in the cache.
3070 See also the "insert" and "nocache" options.
3071
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003072 preserve This option may only be used with "insert" and/or "indirect". It
3073 allows the server to emit the persistence cookie itself. In this
3074 case, if a cookie is found in the response, haproxy will leave it
3075 untouched. This is useful in order to end persistence after a
3076 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003077 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003078 the past. By combining this mechanism with the "disable-on-404"
3079 check option, it is possible to perform a completely graceful
3080 shutdown because users will definitely leave the server after
3081 they logout.
3082
Willy Tarreau4992dd22012-05-31 21:02:17 +02003083 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3084 when a cookie is inserted. This attribute is used so that a
3085 user agent doesn't share the cookie with non-HTTP components.
3086 Please check RFC6265 for more information on this attribute.
3087
3088 secure This option tells haproxy to add a "Secure" cookie attribute when
3089 a cookie is inserted. This attribute is used so that a user agent
3090 never emits this cookie over non-secure channels, which means
3091 that a cookie learned with this flag will be presented only over
3092 SSL/TLS connections. Please check RFC6265 for more information on
3093 this attribute.
3094
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003095 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003096 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003097 name. If the domain begins with a dot, the browser is allowed to
3098 use it for any host ending with that name. It is also possible to
3099 specify several domain names by invoking this option multiple
3100 times. Some browsers might have small limits on the number of
3101 domains, so be careful when doing that. For the record, sending
3102 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003103
Willy Tarreau996a92c2010-10-13 19:30:47 +02003104 maxidle This option allows inserted cookies to be ignored after some idle
3105 time. It only works with insert-mode cookies. When a cookie is
3106 sent to the client, the date this cookie was emitted is sent too.
3107 Upon further presentations of this cookie, if the date is older
3108 than the delay indicated by the parameter (in seconds), it will
3109 be ignored. Otherwise, it will be refreshed if needed when the
3110 response is sent to the client. This is particularly useful to
3111 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003112 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003113 this option is set and a cookie has no date, it is always
3114 accepted, but gets refreshed in the response. This maintains the
3115 ability for admins to access their sites. Cookies that have a
3116 date in the future further than 24 hours are ignored. Doing so
3117 lets admins fix timezone issues without risking kicking users off
3118 the site.
3119
3120 maxlife This option allows inserted cookies to be ignored after some life
3121 time, whether they're in use or not. It only works with insert
3122 mode cookies. When a cookie is first sent to the client, the date
3123 this cookie was emitted is sent too. Upon further presentations
3124 of this cookie, if the date is older than the delay indicated by
3125 the parameter (in seconds), it will be ignored. If the cookie in
3126 the request has no date, it is accepted and a date will be set.
3127 Cookies that have a date in the future further than 24 hours are
3128 ignored. Doing so lets admins fix timezone issues without risking
3129 kicking users off the site. Contrary to maxidle, this value is
3130 not refreshed, only the first visit date counts. Both maxidle and
3131 maxlife may be used at the time. This is particularly useful to
3132 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003133 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003134 is stronger than the maxidle method in that it forces a
3135 redispatch after some absolute delay.
3136
Olivier Houchard4e694042017-03-14 20:01:29 +01003137 dynamic Activate dynamic cookies. When used, a session cookie is
3138 dynamically created for each server, based on the IP and port
3139 of the server, and a secret key, specified in the
3140 "dynamic-cookie-key" backend directive.
3141 The cookie will be regenerated each time the IP address change,
3142 and is only generated for IPv4/IPv6.
3143
Willy Tarreau0ba27502007-12-24 16:55:16 +01003144 There can be only one persistence cookie per HTTP backend, and it can be
3145 declared in a defaults section. The value of the cookie will be the value
3146 indicated after the "cookie" keyword in a "server" statement. If no cookie
3147 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003148
Willy Tarreau0ba27502007-12-24 16:55:16 +01003149 Examples :
3150 cookie JSESSIONID prefix
3151 cookie SRV insert indirect nocache
3152 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003153 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003154
Willy Tarreau294d0f02015-08-10 19:40:12 +02003155 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003156
Willy Tarreau983e01e2010-01-11 18:42:06 +01003157
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003158declare capture [ request | response ] len <length>
3159 Declares a capture slot.
3160 May be used in sections : defaults | frontend | listen | backend
3161 no | yes | yes | no
3162 Arguments:
3163 <length> is the length allowed for the capture.
3164
3165 This declaration is only available in the frontend or listen section, but the
3166 reserved slot can be used in the backends. The "request" keyword allocates a
3167 capture slot for use in the request, and "response" allocates a capture slot
3168 for use in the response.
3169
3170 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003171 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003172 "http-request capture" and "http-response capture".
3173
3174
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003175default-server [param*]
3176 Change default options for a server in a backend
3177 May be used in sections : defaults | frontend | listen | backend
3178 yes | no | yes | yes
3179 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003180 <param*> is a list of parameters for this server. The "default-server"
3181 keyword accepts an important number of options and has a complete
3182 section dedicated to it. Please refer to section 5 for more
3183 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003184
Willy Tarreau983e01e2010-01-11 18:42:06 +01003185 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003186 default-server inter 1000 weight 13
3187
3188 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003189
Willy Tarreau983e01e2010-01-11 18:42:06 +01003190
Willy Tarreau0ba27502007-12-24 16:55:16 +01003191default_backend <backend>
3192 Specify the backend to use when no "use_backend" rule has been matched.
3193 May be used in sections : defaults | frontend | listen | backend
3194 yes | yes | yes | no
3195 Arguments :
3196 <backend> is the name of the backend to use.
3197
3198 When doing content-switching between frontend and backends using the
3199 "use_backend" keyword, it is often useful to indicate which backend will be
3200 used when no rule has matched. It generally is the dynamic backend which
3201 will catch all undetermined requests.
3202
Willy Tarreau0ba27502007-12-24 16:55:16 +01003203 Example :
3204
3205 use_backend dynamic if url_dyn
3206 use_backend static if url_css url_img extension_img
3207 default_backend dynamic
3208
Willy Tarreau98d04852015-05-26 12:18:29 +02003209 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003210
Willy Tarreau0ba27502007-12-24 16:55:16 +01003211
Baptiste Assmann27f51342013-10-09 06:51:49 +02003212description <string>
3213 Describe a listen, frontend or backend.
3214 May be used in sections : defaults | frontend | listen | backend
3215 no | yes | yes | yes
3216 Arguments : string
3217
3218 Allows to add a sentence to describe the related object in the HAProxy HTML
3219 stats page. The description will be printed on the right of the object name
3220 it describes.
3221 No need to backslash spaces in the <string> arguments.
3222
3223
Willy Tarreau0ba27502007-12-24 16:55:16 +01003224disabled
3225 Disable a proxy, frontend or backend.
3226 May be used in sections : defaults | frontend | listen | backend
3227 yes | yes | yes | yes
3228 Arguments : none
3229
3230 The "disabled" keyword is used to disable an instance, mainly in order to
3231 liberate a listening port or to temporarily disable a service. The instance
3232 will still be created and its configuration will be checked, but it will be
3233 created in the "stopped" state and will appear as such in the statistics. It
3234 will not receive any traffic nor will it send any health-checks or logs. It
3235 is possible to disable many instances at once by adding the "disabled"
3236 keyword in a "defaults" section.
3237
3238 See also : "enabled"
3239
3240
Willy Tarreau5ce94572010-06-07 14:35:41 +02003241dispatch <address>:<port>
3242 Set a default server address
3243 May be used in sections : defaults | frontend | listen | backend
3244 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003245 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003246
3247 <address> is the IPv4 address of the default server. Alternatively, a
3248 resolvable hostname is supported, but this name will be resolved
3249 during start-up.
3250
3251 <ports> is a mandatory port specification. All connections will be sent
3252 to this port, and it is not permitted to use port offsets as is
3253 possible with normal servers.
3254
Willy Tarreau787aed52011-04-15 06:45:37 +02003255 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003256 server can take the connection. In the past it was used to forward non
3257 persistent connections to an auxiliary load balancer. Due to its simple
3258 syntax, it has also been used for simple TCP relays. It is recommended not to
3259 use it for more clarity, and to use the "server" directive instead.
3260
3261 See also : "server"
3262
Olivier Houchard4e694042017-03-14 20:01:29 +01003263
3264dynamic-cookie-key <string>
3265 Set the dynamic cookie secret key for a backend.
3266 May be used in sections : defaults | frontend | listen | backend
3267 yes | no | yes | yes
3268 Arguments : The secret key to be used.
3269
3270 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003271 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003272 specified on the "server" line), using a hash of the IP address of the
3273 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003274 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003275 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003276
Willy Tarreau0ba27502007-12-24 16:55:16 +01003277enabled
3278 Enable a proxy, frontend or backend.
3279 May be used in sections : defaults | frontend | listen | backend
3280 yes | yes | yes | yes
3281 Arguments : none
3282
3283 The "enabled" keyword is used to explicitly enable an instance, when the
3284 defaults has been set to "disabled". This is very rarely used.
3285
3286 See also : "disabled"
3287
3288
3289errorfile <code> <file>
3290 Return a file contents instead of errors generated by HAProxy
3291 May be used in sections : defaults | frontend | listen | backend
3292 yes | yes | yes | yes
3293 Arguments :
3294 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003295 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3296 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003297
3298 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003299 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003300 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003301 error pages, and to use absolute paths, since files are read
3302 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003303
3304 It is important to understand that this keyword is not meant to rewrite
3305 errors returned by the server, but errors detected and returned by HAProxy.
3306 This is why the list of supported errors is limited to a small set.
3307
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003308 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3309
Willy Tarreau0ba27502007-12-24 16:55:16 +01003310 The files are returned verbatim on the TCP socket. This allows any trick such
3311 as redirections to another URL or site, as well as tricks to clean cookies,
3312 force enable or disable caching, etc... The package provides default error
3313 files returning the same contents as default errors.
3314
Willy Tarreau59140a22009-02-22 12:02:30 +01003315 The files should not exceed the configured buffer size (BUFSIZE), which
3316 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003317 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003318 loops between the client and HAProxy when all servers are down, causing an
3319 error to be returned instead of an image. For better HTTP compliance, it is
3320 recommended that all header lines end with CR-LF and not LF alone.
3321
Willy Tarreau0ba27502007-12-24 16:55:16 +01003322 The files are read at the same time as the configuration and kept in memory.
3323 For this reason, the errors continue to be returned even when the process is
3324 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003325 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003326 403 status code and interrogating a blocked URL.
3327
3328 See also : "errorloc", "errorloc302", "errorloc303"
3329
Willy Tarreau59140a22009-02-22 12:02:30 +01003330 Example :
3331 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003332 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003333 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3334 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3335
Willy Tarreau2769aa02007-12-27 18:26:09 +01003336
3337errorloc <code> <url>
3338errorloc302 <code> <url>
3339 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3340 May be used in sections : defaults | frontend | listen | backend
3341 yes | yes | yes | yes
3342 Arguments :
3343 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003344 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3345 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003346
3347 <url> it is the exact contents of the "Location" header. It may contain
3348 either a relative URI to an error page hosted on the same site,
3349 or an absolute URI designating an error page on another site.
3350 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003351 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003352
3353 It is important to understand that this keyword is not meant to rewrite
3354 errors returned by the server, but errors detected and returned by HAProxy.
3355 This is why the list of supported errors is limited to a small set.
3356
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003357 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3358
Willy Tarreau2769aa02007-12-27 18:26:09 +01003359 Note that both keyword return the HTTP 302 status code, which tells the
3360 client to fetch the designated URL using the same HTTP method. This can be
3361 quite problematic in case of non-GET methods such as POST, because the URL
3362 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003363 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003364 status code, indicating to the client that the URL must be fetched with a GET
3365 request.
3366
3367 See also : "errorfile", "errorloc303"
3368
3369
3370errorloc303 <code> <url>
3371 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3372 May be used in sections : defaults | frontend | listen | backend
3373 yes | yes | yes | yes
3374 Arguments :
3375 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003376 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3377 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003378
3379 <url> it is the exact contents of the "Location" header. It may contain
3380 either a relative URI to an error page hosted on the same site,
3381 or an absolute URI designating an error page on another site.
3382 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003383 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003384
3385 It is important to understand that this keyword is not meant to rewrite
3386 errors returned by the server, but errors detected and returned by HAProxy.
3387 This is why the list of supported errors is limited to a small set.
3388
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003389 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3390
Willy Tarreau2769aa02007-12-27 18:26:09 +01003391 Note that both keyword return the HTTP 303 status code, which tells the
3392 client to fetch the designated URL using the same HTTP GET method. This
3393 solves the usual problems associated with "errorloc" and the 302 code. It is
3394 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003395 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003396
3397 See also : "errorfile", "errorloc", "errorloc302"
3398
3399
Simon Horman51a1cf62015-02-03 13:00:44 +09003400email-alert from <emailaddr>
3401 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003402 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003403 May be used in sections: defaults | frontend | listen | backend
3404 yes | yes | yes | yes
3405
3406 Arguments :
3407
3408 <emailaddr> is the from email address to use when sending email alerts
3409
3410 Also requires "email-alert mailers" and "email-alert to" to be set
3411 and if so sending email alerts is enabled for the proxy.
3412
Simon Horman64e34162015-02-06 11:11:57 +09003413 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003414 "email-alert myhostname", "email-alert to", section 3.6 about
3415 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003416
3417
3418email-alert level <level>
3419 Declare the maximum log level of messages for which email alerts will be
3420 sent. This acts as a filter on the sending of email alerts.
3421 May be used in sections: defaults | frontend | listen | backend
3422 yes | yes | yes | yes
3423
3424 Arguments :
3425
3426 <level> One of the 8 syslog levels:
3427 emerg alert crit err warning notice info debug
3428 The above syslog levels are ordered from lowest to highest.
3429
3430 By default level is alert
3431
3432 Also requires "email-alert from", "email-alert mailers" and
3433 "email-alert to" to be set and if so sending email alerts is enabled
3434 for the proxy.
3435
Simon Horman1421e212015-04-30 13:10:35 +09003436 Alerts are sent when :
3437
3438 * An un-paused server is marked as down and <level> is alert or lower
3439 * A paused server is marked as down and <level> is notice or lower
3440 * A server is marked as up or enters the drain state and <level>
3441 is notice or lower
3442 * "option log-health-checks" is enabled, <level> is info or lower,
3443 and a health check status update occurs
3444
Simon Horman64e34162015-02-06 11:11:57 +09003445 See also : "email-alert from", "email-alert mailers",
3446 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003447 section 3.6 about mailers.
3448
3449
3450email-alert mailers <mailersect>
3451 Declare the mailers to be used when sending email alerts
3452 May be used in sections: defaults | frontend | listen | backend
3453 yes | yes | yes | yes
3454
3455 Arguments :
3456
3457 <mailersect> is the name of the mailers section to send email alerts.
3458
3459 Also requires "email-alert from" and "email-alert to" to be set
3460 and if so sending email alerts is enabled for the proxy.
3461
Simon Horman64e34162015-02-06 11:11:57 +09003462 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3463 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003464
3465
3466email-alert myhostname <hostname>
3467 Declare the to hostname address to be used when communicating with
3468 mailers.
3469 May be used in sections: defaults | frontend | listen | backend
3470 yes | yes | yes | yes
3471
3472 Arguments :
3473
Baptiste Assmann738bad92015-12-21 15:27:53 +01003474 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003475
3476 By default the systems hostname is used.
3477
3478 Also requires "email-alert from", "email-alert mailers" and
3479 "email-alert to" to be set and if so sending email alerts is enabled
3480 for the proxy.
3481
Simon Horman64e34162015-02-06 11:11:57 +09003482 See also : "email-alert from", "email-alert level", "email-alert mailers",
3483 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003484
3485
3486email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003487 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003488 header of email alerts. This is the address that email alerts are sent to.
3489 May be used in sections: defaults | frontend | listen | backend
3490 yes | yes | yes | yes
3491
3492 Arguments :
3493
3494 <emailaddr> is the to email address to use when sending email alerts
3495
3496 Also requires "email-alert mailers" and "email-alert to" to be set
3497 and if so sending email alerts is enabled for the proxy.
3498
Simon Horman64e34162015-02-06 11:11:57 +09003499 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003500 "email-alert myhostname", section 3.6 about mailers.
3501
3502
Willy Tarreau4de91492010-01-22 19:10:05 +01003503force-persist { if | unless } <condition>
3504 Declare a condition to force persistence on down servers
3505 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003506 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003507
3508 By default, requests are not dispatched to down servers. It is possible to
3509 force this using "option persist", but it is unconditional and redispatches
3510 to a valid server if "option redispatch" is set. That leaves with very little
3511 possibilities to force some requests to reach a server which is artificially
3512 marked down for maintenance operations.
3513
3514 The "force-persist" statement allows one to declare various ACL-based
3515 conditions which, when met, will cause a request to ignore the down status of
3516 a server and still try to connect to it. That makes it possible to start a
3517 server, still replying an error to the health checks, and run a specially
3518 configured browser to test the service. Among the handy methods, one could
3519 use a specific source IP address, or a specific cookie. The cookie also has
3520 the advantage that it can easily be added/removed on the browser from a test
3521 page. Once the service is validated, it is then possible to open the service
3522 to the world by returning a valid response to health checks.
3523
3524 The forced persistence is enabled when an "if" condition is met, or unless an
3525 "unless" condition is met. The final redispatch is always disabled when this
3526 is used.
3527
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003528 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003529 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003530
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003531
3532filter <name> [param*]
3533 Add the filter <name> in the filter list attached to the proxy.
3534 May be used in sections : defaults | frontend | listen | backend
3535 no | yes | yes | yes
3536 Arguments :
3537 <name> is the name of the filter. Officially supported filters are
3538 referenced in section 9.
3539
Tim Düsterhus4896c442016-11-29 02:15:19 +01003540 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003541 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003542 filter. Please refer to the documentation of the corresponding
3543 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003544
3545 Multiple occurrences of the filter line can be used for the same proxy. The
3546 same filter can be referenced many times if needed.
3547
3548 Example:
3549 listen
3550 bind *:80
3551
3552 filter trace name BEFORE-HTTP-COMP
3553 filter compression
3554 filter trace name AFTER-HTTP-COMP
3555
3556 compression algo gzip
3557 compression offload
3558
3559 server srv1 192.168.0.1:80
3560
3561 See also : section 9.
3562
Willy Tarreau4de91492010-01-22 19:10:05 +01003563
Willy Tarreau2769aa02007-12-27 18:26:09 +01003564fullconn <conns>
3565 Specify at what backend load the servers will reach their maxconn
3566 May be used in sections : defaults | frontend | listen | backend
3567 yes | no | yes | yes
3568 Arguments :
3569 <conns> is the number of connections on the backend which will make the
3570 servers use the maximal number of connections.
3571
Willy Tarreau198a7442008-01-17 12:05:32 +01003572 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003573 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003574 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003575 load. The server will then always accept at least <minconn> connections,
3576 never more than <maxconn>, and the limit will be on the ramp between both
3577 values when the backend has less than <conns> concurrent connections. This
3578 makes it possible to limit the load on the servers during normal loads, but
3579 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003580 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003581
Willy Tarreaufbb78422011-06-05 15:38:35 +02003582 Since it's hard to get this value right, haproxy automatically sets it to
3583 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003584 backend (based on "use_backend" and "default_backend" rules). That way it's
3585 safe to leave it unset. However, "use_backend" involving dynamic names are
3586 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003587
Willy Tarreau2769aa02007-12-27 18:26:09 +01003588 Example :
3589 # The servers will accept between 100 and 1000 concurrent connections each
3590 # and the maximum of 1000 will be reached when the backend reaches 10000
3591 # connections.
3592 backend dynamic
3593 fullconn 10000
3594 server srv1 dyn1:80 minconn 100 maxconn 1000
3595 server srv2 dyn2:80 minconn 100 maxconn 1000
3596
3597 See also : "maxconn", "server"
3598
3599
3600grace <time>
3601 Maintain a proxy operational for some time after a soft stop
3602 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003603 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003604 Arguments :
3605 <time> is the time (by default in milliseconds) for which the instance
3606 will remain operational with the frontend sockets still listening
3607 when a soft-stop is received via the SIGUSR1 signal.
3608
3609 This may be used to ensure that the services disappear in a certain order.
3610 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003611 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003612 needed by the equipment to detect the failure.
3613
3614 Note that currently, there is very little benefit in using this parameter,
3615 and it may in fact complicate the soft-reconfiguration process more than
3616 simplify it.
3617
Willy Tarreau0ba27502007-12-24 16:55:16 +01003618
Andrew Rodland17be45e2016-10-25 17:04:12 -04003619hash-balance-factor <factor>
3620 Specify the balancing factor for bounded-load consistent hashing
3621 May be used in sections : defaults | frontend | listen | backend
3622 yes | no | no | yes
3623 Arguments :
3624 <factor> is the control for the maximum number of concurrent requests to
3625 send to a server, expressed as a percentage of the average number
3626 of concurrent requests across all of the active servers.
3627
3628 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3629 enables an algorithm that prevents any one server from getting too many
3630 requests at once, even if some hash buckets receive many more requests than
3631 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3632 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3633 then no server will be allowed to have a load more than 1.5 times the average.
3634 If server weights are used, they will be respected.
3635
3636 If the first-choice server is disqualified, the algorithm will choose another
3637 server based on the request hash, until a server with additional capacity is
3638 found. A higher <factor> allows more imbalance between the servers, while a
3639 lower <factor> means that more servers will be checked on average, affecting
3640 performance. Reasonable values are from 125 to 200.
3641
3642 See also : "balance" and "hash-type".
3643
3644
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003645hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003646 Specify a method to use for mapping hashes to servers
3647 May be used in sections : defaults | frontend | listen | backend
3648 yes | no | yes | yes
3649 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003650 <method> is the method used to select a server from the hash computed by
3651 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003652
Bhaskar98634f02013-10-29 23:30:51 -04003653 map-based the hash table is a static array containing all alive servers.
3654 The hashes will be very smooth, will consider weights, but
3655 will be static in that weight changes while a server is up
3656 will be ignored. This means that there will be no slow start.
3657 Also, since a server is selected by its position in the array,
3658 most mappings are changed when the server count changes. This
3659 means that when a server goes up or down, or when a server is
3660 added to a farm, most connections will be redistributed to
3661 different servers. This can be inconvenient with caches for
3662 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003663
Bhaskar98634f02013-10-29 23:30:51 -04003664 consistent the hash table is a tree filled with many occurrences of each
3665 server. The hash key is looked up in the tree and the closest
3666 server is chosen. This hash is dynamic, it supports changing
3667 weights while the servers are up, so it is compatible with the
3668 slow start feature. It has the advantage that when a server
3669 goes up or down, only its associations are moved. When a
3670 server is added to the farm, only a few part of the mappings
3671 are redistributed, making it an ideal method for caches.
3672 However, due to its principle, the distribution will never be
3673 very smooth and it may sometimes be necessary to adjust a
3674 server's weight or its ID to get a more balanced distribution.
3675 In order to get the same distribution on multiple load
3676 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003677 same IDs. Note: consistent hash uses sdbm and avalanche if no
3678 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003679
3680 <function> is the hash function to be used :
3681
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003682 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003683 reimplementation of ndbm) database library. It was found to do
3684 well in scrambling bits, causing better distribution of the keys
3685 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003686 function with good distribution, unless the total server weight
3687 is a multiple of 64, in which case applying the avalanche
3688 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003689
3690 djb2 this function was first proposed by Dan Bernstein many years ago
3691 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003692 function provides a better distribution than sdbm. It generally
3693 works well with text-based inputs though it can perform extremely
3694 poorly with numeric-only input or when the total server weight is
3695 a multiple of 33, unless the avalanche modifier is also used.
3696
Willy Tarreaua0f42712013-11-14 14:30:35 +01003697 wt6 this function was designed for haproxy while testing other
3698 functions in the past. It is not as smooth as the other ones, but
3699 is much less sensible to the input data set or to the number of
3700 servers. It can make sense as an alternative to sdbm+avalanche or
3701 djb2+avalanche for consistent hashing or when hashing on numeric
3702 data such as a source IP address or a visitor identifier in a URL
3703 parameter.
3704
Willy Tarreau324f07f2015-01-20 19:44:50 +01003705 crc32 this is the most common CRC32 implementation as used in Ethernet,
3706 gzip, PNG, etc. It is slower than the other ones but may provide
3707 a better distribution or less predictable results especially when
3708 used on strings.
3709
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003710 <modifier> indicates an optional method applied after hashing the key :
3711
3712 avalanche This directive indicates that the result from the hash
3713 function above should not be used in its raw form but that
3714 a 4-byte full avalanche hash must be applied first. The
3715 purpose of this step is to mix the resulting bits from the
3716 previous hash in order to avoid any undesired effect when
3717 the input contains some limited values or when the number of
3718 servers is a multiple of one of the hash's components (64
3719 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3720 result less predictable, but it's also not as smooth as when
3721 using the original function. Some testing might be needed
3722 with some workloads. This hash is one of the many proposed
3723 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003724
Bhaskar98634f02013-10-29 23:30:51 -04003725 The default hash type is "map-based" and is recommended for most usages. The
3726 default function is "sdbm", the selection of a function should be based on
3727 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003728
Andrew Rodland17be45e2016-10-25 17:04:12 -04003729 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003730
3731
Willy Tarreau0ba27502007-12-24 16:55:16 +01003732http-check disable-on-404
3733 Enable a maintenance mode upon HTTP/404 response to health-checks
3734 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003735 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003736 Arguments : none
3737
3738 When this option is set, a server which returns an HTTP code 404 will be
3739 excluded from further load-balancing, but will still receive persistent
3740 connections. This provides a very convenient method for Web administrators
3741 to perform a graceful shutdown of their servers. It is also important to note
3742 that a server which is detected as failed while it was in this mode will not
3743 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3744 will immediately be reinserted into the farm. The status on the stats page
3745 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003746 option only works in conjunction with the "httpchk" option. If this option
3747 is used with "http-check expect", then it has precedence over it so that 404
3748 responses will still be considered as soft-stop.
3749
3750 See also : "option httpchk", "http-check expect"
3751
3752
3753http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003754 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003755 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003756 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003757 Arguments :
3758 <match> is a keyword indicating how to look for a specific pattern in the
3759 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003760 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003761 exclamation mark ("!") to negate the match. Spaces are allowed
3762 between the exclamation mark and the keyword. See below for more
3763 details on the supported keywords.
3764
3765 <pattern> is the pattern to look for. It may be a string or a regular
3766 expression. If the pattern contains spaces, they must be escaped
3767 with the usual backslash ('\').
3768
3769 By default, "option httpchk" considers that response statuses 2xx and 3xx
3770 are valid, and that others are invalid. When "http-check expect" is used,
3771 it defines what is considered valid or invalid. Only one "http-check"
3772 statement is supported in a backend. If a server fails to respond or times
3773 out, the check obviously fails. The available matches are :
3774
3775 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003776 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003777 response's status code is exactly this string. If the
3778 "status" keyword is prefixed with "!", then the response
3779 will be considered invalid if the status code matches.
3780
3781 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003782 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003783 response's status code matches the expression. If the
3784 "rstatus" keyword is prefixed with "!", then the response
3785 will be considered invalid if the status code matches.
3786 This is mostly used to check for multiple codes.
3787
3788 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003789 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003790 response's body contains this exact string. If the
3791 "string" keyword is prefixed with "!", then the response
3792 will be considered invalid if the body contains this
3793 string. This can be used to look for a mandatory word at
3794 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003795 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003796 trace).
3797
3798 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003799 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003800 response's body matches this expression. If the "rstring"
3801 keyword is prefixed with "!", then the response will be
3802 considered invalid if the body matches the expression.
3803 This can be used to look for a mandatory word at the end
3804 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003805 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003806
3807 It is important to note that the responses will be limited to a certain size
3808 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3809 Thus, too large responses may not contain the mandatory pattern when using
3810 "string" or "rstring". If a large response is absolutely required, it is
3811 possible to change the default max size by setting the global variable.
3812 However, it is worth keeping in mind that parsing very large responses can
3813 waste some CPU cycles, especially when regular expressions are used, and that
3814 it is always better to focus the checks on smaller resources.
3815
Cyril Bonté32602d22015-01-30 00:07:07 +01003816 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3817 will automatically append a "Connection: close" header, meaning that this
3818 header should not be present in the request provided by "option httpchk".
3819
Willy Tarreaubd741542010-03-16 18:46:54 +01003820 Last, if "http-check expect" is combined with "http-check disable-on-404",
3821 then this last one has precedence when the server responds with 404.
3822
3823 Examples :
3824 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003825 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003826
3827 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003828 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003829
3830 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003831 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003832
3833 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003834 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003835
Willy Tarreaubd741542010-03-16 18:46:54 +01003836 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003837
3838
Willy Tarreauef781042010-01-27 11:53:01 +01003839http-check send-state
3840 Enable emission of a state header with HTTP health checks
3841 May be used in sections : defaults | frontend | listen | backend
3842 yes | no | yes | yes
3843 Arguments : none
3844
3845 When this option is set, haproxy will systematically send a special header
3846 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3847 how they are seen by haproxy. This can be used for instance when a server is
3848 manipulated without access to haproxy and the operator needs to know whether
3849 haproxy still sees it up or not, or if the server is the last one in a farm.
3850
3851 The header is composed of fields delimited by semi-colons, the first of which
3852 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3853 checks on the total number before transition, just as appears in the stats
3854 interface. Next headers are in the form "<variable>=<value>", indicating in
3855 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003856 - a variable "address", containing the address of the backend server.
3857 This corresponds to the <address> field in the server declaration. For
3858 unix domain sockets, it will read "unix".
3859
3860 - a variable "port", containing the port of the backend server. This
3861 corresponds to the <port> field in the server declaration. For unix
3862 domain sockets, it will read "unix".
3863
Willy Tarreauef781042010-01-27 11:53:01 +01003864 - a variable "name", containing the name of the backend followed by a slash
3865 ("/") then the name of the server. This can be used when a server is
3866 checked in multiple backends.
3867
3868 - a variable "node" containing the name of the haproxy node, as set in the
3869 global "node" variable, otherwise the system's hostname if unspecified.
3870
3871 - a variable "weight" indicating the weight of the server, a slash ("/")
3872 and the total weight of the farm (just counting usable servers). This
3873 helps to know if other servers are available to handle the load when this
3874 one fails.
3875
3876 - a variable "scur" indicating the current number of concurrent connections
3877 on the server, followed by a slash ("/") then the total number of
3878 connections on all servers of the same backend.
3879
3880 - a variable "qcur" indicating the current number of requests in the
3881 server's queue.
3882
3883 Example of a header received by the application server :
3884 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
3885 scur=13/22; qcur=0
3886
3887 See also : "option httpchk", "http-check disable-on-404"
3888
Willy Tarreau53275e82017-11-24 07:52:01 +01003889http-request { allow | auth [realm <realm>] | redirect <rule> | reject |
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003890 tarpit [deny_status <status>] | deny [deny_status <status>] |
Willy Tarreauf4c43c12013-06-11 17:01:13 +02003891 add-header <name> <fmt> | set-header <name> <fmt> |
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02003892 capture <sample> [ len <length> | id <id> ] |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003893 del-header <name> | set-nice <nice> | set-log-level <level> |
Sasha Pachev218f0642014-06-16 12:05:59 -06003894 replace-header <name> <match-regex> <replace-fmt> |
3895 replace-value <name> <match-regex> <replace-fmt> |
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01003896 set-method <fmt> | set-path <fmt> | set-query <fmt> |
3897 set-uri <fmt> | set-tos <tos> | set-mark <mark> |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003898 add-acl(<file name>) <key fmt> |
3899 del-acl(<file name>) <key fmt> |
3900 del-map(<file name>) <key fmt> |
Baptiste Assmannbb7e86a2014-09-03 18:29:47 +02003901 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02003902 set-var(<var name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01003903 unset-var(<var name>) |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01003904 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02003905 sc-inc-gpc0(<sc-id>) |
Frédéric Lécaille6778b272018-01-29 15:22:53 +01003906 sc-inc-gpc1(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02003907 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02003908 silent-drop |
William Lallemand86d0df02017-11-24 21:36:45 +01003909 send-spoe-group |
3910 cache-use
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02003911 }
Cyril Bontéf0c60612010-02-06 14:44:47 +01003912 [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003913 Access control for Layer 7 requests
3914
3915 May be used in sections: defaults | frontend | listen | backend
3916 no | yes | yes | yes
3917
Willy Tarreau20b0de52012-12-24 15:45:22 +01003918 The http-request statement defines a set of rules which apply to layer 7
3919 processing. The rules are evaluated in their declaration order when they are
3920 met in a frontend, listen or backend section. Any rule may optionally be
3921 followed by an ACL-based condition, in which case it will only be evaluated
3922 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01003923
Willy Tarreau20b0de52012-12-24 15:45:22 +01003924 The first keyword is the rule's action. Currently supported actions include :
3925 - "allow" : this stops the evaluation of the rules and lets the request
3926 pass the check. No further "http-request" rules are evaluated.
3927
3928 - "deny" : this stops the evaluation of the rules and immediately rejects
Willy Tarreaube1d34d2016-06-26 19:37:59 +02003929 the request and emits an HTTP 403 error, or optionally the status code
3930 specified as an argument to "deny_status". The list of permitted status
3931 codes is limited to those that can be overridden by the "errorfile"
3932 directive. No further "http-request" rules are evaluated.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003933
Willy Tarreau53275e82017-11-24 07:52:01 +01003934 - "reject" : this stops the evaluation of the rules and immediately closes
3935 the connection without sending any response. It acts similarly to the
3936 "tcp-request content reject" rules. It can be useful to force an
3937 immediate connection closure on HTTP/2 connections.
3938
Willy Tarreauccbcc372012-12-27 12:37:57 +01003939 - "tarpit" : this stops the evaluation of the rules and immediately blocks
3940 the request without responding for a delay specified by "timeout tarpit"
3941 or "timeout connect" if the former is not set. After that delay, if the
Jarno Huuskonen800d1762017-03-06 14:56:36 +02003942 client is still connected, an HTTP error 500 (or optionally the status
3943 code specified as an argument to "deny_status") is returned so that the
Willy Tarreauccbcc372012-12-27 12:37:57 +01003944 client does not suspect it has been tarpitted. Logs will report the flags
3945 "PT". The goal of the tarpit rule is to slow down robots during an attack
3946 when they're limited on the number of concurrent requests. It can be very
3947 efficient against very dumb robots, and will significantly reduce the
3948 load on firewalls compared to a "deny" rule. But when facing "correctly"
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003949 developed robots, it can make things worse by forcing haproxy and the
Willy Tarreau2d392c22015-08-24 01:43:45 +02003950 front firewall to support insane number of concurrent connections. See
3951 also the "silent-drop" action below.
Willy Tarreauccbcc372012-12-27 12:37:57 +01003952
Willy Tarreau20b0de52012-12-24 15:45:22 +01003953 - "auth" : this stops the evaluation of the rules and immediately responds
3954 with an HTTP 401 or 407 error code to invite the user to present a valid
3955 user name and password. No further "http-request" rules are evaluated. An
3956 optional "realm" parameter is supported, it sets the authentication realm
3957 that is returned with the response (typically the application's name).
3958
Willy Tarreau81499eb2012-12-27 12:19:02 +01003959 - "redirect" : this performs an HTTP redirection based on a redirect rule.
3960 This is exactly the same as the "redirect" statement except that it
3961 inserts a redirect rule which can be processed in the middle of other
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01003962 "http-request" rules and that these rules use the "log-format" strings.
3963 See the "redirect" keyword for the rule's syntax.
Willy Tarreau81499eb2012-12-27 12:19:02 +01003964
Willy Tarreau20b0de52012-12-24 15:45:22 +01003965 - "add-header" appends an HTTP header field whose name is specified in
3966 <name> and whose value is defined by <fmt> which follows the log-format
3967 rules (see Custom Log Format in section 8.2.4). This is particularly
Davor Ocelice9ed2812017-12-25 17:49:28 +01003968 useful to pass connection-specific information to the server (e.g. the
Willy Tarreau20b0de52012-12-24 15:45:22 +01003969 client's SSL certificate), or to combine several headers into one. This
3970 rule is not final, so it is possible to add other similar rules. Note
3971 that header addition is performed immediately, so one rule might reuse
3972 the resulting header from a previous rule.
3973
3974 - "set-header" does the same as "add-header" except that the header name
3975 is first removed if it existed. This is useful when passing security
3976 information to the server, where the header must not be manipulated by
Willy Tarreau85603282015-01-21 20:39:27 +01003977 external users. Note that the new value is computed before the removal so
3978 it is possible to concatenate a value to an existing header.
Willy Tarreau20b0de52012-12-24 15:45:22 +01003979
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02003980 - "del-header" removes all HTTP header fields whose name is specified in
3981 <name>.
3982
Sasha Pachev218f0642014-06-16 12:05:59 -06003983 - "replace-header" matches the regular expression in all occurrences of
3984 header field <name> according to <match-regex>, and replaces them with
3985 the <replace-fmt> argument. Format characters are allowed in replace-fmt
3986 and work like in <fmt> arguments in "add-header". The match is only
3987 case-sensitive. It is important to understand that this action only
3988 considers whole header lines, regardless of the number of values they
3989 may contain. This usage is suited to headers naturally containing commas
3990 in their value, such as If-Modified-Since and so on.
3991
3992 Example:
3993
3994 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
3995
3996 applied to:
3997
3998 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
3999
4000 outputs:
4001
4002 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4003
4004 assuming the backend IP is 192.168.1.20
4005
4006 - "replace-value" works like "replace-header" except that it matches the
4007 regex against every comma-delimited value of the header field <name>
4008 instead of the entire header. This is suited for all headers which are
4009 allowed to carry more than one value. An example could be the Accept
4010 header.
4011
4012 Example:
4013
4014 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
4015
4016 applied to:
4017
4018 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
4019
4020 outputs:
4021
4022 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
4023
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004024 - "set-method" rewrites the request method with the result of the
4025 evaluation of format string <fmt>. There should be very few valid reasons
4026 for having to do so as this is more likely to break something than to fix
4027 it.
4028
4029 - "set-path" rewrites the request path with the result of the evaluation of
4030 format string <fmt>. The query string, if any, is left intact. If a
4031 scheme and authority is found before the path, they are left intact as
4032 well. If the request doesn't have a path ("*"), this one is replaced with
4033 the format. This can be used to prepend a directory component in front of
4034 a path for example. See also "set-query" and "set-uri".
4035
4036 Example :
4037 # prepend the host name before the path
4038 http-request set-path /%[hdr(host)]%[path]
4039
4040 - "set-query" rewrites the request's query string which appears after the
4041 first question mark ("?") with the result of the evaluation of format
4042 string <fmt>. The part prior to the question mark is left intact. If the
4043 request doesn't contain a question mark and the new value is not empty,
4044 then one is added at the end of the URI, followed by the new value. If
4045 a question mark was present, it will never be removed even if the value
4046 is empty. This can be used to add or remove parameters from the query
4047 string. See also "set-query" and "set-uri".
4048
4049 Example :
4050 # replace "%3D" with "=" in the query string
4051 http-request set-query %[query,regsub(%3D,=,g)]
4052
4053 - "set-uri" rewrites the request URI with the result of the evaluation of
4054 format string <fmt>. The scheme, authority, path and query string are all
4055 replaced at once. This can be used to rewrite hosts in front of proxies,
4056 or to perform complex modifications to the URI such as moving parts
4057 between the path and the query string. See also "set-path" and
4058 "set-query".
4059
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004060 - "set-nice" sets the "nice" factor of the current request being processed.
4061 It only has effect against the other requests being processed at the same
4062 time. The default value is 0, unless altered by the "nice" setting on the
4063 "bind" line. The accepted range is -1024..1024. The higher the value, the
4064 nicest the request will be. Lower values will make the request more
4065 important than other ones. This can be useful to improve the speed of
4066 some requests, or lower the priority of non-important requests. Using
4067 this setting without prior experimentation can cause some major slowdown.
4068
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004069 - "set-log-level" is used to change the log level of the current request
4070 when a certain condition is met. Valid levels are the 8 syslog levels
4071 (see the "log" keyword) plus the special level "silent" which disables
4072 logging for this request. This rule is not final so the last matching
4073 rule wins. This rule can be useful to disable health checks coming from
4074 another equipment.
4075
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004076 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
4077 the client to the value passed in <tos> on platforms which support this.
4078 This value represents the whole 8 bits of the IP TOS field, and can be
4079 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
4080 that only the 6 higher bits are used in DSCP or TOS, and the two lower
Davor Ocelice9ed2812017-12-25 17:49:28 +01004081 bits are always 0. This can be used to adjust some routing behavior on
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004082 border routers based on some information from the request. See RFC 2474,
4083 2597, 3260 and 4594 for more information.
4084
Willy Tarreau51347ed2013-06-11 19:34:13 +02004085 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
4086 client to the value passed in <mark> on platforms which support it. This
4087 value is an unsigned 32 bit value which can be matched by netfilter and
4088 by the routing table. It can be expressed both in decimal or hexadecimal
4089 format (prefixed by "0x"). This can be useful to force certain packets to
4090 take a different route (for example a cheaper network path for bulk
4091 downloads). This works on Linux kernels 2.6.32 and above and requires
4092 admin privileges.
4093
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004094 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
4095 from a file (even a dummy empty file). The file name of the ACL to be
4096 updated is passed between parentheses. It takes one argument: <key fmt>,
4097 which follows log-format rules, to collect content of the new entry. It
4098 performs a lookup in the ACL before insertion, to avoid duplicated (or
4099 more) values. This lookup is done by a linear search and can be expensive
4100 with large lists! It is the equivalent of the "add acl" command from the
4101 stats socket, but can be triggered by an HTTP request.
4102
4103 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
4104 from a file (even a dummy empty file). The file name of the ACL to be
4105 updated is passed between parentheses. It takes one argument: <key fmt>,
4106 which follows log-format rules, to collect content of the entry to delete.
4107 It is the equivalent of the "del acl" command from the stats socket, but
4108 can be triggered by an HTTP request.
4109
4110 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
4111 from a file (even a dummy empty file). The file name of the MAP to be
4112 updated is passed between parentheses. It takes one argument: <key fmt>,
4113 which follows log-format rules, to collect content of the entry to delete.
4114 It takes one argument: "file name" It is the equivalent of the "del map"
4115 command from the stats socket, but can be triggered by an HTTP request.
4116
4117 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
4118 from a file (even a dummy empty file). The file name of the MAP to be
4119 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
4120 which follows log-format rules, used to collect MAP key, and <value fmt>,
4121 which follows log-format rules, used to collect content for the new entry.
4122 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4123 more) values. This lookup is done by a linear search and can be expensive
4124 with large lists! It is the equivalent of the "set map" command from the
4125 stats socket, but can be triggered by an HTTP request.
4126
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004127 - capture <sample> [ len <length> | id <id> ] :
Willy Tarreaua9083d02015-05-08 15:27:59 +02004128 captures sample expression <sample> from the request buffer, and converts
4129 it to a string of at most <len> characters. The resulting string is
4130 stored into the next request "capture" slot, so it will possibly appear
4131 next to some captured HTTP headers. It will then automatically appear in
4132 the logs, and it will be possible to extract it using sample fetch rules
4133 to feed it into headers or anything. The length should be limited given
4134 that this size will be allocated for each capture during the whole
4135 session life. Please check section 7.3 (Fetching samples) and "capture
4136 request header" for more information.
4137
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004138 If the keyword "id" is used instead of "len", the action tries to store
4139 the captured string in a previously declared capture slot. This is useful
4140 to run captures in backends. The slot id can be declared by a previous
4141 directive "http-request capture" or with the "declare capture" keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01004142 If the slot <id> doesn't exist, then HAProxy fails parsing the
4143 configuration to prevent unexpected behavior at run time.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004144
William Lallemand86d0df02017-11-24 21:36:45 +01004145 - cache-use <name> :
4146 See section 10.2 about cache setup.
4147
Willy Tarreau09448f72014-06-25 18:12:15 +02004148 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4149 enables tracking of sticky counters from current request. These rules
4150 do not stop evaluation and do not change default action. Three sets of
4151 counters may be simultaneously tracked by the same connection. The first
4152 "track-sc0" rule executed enables tracking of the counters of the
4153 specified table as the first set. The first "track-sc1" rule executed
4154 enables tracking of the counters of the specified table as the second
4155 set. The first "track-sc2" rule executed enables tracking of the
4156 counters of the specified table as the third set. It is a recommended
4157 practice to use the first set of counters for the per-frontend counters
4158 and the second set for the per-backend ones. But this is just a
4159 guideline, all may be used everywhere.
4160
4161 These actions take one or two arguments :
4162 <key> is mandatory, and is a sample expression rule as described
4163 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01004164 request or connection will be analyzed, extracted, combined,
Willy Tarreau09448f72014-06-25 18:12:15 +02004165 and used to select which table entry to update the counters.
4166
4167 <table> is an optional table to be used instead of the default one,
4168 which is the stick-table declared in the current proxy. All
4169 the counters for the matches and updates for the key will
4170 then be performed in that table until the session ends.
4171
4172 Once a "track-sc*" rule is executed, the key is looked up in the table
4173 and if it is not found, an entry is allocated for it. Then a pointer to
4174 that entry is kept during all the session's life, and this entry's
4175 counters are updated as often as possible, every time the session's
4176 counters are updated, and also systematically when the session ends.
4177 Counters are only updated for events that happen after the tracking has
4178 been started. As an exception, connection counters and request counters
4179 are systematically updated so that they reflect useful information.
4180
4181 If the entry tracks concurrent connection counters, one connection is
4182 counted for as long as the entry is tracked, and the entry will not
4183 expire during that time. Tracking counters also provides a performance
4184 advantage over just checking the keys, because only one table lookup is
4185 performed for all ACL checks that make use of it.
4186
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004187 - sc-set-gpt0(<sc-id>) <int> :
4188 This action sets the GPT0 tag according to the sticky counter designated
4189 by <sc-id> and the value of <int>. The expected result is a boolean. If
4190 an error occurs, this action silently fails and the actions evaluation
4191 continues.
4192
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004193 - sc-inc-gpc0(<sc-id>):
4194 This action increments the GPC0 counter according with the sticky counter
4195 designated by <sc-id>. If an error occurs, this action silently fails and
4196 the actions evaluation continues.
4197
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004198 - sc-inc-gpc1(<sc-id>):
4199 Same as "sc-inc-gpc0" action above but for GPC1 counter.
4200
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004201 - set-var(<var-name>) <expr> :
4202 Is used to set the contents of a variable. The variable is declared
4203 inline.
4204
Daniel Schneller0b547052016-03-21 20:46:57 +01004205 <var-name> The name of the variable starts with an indication about
4206 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004207 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004208 "sess" : the variable is shared with the whole session
4209 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004210 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004211 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004212 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004213 "res" : the variable is shared only during response
4214 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004215 This prefix is followed by a name. The separator is a '.'.
Daniel Schneller0b547052016-03-21 20:46:57 +01004216 The name may only contain characters 'a-z', 'A-Z', '0-9'
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004217 and '_'.
4218
4219 <expr> Is a standard HAProxy expression formed by a sample-fetch
4220 followed by some converters.
4221
4222 Example:
4223
4224 http-request set-var(req.my_var) req.fhdr(user-agent),lower
4225
Christopher Faulet85d79c92016-11-09 16:54:56 +01004226 - unset-var(<var-name>) :
4227 Is used to unset a variable. See above for details about <var-name>.
4228
4229 Example:
4230
4231 http-request unset-var(req.my_var)
4232
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004233 - set-src <expr> :
4234 Is used to set the source IP address to the value of specified
4235 expression. Useful when a proxy in front of HAProxy rewrites source IP,
4236 but provides the correct IP in a HTTP header; or you want to mask
4237 source IP for privacy.
4238
4239 <expr> Is a standard HAProxy expression formed by a sample-fetch
4240 followed by some converters.
4241
4242 Example:
4243
4244 http-request set-src hdr(x-forwarded-for)
4245 http-request set-src src,ipmask(24)
4246
Willy Tarreau00005ce2016-10-21 15:07:45 +02004247 When possible, set-src preserves the original source port as long as the
4248 address family allows it, otherwise the source port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004249
William Lallemand44be6402016-05-25 01:51:35 +02004250 - set-src-port <expr> :
4251 Is used to set the source port address to the value of specified
4252 expression.
4253
4254 <expr> Is a standard HAProxy expression formed by a sample-fetch
4255 followed by some converters.
4256
4257 Example:
4258
4259 http-request set-src-port hdr(x-port)
4260 http-request set-src-port int(4000)
4261
Willy Tarreau00005ce2016-10-21 15:07:45 +02004262 When possible, set-src-port preserves the original source address as long
4263 as the address family supports a port, otherwise it forces the source
4264 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004265
William Lallemand13e9b0c2016-05-25 02:34:07 +02004266 - set-dst <expr> :
4267 Is used to set the destination IP address to the value of specified
4268 expression. Useful when a proxy in front of HAProxy rewrites destination
4269 IP, but provides the correct IP in a HTTP header; or you want to mask
4270 the IP for privacy. If you want to connect to the new address/port, use
4271 '0.0.0.0:0' as a server address in the backend.
4272
4273 <expr> Is a standard HAProxy expression formed by a sample-fetch
4274 followed by some converters.
4275
4276 Example:
4277
4278 http-request set-dst hdr(x-dst)
4279 http-request set-dst dst,ipmask(24)
4280
Willy Tarreau00005ce2016-10-21 15:07:45 +02004281 When possible, set-dst preserves the original destination port as long as
4282 the address family allows it, otherwise the destination port is set to 0.
4283
William Lallemand13e9b0c2016-05-25 02:34:07 +02004284 - set-dst-port <expr> :
4285 Is used to set the destination port address to the value of specified
4286 expression. If you want to connect to the new address/port, use
4287 '0.0.0.0:0' as a server address in the backend.
4288
4289 <expr> Is a standard HAProxy expression formed by a sample-fetch
4290 followed by some converters.
4291
4292 Example:
4293
4294 http-request set-dst-port hdr(x-port)
4295 http-request set-dst-port int(4000)
4296
Willy Tarreau00005ce2016-10-21 15:07:45 +02004297 When possible, set-dst-port preserves the original destination address as
4298 long as the address family supports a port, otherwise it forces the
4299 destination address to IPv4 "0.0.0.0" before rewriting the port.
4300
Willy Tarreau2d392c22015-08-24 01:43:45 +02004301 - "silent-drop" : this stops the evaluation of the rules and makes the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004302 client-facing connection suddenly disappear using a system-dependent way
Willy Tarreau2d392c22015-08-24 01:43:45 +02004303 that tries to prevent the client from being notified. The effect it then
4304 that the client still sees an established connection while there's none
4305 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4306 except that it doesn't use any local resource at all on the machine
4307 running HAProxy. It can resist much higher loads than "tarpit", and slow
Davor Ocelice9ed2812017-12-25 17:49:28 +01004308 down stronger attackers. It is important to understand the impact of using
4309 this mechanism. All stateful equipment placed between the client and
Willy Tarreau2d392c22015-08-24 01:43:45 +02004310 HAProxy (firewalls, proxies, load balancers) will also keep the
4311 established connection for a long time and may suffer from this action.
4312 On modern Linux systems running with enough privileges, the TCP_REPAIR
4313 socket option is used to block the emission of a TCP reset. On other
4314 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4315 pass the first router, though it's still delivered to local networks. Do
4316 not use it unless you fully understand how it works.
4317
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004318
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004319 - "wait-for-handshake" : this will delay the processing of the request
4320 until the SSL handshake happened. This is mostly useful to delay
4321 processing early data until we're sure they are valid.
4322
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004323 - send-spoe-group <engine-name> <group-name> :
4324 This action is used to trigger sending of a group of SPOE messages. To do
4325 so, the SPOE engine used to send messages must be defined, as well as the
4326 SPOE group to send. Of course, the SPOE engine must refer to an existing
4327 SPOE filter. If not engine name is provided on the SPOE filter line, the
4328 SPOE agent name must be used.
4329
4330 <engine-name> The SPOE engine name.
4331
4332 <group-name> The SPOE group name as specified in the engine
4333 configuration.
4334
Willy Tarreau20b0de52012-12-24 15:45:22 +01004335 There is no limit to the number of http-request statements per instance.
4336
4337 It is important to know that http-request rules are processed very early in
4338 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004339 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4340 visible by almost all further ACL rules.
4341
4342 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4343 in newer versions (>= 1.5). But if you need to use regular expression to
4344 delete headers, you can still use "reqdel". Also please use
4345 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004346
4347 Example:
Cyril Bonté78caf842010-03-10 22:41:43 +01004348 acl nagios src 192.168.129.3
4349 acl local_net src 192.168.0.0/16
4350 acl auth_ok http_auth(L1)
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004351
Cyril Bonté78caf842010-03-10 22:41:43 +01004352 http-request allow if nagios
4353 http-request allow if local_net auth_ok
4354 http-request auth realm Gimme if local_net auth_ok
4355 http-request deny
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004356
Cyril Bonté78caf842010-03-10 22:41:43 +01004357 Example:
4358 acl auth_ok http_auth_group(L1) G1
Cyril Bonté78caf842010-03-10 22:41:43 +01004359 http-request auth unless auth_ok
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004360
Willy Tarreau20b0de52012-12-24 15:45:22 +01004361 Example:
4362 http-request set-header X-Haproxy-Current-Date %T
4363 http-request set-header X-SSL %[ssl_fc]
Willy Tarreaufca42612015-08-27 17:15:05 +02004364 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004365 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4366 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4367 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4368 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4369 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4370 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
4371
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004372 Example:
4373 acl key req.hdr(X-Add-Acl-Key) -m found
4374 acl add path /addacl
4375 acl del path /delacl
4376
4377 acl myhost hdr(Host) -f myhost.lst
4378
4379 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4380 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
4381
4382 Example:
4383 acl value req.hdr(X-Value) -m found
4384 acl setmap path /setmap
4385 acl delmap path /delmap
4386
4387 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4388
4389 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4390 http-request del-map(map.lst) %[src] if delmap
4391
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02004392 See also : "stats http-request", section 3.4 about userlists and section 7
4393 about ACL usage.
Willy Tarreauef781042010-01-27 11:53:01 +01004394
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004395http-response { allow | deny | add-header <name> <fmt> | set-nice <nice> |
Willy Tarreau51d861a2015-05-22 17:30:48 +02004396 capture <sample> id <id> | redirect <rule> |
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004397 set-header <name> <fmt> | del-header <name> |
Sasha Pachev218f0642014-06-16 12:05:59 -06004398 replace-header <name> <regex-match> <replace-fmt> |
4399 replace-value <name> <regex-match> <replace-fmt> |
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004400 set-status <status> [reason <str>] |
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004401 set-log-level <level> | set-mark <mark> | set-tos <tos> |
4402 add-acl(<file name>) <key fmt> |
4403 del-acl(<file name>) <key fmt> |
4404 del-map(<file name>) <key fmt> |
Thierry FOURNIER90da1912015-03-05 11:17:06 +01004405 set-map(<file name>) <key fmt> <value fmt> |
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004406 set-var(<var-name>) <expr> |
Christopher Faulet85d79c92016-11-09 16:54:56 +01004407 unset-var(<var-name>) |
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004408 { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] |
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004409 sc-inc-gpc0(<sc-id>) |
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004410 sc-inc-gpc1(<sc-id>) |
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004411 sc-set-gpt0(<sc-id>) <int> |
Willy Tarreau2d392c22015-08-24 01:43:45 +02004412 silent-drop |
William Lallemand86d0df02017-11-24 21:36:45 +01004413 send-spoe-group |
4414 cache-store
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004415 }
Lukas Tribus2dd1d1a2013-06-19 23:34:41 +02004416 [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004417 Access control for Layer 7 responses
4418
4419 May be used in sections: defaults | frontend | listen | backend
4420 no | yes | yes | yes
4421
4422 The http-response statement defines a set of rules which apply to layer 7
4423 processing. The rules are evaluated in their declaration order when they are
4424 met in a frontend, listen or backend section. Any rule may optionally be
4425 followed by an ACL-based condition, in which case it will only be evaluated
4426 if the condition is true. Since these rules apply on responses, the backend
4427 rules are applied first, followed by the frontend's rules.
4428
4429 The first keyword is the rule's action. Currently supported actions include :
4430 - "allow" : this stops the evaluation of the rules and lets the response
4431 pass the check. No further "http-response" rules are evaluated for the
4432 current section.
4433
4434 - "deny" : this stops the evaluation of the rules and immediately rejects
4435 the response and emits an HTTP 502 error. No further "http-response"
4436 rules are evaluated.
4437
4438 - "add-header" appends an HTTP header field whose name is specified in
4439 <name> and whose value is defined by <fmt> which follows the log-format
4440 rules (see Custom Log Format in section 8.2.4). This may be used to send
4441 a cookie to a client for example, or to pass some internal information.
4442 This rule is not final, so it is possible to add other similar rules.
4443 Note that header addition is performed immediately, so one rule might
4444 reuse the resulting header from a previous rule.
4445
4446 - "set-header" does the same as "add-header" except that the header name
4447 is first removed if it existed. This is useful when passing security
4448 information to the server, where the header must not be manipulated by
4449 external users.
4450
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004451 - "del-header" removes all HTTP header fields whose name is specified in
4452 <name>.
4453
Sasha Pachev218f0642014-06-16 12:05:59 -06004454 - "replace-header" matches the regular expression in all occurrences of
4455 header field <name> according to <match-regex>, and replaces them with
4456 the <replace-fmt> argument. Format characters are allowed in replace-fmt
4457 and work like in <fmt> arguments in "add-header". The match is only
4458 case-sensitive. It is important to understand that this action only
4459 considers whole header lines, regardless of the number of values they
4460 may contain. This usage is suited to headers naturally containing commas
4461 in their value, such as Set-Cookie, Expires and so on.
4462
4463 Example:
4464
4465 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4466
4467 applied to:
4468
4469 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4470
4471 outputs:
4472
4473 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4474
4475 assuming the backend IP is 192.168.1.20.
4476
4477 - "replace-value" works like "replace-header" except that it matches the
4478 regex against every comma-delimited value of the header field <name>
4479 instead of the entire header. This is suited for all headers which are
4480 allowed to carry more than one value. An example could be the Accept
4481 header.
4482
4483 Example:
4484
4485 http-response replace-value Cache-control ^public$ private
4486
4487 applied to:
4488
4489 Cache-Control: max-age=3600, public
4490
4491 outputs:
4492
4493 Cache-Control: max-age=3600, private
4494
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004495 - "set-status" replaces the response status code with <status> which must
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004496 be an integer between 100 and 999. Optionally, a custom reason text can be
4497 provided defined by <str>, or the default reason for the specified code
4498 will be used as a fallback.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004499
4500 Example:
4501
4502 # return "431 Request Header Fields Too Large"
4503 http-response set-status 431
Robin H. Johnson52f5db22017-01-01 13:10:52 -08004504 # return "503 Slow Down", custom reason
4505 http-response set-status 503 reason "Slow Down".
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004506
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004507 - "set-nice" sets the "nice" factor of the current request being processed.
4508 It only has effect against the other requests being processed at the same
4509 time. The default value is 0, unless altered by the "nice" setting on the
4510 "bind" line. The accepted range is -1024..1024. The higher the value, the
4511 nicest the request will be. Lower values will make the request more
4512 important than other ones. This can be useful to improve the speed of
4513 some requests, or lower the priority of non-important requests. Using
4514 this setting without prior experimentation can cause some major slowdown.
4515
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004516 - "set-log-level" is used to change the log level of the current request
4517 when a certain condition is met. Valid levels are the 8 syslog levels
4518 (see the "log" keyword) plus the special level "silent" which disables
4519 logging for this request. This rule is not final so the last matching
4520 rule wins. This rule can be useful to disable health checks coming from
4521 another equipment.
4522
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004523 - "set-tos" is used to set the TOS or DSCP field value of packets sent to
4524 the client to the value passed in <tos> on platforms which support this.
4525 This value represents the whole 8 bits of the IP TOS field, and can be
4526 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note
4527 that only the 6 higher bits are used in DSCP or TOS, and the two lower
Davor Ocelice9ed2812017-12-25 17:49:28 +01004528 bits are always 0. This can be used to adjust some routing behavior on
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004529 border routers based on some information from the request. See RFC 2474,
4530 2597, 3260 and 4594 for more information.
4531
Willy Tarreau51347ed2013-06-11 19:34:13 +02004532 - "set-mark" is used to set the Netfilter MARK on all packets sent to the
4533 client to the value passed in <mark> on platforms which support it. This
4534 value is an unsigned 32 bit value which can be matched by netfilter and
4535 by the routing table. It can be expressed both in decimal or hexadecimal
4536 format (prefixed by "0x"). This can be useful to force certain packets to
4537 take a different route (for example a cheaper network path for bulk
4538 downloads). This works on Linux kernels 2.6.32 and above and requires
4539 admin privileges.
4540
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004541 - "add-acl" is used to add a new entry into an ACL. The ACL must be loaded
4542 from a file (even a dummy empty file). The file name of the ACL to be
4543 updated is passed between parentheses. It takes one argument: <key fmt>,
4544 which follows log-format rules, to collect content of the new entry. It
4545 performs a lookup in the ACL before insertion, to avoid duplicated (or
4546 more) values. This lookup is done by a linear search and can be expensive
4547 with large lists! It is the equivalent of the "add acl" command from the
4548 stats socket, but can be triggered by an HTTP response.
4549
4550 - "del-acl" is used to delete an entry from an ACL. The ACL must be loaded
4551 from a file (even a dummy empty file). The file name of the ACL to be
4552 updated is passed between parentheses. It takes one argument: <key fmt>,
4553 which follows log-format rules, to collect content of the entry to delete.
4554 It is the equivalent of the "del acl" command from the stats socket, but
4555 can be triggered by an HTTP response.
4556
4557 - "del-map" is used to delete an entry from a MAP. The MAP must be loaded
4558 from a file (even a dummy empty file). The file name of the MAP to be
4559 updated is passed between parentheses. It takes one argument: <key fmt>,
4560 which follows log-format rules, to collect content of the entry to delete.
4561 It takes one argument: "file name" It is the equivalent of the "del map"
4562 command from the stats socket, but can be triggered by an HTTP response.
4563
4564 - "set-map" is used to add a new entry into a MAP. The MAP must be loaded
4565 from a file (even a dummy empty file). The file name of the MAP to be
4566 updated is passed between parentheses. It takes 2 arguments: <key fmt>,
4567 which follows log-format rules, used to collect MAP key, and <value fmt>,
4568 which follows log-format rules, used to collect content for the new entry.
4569 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4570 more) values. This lookup is done by a linear search and can be expensive
4571 with large lists! It is the equivalent of the "set map" command from the
4572 stats socket, but can be triggered by an HTTP response.
4573
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004574 - capture <sample> id <id> :
4575 captures sample expression <sample> from the response buffer, and converts
4576 it to a string. The resulting string is stored into the next request
4577 "capture" slot, so it will possibly appear next to some captured HTTP
4578 headers. It will then automatically appear in the logs, and it will be
4579 possible to extract it using sample fetch rules to feed it into headers or
4580 anything. Please check section 7.3 (Fetching samples) and "capture
4581 response header" for more information.
4582
4583 The keyword "id" is the id of the capture slot which is used for storing
4584 the string. The capture slot must be defined in an associated frontend.
4585 This is useful to run captures in backends. The slot id can be declared by
4586 a previous directive "http-response capture" or with the "declare capture"
4587 keyword.
Baptiste Assmanne9544932015-11-03 23:31:35 +01004588 If the slot <id> doesn't exist, then HAProxy fails parsing the
4589 configuration to prevent unexpected behavior at run time.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004590
William Lallemand86d0df02017-11-24 21:36:45 +01004591 - cache-store <name> :
4592 See section 10.2 about cache setup.
4593
Willy Tarreau51d861a2015-05-22 17:30:48 +02004594 - "redirect" : this performs an HTTP redirection based on a redirect rule.
4595 This supports a format string similarly to "http-request redirect" rules,
4596 with the exception that only the "location" type of redirect is possible
4597 on the response. See the "redirect" keyword for the rule's syntax. When
4598 a redirect rule is applied during a response, connections to the server
4599 are closed so that no data can be forwarded from the server to the client.
4600
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004601 - set-var(<var-name>) expr:
4602 Is used to set the contents of a variable. The variable is declared
4603 inline.
4604
Daniel Schneller0b547052016-03-21 20:46:57 +01004605 <var-name> The name of the variable starts with an indication about
4606 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01004607 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01004608 "sess" : the variable is shared with the whole session
4609 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004610 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01004611 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004612 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01004613 "res" : the variable is shared only during response
4614 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004615 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01004616 The name may only contain characters 'a-z', 'A-Z', '0-9',
4617 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004618
4619 <expr> Is a standard HAProxy expression formed by a sample-fetch
4620 followed by some converters.
4621
4622 Example:
4623
4624 http-response set-var(sess.last_redir) res.hdr(location)
4625
Christopher Faulet85d79c92016-11-09 16:54:56 +01004626 - unset-var(<var-name>) :
4627 Is used to unset a variable. See above for details about <var-name>.
4628
4629 Example:
4630
4631 http-response unset-var(sess.last_redir)
4632
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004633 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
4634 enables tracking of sticky counters from current response. Please refer to
4635 "http-request track-sc" for a complete description. The only difference
4636 from "http-request track-sc" is the <key> sample expression can only make
Davor Ocelice9ed2812017-12-25 17:49:28 +01004637 use of samples in response (e.g. res.*, status etc.) and samples below
4638 Layer 6 (e.g. SSL-related samples, see section 7.3.4). If the sample is
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004639 not supported, haproxy will fail and warn while parsing the config.
4640
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004641 - sc-set-gpt0(<sc-id>) <int> :
4642 This action sets the GPT0 tag according to the sticky counter designated
4643 by <sc-id> and the value of <int>. The expected result is a boolean. If
4644 an error occurs, this action silently fails and the actions evaluation
4645 continues.
4646
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004647 - sc-inc-gpc0(<sc-id>):
4648 This action increments the GPC0 counter according with the sticky counter
4649 designated by <sc-id>. If an error occurs, this action silently fails and
4650 the actions evaluation continues.
4651
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004652 - sc-inc-gpc1(<sc-id>):
4653 Same as "sc-inc-gpc0" action above but for GPC1 counter.
4654
Willy Tarreau2d392c22015-08-24 01:43:45 +02004655 - "silent-drop" : this stops the evaluation of the rules and makes the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004656 client-facing connection suddenly disappear using a system-dependent way
Willy Tarreau2d392c22015-08-24 01:43:45 +02004657 that tries to prevent the client from being notified. The effect it then
4658 that the client still sees an established connection while there's none
4659 on HAProxy. The purpose is to achieve a comparable effect to "tarpit"
4660 except that it doesn't use any local resource at all on the machine
4661 running HAProxy. It can resist much higher loads than "tarpit", and slow
Davor Ocelice9ed2812017-12-25 17:49:28 +01004662 down stronger attackers. It is important to understand the impact of using
4663 this mechanism. All stateful equipment placed between the client and
Willy Tarreau2d392c22015-08-24 01:43:45 +02004664 HAProxy (firewalls, proxies, load balancers) will also keep the
4665 established connection for a long time and may suffer from this action.
4666 On modern Linux systems running with enough privileges, the TCP_REPAIR
4667 socket option is used to block the emission of a TCP reset. On other
4668 systems, the socket's TTL is reduced to 1 so that the TCP reset doesn't
4669 pass the first router, though it's still delivered to local networks. Do
4670 not use it unless you fully understand how it works.
4671
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004672 - send-spoe-group <engine-name> <group-name> :
4673 This action is used to trigger sending of a group of SPOE messages. To do
4674 so, the SPOE engine used to send messages must be defined, as well as the
4675 SPOE group to send. Of course, the SPOE engine must refer to an existing
4676 SPOE filter. If not engine name is provided on the SPOE filter line, the
4677 SPOE agent name must be used.
4678
4679 <engine-name> The SPOE engine name.
4680
4681 <group-name> The SPOE group name as specified in the engine
4682 configuration.
4683
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004684 There is no limit to the number of http-response statements per instance.
4685
Godbach09250262013-07-02 01:19:15 +08004686 It is important to know that http-response rules are processed very early in
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004687 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
Davor Ocelice9ed2812017-12-25 17:49:28 +01004688 headers added by "add-header"/"set-header" are visible by almost all further
4689 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004690
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004691 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4692 in newer versions (>= 1.5). But if you need to use regular expression to
4693 delete headers, you can still use "rspdel". Also please use
4694 "http-response deny" instead of "rspdeny".
4695
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004696 Example:
4697 acl key_acl res.hdr(X-Acl-Key) -m found
4698
4699 acl myhost hdr(Host) -f myhost.lst
4700
4701 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4702 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4703
4704 Example:
4705 acl value res.hdr(X-Value) -m found
4706
4707 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
4708
4709 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4710 http-response del-map(map.lst) %[src] if ! value
4711
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004712 See also : "http-request", section 3.4 about userlists and section 7 about
4713 ACL usage.
4714
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004715
Willy Tarreau30631952015-08-06 15:05:24 +02004716http-reuse { never | safe | aggressive | always }
4717 Declare how idle HTTP connections may be shared between requests
4718
4719 May be used in sections: defaults | frontend | listen | backend
4720 yes | no | yes | yes
4721
4722 By default, a connection established between haproxy and the backend server
4723 belongs to the session that initiated it. The downside is that between the
4724 response and the next request, the connection remains idle and is not used.
4725 In many cases for performance reasons it is desirable to make it possible to
4726 reuse these idle connections to serve other requests from different sessions.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004727 This directive allows to tune this behavior.
Willy Tarreau30631952015-08-06 15:05:24 +02004728
4729 The argument indicates the desired connection reuse strategy :
4730
4731 - "never" : idle connections are never shared between sessions. This is
4732 the default choice. It may be enforced to cancel a different
4733 strategy inherited from a defaults section or for
4734 troubleshooting. For example, if an old bogus application
4735 considers that multiple requests over the same connection come
4736 from the same client and it is not possible to fix the
4737 application, it may be desirable to disable connection sharing
4738 in a single backend. An example of such an application could
4739 be an old haproxy using cookie insertion in tunnel mode and
4740 not checking any request past the first one.
4741
4742 - "safe" : this is the recommended strategy. The first request of a
4743 session is always sent over its own connection, and only
4744 subsequent requests may be dispatched over other existing
4745 connections. This ensures that in case the server closes the
4746 connection when the request is being sent, the browser can
4747 decide to silently retry it. Since it is exactly equivalent to
4748 regular keep-alive, there should be no side effects.
4749
4750 - "aggressive" : this mode may be useful in webservices environments where
4751 all servers are not necessarily known and where it would be
4752 appreciable to deliver most first requests over existing
4753 connections. In this case, first requests are only delivered
4754 over existing connections that have been reused at least once,
4755 proving that the server correctly supports connection reuse.
4756 It should only be used when it's sure that the client can
4757 retry a failed request once in a while and where the benefit
4758 of aggressive connection reuse significantly outweights the
4759 downsides of rare connection failures.
4760
4761 - "always" : this mode is only recommended when the path to the server is
4762 known for never breaking existing connections quickly after
4763 releasing them. It allows the first request of a session to be
4764 sent to an existing connection. This can provide a significant
4765 performance increase over the "safe" strategy when the backend
4766 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004767 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004768 sharing. It is recommended that the "http-keep-alive" timeout
4769 remains low in this mode so that no dead connections remain
4770 usable. In most cases, this will lead to the same performance
4771 gains as "aggressive" but with more risks. It should only be
4772 used when it improves the situation over "aggressive".
4773
4774 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004775 connection properties and compatibility. Specifically :
4776 - connections made with "usesrc" followed by a client-dependent value
4777 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004778
4779 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004780 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004781
4782 - connections receiving a status code 401 or 407 expect some authentication
4783 to be sent in return. Due to certain bogus authentication schemes (such
4784 as NTLM) relying on the connection, these connections are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004785 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004786
4787 No connection pool is involved, once a session dies, the last idle connection
4788 it was attached to is deleted at the same time. This ensures that connections
4789 may not last after all sessions are closed.
4790
4791 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4792 because almost no new connection will be established while idle connections
4793 remain available. This is particularly true with the "always" strategy.
4794
4795 See also : "option http-keep-alive", "server maxconn"
4796
4797
Mark Lamourinec2247f02012-01-04 13:02:01 -05004798http-send-name-header [<header>]
4799 Add the server name to a request. Use the header string given by <header>
4800
4801 May be used in sections: defaults | frontend | listen | backend
4802 yes | no | yes | yes
4803
4804 Arguments :
4805
4806 <header> The header string to use to send the server name
4807
4808 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01004809 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05004810 is added with the header string proved.
4811
4812 See also : "server"
4813
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004814id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02004815 Set a persistent ID to a proxy.
4816 May be used in sections : defaults | frontend | listen | backend
4817 no | yes | yes | yes
4818 Arguments : none
4819
4820 Set a persistent ID for the proxy. This ID must be unique and positive.
4821 An unused ID will automatically be assigned if unset. The first assigned
4822 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01004823
4824
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004825ignore-persist { if | unless } <condition>
4826 Declare a condition to ignore persistence
4827 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004828 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004829
4830 By default, when cookie persistence is enabled, every requests containing
4831 the cookie are unconditionally persistent (assuming the target server is up
4832 and running).
4833
4834 The "ignore-persist" statement allows one to declare various ACL-based
4835 conditions which, when met, will cause a request to ignore persistence.
4836 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004837 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004838 persistence for a specific User-Agent (for example, some web crawler bots).
4839
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004840 The persistence is ignored when an "if" condition is met, or unless an
4841 "unless" condition is met.
4842
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004843 Example:
4844 acl url_static path_beg /static /images /img /css
4845 acl url_static path_end .gif .png .jpg .css .js
4846 ignore-persist if url_static
4847
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004848 See also : "force-persist", "cookie", and section 7 about ACL usage.
4849
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004850load-server-state-from-file { global | local | none }
4851 Allow seamless reload of HAProxy
4852 May be used in sections: defaults | frontend | listen | backend
4853 yes | no | yes | yes
4854
4855 This directive points HAProxy to a file where server state from previous
4856 running process has been saved. That way, when starting up, before handling
4857 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01004858 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004859 to tell haproxy which file to use. For now, only 2 arguments to either prevent
4860 loading state or load states from a file containing all backends and servers.
4861 The state file can be generated by running the command "show servers state"
4862 over the stats socket and redirect output.
4863
Davor Ocelice9ed2812017-12-25 17:49:28 +01004864 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004865 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02004866 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004867
4868 Arguments:
4869 global load the content of the file pointed by the global directive
4870 named "server-state-file".
4871
4872 local load the content of the file pointed by the directive
4873 "server-state-file-name" if set. If not set, then the backend
4874 name is used as a file name.
4875
4876 none don't load any stat for this backend
4877
4878 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01004879 - server's IP address is preserved across reloads by default, but the
4880 order can be changed thanks to the server's "init-addr" setting. This
4881 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01004882 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01004883 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004884
4885 - server's weight is applied from previous running process unless it has
4886 has changed between previous and new configuration files.
4887
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004888 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004889
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004890 global
4891 stats socket /tmp/socket
4892 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004893
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004894 defaults
4895 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004896
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004897 backend bk
4898 server s1 127.0.0.1:22 check weight 11
4899 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004900
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004901
4902 Then one can run :
4903
4904 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
4905
4906 Content of the file /tmp/server_state would be like this:
4907
4908 1
4909 # <field names skipped for the doc example>
4910 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
4911 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
4912
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004913 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004914
4915 global
4916 stats socket /tmp/socket
4917 server-state-base /etc/haproxy/states
4918
4919 defaults
4920 load-server-state-from-file local
4921
4922 backend bk
4923 server s1 127.0.0.1:22 check weight 11
4924 server s2 127.0.0.1:22 check weight 12
4925
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02004926
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02004927 Then one can run :
4928
4929 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
4930
4931 Content of the file /etc/haproxy/states/bk would be like this:
4932
4933 1
4934 # <field names skipped for the doc example>
4935 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
4936 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
4937
4938 See also: "server-state-file", "server-state-file-name", and
4939 "show servers state"
4940
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004941
Willy Tarreau2769aa02007-12-27 18:26:09 +01004942log global
Willy Tarreau18324f52014-06-27 18:10:07 +02004943log <address> [len <length>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02004944no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01004945 Enable per-instance logging of events and traffic.
4946 May be used in sections : defaults | frontend | listen | backend
4947 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02004948
4949 Prefix :
4950 no should be used when the logger list must be flushed. For example,
4951 if you don't want to inherit from the default logger list. This
4952 prefix does not allow arguments.
4953
Willy Tarreau2769aa02007-12-27 18:26:09 +01004954 Arguments :
4955 global should be used when the instance's logging parameters are the
4956 same as the global ones. This is the most common usage. "global"
4957 replaces <address>, <facility> and <level> with those of the log
4958 entries found in the "global" section. Only one "log global"
4959 statement may be used per instance, and this form takes no other
4960 parameter.
4961
4962 <address> indicates where to send the logs. It takes the same format as
4963 for the "global" section's logs, and can be one of :
4964
4965 - An IPv4 address optionally followed by a colon (':') and a UDP
4966 port. If no port is specified, 514 is used by default (the
4967 standard syslog port).
4968
David du Colombier24bb5f52011-03-17 10:40:23 +01004969 - An IPv6 address followed by a colon (':') and optionally a UDP
4970 port. If no port is specified, 514 is used by default (the
4971 standard syslog port).
4972
Willy Tarreau2769aa02007-12-27 18:26:09 +01004973 - A filesystem path to a UNIX domain socket, keeping in mind
4974 considerations for chroot (be sure the path is accessible
4975 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01004976 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004977
William Lallemandb2f07452015-05-12 14:27:13 +02004978 You may want to reference some environment variables in the
4979 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01004980
Willy Tarreau18324f52014-06-27 18:10:07 +02004981 <length> is an optional maximum line length. Log lines larger than this
4982 value will be truncated before being sent. The reason is that
4983 syslog servers act differently on log line length. All servers
4984 support the default value of 1024, but some servers simply drop
4985 larger lines while others do log them. If a server supports long
4986 lines, it may make sense to set this value here in order to avoid
4987 truncating long lines. Similarly, if a server drops long lines,
4988 it is preferable to truncate them before sending them. Accepted
4989 values are 80 to 65535 inclusive. The default value of 1024 is
4990 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01004991 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02004992
Willy Tarreau2769aa02007-12-27 18:26:09 +01004993 <facility> must be one of the 24 standard syslog facilities :
4994
4995 kern user mail daemon auth syslog lpr news
4996 uucp cron auth2 ftp ntp audit alert cron2
4997 local0 local1 local2 local3 local4 local5 local6 local7
4998
4999 <level> is optional and can be specified to filter outgoing messages. By
5000 default, all messages are sent. If a level is specified, only
5001 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005002 will be sent. An optional minimum level can be specified. If it
5003 is set, logs emitted with a more severe level than this one will
5004 be capped to this level. This is used to avoid sending "emerg"
5005 messages on all terminals on some default syslog configurations.
5006 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005007
5008 emerg alert crit err warning notice info debug
5009
William Lallemand0f99e342011-10-12 17:50:54 +02005010 It is important to keep in mind that it is the frontend which decides what to
5011 log from a connection, and that in case of content switching, the log entries
5012 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005013
5014 However, backend log declaration define how and where servers status changes
5015 will be logged. Level "notice" will be used to indicate a server going up,
5016 "warning" will be used for termination signals and definitive service
5017 termination, and "alert" will be used for when a server goes down.
5018
5019 Note : According to RFC3164, messages are truncated to 1024 bytes before
5020 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005021
5022 Example :
5023 log global
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005024 log 127.0.0.1:514 local0 notice # only send important events
5025 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005026 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005027
Willy Tarreau2769aa02007-12-27 18:26:09 +01005028
William Lallemand48940402012-01-30 16:47:22 +01005029log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005030 Specifies the log format string to use for traffic logs
5031 May be used in sections: defaults | frontend | listen | backend
5032 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005033
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005034 This directive specifies the log format string that will be used for all logs
5035 resulting from traffic passing through the frontend using this line. If the
5036 directive is used in a defaults section, all subsequent frontends will use
5037 the same log format. Please see section 8.2.4 which covers the log format
5038 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005039
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005040 "log-format" directive overrides previous "option tcplog", "log-format" and
5041 "option httplog" directives.
5042
Dragan Dosen7ad31542015-09-28 17:16:47 +02005043log-format-sd <string>
5044 Specifies the RFC5424 structured-data log format string
5045 May be used in sections: defaults | frontend | listen | backend
5046 yes | yes | yes | no
5047
5048 This directive specifies the RFC5424 structured-data log format string that
5049 will be used for all logs resulting from traffic passing through the frontend
5050 using this line. If the directive is used in a defaults section, all
5051 subsequent frontends will use the same log format. Please see section 8.2.4
5052 which covers the log format string in depth.
5053
5054 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5055 about the RFC5424 structured-data part.
5056
5057 Note : This log format string will be used only for loggers that have set
5058 log format to "rfc5424".
5059
5060 Example :
5061 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5062
5063
Willy Tarreau094af4e2015-01-07 15:03:42 +01005064log-tag <string>
5065 Specifies the log tag to use for all outgoing logs
5066 May be used in sections: defaults | frontend | listen | backend
5067 yes | yes | yes | yes
5068
5069 Sets the tag field in the syslog header to this string. It defaults to the
5070 log-tag set in the global section, otherwise the program name as launched
5071 from the command line, which usually is "haproxy". Sometimes it can be useful
5072 to differentiate between multiple processes running on the same host, or to
5073 differentiate customer instances running in the same process. In the backend,
5074 logs about servers up/down will use this tag. As a hint, it can be convenient
5075 to set a log-tag related to a hosted customer in a defaults section then put
5076 all the frontends and backends for that customer, then start another customer
5077 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005078
Willy Tarreauc35362a2014-04-25 13:58:37 +02005079max-keep-alive-queue <value>
5080 Set the maximum server queue size for maintaining keep-alive connections
5081 May be used in sections: defaults | frontend | listen | backend
5082 yes | no | yes | yes
5083
5084 HTTP keep-alive tries to reuse the same server connection whenever possible,
5085 but sometimes it can be counter-productive, for example if a server has a lot
5086 of connections while other ones are idle. This is especially true for static
5087 servers.
5088
5089 The purpose of this setting is to set a threshold on the number of queued
5090 connections at which haproxy stops trying to reuse the same server and prefers
5091 to find another one. The default value, -1, means there is no limit. A value
5092 of zero means that keep-alive requests will never be queued. For very close
5093 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005094 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005095 use a value of 10 or less). For remote servers suffering from a high latency,
5096 higher values might be needed to cover for the latency and/or the cost of
5097 picking a different server.
5098
5099 Note that this has no impact on responses which are maintained to the same
5100 server consecutively to a 401 response. They will still go to the same server
5101 even if they have to be queued.
5102
5103 See also : "option http-server-close", "option prefer-last-server", server
5104 "maxconn" and cookie persistence.
5105
5106
Willy Tarreau2769aa02007-12-27 18:26:09 +01005107maxconn <conns>
5108 Fix the maximum number of concurrent connections on a frontend
5109 May be used in sections : defaults | frontend | listen | backend
5110 yes | yes | yes | no
5111 Arguments :
5112 <conns> is the maximum number of concurrent connections the frontend will
5113 accept to serve. Excess connections will be queued by the system
5114 in the socket's listen queue and will be served once a connection
5115 closes.
5116
5117 If the system supports it, it can be useful on big sites to raise this limit
5118 very high so that haproxy manages connection queues, instead of leaving the
5119 clients with unanswered connection attempts. This value should not exceed the
5120 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005121 of tune.bufsize (16kB by default) each, as well as some other data resulting
5122 in about 33 kB of RAM being consumed per established connection. That means
5123 that a medium system equipped with 1GB of RAM can withstand around
5124 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005125
5126 Also, when <conns> is set to large values, it is possible that the servers
5127 are not sized to accept such loads, and for this reason it is generally wise
5128 to assign them some reasonable connection limits.
5129
Vincent Bernat6341be52012-06-27 17:18:30 +02005130 By default, this value is set to 2000.
5131
Willy Tarreau2769aa02007-12-27 18:26:09 +01005132 See also : "server", global section's "maxconn", "fullconn"
5133
5134
5135mode { tcp|http|health }
5136 Set the running mode or protocol of the instance
5137 May be used in sections : defaults | frontend | listen | backend
5138 yes | yes | yes | yes
5139 Arguments :
5140 tcp The instance will work in pure TCP mode. A full-duplex connection
5141 will be established between clients and servers, and no layer 7
5142 examination will be performed. This is the default mode. It
5143 should be used for SSL, SSH, SMTP, ...
5144
5145 http The instance will work in HTTP mode. The client request will be
5146 analyzed in depth before connecting to any server. Any request
5147 which is not RFC-compliant will be rejected. Layer 7 filtering,
5148 processing and switching will be possible. This is the mode which
5149 brings HAProxy most of its value.
5150
5151 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005152 to incoming connections and close the connection. Alternatively,
5153 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5154 instead. Nothing will be logged in either case. This mode is used
5155 to reply to external components health checks. This mode is
5156 deprecated and should not be used anymore as it is possible to do
5157 the same and even better by combining TCP or HTTP modes with the
5158 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005159
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005160 When doing content switching, it is mandatory that the frontend and the
5161 backend are in the same mode (generally HTTP), otherwise the configuration
5162 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005163
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005164 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005165 defaults http_instances
5166 mode http
5167
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005168 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005169
Willy Tarreau0ba27502007-12-24 16:55:16 +01005170
Cyril Bontéf0c60612010-02-06 14:44:47 +01005171monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005172 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005173 May be used in sections : defaults | frontend | listen | backend
5174 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005175 Arguments :
5176 if <cond> the monitor request will fail if the condition is satisfied,
5177 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005178 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005179 are met, for instance a low number of servers both in a
5180 backend and its backup.
5181
5182 unless <cond> the monitor request will succeed only if the condition is
5183 satisfied, and will fail otherwise. Such a condition may be
5184 based on a test on the presence of a minimum number of active
5185 servers in a list of backends.
5186
5187 This statement adds a condition which can force the response to a monitor
5188 request to report a failure. By default, when an external component queries
5189 the URI dedicated to monitoring, a 200 response is returned. When one of the
5190 conditions above is met, haproxy will return 503 instead of 200. This is
5191 very useful to report a site failure to an external component which may base
5192 routing advertisements between multiple sites on the availability reported by
5193 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005194 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5195 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005196
5197 Example:
5198 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005199 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005200 acl site_dead nbsrv(dynamic) lt 2
5201 acl site_dead nbsrv(static) lt 2
5202 monitor-uri /site_alive
5203 monitor fail if site_dead
5204
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005205 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005206
5207
5208monitor-net <source>
5209 Declare a source network which is limited to monitor requests
5210 May be used in sections : defaults | frontend | listen | backend
5211 yes | yes | yes | no
5212 Arguments :
5213 <source> is the source IPv4 address or network which will only be able to
5214 get monitor responses to any request. It can be either an IPv4
5215 address, a host name, or an address followed by a slash ('/')
5216 followed by a mask.
5217
5218 In TCP mode, any connection coming from a source matching <source> will cause
5219 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005220 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005221 forwarding the connection to a remote server.
5222
5223 In HTTP mode, a connection coming from a source matching <source> will be
5224 accepted, the following response will be sent without waiting for a request,
5225 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5226 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005227 running without forwarding the request to a backend server. Note that this
5228 response is sent in raw format, without any transformation. This is important
5229 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005230
Willy Tarreau82569f92012-09-27 23:48:56 +02005231 Monitor requests are processed very early, just after tcp-request connection
5232 ACLs which are the only ones able to block them. These connections are short
5233 lived and never wait for any data from the client. They cannot be logged, and
5234 it is the intended purpose. They are only used to report HAProxy's health to
5235 an upper component, nothing more. Please note that "monitor fail" rules do
5236 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005237
Willy Tarreau95cd2832010-03-04 23:36:33 +01005238 Last, please note that only one "monitor-net" statement can be specified in
5239 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005240
Willy Tarreau2769aa02007-12-27 18:26:09 +01005241 Example :
5242 # addresses .252 and .253 are just probing us.
5243 frontend www
5244 monitor-net 192.168.0.252/31
5245
5246 See also : "monitor fail", "monitor-uri"
5247
5248
5249monitor-uri <uri>
5250 Intercept a URI used by external components' monitor requests
5251 May be used in sections : defaults | frontend | listen | backend
5252 yes | yes | yes | no
5253 Arguments :
5254 <uri> is the exact URI which we want to intercept to return HAProxy's
5255 health status instead of forwarding the request.
5256
5257 When an HTTP request referencing <uri> will be received on a frontend,
5258 HAProxy will not forward it nor log it, but instead will return either
5259 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5260 conditions defined with "monitor fail". This is normally enough for any
5261 front-end HTTP probe to detect that the service is UP and running without
5262 forwarding the request to a backend server. Note that the HTTP method, the
5263 version and all headers are ignored, but the request must at least be valid
5264 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5265
Willy Tarreau721d8e02017-12-01 18:25:08 +01005266 Monitor requests are processed very early, just after the request is parsed
5267 and even before any "http-request" or "block" rulesets. The only rulesets
5268 applied before are the tcp-request ones. They cannot be logged either, and it
5269 is the intended purpose. They are only used to report HAProxy's health to an
5270 upper component, nothing more. However, it is possible to add any number of
5271 conditions using "monitor fail" and ACLs so that the result can be adjusted
5272 to whatever check can be imagined (most often the number of available servers
5273 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005274
5275 Example :
5276 # Use /haproxy_test to report haproxy's status
5277 frontend www
5278 mode http
5279 monitor-uri /haproxy_test
5280
5281 See also : "monitor fail", "monitor-net"
5282
Willy Tarreau0ba27502007-12-24 16:55:16 +01005283
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005284option abortonclose
5285no option abortonclose
5286 Enable or disable early dropping of aborted requests pending in queues.
5287 May be used in sections : defaults | frontend | listen | backend
5288 yes | no | yes | yes
5289 Arguments : none
5290
5291 In presence of very high loads, the servers will take some time to respond.
5292 The per-instance connection queue will inflate, and the response time will
5293 increase respective to the size of the queue times the average per-session
5294 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005295 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005296 the queue, and slowing down other users, and the servers as well, because the
5297 request will eventually be served, then aborted at the first error
5298 encountered while delivering the response.
5299
5300 As there is no way to distinguish between a full STOP and a simple output
5301 close on the client side, HTTP agents should be conservative and consider
5302 that the client might only have closed its output channel while waiting for
5303 the response. However, this introduces risks of congestion when lots of users
5304 do the same, and is completely useless nowadays because probably no client at
5305 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005306 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005307 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005308 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005309 of being the single component to break rare but valid traffic is extremely
5310 low, which adds to the temptation to be able to abort a session early while
5311 still not served and not pollute the servers.
5312
Davor Ocelice9ed2812017-12-25 17:49:28 +01005313 In HAProxy, the user can choose the desired behavior using the option
5314 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005315 compliant and aborted requests will be served. But when the option is
5316 specified, a session with an incoming channel closed will be aborted while
5317 it is still possible, either pending in the queue for a connection slot, or
5318 during the connection establishment if the server has not yet acknowledged
5319 the connection request. This considerably reduces the queue size and the load
5320 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005321 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005322
5323 If this option has been enabled in a "defaults" section, it can be disabled
5324 in a specific instance by prepending the "no" keyword before it.
5325
5326 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5327
5328
Willy Tarreau4076a152009-04-02 15:18:36 +02005329option accept-invalid-http-request
5330no option accept-invalid-http-request
5331 Enable or disable relaxing of HTTP request parsing
5332 May be used in sections : defaults | frontend | listen | backend
5333 yes | yes | yes | no
5334 Arguments : none
5335
Willy Tarreau91852eb2015-05-01 13:26:00 +02005336 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005337 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005338 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005339 forbidden characters are essentially used to build attacks exploiting server
5340 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5341 server will emit invalid header names for whatever reason (configuration,
5342 implementation) and the issue will not be immediately fixed. In such a case,
5343 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005344 even if that does not make sense, by specifying this option. Similarly, the
5345 list of characters allowed to appear in a URI is well defined by RFC3986, and
5346 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5347 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005348 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005349 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005350 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5351 to pass through (no version specified) and multiple digits for both the major
5352 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005353
5354 This option should never be enabled by default as it hides application bugs
5355 and open security breaches. It should only be deployed after a problem has
5356 been confirmed.
5357
5358 When this option is enabled, erroneous header names will still be accepted in
5359 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005360 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5361 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005362 also helps confirming that the issue has been solved.
5363
5364 If this option has been enabled in a "defaults" section, it can be disabled
5365 in a specific instance by prepending the "no" keyword before it.
5366
5367 See also : "option accept-invalid-http-response" and "show errors" on the
5368 stats socket.
5369
5370
5371option accept-invalid-http-response
5372no option accept-invalid-http-response
5373 Enable or disable relaxing of HTTP response parsing
5374 May be used in sections : defaults | frontend | listen | backend
5375 yes | no | yes | yes
5376 Arguments : none
5377
Willy Tarreau91852eb2015-05-01 13:26:00 +02005378 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005379 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005380 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005381 forbidden characters are essentially used to build attacks exploiting server
5382 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5383 server will emit invalid header names for whatever reason (configuration,
5384 implementation) and the issue will not be immediately fixed. In such a case,
5385 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005386 even if that does not make sense, by specifying this option. This option also
5387 relaxes the test on the HTTP version format, it allows multiple digits for
5388 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005389
5390 This option should never be enabled by default as it hides application bugs
5391 and open security breaches. It should only be deployed after a problem has
5392 been confirmed.
5393
5394 When this option is enabled, erroneous header names will still be accepted in
5395 responses, but the complete response will be captured in order to permit
5396 later analysis using the "show errors" request on the UNIX stats socket.
5397 Doing this also helps confirming that the issue has been solved.
5398
5399 If this option has been enabled in a "defaults" section, it can be disabled
5400 in a specific instance by prepending the "no" keyword before it.
5401
5402 See also : "option accept-invalid-http-request" and "show errors" on the
5403 stats socket.
5404
5405
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005406option allbackups
5407no option allbackups
5408 Use either all backup servers at a time or only the first one
5409 May be used in sections : defaults | frontend | listen | backend
5410 yes | no | yes | yes
5411 Arguments : none
5412
5413 By default, the first operational backup server gets all traffic when normal
5414 servers are all down. Sometimes, it may be preferred to use multiple backups
5415 at once, because one will not be enough. When "option allbackups" is enabled,
5416 the load balancing will be performed among all backup servers when all normal
5417 ones are unavailable. The same load balancing algorithm will be used and the
5418 servers' weights will be respected. Thus, there will not be any priority
5419 order between the backup servers anymore.
5420
5421 This option is mostly used with static server farms dedicated to return a
5422 "sorry" page when an application is completely offline.
5423
5424 If this option has been enabled in a "defaults" section, it can be disabled
5425 in a specific instance by prepending the "no" keyword before it.
5426
5427
5428option checkcache
5429no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005430 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005431 May be used in sections : defaults | frontend | listen | backend
5432 yes | no | yes | yes
5433 Arguments : none
5434
5435 Some high-level frameworks set application cookies everywhere and do not
5436 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005437 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005438 high risk of session crossing or stealing between users traversing the same
5439 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005440 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005441
5442 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005443 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005444 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005445 response to check if there's a risk of caching a cookie on a client-side
5446 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005447 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005448 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005449 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5450 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005451 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005452 - all those that result from a request using a method other than GET, HEAD,
5453 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005454 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005455 - those with a 'Pragma: no-cache' header
5456 - those with a 'Cache-control: private' header
5457 - those with a 'Cache-control: no-store' header
5458 - those with a 'Cache-control: max-age=0' header
5459 - those with a 'Cache-control: s-maxage=0' header
5460 - those with a 'Cache-control: no-cache' header
5461 - those with a 'Cache-control: no-cache="set-cookie"' header
5462 - those with a 'Cache-control: no-cache="set-cookie,' header
5463 (allowing other fields after set-cookie)
5464
5465 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005466 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005467 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005468 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005469 that admins are informed that there's something to be fixed.
5470
5471 Due to the high impact on the application, the application should be tested
5472 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005473 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005474 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005475
5476 If this option has been enabled in a "defaults" section, it can be disabled
5477 in a specific instance by prepending the "no" keyword before it.
5478
5479
5480option clitcpka
5481no option clitcpka
5482 Enable or disable the sending of TCP keepalive packets on the client side
5483 May be used in sections : defaults | frontend | listen | backend
5484 yes | yes | yes | no
5485 Arguments : none
5486
5487 When there is a firewall or any session-aware component between a client and
5488 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005489 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005490 components decides to expire a session which has remained idle for too long.
5491
5492 Enabling socket-level TCP keep-alives makes the system regularly send packets
5493 to the other end of the connection, leaving it active. The delay between
5494 keep-alive probes is controlled by the system only and depends both on the
5495 operating system and its tuning parameters.
5496
5497 It is important to understand that keep-alive packets are neither emitted nor
5498 received at the application level. It is only the network stacks which sees
5499 them. For this reason, even if one side of the proxy already uses keep-alives
5500 to maintain its connection alive, those keep-alive packets will not be
5501 forwarded to the other side of the proxy.
5502
5503 Please note that this has nothing to do with HTTP keep-alive.
5504
5505 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5506 client side of a connection, which should help when session expirations are
5507 noticed between HAProxy and a client.
5508
5509 If this option has been enabled in a "defaults" section, it can be disabled
5510 in a specific instance by prepending the "no" keyword before it.
5511
5512 See also : "option srvtcpka", "option tcpka"
5513
5514
Willy Tarreau0ba27502007-12-24 16:55:16 +01005515option contstats
5516 Enable continuous traffic statistics updates
5517 May be used in sections : defaults | frontend | listen | backend
5518 yes | yes | yes | no
5519 Arguments : none
5520
5521 By default, counters used for statistics calculation are incremented
5522 only when a session finishes. It works quite well when serving small
5523 objects, but with big ones (for example large images or archives) or
5524 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005525 a hedgehog. With this option enabled counters get incremented frequently
5526 along the session, typically every 5 seconds, which is often enough to
5527 produce clean graphs. Recounting touches a hotpath directly so it is not
5528 not enabled by default, as it can cause a lot of wakeups for very large
5529 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005530
5531
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005532option dontlog-normal
5533no option dontlog-normal
5534 Enable or disable logging of normal, successful connections
5535 May be used in sections : defaults | frontend | listen | backend
5536 yes | yes | yes | no
5537 Arguments : none
5538
5539 There are large sites dealing with several thousand connections per second
5540 and for which logging is a major pain. Some of them are even forced to turn
5541 logs off and cannot debug production issues. Setting this option ensures that
5542 normal connections, those which experience no error, no timeout, no retry nor
5543 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5544 mode, the response status code is checked and return codes 5xx will still be
5545 logged.
5546
5547 It is strongly discouraged to use this option as most of the time, the key to
5548 complex issues is in the normal logs which will not be logged here. If you
5549 need to separate logs, see the "log-separate-errors" option instead.
5550
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005551 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005552 logging.
5553
5554
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005555option dontlognull
5556no option dontlognull
5557 Enable or disable logging of null connections
5558 May be used in sections : defaults | frontend | listen | backend
5559 yes | yes | yes | no
5560 Arguments : none
5561
5562 In certain environments, there are components which will regularly connect to
5563 various systems to ensure that they are still alive. It can be the case from
5564 another load balancer as well as from monitoring systems. By default, even a
5565 simple port probe or scan will produce a log. If those connections pollute
5566 the logs too much, it is possible to enable option "dontlognull" to indicate
5567 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005568 which typically corresponds to those probes. Note that errors will still be
5569 returned to the client and accounted for in the stats. If this is not what is
5570 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005571
5572 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005573 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005574 would not be logged.
5575
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
Willy Tarreau0f228a02015-05-01 15:37:53 +02005579 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5580 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005581
5582
5583option forceclose
5584no option forceclose
5585 Enable or disable active connection closing after response is transferred.
5586 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaua31e5df2009-12-30 01:10:35 +01005587 yes | yes | yes | yes
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005588 Arguments : none
5589
5590 Some HTTP servers do not necessarily close the connections when they receive
5591 the "Connection: close" set by "option httpclose", and if the client does not
5592 close either, then the connection remains open till the timeout expires. This
5593 causes high number of simultaneous connections on the servers and shows high
5594 global session times in the logs.
5595
5596 When this happens, it is possible to use "option forceclose". It will
Willy Tarreau82eeaf22009-12-29 12:09:05 +01005597 actively close the outgoing server channel as soon as the server has finished
Cyril Bonté653dcd62014-02-20 00:13:15 +01005598 to respond and release some resources earlier than with "option httpclose".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005599
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005600 This option may also be combined with "option http-pretend-keepalive", which
5601 will disable sending of the "Connection: close" header, but will still cause
5602 the connection to be closed once the whole response is received.
5603
Cyril Bonté653dcd62014-02-20 00:13:15 +01005604 This option disables and replaces any previous "option httpclose", "option
5605 http-server-close", "option http-keep-alive", or "option http-tunnel".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005606
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005607 If this option has been enabled in a "defaults" section, it can be disabled
5608 in a specific instance by prepending the "no" keyword before it.
5609
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005610 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005611
5612
Willy Tarreau87cf5142011-08-19 22:57:24 +02005613option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005614 Enable insertion of the X-Forwarded-For header to requests sent to servers
5615 May be used in sections : defaults | frontend | listen | backend
5616 yes | yes | yes | yes
5617 Arguments :
5618 <network> is an optional argument used to disable this option for sources
5619 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005620 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005621 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005622
5623 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5624 their client address. This is sometimes annoying when the client's IP address
5625 is expected in server logs. To solve this problem, the well-known HTTP header
5626 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5627 This header contains a value representing the client's IP address. Since this
5628 header is always appended at the end of the existing header list, the server
5629 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005630 the server's manual to find how to enable use of this standard header. Note
5631 that only the last occurrence of the header must be used, since it is really
5632 possible that the client has already brought one.
5633
Willy Tarreaud72758d2010-01-12 10:42:19 +01005634 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005635 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005636 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005637 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005638 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005639 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005640
5641 Sometimes, a same HAProxy instance may be shared between a direct client
5642 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5643 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5644 header for a known source address or network by adding the "except" keyword
5645 followed by the network address. In this case, any source IP matching the
5646 network will not cause an addition of this header. Most common uses are with
5647 private networks or 127.0.0.1.
5648
Willy Tarreau87cf5142011-08-19 22:57:24 +02005649 Alternatively, the keyword "if-none" states that the header will only be
5650 added if it is not present. This should only be used in perfectly trusted
5651 environment, as this might cause a security issue if headers reaching haproxy
5652 are under the control of the end-user.
5653
Willy Tarreauc27debf2008-01-06 08:57:02 +01005654 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005655 least one of them uses it, the header will be added. Note that the backend's
5656 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005657 both are defined. In the case of the "if-none" argument, if at least one of
5658 the frontend or the backend does not specify it, it wants the addition to be
5659 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005660
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005661 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005662 # Public HTTP address also used by stunnel on the same machine
5663 frontend www
5664 mode http
5665 option forwardfor except 127.0.0.1 # stunnel already adds the header
5666
Ross Westaf72a1d2008-08-03 10:51:45 +02005667 # Those servers want the IP Address in X-Client
5668 backend www
5669 mode http
5670 option forwardfor header X-Client
5671
Willy Tarreau87cf5142011-08-19 22:57:24 +02005672 See also : "option httpclose", "option http-server-close",
Willy Tarreau70dffda2014-01-30 03:07:23 +01005673 "option forceclose", "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005674
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005675
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005676option http-buffer-request
5677no option http-buffer-request
5678 Enable or disable waiting for whole HTTP request body before proceeding
5679 May be used in sections : defaults | frontend | listen | backend
5680 yes | yes | yes | yes
5681 Arguments : none
5682
5683 It is sometimes desirable to wait for the body of an HTTP request before
5684 taking a decision. This is what is being done by "balance url_param" for
5685 example. The first use case is to buffer requests from slow clients before
5686 connecting to the server. Another use case consists in taking the routing
5687 decision based on the request body's contents. This option placed in a
5688 frontend or backend forces the HTTP processing to wait until either the whole
5689 body is received, or the request buffer is full, or the first chunk is
5690 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005691 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005692 the frontend and the backend, so this should definitely not be used by
5693 default.
5694
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005695 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005696
5697
Willy Tarreau0f228a02015-05-01 15:37:53 +02005698option http-ignore-probes
5699no option http-ignore-probes
5700 Enable or disable logging of null connections and request timeouts
5701 May be used in sections : defaults | frontend | listen | backend
5702 yes | yes | yes | no
5703 Arguments : none
5704
5705 Recently some browsers started to implement a "pre-connect" feature
5706 consisting in speculatively connecting to some recently visited web sites
5707 just in case the user would like to visit them. This results in many
5708 connections being established to web sites, which end up in 408 Request
5709 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5710 decides to close them first. These ones pollute the log and feed the error
5711 counters. There was already "option dontlognull" but it's insufficient in
5712 this case. Instead, this option does the following things :
5713 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005714 was received over a connection before it was closed;
5715 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005716 - prevent any error counter from being incremented
5717
5718 That way the empty connection is silently ignored. Note that it is better
5719 not to use this unless it is clear that it is needed, because it will hide
5720 real problems. The most common reason for not receiving a request and seeing
5721 a 408 is due to an MTU inconsistency between the client and an intermediary
5722 element such as a VPN, which blocks too large packets. These issues are
5723 generally seen with POST requests as well as GET with large cookies. The logs
5724 are often the only way to detect them.
5725
5726 If this option has been enabled in a "defaults" section, it can be disabled
5727 in a specific instance by prepending the "no" keyword before it.
5728
5729 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5730
5731
Willy Tarreau16bfb022010-01-16 19:48:41 +01005732option http-keep-alive
5733no option http-keep-alive
5734 Enable or disable HTTP keep-alive from client to server
5735 May be used in sections : defaults | frontend | listen | backend
5736 yes | yes | yes | yes
5737 Arguments : none
5738
Willy Tarreau70dffda2014-01-30 03:07:23 +01005739 By default HAProxy operates in keep-alive mode with regards to persistent
5740 connections: for each connection it processes each request and response, and
5741 leaves the connection idle on both sides between the end of a response and the
5742 start of a new request. This mode may be changed by several options such as
5743 "option http-server-close", "option forceclose", "option httpclose" or
5744 "option http-tunnel". This option allows to set back the keep-alive mode,
5745 which can be useful when another mode was used in a defaults section.
5746
5747 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5748 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005749 network) and the fastest session reuse on the server side at the expense
5750 of maintaining idle connections to the servers. In general, it is possible
5751 with this option to achieve approximately twice the request rate that the
5752 "http-server-close" option achieves on small objects. There are mainly two
5753 situations where this option may be useful :
5754
5755 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005756 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005757
5758 - when the cost of establishing the connection to the server is significant
5759 compared to the cost of retrieving the associated object from the server.
5760
5761 This last case can happen when the server is a fast static server of cache.
5762 In this case, the server will need to be properly tuned to support high enough
5763 connection counts because connections will last until the client sends another
5764 request.
5765
5766 If the client request has to go to another backend or another server due to
5767 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01005768 immediately be closed and a new one re-opened. Option "prefer-last-server" is
5769 available to try optimize server selection so that if the server currently
5770 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01005771
Willy Tarreau16bfb022010-01-16 19:48:41 +01005772 At the moment, logs will not indicate whether requests came from the same
5773 session or not. The accept date reported in the logs corresponds to the end
5774 of the previous request, and the request time corresponds to the time spent
5775 waiting for a new request. The keep-alive request time is still bound to the
5776 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5777 not set.
5778
Cyril Bonté653dcd62014-02-20 00:13:15 +01005779 This option disables and replaces any previous "option httpclose", "option
5780 http-server-close", "option forceclose" or "option http-tunnel". When backend
Willy Tarreau70dffda2014-01-30 03:07:23 +01005781 and frontend options differ, all of these 4 options have precedence over
Cyril Bonté653dcd62014-02-20 00:13:15 +01005782 "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005783
5784 See also : "option forceclose", "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01005785 "option prefer-last-server", "option http-pretend-keepalive",
5786 "option httpclose", and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01005787
5788
Willy Tarreau96e31212011-05-30 18:10:30 +02005789option http-no-delay
5790no option http-no-delay
5791 Instruct the system to favor low interactive delays over performance in HTTP
5792 May be used in sections : defaults | frontend | listen | backend
5793 yes | yes | yes | yes
5794 Arguments : none
5795
5796 In HTTP, each payload is unidirectional and has no notion of interactivity.
5797 Any agent is expected to queue data somewhat for a reasonably low delay.
5798 There are some very rare server-to-server applications that abuse the HTTP
5799 protocol and expect the payload phase to be highly interactive, with many
5800 interleaved data chunks in both directions within a single request. This is
5801 absolutely not supported by the HTTP specification and will not work across
5802 most proxies or servers. When such applications attempt to do this through
5803 haproxy, it works but they will experience high delays due to the network
5804 optimizations which favor performance by instructing the system to wait for
5805 enough data to be available in order to only send full packets. Typical
5806 delays are around 200 ms per round trip. Note that this only happens with
5807 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
5808 affected.
5809
5810 When "option http-no-delay" is present in either the frontend or the backend
5811 used by a connection, all such optimizations will be disabled in order to
5812 make the exchanges as fast as possible. Of course this offers no guarantee on
5813 the functionality, as it may break at any other place. But if it works via
5814 HAProxy, it will work as fast as possible. This option should never be used
5815 by default, and should never be used at all unless such a buggy application
5816 is discovered. The impact of using this option is an increase of bandwidth
5817 usage and CPU usage, which may significantly lower performance in high
5818 latency environments.
5819
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005820 See also : "option http-buffer-request"
5821
Willy Tarreau96e31212011-05-30 18:10:30 +02005822
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005823option http-pretend-keepalive
5824no option http-pretend-keepalive
5825 Define whether haproxy will announce keepalive to the server or not
5826 May be used in sections : defaults | frontend | listen | backend
5827 yes | yes | yes | yes
5828 Arguments : none
5829
5830 When running with "option http-server-close" or "option forceclose", haproxy
5831 adds a "Connection: close" header to the request forwarded to the server.
5832 Unfortunately, when some servers see this header, they automatically refrain
5833 from using the chunked encoding for responses of unknown length, while this
5834 is totally unrelated. The immediate effect is that this prevents haproxy from
5835 maintaining the client connection alive. A second effect is that a client or
5836 a cache could receive an incomplete response without being aware of it, and
5837 consider the response complete.
5838
5839 By setting "option http-pretend-keepalive", haproxy will make the server
5840 believe it will keep the connection alive. The server will then not fall back
5841 to the abnormal undesired above. When haproxy gets the whole response, it
5842 will close the connection with the server just as it would do with the
5843 "forceclose" option. That way the client gets a normal response and the
5844 connection is correctly closed on the server side.
5845
5846 It is recommended not to enable this option by default, because most servers
5847 will more efficiently close the connection themselves after the last packet,
5848 and release its buffers slightly earlier. Also, the added packet on the
5849 network could slightly reduce the overall peak performance. However it is
5850 worth noting that when this option is enabled, haproxy will have slightly
5851 less work to do. So if haproxy is the bottleneck on the whole architecture,
5852 enabling this option might save a few CPU cycles.
5853
5854 This option may be set both in a frontend and in a backend. It is enabled if
5855 at least one of the frontend or backend holding a connection has it enabled.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005856 This option may be combined with "option httpclose", which will cause
Willy Tarreau22a95342010-09-29 14:31:41 +02005857 keepalive to be announced to the server and close to be announced to the
5858 client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005859
5860 If this option has been enabled in a "defaults" section, it can be disabled
5861 in a specific instance by prepending the "no" keyword before it.
5862
Willy Tarreau16bfb022010-01-16 19:48:41 +01005863 See also : "option forceclose", "option http-server-close", and
5864 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005865
Willy Tarreauc27debf2008-01-06 08:57:02 +01005866
Willy Tarreaub608feb2010-01-02 22:47:18 +01005867option http-server-close
5868no option http-server-close
5869 Enable or disable HTTP connection closing on the server side
5870 May be used in sections : defaults | frontend | listen | backend
5871 yes | yes | yes | yes
5872 Arguments : none
5873
Willy Tarreau70dffda2014-01-30 03:07:23 +01005874 By default HAProxy operates in keep-alive mode with regards to persistent
5875 connections: for each connection it processes each request and response, and
5876 leaves the connection idle on both sides between the end of a response and
5877 the start of a new request. This mode may be changed by several options such
5878 as "option http-server-close", "option forceclose", "option httpclose" or
5879 "option http-tunnel". Setting "option http-server-close" enables HTTP
5880 connection-close mode on the server side while keeping the ability to support
Davor Ocelice9ed2812017-12-25 17:49:28 +01005881 HTTP keep-alive and pipelining on the client side. This provides the lowest
Willy Tarreau70dffda2014-01-30 03:07:23 +01005882 latency on the client side (slow network) and the fastest session reuse on
5883 the server side to save server resources, similarly to "option forceclose".
5884 It also permits non-keepalive capable servers to be served in keep-alive mode
Lukas Tribus23953682017-04-28 13:24:30 +00005885 to the clients if they conform to the requirements of RFC7230. Please note
Willy Tarreau70dffda2014-01-30 03:07:23 +01005886 that some servers do not always conform to those requirements when they see
5887 "Connection: close" in the request. The effect will be that keep-alive will
5888 never be used. A workaround consists in enabling "option
5889 http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005890
5891 At the moment, logs will not indicate whether requests came from the same
5892 session or not. The accept date reported in the logs corresponds to the end
5893 of the previous request, and the request time corresponds to the time spent
5894 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01005895 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
5896 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005897
5898 This option may be set both in a frontend and in a backend. It is enabled if
5899 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01005900 It disables and replaces any previous "option httpclose", "option forceclose",
5901 "option http-tunnel" or "option http-keep-alive". Please check section 4
Willy Tarreau70dffda2014-01-30 03:07:23 +01005902 ("Proxies") to see how this option combines with others when frontend and
5903 backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01005904
5905 If this option has been enabled in a "defaults" section, it can be disabled
5906 in a specific instance by prepending the "no" keyword before it.
5907
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02005908 See also : "option forceclose", "option http-pretend-keepalive",
Willy Tarreau16bfb022010-01-16 19:48:41 +01005909 "option httpclose", "option http-keep-alive", and
5910 "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01005911
5912
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005913option http-tunnel
5914no option http-tunnel
5915 Disable or enable HTTP connection processing after first transaction
5916 May be used in sections : defaults | frontend | listen | backend
5917 yes | yes | yes | yes
5918 Arguments : none
5919
Willy Tarreau70dffda2014-01-30 03:07:23 +01005920 By default HAProxy operates in keep-alive mode with regards to persistent
5921 connections: for each connection it processes each request and response, and
5922 leaves the connection idle on both sides between the end of a response and
5923 the start of a new request. This mode may be changed by several options such
5924 as "option http-server-close", "option forceclose", "option httpclose" or
5925 "option http-tunnel".
5926
5927 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005928 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01005929 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
5930 is normally not needed anymore unless in very specific cases such as when
5931 using an in-house protocol that looks like HTTP but is not compatible, or
5932 just to log one request per client in order to reduce log size. Note that
5933 everything which works at the HTTP level, including header parsing/addition,
5934 cookie processing or content switching will only work for the first request
5935 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01005936
5937 If this option has been enabled in a "defaults" section, it can be disabled
5938 in a specific instance by prepending the "no" keyword before it.
5939
5940 See also : "option forceclose", "option http-server-close",
5941 "option httpclose", "option http-keep-alive", and
5942 "1.1. The HTTP transaction model".
5943
5944
Willy Tarreau88d349d2010-01-25 12:15:43 +01005945option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01005946no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01005947 Make use of non-standard Proxy-Connection header instead of Connection
5948 May be used in sections : defaults | frontend | listen | backend
5949 yes | yes | yes | no
5950 Arguments : none
5951
Lukas Tribus23953682017-04-28 13:24:30 +00005952 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01005953 Connection header to indicate their wish of persistent or non-persistent
5954 connections, both browsers and proxies ignore this header for proxied
5955 connections and make use of the undocumented, non-standard Proxy-Connection
5956 header instead. The issue begins when trying to put a load balancer between
5957 browsers and such proxies, because there will be a difference between what
5958 haproxy understands and what the client and the proxy agree on.
5959
5960 By setting this option in a frontend, haproxy can automatically switch to use
5961 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01005962 defined here as one where the URI begins with neither a '/' nor a '*'. This
5963 is incompatible with the HTTP tunnel mode. Note that this option can only be
5964 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01005965
Willy Tarreau844a7e72010-01-31 21:46:18 +01005966 Also, when this option is set, a request which requires authentication will
5967 automatically switch to use proxy authentication headers if it is itself a
5968 proxied request. That makes it possible to check or enforce authentication in
5969 front of an existing proxy.
5970
Willy Tarreau88d349d2010-01-25 12:15:43 +01005971 This option should normally never be used, except in front of a proxy.
5972
5973 See also : "option httpclose", "option forceclose" and "option
5974 http-server-close".
5975
5976
Willy Tarreaud63335a2010-02-26 12:56:52 +01005977option httpchk
5978option httpchk <uri>
5979option httpchk <method> <uri>
5980option httpchk <method> <uri> <version>
5981 Enable HTTP protocol to check on the servers health
5982 May be used in sections : defaults | frontend | listen | backend
5983 yes | no | yes | yes
5984 Arguments :
5985 <method> is the optional HTTP method used with the requests. When not set,
5986 the "OPTIONS" method is used, as it generally requires low server
5987 processing and is easy to filter out from the logs. Any method
5988 may be used, though it is not recommended to invent non-standard
5989 ones.
5990
5991 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
5992 which is accessible by default on almost any server, but may be
5993 changed to any other URI. Query strings are permitted.
5994
5995 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
5996 but some servers might behave incorrectly in HTTP 1.0, so turning
5997 it to HTTP/1.1 may sometimes help. Note that the Host field is
5998 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
5999 after "\r\n" following the version string.
6000
6001 By default, server health checks only consist in trying to establish a TCP
6002 connection. When "option httpchk" is specified, a complete HTTP request is
6003 sent once the TCP connection is established, and responses 2xx and 3xx are
6004 considered valid, while all other ones indicate a server failure, including
6005 the lack of any response.
6006
6007 The port and interval are specified in the server configuration.
6008
6009 This option does not necessarily require an HTTP backend, it also works with
6010 plain TCP backends. This is particularly useful to check simple scripts bound
6011 to some dedicated ports using the inetd daemon.
6012
6013 Examples :
6014 # Relay HTTPS traffic to Apache instance and check service availability
6015 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6016 backend https_relay
6017 mode tcp
6018 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6019 server apache1 192.168.1.1:443 check port 80
6020
Simon Hormanafc47ee2013-11-25 10:46:35 +09006021 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6022 "option pgsql-check", "http-check" and the "check", "port" and
6023 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006024
6025
Willy Tarreauc27debf2008-01-06 08:57:02 +01006026option httpclose
6027no option httpclose
6028 Enable or disable passive HTTP connection closing
6029 May be used in sections : defaults | frontend | listen | backend
6030 yes | yes | yes | yes
6031 Arguments : none
6032
Willy Tarreau70dffda2014-01-30 03:07:23 +01006033 By default HAProxy operates in keep-alive mode with regards to persistent
6034 connections: for each connection it processes each request and response, and
6035 leaves the connection idle on both sides between the end of a response and
6036 the start of a new request. This mode may be changed by several options such
Cyril Bonté653dcd62014-02-20 00:13:15 +01006037 as "option http-server-close", "option forceclose", "option httpclose" or
Willy Tarreau70dffda2014-01-30 03:07:23 +01006038 "option http-tunnel".
6039
6040 If "option httpclose" is set, HAProxy will work in HTTP tunnel mode and check
6041 if a "Connection: close" header is already set in each direction, and will
6042 add one if missing. Each end should react to this by actively closing the TCP
6043 connection after each transfer, thus resulting in a switch to the HTTP close
6044 mode. Any "Connection" header different from "close" will also be removed.
6045 Note that this option is deprecated since what it does is very cheap but not
6046 reliable. Using "option http-server-close" or "option forceclose" is strongly
6047 recommended instead.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006048
6049 It seldom happens that some servers incorrectly ignore this header and do not
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006050 close the connection even though they reply "Connection: close". For this
Willy Tarreau0dfdf192010-01-05 11:33:11 +01006051 reason, they are not compatible with older HTTP 1.0 browsers. If this happens
6052 it is possible to use the "option forceclose" which actively closes the
6053 request connection once the server responds. Option "forceclose" also
6054 releases the server connection earlier because it does not have to wait for
6055 the client to acknowledge it.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006056
6057 This option may be set both in a frontend and in a backend. It is enabled if
6058 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006059 It disables and replaces any previous "option http-server-close",
6060 "option forceclose", "option http-keep-alive" or "option http-tunnel". Please
Willy Tarreau70dffda2014-01-30 03:07:23 +01006061 check section 4 ("Proxies") to see how this option combines with others when
6062 frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006063
6064 If this option has been enabled in a "defaults" section, it can be disabled
6065 in a specific instance by prepending the "no" keyword before it.
6066
Patrick Mezard9ec2ec42010-06-12 17:02:45 +02006067 See also : "option forceclose", "option http-server-close" and
6068 "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006069
6070
Emeric Brun3a058f32009-06-30 18:26:00 +02006071option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006072 Enable logging of HTTP request, session state and timers
6073 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006074 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006075 Arguments :
6076 clf if the "clf" argument is added, then the output format will be
6077 the CLF format instead of HAProxy's default HTTP format. You can
6078 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006079 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006080 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006081
6082 By default, the log output format is very poor, as it only contains the
6083 source and destination addresses, and the instance name. By specifying
6084 "option httplog", each log line turns into a much richer format including,
6085 but not limited to, the HTTP request, the connection timers, the session
6086 status, the connections numbers, the captured headers and cookies, the
6087 frontend, backend and server name, and of course the source address and
6088 ports.
6089
PiBa-NLbd556bf2014-12-11 21:31:54 +01006090 Specifying only "option httplog" will automatically clear the 'clf' mode
6091 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006092
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006093 "option httplog" overrides any previous "log-format" directive.
6094
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006095 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006096
Willy Tarreau55165fe2009-05-10 12:02:55 +02006097
6098option http_proxy
6099no option http_proxy
6100 Enable or disable plain HTTP proxy mode
6101 May be used in sections : defaults | frontend | listen | backend
6102 yes | yes | yes | yes
6103 Arguments : none
6104
6105 It sometimes happens that people need a pure HTTP proxy which understands
6106 basic proxy requests without caching nor any fancy feature. In this case,
6107 it may be worth setting up an HAProxy instance with the "option http_proxy"
6108 set. In this mode, no server is declared, and the connection is forwarded to
6109 the IP address and port found in the URL after the "http://" scheme.
6110
6111 No host address resolution is performed, so this only works when pure IP
6112 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006113 it will probably be used only by experts who know they need exactly it. This
6114 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006115
6116 If this option has been enabled in a "defaults" section, it can be disabled
6117 in a specific instance by prepending the "no" keyword before it.
6118
6119 Example :
6120 # this backend understands HTTP proxy requests and forwards them directly.
6121 backend direct_forward
6122 option httpclose
6123 option http_proxy
6124
6125 See also : "option httpclose"
6126
Willy Tarreau211ad242009-10-03 21:45:07 +02006127
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006128option independent-streams
6129no option independent-streams
6130 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006131 May be used in sections : defaults | frontend | listen | backend
6132 yes | yes | yes | yes
6133 Arguments : none
6134
6135 By default, when data is sent over a socket, both the write timeout and the
6136 read timeout for that socket are refreshed, because we consider that there is
6137 activity on that socket, and we have no other means of guessing if we should
6138 receive data or not.
6139
Davor Ocelice9ed2812017-12-25 17:49:28 +01006140 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006141 exists a situation where it is desirable to disable it, and only refresh the
6142 read timeout if there are incoming data. This happens on sessions with large
6143 timeouts and low amounts of exchanged data such as telnet session. If the
6144 server suddenly disappears, the output data accumulates in the system's
6145 socket buffers, both timeouts are correctly refreshed, and there is no way
6146 to know the server does not receive them, so we don't timeout. However, when
6147 the underlying protocol always echoes sent data, it would be enough by itself
6148 to detect the issue using the read timeout. Note that this problem does not
6149 happen with more verbose protocols because data won't accumulate long in the
6150 socket buffers.
6151
6152 When this option is set on the frontend, it will disable read timeout updates
6153 on data sent to the client. There probably is little use of this case. When
6154 the option is set on the backend, it will disable read timeout updates on
6155 data sent to the server. Doing so will typically break large HTTP posts from
6156 slow lines, so use it with caution.
6157
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006158 Note: older versions used to call this setting "option independent-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006159 with a spelling mistake. This spelling is still supported but
6160 deprecated.
6161
Willy Tarreauce887fd2012-05-12 12:50:00 +02006162 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006163
6164
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006165option ldap-check
6166 Use LDAPv3 health checks for server testing
6167 May be used in sections : defaults | frontend | listen | backend
6168 yes | no | yes | yes
6169 Arguments : none
6170
6171 It is possible to test that the server correctly talks LDAPv3 instead of just
6172 testing that it accepts the TCP connection. When this option is set, an
6173 LDAPv3 anonymous simple bind message is sent to the server, and the response
6174 is analyzed to find an LDAPv3 bind response message.
6175
6176 The server is considered valid only when the LDAP response contains success
6177 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6178
6179 Logging of bind requests is server dependent see your documentation how to
6180 configure it.
6181
6182 Example :
6183 option ldap-check
6184
6185 See also : "option httpchk"
6186
6187
Simon Horman98637e52014-06-20 12:30:16 +09006188option external-check
6189 Use external processes for server health checks
6190 May be used in sections : defaults | frontend | listen | backend
6191 yes | no | yes | yes
6192
6193 It is possible to test the health of a server using an external command.
6194 This is achieved by running the executable set using "external-check
6195 command".
6196
6197 Requires the "external-check" global to be set.
6198
6199 See also : "external-check", "external-check command", "external-check path"
6200
6201
Willy Tarreau211ad242009-10-03 21:45:07 +02006202option log-health-checks
6203no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006204 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006205 May be used in sections : defaults | frontend | listen | backend
6206 yes | no | yes | yes
6207 Arguments : none
6208
Willy Tarreaubef1b322014-05-13 21:01:39 +02006209 By default, failed health check are logged if server is UP and successful
6210 health checks are logged if server is DOWN, so the amount of additional
6211 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006212
Willy Tarreaubef1b322014-05-13 21:01:39 +02006213 When this option is enabled, any change of the health check status or to
6214 the server's health will be logged, so that it becomes possible to know
6215 that a server was failing occasional checks before crashing, or exactly when
6216 it failed to respond a valid HTTP status, then when the port started to
6217 reject connections, then when the server stopped responding at all.
6218
Davor Ocelice9ed2812017-12-25 17:49:28 +01006219 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006220 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006221
Willy Tarreaubef1b322014-05-13 21:01:39 +02006222 See also: "option httpchk", "option ldap-check", "option mysql-check",
6223 "option pgsql-check", "option redis-check", "option smtpchk",
6224 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006225
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006226
6227option log-separate-errors
6228no option log-separate-errors
6229 Change log level for non-completely successful connections
6230 May be used in sections : defaults | frontend | listen | backend
6231 yes | yes | yes | no
6232 Arguments : none
6233
6234 Sometimes looking for errors in logs is not easy. This option makes haproxy
6235 raise the level of logs containing potentially interesting information such
6236 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6237 level changes from "info" to "err". This makes it possible to log them
6238 separately to a different file with most syslog daemons. Be careful not to
6239 remove them from the original file, otherwise you would lose ordering which
6240 provides very important information.
6241
6242 Using this option, large sites dealing with several thousand connections per
6243 second may log normal traffic to a rotating buffer and only archive smaller
6244 error logs.
6245
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006246 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006247 logging.
6248
Willy Tarreauc27debf2008-01-06 08:57:02 +01006249
6250option logasap
6251no option logasap
6252 Enable or disable early logging of HTTP requests
6253 May be used in sections : defaults | frontend | listen | backend
6254 yes | yes | yes | no
6255 Arguments : none
6256
6257 By default, HTTP requests are logged upon termination so that the total
6258 transfer time and the number of bytes appear in the logs. When large objects
6259 are being transferred, it may take a while before the request appears in the
6260 logs. Using "option logasap", the request gets logged as soon as the server
6261 sends the complete headers. The only missing information in the logs will be
6262 the total number of bytes which will indicate everything except the amount
6263 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006264 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006265 "Content-Length" response header so that the logs at least indicate how many
6266 bytes are expected to be transferred.
6267
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006268 Examples :
6269 listen http_proxy 0.0.0.0:80
6270 mode http
6271 option httplog
6272 option logasap
6273 log 192.168.2.200 local3
6274
6275 >>> Feb 6 12:14:14 localhost \
6276 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6277 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6278 "GET /image.iso HTTP/1.0"
6279
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006280 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006281 logging.
6282
6283
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006284option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006285 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006286 May be used in sections : defaults | frontend | listen | backend
6287 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006288 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006289 <username> This is the username which will be used when connecting to MySQL
6290 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006291 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006292
6293 If you specify a username, the check consists of sending two MySQL packet,
6294 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006295 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006296 Error packet. It is a basic but useful test which does not produce error nor
6297 aborted connect on the server. However, it requires adding an authorization
6298 in the MySQL table, like this :
6299
6300 USE mysql;
6301 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6302 FLUSH PRIVILEGES;
6303
6304 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006305 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006306 Error packet, we don't send anything in this mode. It was reported that it
6307 can generate lockout if check is too frequent and/or if there is not enough
6308 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6309 value as if a connection is established successfully within fewer than MySQL
6310 "max_connect_errors" attempts after a previous connection was interrupted,
6311 the error count for the host is cleared to zero. If HAProxy's server get
6312 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6313
6314 Remember that this does not check database presence nor database consistency.
6315 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006316
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006317 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006318
6319 Most often, an incoming MySQL server needs to see the client's IP address for
6320 various purposes, including IP privilege matching and connection logging.
6321 When possible, it is often wise to masquerade the client's IP address when
6322 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006323 which requires the transparent proxy feature to be compiled in, and the MySQL
6324 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006325
6326 See also: "option httpchk"
6327
6328
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006329option nolinger
6330no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006331 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006332 May be used in sections: defaults | frontend | listen | backend
6333 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006334 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006335
Davor Ocelice9ed2812017-12-25 17:49:28 +01006336 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006337 physically disconnected), the session timeouts triggers and the session is
6338 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6339 using some resources and possibly limiting the ability to establish newer
6340 connections.
6341
6342 When this happens, it is possible to activate "option nolinger" which forces
6343 the system to immediately remove any socket's pending data on close. Thus,
6344 the session is instantly purged from the system's tables. This usually has
6345 side effects such as increased number of TCP resets due to old retransmits
6346 getting immediately rejected. Some firewalls may sometimes complain about
6347 this too.
6348
6349 For this reason, it is not recommended to use this option when not absolutely
6350 needed. You know that you need it when you have thousands of FIN_WAIT1
6351 sessions on your system (TIME_WAIT ones do not count).
6352
6353 This option may be used both on frontends and backends, depending on the side
6354 where it is required. Use it on the frontend for clients, and on the backend
6355 for servers.
6356
6357 If this option has been enabled in a "defaults" section, it can be disabled
6358 in a specific instance by prepending the "no" keyword before it.
6359
6360
Willy Tarreau55165fe2009-05-10 12:02:55 +02006361option originalto [ except <network> ] [ header <name> ]
6362 Enable insertion of the X-Original-To header to requests sent to servers
6363 May be used in sections : defaults | frontend | listen | backend
6364 yes | yes | yes | yes
6365 Arguments :
6366 <network> is an optional argument used to disable this option for sources
6367 matching <network>
6368 <name> an optional argument to specify a different "X-Original-To"
6369 header name.
6370
6371 Since HAProxy can work in transparent mode, every request from a client can
6372 be redirected to the proxy and HAProxy itself can proxy every request to a
6373 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6374 be lost. This is annoying when you want access rules based on destination ip
6375 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6376 added by HAProxy to all requests sent to the server. This header contains a
6377 value representing the original destination IP address. Since this must be
6378 configured to always use the last occurrence of this header only. Note that
6379 only the last occurrence of the header must be used, since it is really
6380 possible that the client has already brought one.
6381
6382 The keyword "header" may be used to supply a different header name to replace
6383 the default "X-Original-To". This can be useful where you might already
6384 have a "X-Original-To" header from a different application, and you need
6385 preserve it. Also if your backend server doesn't use the "X-Original-To"
6386 header and requires different one.
6387
6388 Sometimes, a same HAProxy instance may be shared between a direct client
6389 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6390 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6391 header for a known source address or network by adding the "except" keyword
6392 followed by the network address. In this case, any source IP matching the
6393 network will not cause an addition of this header. Most common uses are with
6394 private networks or 127.0.0.1.
6395
6396 This option may be specified either in the frontend or in the backend. If at
6397 least one of them uses it, the header will be added. Note that the backend's
6398 setting of the header subargument takes precedence over the frontend's if
6399 both are defined.
6400
Willy Tarreau55165fe2009-05-10 12:02:55 +02006401 Examples :
6402 # Original Destination address
6403 frontend www
6404 mode http
6405 option originalto except 127.0.0.1
6406
6407 # Those servers want the IP Address in X-Client-Dst
6408 backend www
6409 mode http
6410 option originalto header X-Client-Dst
6411
Willy Tarreau87cf5142011-08-19 22:57:24 +02006412 See also : "option httpclose", "option http-server-close",
6413 "option forceclose"
Willy Tarreau55165fe2009-05-10 12:02:55 +02006414
6415
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006416option persist
6417no option persist
6418 Enable or disable forced persistence on down servers
6419 May be used in sections: defaults | frontend | listen | backend
6420 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006421 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006422
6423 When an HTTP request reaches a backend with a cookie which references a dead
6424 server, by default it is redispatched to another server. It is possible to
6425 force the request to be sent to the dead server first using "option persist"
6426 if absolutely needed. A common use case is when servers are under extreme
6427 load and spend their time flapping. In this case, the users would still be
6428 directed to the server they opened the session on, in the hope they would be
6429 correctly served. It is recommended to use "option redispatch" in conjunction
6430 with this option so that in the event it would not be possible to connect to
6431 the server at all (server definitely dead), the client would finally be
6432 redirected to another valid server.
6433
6434 If this option has been enabled in a "defaults" section, it can be disabled
6435 in a specific instance by prepending the "no" keyword before it.
6436
Willy Tarreau4de91492010-01-22 19:10:05 +01006437 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006438
6439
Willy Tarreau0c122822013-12-15 18:49:01 +01006440option pgsql-check [ user <username> ]
6441 Use PostgreSQL health checks for server testing
6442 May be used in sections : defaults | frontend | listen | backend
6443 yes | no | yes | yes
6444 Arguments :
6445 <username> This is the username which will be used when connecting to
6446 PostgreSQL server.
6447
6448 The check sends a PostgreSQL StartupMessage and waits for either
6449 Authentication request or ErrorResponse message. It is a basic but useful
6450 test which does not produce error nor aborted connect on the server.
6451 This check is identical with the "mysql-check".
6452
6453 See also: "option httpchk"
6454
6455
Willy Tarreau9420b122013-12-15 18:58:25 +01006456option prefer-last-server
6457no option prefer-last-server
6458 Allow multiple load balanced requests to remain on the same server
6459 May be used in sections: defaults | frontend | listen | backend
6460 yes | no | yes | yes
6461 Arguments : none
6462
6463 When the load balancing algorithm in use is not deterministic, and a previous
6464 request was sent to a server to which haproxy still holds a connection, it is
6465 sometimes desirable that subsequent requests on a same session go to the same
6466 server as much as possible. Note that this is different from persistence, as
6467 we only indicate a preference which haproxy tries to apply without any form
6468 of warranty. The real use is for keep-alive connections sent to servers. When
6469 this option is used, haproxy will try to reuse the same connection that is
6470 attached to the server instead of rebalancing to another server, causing a
6471 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006472 not make much sense to use this in combination with hashing algorithms. Note,
6473 haproxy already automatically tries to stick to a server which sends a 401 or
6474 to a proxy which sends a 407 (authentication required). This is mandatory for
6475 use with the broken NTLM authentication challenge, and significantly helps in
6476 troubleshooting some faulty applications. Option prefer-last-server might be
6477 desirable in these environments as well, to avoid redistributing the traffic
6478 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006479
6480 If this option has been enabled in a "defaults" section, it can be disabled
6481 in a specific instance by prepending the "no" keyword before it.
6482
6483 See also: "option http-keep-alive"
6484
6485
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006486option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006487option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006488no option redispatch
6489 Enable or disable session redistribution in case of connection failure
6490 May be used in sections: defaults | frontend | listen | backend
6491 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006492 Arguments :
6493 <interval> The optional integer value that controls how often redispatches
6494 occur when retrying connections. Positive value P indicates a
6495 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006496 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006497 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006498 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006499 positive value of 1 would indicate a redispatch on every retry,
6500 and a positive value of 3 would indicate a redispatch on every
6501 third retry. You can disable redispatches with a value of 0.
6502
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006503
6504 In HTTP mode, if a server designated by a cookie is down, clients may
6505 definitely stick to it because they cannot flush the cookie, so they will not
6506 be able to access the service anymore.
6507
6508 Specifying "option redispatch" will allow the proxy to break their
6509 persistence and redistribute them to a working server.
6510
Joseph Lynch726ab712015-05-11 23:25:34 -07006511 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006512 connection failures. Of course, it requires having "retries" set to a nonzero
6513 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006514
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006515 This form is the preferred form, which replaces both the "redispatch" and
6516 "redisp" keywords.
6517
6518 If this option has been enabled in a "defaults" section, it can be disabled
6519 in a specific instance by prepending the "no" keyword before it.
6520
Willy Tarreau4de91492010-01-22 19:10:05 +01006521 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006522
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006523
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006524option redis-check
6525 Use redis health checks for server testing
6526 May be used in sections : defaults | frontend | listen | backend
6527 yes | no | yes | yes
6528 Arguments : none
6529
6530 It is possible to test that the server correctly talks REDIS protocol instead
6531 of just testing that it accepts the TCP connection. When this option is set,
6532 a PING redis command is sent to the server, and the response is analyzed to
6533 find the "+PONG" response message.
6534
6535 Example :
6536 option redis-check
6537
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006538 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006539
6540
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006541option smtpchk
6542option smtpchk <hello> <domain>
6543 Use SMTP health checks for server testing
6544 May be used in sections : defaults | frontend | listen | backend
6545 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006546 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006547 <hello> is an optional argument. It is the "hello" command to use. It can
6548 be either "HELO" (for SMTP) or "EHLO" (for ESTMP). All other
6549 values will be turned into the default command ("HELO").
6550
6551 <domain> is the domain name to present to the server. It may only be
6552 specified (and is mandatory) if the hello command has been
6553 specified. By default, "localhost" is used.
6554
6555 When "option smtpchk" is set, the health checks will consist in TCP
6556 connections followed by an SMTP command. By default, this command is
6557 "HELO localhost". The server's return code is analyzed and only return codes
6558 starting with a "2" will be considered as valid. All other responses,
6559 including a lack of response will constitute an error and will indicate a
6560 dead server.
6561
6562 This test is meant to be used with SMTP servers or relays. Depending on the
6563 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006564 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006565 25 is often easier than adjusting the configuration.
6566
6567 Most often, an incoming SMTP server needs to see the client's IP address for
6568 various purposes, including spam filtering, anti-spoofing and logging. When
6569 possible, it is often wise to masquerade the client's IP address when
6570 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006571 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006572
6573 Example :
6574 option smtpchk HELO mydomain.org
6575
6576 See also : "option httpchk", "source"
6577
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006578
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006579option socket-stats
6580no option socket-stats
6581
6582 Enable or disable collecting & providing separate statistics for each socket.
6583 May be used in sections : defaults | frontend | listen | backend
6584 yes | yes | yes | no
6585
6586 Arguments : none
6587
6588
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006589option splice-auto
6590no option splice-auto
6591 Enable or disable automatic kernel acceleration on sockets in both directions
6592 May be used in sections : defaults | frontend | listen | backend
6593 yes | yes | yes | yes
6594 Arguments : none
6595
6596 When this option is enabled either on a frontend or on a backend, haproxy
6597 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006598 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006599 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006600 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006601 are not much aggressive in order to limit excessive use of splicing. This
6602 option requires splicing to be enabled at compile time, and may be globally
6603 disabled with the global option "nosplice". Since splice uses pipes, using it
6604 requires that there are enough spare pipes.
6605
6606 Important note: kernel-based TCP splicing is a Linux-specific feature which
6607 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6608 transfer data between sockets without copying these data to user-space, thus
6609 providing noticeable performance gains and CPU cycles savings. Since many
6610 early implementations are buggy, corrupt data and/or are inefficient, this
6611 feature is not enabled by default, and it should be used with extreme care.
6612 While it is not possible to detect the correctness of an implementation,
6613 2.6.29 is the first version offering a properly working implementation. In
6614 case of doubt, splicing may be globally disabled using the global "nosplice"
6615 keyword.
6616
6617 Example :
6618 option splice-auto
6619
6620 If this option has been enabled in a "defaults" section, it can be disabled
6621 in a specific instance by prepending the "no" keyword before it.
6622
6623 See also : "option splice-request", "option splice-response", and global
6624 options "nosplice" and "maxpipes"
6625
6626
6627option splice-request
6628no option splice-request
6629 Enable or disable automatic kernel acceleration on sockets for requests
6630 May be used in sections : defaults | frontend | listen | backend
6631 yes | yes | yes | yes
6632 Arguments : none
6633
6634 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006635 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006636 the client to the server. It might still use the recv/send scheme if there
6637 are no spare pipes left. This option requires splicing to be enabled at
6638 compile time, and may be globally disabled with the global option "nosplice".
6639 Since splice uses pipes, using it requires that there are enough spare pipes.
6640
6641 Important note: see "option splice-auto" for usage limitations.
6642
6643 Example :
6644 option splice-request
6645
6646 If this option has been enabled in a "defaults" section, it can be disabled
6647 in a specific instance by prepending the "no" keyword before it.
6648
6649 See also : "option splice-auto", "option splice-response", and global options
6650 "nosplice" and "maxpipes"
6651
6652
6653option splice-response
6654no option splice-response
6655 Enable or disable automatic kernel acceleration on sockets for responses
6656 May be used in sections : defaults | frontend | listen | backend
6657 yes | yes | yes | yes
6658 Arguments : none
6659
6660 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006661 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006662 the server to the client. It might still use the recv/send scheme if there
6663 are no spare pipes left. This option requires splicing to be enabled at
6664 compile time, and may be globally disabled with the global option "nosplice".
6665 Since splice uses pipes, using it requires that there are enough spare pipes.
6666
6667 Important note: see "option splice-auto" for usage limitations.
6668
6669 Example :
6670 option splice-response
6671
6672 If this option has been enabled in a "defaults" section, it can be disabled
6673 in a specific instance by prepending the "no" keyword before it.
6674
6675 See also : "option splice-auto", "option splice-request", and global options
6676 "nosplice" and "maxpipes"
6677
6678
Christopher Fauletba7bc162016-11-07 21:07:38 +01006679option spop-check
6680 Use SPOP health checks for server testing
6681 May be used in sections : defaults | frontend | listen | backend
6682 no | no | no | yes
6683 Arguments : none
6684
6685 It is possible to test that the server correctly talks SPOP protocol instead
6686 of just testing that it accepts the TCP connection. When this option is set,
6687 a HELLO handshake is performed between HAProxy and the server, and the
6688 response is analyzed to check no error is reported.
6689
6690 Example :
6691 option spop-check
6692
6693 See also : "option httpchk"
6694
6695
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006696option srvtcpka
6697no option srvtcpka
6698 Enable or disable the sending of TCP keepalive packets on the server side
6699 May be used in sections : defaults | frontend | listen | backend
6700 yes | no | yes | yes
6701 Arguments : none
6702
6703 When there is a firewall or any session-aware component between a client and
6704 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006705 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006706 components decides to expire a session which has remained idle for too long.
6707
6708 Enabling socket-level TCP keep-alives makes the system regularly send packets
6709 to the other end of the connection, leaving it active. The delay between
6710 keep-alive probes is controlled by the system only and depends both on the
6711 operating system and its tuning parameters.
6712
6713 It is important to understand that keep-alive packets are neither emitted nor
6714 received at the application level. It is only the network stacks which sees
6715 them. For this reason, even if one side of the proxy already uses keep-alives
6716 to maintain its connection alive, those keep-alive packets will not be
6717 forwarded to the other side of the proxy.
6718
6719 Please note that this has nothing to do with HTTP keep-alive.
6720
6721 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6722 server side of a connection, which should help when session expirations are
6723 noticed between HAProxy and a server.
6724
6725 If this option has been enabled in a "defaults" section, it can be disabled
6726 in a specific instance by prepending the "no" keyword before it.
6727
6728 See also : "option clitcpka", "option tcpka"
6729
6730
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006731option ssl-hello-chk
6732 Use SSLv3 client hello health checks for server testing
6733 May be used in sections : defaults | frontend | listen | backend
6734 yes | no | yes | yes
6735 Arguments : none
6736
6737 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6738 possible to test that the server correctly talks SSL instead of just testing
6739 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
6740 SSLv3 client hello messages are sent once the connection is established to
6741 the server, and the response is analyzed to find an SSL server hello message.
6742 The server is considered valid only when the response contains this server
6743 hello message.
6744
6745 All servers tested till there correctly reply to SSLv3 client hello messages,
6746 and most servers tested do not even log the requests containing only hello
6747 messages, which is appreciable.
6748
Willy Tarreau763a95b2012-10-04 23:15:39 +02006749 Note that this check works even when SSL support was not built into haproxy
6750 because it forges the SSL message. When SSL support is available, it is best
6751 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006752
Willy Tarreau763a95b2012-10-04 23:15:39 +02006753 See also: "option httpchk", "check-ssl"
6754
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006755
Willy Tarreaued179852013-12-16 01:07:00 +01006756option tcp-check
6757 Perform health checks using tcp-check send/expect sequences
6758 May be used in sections: defaults | frontend | listen | backend
6759 yes | no | yes | yes
6760
6761 This health check method is intended to be combined with "tcp-check" command
6762 lists in order to support send/expect types of health check sequences.
6763
6764 TCP checks currently support 4 modes of operations :
6765 - no "tcp-check" directive : the health check only consists in a connection
6766 attempt, which remains the default mode.
6767
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006768 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01006769 used to send a string along with a connection opening. With some
6770 protocols, it helps sending a "QUIT" message for example that prevents
6771 the server from logging a connection error for each health check. The
6772 check result will still be based on the ability to open the connection
6773 only.
6774
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006775 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01006776 The connection is opened and haproxy waits for the server to present some
6777 contents which must validate some rules. The check result will be based
6778 on the matching between the contents and the rules. This is suited for
6779 POP, IMAP, SMTP, FTP, SSH, TELNET.
6780
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006781 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01006782 used to test a hello-type protocol. HAProxy sends a message, the server
6783 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006784 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01006785 suited for protocols which require a binding or a request/response model.
6786 LDAP, MySQL, Redis and SSL are example of such protocols, though they
6787 already all have their dedicated checks with a deeper understanding of
6788 the respective protocols.
6789 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01006790 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01006791
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006792 A fifth mode can be used to insert comments in different steps of the
6793 script.
6794
6795 For each tcp-check rule you create, you can add a "comment" directive,
6796 followed by a string. This string will be reported in the log and stderr
6797 in debug mode. It is useful to make user-friendly error reporting.
6798 The "comment" is of course optional.
6799
6800
Willy Tarreaued179852013-12-16 01:07:00 +01006801 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006802 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01006803 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006804 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006805
Davor Ocelice9ed2812017-12-25 17:49:28 +01006806 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01006807 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006808 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01006809
6810 # look for the redis master server after ensuring it speaks well
6811 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01006812 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01006813 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006814 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006815 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02006816 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006817 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01006818 tcp-check send info\ replication\r\n
6819 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006820 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01006821 tcp-check send QUIT\r\n
6822 tcp-check expect string +OK
6823
Davor Ocelice9ed2812017-12-25 17:49:28 +01006824 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01006825 (send many headers before analyzing)
6826 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006827 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01006828 tcp-check send HEAD\ /\ HTTP/1.1\r\n
6829 tcp-check send Host:\ www.mydomain.com\r\n
6830 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
6831 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02006832 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01006833
6834
6835 See also : "tcp-check expect", "tcp-check send"
6836
6837
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006838option tcp-smart-accept
6839no option tcp-smart-accept
6840 Enable or disable the saving of one ACK packet during the accept sequence
6841 May be used in sections : defaults | frontend | listen | backend
6842 yes | yes | yes | no
6843 Arguments : none
6844
6845 When an HTTP connection request comes in, the system acknowledges it on
6846 behalf of HAProxy, then the client immediately sends its request, and the
6847 system acknowledges it too while it is notifying HAProxy about the new
6848 connection. HAProxy then reads the request and responds. This means that we
6849 have one TCP ACK sent by the system for nothing, because the request could
6850 very well be acknowledged by HAProxy when it sends its response.
6851
6852 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
6853 sending this useless ACK on platforms which support it (currently at least
6854 Linux). It must not cause any problem, because the system will send it anyway
6855 after 40 ms if the response takes more time than expected to come.
6856
6857 During complex network debugging sessions, it may be desirable to disable
6858 this optimization because delayed ACKs can make troubleshooting more complex
6859 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006860 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006861
6862 It is also possible to force it for non-HTTP proxies by simply specifying
6863 "option tcp-smart-accept". For instance, it can make sense with some services
6864 such as SMTP where the server speaks first.
6865
6866 It is recommended to avoid forcing this option in a defaults section. In case
6867 of doubt, consider setting it back to automatic values by prepending the
6868 "default" keyword before it, or disabling it using the "no" keyword.
6869
Willy Tarreaud88edf22009-06-14 15:48:17 +02006870 See also : "option tcp-smart-connect"
6871
6872
6873option tcp-smart-connect
6874no option tcp-smart-connect
6875 Enable or disable the saving of one ACK packet during the connect sequence
6876 May be used in sections : defaults | frontend | listen | backend
6877 yes | no | yes | yes
6878 Arguments : none
6879
6880 On certain systems (at least Linux), HAProxy can ask the kernel not to
6881 immediately send an empty ACK upon a connection request, but to directly
6882 send the buffer request instead. This saves one packet on the network and
6883 thus boosts performance. It can also be useful for some servers, because they
6884 immediately get the request along with the incoming connection.
6885
6886 This feature is enabled when "option tcp-smart-connect" is set in a backend.
6887 It is not enabled by default because it makes network troubleshooting more
6888 complex.
6889
6890 It only makes sense to enable it with protocols where the client speaks first
6891 such as HTTP. In other situations, if there is no data to send in place of
6892 the ACK, a normal ACK is sent.
6893
6894 If this option has been enabled in a "defaults" section, it can be disabled
6895 in a specific instance by prepending the "no" keyword before it.
6896
6897 See also : "option tcp-smart-accept"
6898
Willy Tarreau9ea05a72009-06-14 12:07:01 +02006899
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006900option tcpka
6901 Enable or disable the sending of TCP keepalive packets on both sides
6902 May be used in sections : defaults | frontend | listen | backend
6903 yes | yes | yes | yes
6904 Arguments : none
6905
6906 When there is a firewall or any session-aware component between a client and
6907 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006908 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006909 components decides to expire a session which has remained idle for too long.
6910
6911 Enabling socket-level TCP keep-alives makes the system regularly send packets
6912 to the other end of the connection, leaving it active. The delay between
6913 keep-alive probes is controlled by the system only and depends both on the
6914 operating system and its tuning parameters.
6915
6916 It is important to understand that keep-alive packets are neither emitted nor
6917 received at the application level. It is only the network stacks which sees
6918 them. For this reason, even if one side of the proxy already uses keep-alives
6919 to maintain its connection alive, those keep-alive packets will not be
6920 forwarded to the other side of the proxy.
6921
6922 Please note that this has nothing to do with HTTP keep-alive.
6923
6924 Using option "tcpka" enables the emission of TCP keep-alive probes on both
6925 the client and server sides of a connection. Note that this is meaningful
6926 only in "defaults" or "listen" sections. If this option is used in a
6927 frontend, only the client side will get keep-alives, and if this option is
6928 used in a backend, only the server side will get keep-alives. For this
6929 reason, it is strongly recommended to explicitly use "option clitcpka" and
6930 "option srvtcpka" when the configuration is split between frontends and
6931 backends.
6932
6933 See also : "option clitcpka", "option srvtcpka"
6934
Willy Tarreau844e3c52008-01-11 16:28:18 +01006935
6936option tcplog
6937 Enable advanced logging of TCP connections with session state and timers
6938 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006939 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01006940 Arguments : none
6941
6942 By default, the log output format is very poor, as it only contains the
6943 source and destination addresses, and the instance name. By specifying
6944 "option tcplog", each log line turns into a much richer format including, but
6945 not limited to, the connection timers, the session status, the connections
6946 numbers, the frontend, backend and server name, and of course the source
6947 address and ports. This option is useful for pure TCP proxies in order to
6948 find which of the client or server disconnects or times out. For normal HTTP
6949 proxies, it's better to use "option httplog" which is even more complete.
6950
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006951 "option tcplog" overrides any previous "log-format" directive.
6952
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006953 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01006954
6955
Willy Tarreau844e3c52008-01-11 16:28:18 +01006956option transparent
6957no option transparent
6958 Enable client-side transparent proxying
6959 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01006960 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01006961 Arguments : none
6962
6963 This option was introduced in order to provide layer 7 persistence to layer 3
6964 load balancers. The idea is to use the OS's ability to redirect an incoming
6965 connection for a remote address to a local process (here HAProxy), and let
6966 this process know what address was initially requested. When this option is
6967 used, sessions without cookies will be forwarded to the original destination
6968 IP address of the incoming request (which should match that of another
6969 equipment), while requests with cookies will still be forwarded to the
6970 appropriate server.
6971
6972 Note that contrary to a common belief, this option does NOT make HAProxy
6973 present the client's IP to the server when establishing the connection.
6974
Willy Tarreaua1146052011-03-01 09:51:54 +01006975 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006976 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01006977
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006978
Simon Horman98637e52014-06-20 12:30:16 +09006979external-check command <command>
6980 Executable to run when performing an external-check
6981 May be used in sections : defaults | frontend | listen | backend
6982 yes | no | yes | yes
6983
6984 Arguments :
6985 <command> is the external command to run
6986
Simon Horman98637e52014-06-20 12:30:16 +09006987 The arguments passed to the to the command are:
6988
Cyril Bonté777be862014-12-02 21:21:35 +01006989 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09006990
Cyril Bonté777be862014-12-02 21:21:35 +01006991 The <proxy_address> and <proxy_port> are derived from the first listener
6992 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
6993 listener the proxy_address will be the path of the socket and the
6994 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
6995 possible to determine a listener, and both <proxy_address> and <proxy_port>
6996 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09006997
Cyril Bonté72cda2a2014-12-27 22:28:39 +01006998 Some values are also provided through environment variables.
6999
7000 Environment variables :
7001 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7002 applicable, for example in a "backend" section).
7003
7004 HAPROXY_PROXY_ID The backend id.
7005
7006 HAPROXY_PROXY_NAME The backend name.
7007
7008 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7009 applicable, for example in a "backend" section or
7010 for a UNIX socket).
7011
7012 HAPROXY_SERVER_ADDR The server address.
7013
7014 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7015
7016 HAPROXY_SERVER_ID The server id.
7017
7018 HAPROXY_SERVER_MAXCONN The server max connections.
7019
7020 HAPROXY_SERVER_NAME The server name.
7021
7022 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7023 socket).
7024
7025 PATH The PATH environment variable used when executing
7026 the command may be set using "external-check path".
7027
Simon Horman98637e52014-06-20 12:30:16 +09007028 If the command executed and exits with a zero status then the check is
7029 considered to have passed, otherwise the check is considered to have
7030 failed.
7031
7032 Example :
7033 external-check command /bin/true
7034
7035 See also : "external-check", "option external-check", "external-check path"
7036
7037
7038external-check path <path>
7039 The value of the PATH environment variable used when running an external-check
7040 May be used in sections : defaults | frontend | listen | backend
7041 yes | no | yes | yes
7042
7043 Arguments :
7044 <path> is the path used when executing external command to run
7045
7046 The default path is "".
7047
7048 Example :
7049 external-check path "/usr/bin:/bin"
7050
7051 See also : "external-check", "option external-check",
7052 "external-check command"
7053
7054
Emeric Brun647caf12009-06-30 17:57:00 +02007055persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007056persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007057 Enable RDP cookie-based persistence
7058 May be used in sections : defaults | frontend | listen | backend
7059 yes | no | yes | yes
7060 Arguments :
7061 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007062 default cookie name "msts" will be used. There currently is no
7063 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007064
7065 This statement enables persistence based on an RDP cookie. The RDP cookie
7066 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007067 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007068 and if an RDP cookie is found, it is decoded. If it matches a known server
7069 which is still UP (or if "option persist" is set), then the connection is
7070 forwarded to this server.
7071
7072 Note that this only makes sense in a TCP backend, but for this to work, the
7073 frontend must have waited long enough to ensure that an RDP cookie is present
7074 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007075 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007076 a single "listen" section.
7077
Willy Tarreau61e28f22010-05-16 22:31:05 +02007078 Also, it is important to understand that the terminal server will emit this
7079 RDP cookie only if it is configured for "token redirection mode", which means
7080 that the "IP address redirection" option is disabled.
7081
Emeric Brun647caf12009-06-30 17:57:00 +02007082 Example :
7083 listen tse-farm
7084 bind :3389
7085 # wait up to 5s for an RDP cookie in the request
7086 tcp-request inspect-delay 5s
7087 tcp-request content accept if RDP_COOKIE
7088 # apply RDP cookie persistence
7089 persist rdp-cookie
7090 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007091 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007092 balance rdp-cookie
7093 server srv1 1.1.1.1:3389
7094 server srv2 1.1.1.2:3389
7095
Simon Hormanab814e02011-06-24 14:50:20 +09007096 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7097 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007098
7099
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007100rate-limit sessions <rate>
7101 Set a limit on the number of new sessions accepted per second on a frontend
7102 May be used in sections : defaults | frontend | listen | backend
7103 yes | yes | yes | no
7104 Arguments :
7105 <rate> The <rate> parameter is an integer designating the maximum number
7106 of new sessions per second to accept on the frontend.
7107
7108 When the frontend reaches the specified number of new sessions per second, it
7109 stops accepting new connections until the rate drops below the limit again.
7110 During this time, the pending sessions will be kept in the socket's backlog
7111 (in system buffers) and haproxy will not even be aware that sessions are
7112 pending. When applying very low limit on a highly loaded service, it may make
7113 sense to increase the socket's backlog using the "backlog" keyword.
7114
7115 This feature is particularly efficient at blocking connection-based attacks
7116 or service abuse on fragile servers. Since the session rate is measured every
7117 millisecond, it is extremely accurate. Also, the limit applies immediately,
7118 no delay is needed at all to detect the threshold.
7119
7120 Example : limit the connection rate on SMTP to 10 per second max
7121 listen smtp
7122 mode tcp
7123 bind :25
7124 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007125 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007126
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007127 Note : when the maximum rate is reached, the frontend's status is not changed
7128 but its sockets appear as "WAITING" in the statistics if the
7129 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007130
7131 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7132
7133
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007134redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7135redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7136redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007137 Return an HTTP redirection if/unless a condition is matched
7138 May be used in sections : defaults | frontend | listen | backend
7139 no | yes | yes | yes
7140
7141 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007142 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007143
Willy Tarreau0140f252008-11-19 21:07:09 +01007144 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007145 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007146 the HTTP "Location" header. When used in an "http-request" rule,
7147 <loc> value follows the log-format rules and can include some
7148 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007149
7150 <pfx> With "redirect prefix", the "Location" header is built from the
7151 concatenation of <pfx> and the complete URI path, including the
7152 query string, unless the "drop-query" option is specified (see
7153 below). As a special case, if <pfx> equals exactly "/", then
7154 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007155 redirect to the same URL (for instance, to insert a cookie). When
7156 used in an "http-request" rule, <pfx> value follows the log-format
7157 rules and can include some dynamic values (see Custom Log Format
7158 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007159
7160 <sch> With "redirect scheme", then the "Location" header is built by
7161 concatenating <sch> with "://" then the first occurrence of the
7162 "Host" header, and then the URI path, including the query string
7163 unless the "drop-query" option is specified (see below). If no
7164 path is found or if the path is "*", then "/" is used instead. If
7165 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007166 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007167 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007168 HTTPS. When used in an "http-request" rule, <sch> value follows
7169 the log-format rules and can include some dynamic values (see
7170 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007171
7172 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007173 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7174 with 302 used by default if no code is specified. 301 means
7175 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007176 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007177 cache the redirection. 303 is equivalent to 302 except that the
7178 browser will fetch the location with a GET method. 307 is just
7179 like 302 but makes it clear that the same method must be reused.
7180 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007181
7182 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007183 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007184
7185 - "drop-query"
7186 When this keyword is used in a prefix-based redirection, then the
7187 location will be set without any possible query-string, which is useful
7188 for directing users to a non-secure page for instance. It has no effect
7189 with a location-type redirect.
7190
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007191 - "append-slash"
7192 This keyword may be used in conjunction with "drop-query" to redirect
7193 users who use a URL not ending with a '/' to the same one with the '/'.
7194 It can be useful to ensure that search engines will only see one URL.
7195 For this, a return code 301 is preferred.
7196
Willy Tarreau0140f252008-11-19 21:07:09 +01007197 - "set-cookie NAME[=value]"
7198 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7199 to the response. This is sometimes used to indicate that a user has
7200 been seen, for instance to protect against some types of DoS. No other
7201 cookie option is added, so the cookie will be a session cookie. Note
7202 that for a browser, a sole cookie name without an equal sign is
7203 different from a cookie with an equal sign.
7204
7205 - "clear-cookie NAME[=]"
7206 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7207 with the "Max-Age" attribute set to zero. This will tell the browser to
7208 delete this cookie. It is useful for instance on logout pages. It is
7209 important to note that clearing the cookie "NAME" will not remove a
7210 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7211 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007212
7213 Example: move the login URL only to HTTPS.
7214 acl clear dst_port 80
7215 acl secure dst_port 8080
7216 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007217 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007218 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007219 acl cookie_set hdr_sub(cookie) SEEN=1
7220
7221 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007222 redirect prefix https://mysite.com if login_page !secure
7223 redirect prefix http://mysite.com drop-query if login_page !uid_given
7224 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007225 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007226
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007227 Example: send redirects for request for articles without a '/'.
7228 acl missing_slash path_reg ^/article/[^/]*$
7229 redirect code 301 prefix / drop-query append-slash if missing_slash
7230
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007231 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007232 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007233
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007234 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007235 http-request redirect code 301 location \
7236 http://www.%[hdr(host)]%[capture.req.uri] \
7237 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007238
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007239 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007240
7241
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007242redisp (deprecated)
7243redispatch (deprecated)
7244 Enable or disable session redistribution in case of connection failure
7245 May be used in sections: defaults | frontend | listen | backend
7246 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007247 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007248
7249 In HTTP mode, if a server designated by a cookie is down, clients may
7250 definitely stick to it because they cannot flush the cookie, so they will not
7251 be able to access the service anymore.
7252
7253 Specifying "redispatch" will allow the proxy to break their persistence and
7254 redistribute them to a working server.
7255
7256 It also allows to retry last connection to another server in case of multiple
7257 connection failures. Of course, it requires having "retries" set to a nonzero
7258 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007259
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007260 This form is deprecated, do not use it in any new configuration, use the new
7261 "option redispatch" instead.
7262
7263 See also : "option redispatch"
7264
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007265
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007266reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007267 Add a header at the end of the HTTP request
7268 May be used in sections : defaults | frontend | listen | backend
7269 no | yes | yes | yes
7270 Arguments :
7271 <string> is the complete line to be added. Any space or known delimiter
7272 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007273 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007274
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007275 <cond> is an optional matching condition built from ACLs. It makes it
7276 possible to ignore this rule when other conditions are not met.
7277
Willy Tarreau303c0352008-01-17 19:01:39 +01007278 A new line consisting in <string> followed by a line feed will be added after
7279 the last header of an HTTP request.
7280
7281 Header transformations only apply to traffic which passes through HAProxy,
7282 and not to traffic generated by HAProxy, such as health-checks or error
7283 responses.
7284
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007285 Example : add "X-Proto: SSL" to requests coming via port 81
7286 acl is-ssl dst_port 81
7287 reqadd X-Proto:\ SSL if is-ssl
7288
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007289 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7290 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007291
7292
Willy Tarreau5321c422010-01-28 20:35:13 +01007293reqallow <search> [{if | unless} <cond>]
7294reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007295 Definitely allow an HTTP request if a line matches a regular expression
7296 May be used in sections : defaults | frontend | listen | backend
7297 no | yes | yes | yes
7298 Arguments :
7299 <search> is the regular expression applied to HTTP headers and to the
7300 request line. This is an extended regular expression. Parenthesis
7301 grouping is supported and no preliminary backslash is required.
7302 Any space or known delimiter must be escaped using a backslash
7303 ('\'). The pattern applies to a full line at a time. The
7304 "reqallow" keyword strictly matches case while "reqiallow"
7305 ignores case.
7306
Willy Tarreau5321c422010-01-28 20:35:13 +01007307 <cond> is an optional matching condition built from ACLs. It makes it
7308 possible to ignore this rule when other conditions are not met.
7309
Willy Tarreau303c0352008-01-17 19:01:39 +01007310 A request containing any line which matches extended regular expression
7311 <search> will mark the request as allowed, even if any later test would
7312 result in a deny. The test applies both to the request line and to request
7313 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007314 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007315
7316 It is easier, faster and more powerful to use ACLs to write access policies.
7317 Reqdeny, reqallow and reqpass should be avoided in new designs.
7318
7319 Example :
7320 # allow www.* but refuse *.local
7321 reqiallow ^Host:\ www\.
7322 reqideny ^Host:\ .*\.local
7323
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007324 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7325 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007326
7327
Willy Tarreau5321c422010-01-28 20:35:13 +01007328reqdel <search> [{if | unless} <cond>]
7329reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007330 Delete all headers matching a regular expression in an HTTP request
7331 May be used in sections : defaults | frontend | listen | backend
7332 no | yes | yes | yes
7333 Arguments :
7334 <search> is the regular expression applied to HTTP headers and to the
7335 request line. This is an extended regular expression. Parenthesis
7336 grouping is supported and no preliminary backslash is required.
7337 Any space or known delimiter must be escaped using a backslash
7338 ('\'). The pattern applies to a full line at a time. The "reqdel"
7339 keyword strictly matches case while "reqidel" ignores case.
7340
Willy Tarreau5321c422010-01-28 20:35:13 +01007341 <cond> is an optional matching condition built from ACLs. It makes it
7342 possible to ignore this rule when other conditions are not met.
7343
Willy Tarreau303c0352008-01-17 19:01:39 +01007344 Any header line matching extended regular expression <search> in the request
7345 will be completely deleted. Most common use of this is to remove unwanted
7346 and/or dangerous headers or cookies from a request before passing it to the
7347 next servers.
7348
7349 Header transformations only apply to traffic which passes through HAProxy,
7350 and not to traffic generated by HAProxy, such as health-checks or error
7351 responses. Keep in mind that header names are not case-sensitive.
7352
7353 Example :
7354 # remove X-Forwarded-For header and SERVER cookie
7355 reqidel ^X-Forwarded-For:.*
7356 reqidel ^Cookie:.*SERVER=
7357
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007358 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7359 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007360
7361
Willy Tarreau5321c422010-01-28 20:35:13 +01007362reqdeny <search> [{if | unless} <cond>]
7363reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007364 Deny an HTTP request if a line matches a regular expression
7365 May be used in sections : defaults | frontend | listen | backend
7366 no | yes | yes | yes
7367 Arguments :
7368 <search> is the regular expression applied to HTTP headers and to the
7369 request line. This is an extended regular expression. Parenthesis
7370 grouping is supported and no preliminary backslash is required.
7371 Any space or known delimiter must be escaped using a backslash
7372 ('\'). The pattern applies to a full line at a time. The
7373 "reqdeny" keyword strictly matches case while "reqideny" ignores
7374 case.
7375
Willy Tarreau5321c422010-01-28 20:35:13 +01007376 <cond> is an optional matching condition built from ACLs. It makes it
7377 possible to ignore this rule when other conditions are not met.
7378
Willy Tarreau303c0352008-01-17 19:01:39 +01007379 A request containing any line which matches extended regular expression
7380 <search> will mark the request as denied, even if any later test would
7381 result in an allow. The test applies both to the request line and to request
7382 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007383 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007384
Willy Tarreauced27012008-01-17 20:35:34 +01007385 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007386 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007387 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007388
Willy Tarreau303c0352008-01-17 19:01:39 +01007389 It is easier, faster and more powerful to use ACLs to write access policies.
7390 Reqdeny, reqallow and reqpass should be avoided in new designs.
7391
7392 Example :
7393 # refuse *.local, then allow www.*
7394 reqideny ^Host:\ .*\.local
7395 reqiallow ^Host:\ www\.
7396
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007397 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7398 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007399
7400
Willy Tarreau5321c422010-01-28 20:35:13 +01007401reqpass <search> [{if | unless} <cond>]
7402reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007403 Ignore any HTTP request line matching a regular expression in next rules
7404 May be used in sections : defaults | frontend | listen | backend
7405 no | yes | yes | yes
7406 Arguments :
7407 <search> is the regular expression applied to HTTP headers and to the
7408 request line. This is an extended regular expression. Parenthesis
7409 grouping is supported and no preliminary backslash is required.
7410 Any space or known delimiter must be escaped using a backslash
7411 ('\'). The pattern applies to a full line at a time. The
7412 "reqpass" keyword strictly matches case while "reqipass" ignores
7413 case.
7414
Willy Tarreau5321c422010-01-28 20:35:13 +01007415 <cond> is an optional matching condition built from ACLs. It makes it
7416 possible to ignore this rule when other conditions are not met.
7417
Willy Tarreau303c0352008-01-17 19:01:39 +01007418 A request containing any line which matches extended regular expression
7419 <search> will skip next rules, without assigning any deny or allow verdict.
7420 The test applies both to the request line and to request headers. Keep in
7421 mind that URLs in request line are case-sensitive while header names are not.
7422
7423 It is easier, faster and more powerful to use ACLs to write access policies.
7424 Reqdeny, reqallow and reqpass should be avoided in new designs.
7425
7426 Example :
7427 # refuse *.local, then allow www.*, but ignore "www.private.local"
7428 reqipass ^Host:\ www.private\.local
7429 reqideny ^Host:\ .*\.local
7430 reqiallow ^Host:\ www\.
7431
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007432 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7433 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007434
7435
Willy Tarreau5321c422010-01-28 20:35:13 +01007436reqrep <search> <string> [{if | unless} <cond>]
7437reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007438 Replace a regular expression with a string in an HTTP request line
7439 May be used in sections : defaults | frontend | listen | backend
7440 no | yes | yes | yes
7441 Arguments :
7442 <search> is the regular expression applied to HTTP headers and to the
7443 request line. This is an extended regular expression. Parenthesis
7444 grouping is supported and no preliminary backslash is required.
7445 Any space or known delimiter must be escaped using a backslash
7446 ('\'). The pattern applies to a full line at a time. The "reqrep"
7447 keyword strictly matches case while "reqirep" ignores case.
7448
7449 <string> is the complete line to be added. Any space or known delimiter
7450 must be escaped using a backslash ('\'). References to matched
7451 pattern groups are possible using the common \N form, with N
7452 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007453 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007454
Willy Tarreau5321c422010-01-28 20:35:13 +01007455 <cond> is an optional matching condition built from ACLs. It makes it
7456 possible to ignore this rule when other conditions are not met.
7457
Willy Tarreau303c0352008-01-17 19:01:39 +01007458 Any line matching extended regular expression <search> in the request (both
7459 the request line and header lines) will be completely replaced with <string>.
7460 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7461
7462 Header transformations only apply to traffic which passes through HAProxy,
7463 and not to traffic generated by HAProxy, such as health-checks or error
7464 responses. Note that for increased readability, it is suggested to add enough
7465 spaces between the request and the response. Keep in mind that URLs in
7466 request line are case-sensitive while header names are not.
7467
7468 Example :
7469 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007470 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007471 # replace "www.mydomain.com" with "www" in the host name.
7472 reqirep ^Host:\ www.mydomain.com Host:\ www
7473
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007474 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7475 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007476
7477
Willy Tarreau5321c422010-01-28 20:35:13 +01007478reqtarpit <search> [{if | unless} <cond>]
7479reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007480 Tarpit an HTTP request containing a line matching a regular expression
7481 May be used in sections : defaults | frontend | listen | backend
7482 no | yes | yes | yes
7483 Arguments :
7484 <search> is the regular expression applied to HTTP headers and to the
7485 request line. This is an extended regular expression. Parenthesis
7486 grouping is supported and no preliminary backslash is required.
7487 Any space or known delimiter must be escaped using a backslash
7488 ('\'). The pattern applies to a full line at a time. The
7489 "reqtarpit" keyword strictly matches case while "reqitarpit"
7490 ignores case.
7491
Willy Tarreau5321c422010-01-28 20:35:13 +01007492 <cond> is an optional matching condition built from ACLs. It makes it
7493 possible to ignore this rule when other conditions are not met.
7494
Willy Tarreau303c0352008-01-17 19:01:39 +01007495 A request containing any line which matches extended regular expression
7496 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007497 be kept open for a pre-defined time, then will return an HTTP error 500 so
7498 that the attacker does not suspect it has been tarpitted. The status 500 will
7499 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007500 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7501 not set.
7502
7503 The goal of the tarpit is to slow down robots attacking servers with
7504 identifiable requests. Many robots limit their outgoing number of connections
7505 and stay connected waiting for a reply which can take several minutes to
7506 come. Depending on the environment and attack, it may be particularly
7507 efficient at reducing the load on the network and firewalls.
7508
Willy Tarreau5321c422010-01-28 20:35:13 +01007509 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007510 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007511 # block all others.
7512 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7513 reqitarpit ^User-Agent:
7514
Willy Tarreau5321c422010-01-28 20:35:13 +01007515 # block bad guys
7516 acl badguys src 10.1.0.3 172.16.13.20/28
7517 reqitarpit . if badguys
7518
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007519 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7520 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007521
7522
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007523retries <value>
7524 Set the number of retries to perform on a server after a connection failure
7525 May be used in sections: defaults | frontend | listen | backend
7526 yes | no | yes | yes
7527 Arguments :
7528 <value> is the number of times a connection attempt should be retried on
7529 a server when a connection either is refused or times out. The
7530 default value is 3.
7531
7532 It is important to understand that this value applies to the number of
7533 connection attempts, not full requests. When a connection has effectively
7534 been established to a server, there will be no more retry.
7535
7536 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007537 a turn-around timer of min("timeout connect", one second) is applied before
7538 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007539
7540 When "option redispatch" is set, the last retry may be performed on another
7541 server even if a cookie references a different server.
7542
7543 See also : "option redispatch"
7544
7545
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007546rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007547 Add a header at the end of the HTTP response
7548 May be used in sections : defaults | frontend | listen | backend
7549 no | yes | yes | yes
7550 Arguments :
7551 <string> is the complete line to be added. Any space or known delimiter
7552 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007553 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007554
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007555 <cond> is an optional matching condition built from ACLs. It makes it
7556 possible to ignore this rule when other conditions are not met.
7557
Willy Tarreau303c0352008-01-17 19:01:39 +01007558 A new line consisting in <string> followed by a line feed will be added after
7559 the last header of an HTTP response.
7560
7561 Header transformations only apply to traffic which passes through HAProxy,
7562 and not to traffic generated by HAProxy, such as health-checks or error
7563 responses.
7564
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007565 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7566 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007567
7568
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007569rspdel <search> [{if | unless} <cond>]
7570rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007571 Delete all headers matching a regular expression in an HTTP response
7572 May be used in sections : defaults | frontend | listen | backend
7573 no | yes | yes | yes
7574 Arguments :
7575 <search> is the regular expression applied to HTTP headers and to the
7576 response line. This is an extended regular expression, so
7577 parenthesis grouping is supported and no preliminary backslash
7578 is required. Any space or known delimiter must be escaped using
7579 a backslash ('\'). The pattern applies to a full line at a time.
7580 The "rspdel" keyword strictly matches case while "rspidel"
7581 ignores case.
7582
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007583 <cond> is an optional matching condition built from ACLs. It makes it
7584 possible to ignore this rule when other conditions are not met.
7585
Willy Tarreau303c0352008-01-17 19:01:39 +01007586 Any header line matching extended regular expression <search> in the response
7587 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007588 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007589 client.
7590
7591 Header transformations only apply to traffic which passes through HAProxy,
7592 and not to traffic generated by HAProxy, such as health-checks or error
7593 responses. Keep in mind that header names are not case-sensitive.
7594
7595 Example :
7596 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007597 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007598
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007599 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7600 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007601
7602
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007603rspdeny <search> [{if | unless} <cond>]
7604rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007605 Block an HTTP response if a line matches a regular expression
7606 May be used in sections : defaults | frontend | listen | backend
7607 no | yes | yes | yes
7608 Arguments :
7609 <search> is the regular expression applied to HTTP headers and to the
7610 response line. This is an extended regular expression, so
7611 parenthesis grouping is supported and no preliminary backslash
7612 is required. Any space or known delimiter must be escaped using
7613 a backslash ('\'). The pattern applies to a full line at a time.
7614 The "rspdeny" keyword strictly matches case while "rspideny"
7615 ignores case.
7616
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007617 <cond> is an optional matching condition built from ACLs. It makes it
7618 possible to ignore this rule when other conditions are not met.
7619
Willy Tarreau303c0352008-01-17 19:01:39 +01007620 A response containing any line which matches extended regular expression
7621 <search> will mark the request as denied. The test applies both to the
7622 response line and to response headers. Keep in mind that header names are not
7623 case-sensitive.
7624
7625 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007626 block the response before it reaches the client. If a response is denied, it
7627 will be replaced with an HTTP 502 error so that the client never retrieves
7628 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007629
7630 It is easier, faster and more powerful to use ACLs to write access policies.
7631 Rspdeny should be avoided in new designs.
7632
7633 Example :
7634 # Ensure that no content type matching ms-word will leak
7635 rspideny ^Content-type:\.*/ms-word
7636
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007637 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7638 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007639
7640
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007641rsprep <search> <string> [{if | unless} <cond>]
7642rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007643 Replace a regular expression with a string in an HTTP response line
7644 May be used in sections : defaults | frontend | listen | backend
7645 no | yes | yes | yes
7646 Arguments :
7647 <search> is the regular expression applied to HTTP headers and to the
7648 response line. This is an extended regular expression, so
7649 parenthesis grouping is supported and no preliminary backslash
7650 is required. Any space or known delimiter must be escaped using
7651 a backslash ('\'). The pattern applies to a full line at a time.
7652 The "rsprep" keyword strictly matches case while "rspirep"
7653 ignores case.
7654
7655 <string> is the complete line to be added. Any space or known delimiter
7656 must be escaped using a backslash ('\'). References to matched
7657 pattern groups are possible using the common \N form, with N
7658 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007659 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007660
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007661 <cond> is an optional matching condition built from ACLs. It makes it
7662 possible to ignore this rule when other conditions are not met.
7663
Willy Tarreau303c0352008-01-17 19:01:39 +01007664 Any line matching extended regular expression <search> in the response (both
7665 the response line and header lines) will be completely replaced with
7666 <string>. Most common use of this is to rewrite Location headers.
7667
7668 Header transformations only apply to traffic which passes through HAProxy,
7669 and not to traffic generated by HAProxy, such as health-checks or error
7670 responses. Note that for increased readability, it is suggested to add enough
7671 spaces between the request and the response. Keep in mind that header names
7672 are not case-sensitive.
7673
7674 Example :
7675 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7676 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7677
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007678 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7679 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007680
7681
David du Colombier486df472011-03-17 10:40:26 +01007682server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007683 Declare a server in a backend
7684 May be used in sections : defaults | frontend | listen | backend
7685 no | no | yes | yes
7686 Arguments :
7687 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007688 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007689 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007690
David du Colombier486df472011-03-17 10:40:26 +01007691 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7692 resolvable hostname is supported, but this name will be resolved
7693 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7694 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007695 address as the one from the client connection. This is useful in
7696 transparent proxy architectures where the client's connection is
7697 intercepted and haproxy must forward to the original destination
7698 address. This is more or less what the "transparent" keyword does
7699 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007700 to report statistics. Optionally, an address family prefix may be
7701 used before the address to force the family regardless of the
7702 address format, which can be useful to specify a path to a unix
7703 socket with no slash ('/'). Currently supported prefixes are :
7704 - 'ipv4@' -> address is always IPv4
7705 - 'ipv6@' -> address is always IPv6
7706 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007707 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemandb2f07452015-05-12 14:27:13 +02007708 You may want to reference some environment variables in the
7709 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007710 variables. The "init-addr" setting can be used to modify the way
7711 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007712
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007713 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007714 be sent to this port. If unset, the same port the client
7715 connected to will be used. The port may also be prefixed by a "+"
7716 or a "-". In this case, the server's port will be determined by
7717 adding this value to the client's port.
7718
7719 <param*> is a list of parameters for this server. The "server" keywords
7720 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007721 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007722
7723 Examples :
7724 server first 10.1.1.1:1080 cookie first check inter 1000
7725 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007726 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007727 server backup "${SRV_BACKUP}:1080" backup
7728 server www1_dc1 "${LAN_DC1}.101:80"
7729 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007730
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007731 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7732 sun_path length is used for the address length. Some other programs
7733 such as socat use the string length only by default. Pass the option
7734 ",unix-tightsocklen=0" to any abstract socket definition in socat to
7735 make it compatible with HAProxy's.
7736
Mark Lamourinec2247f02012-01-04 13:02:01 -05007737 See also: "default-server", "http-send-name-header" and section 5 about
7738 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007739
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007740server-state-file-name [<file>]
7741 Set the server state file to read, load and apply to servers available in
7742 this backend. It only applies when the directive "load-server-state-from-file"
7743 is set to "local". When <file> is not provided or if this directive is not
7744 set, then backend name is used. If <file> starts with a slash '/', then it is
7745 considered as an absolute path. Otherwise, <file> is concatenated to the
7746 global directive "server-state-file-base".
7747
7748 Example: the minimal configuration below would make HAProxy look for the
7749 state server file '/etc/haproxy/states/bk':
7750
7751 global
7752 server-state-file-base /etc/haproxy/states
7753
7754 backend bk
7755 load-server-state-from-file
7756
7757 See also: "server-state-file-base", "load-server-state-from-file", and
7758 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007759
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02007760server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
7761 Set a template to initialize servers with shared parameters.
7762 The names of these servers are built from <prefix> and <num | range> parameters.
7763 May be used in sections : defaults | frontend | listen | backend
7764 no | no | yes | yes
7765
7766 Arguments:
7767 <prefix> A prefix for the server names to be built.
7768
7769 <num | range>
7770 If <num> is provided, this template initializes <num> servers
7771 with 1 up to <num> as server name suffixes. A range of numbers
7772 <num_low>-<num_high> may also be used to use <num_low> up to
7773 <num_high> as server name suffixes.
7774
7775 <fqdn> A FQDN for all the servers this template initializes.
7776
7777 <port> Same meaning as "server" <port> argument (see "server" keyword).
7778
7779 <params*>
7780 Remaining server parameters among all those supported by "server"
7781 keyword.
7782
7783 Examples:
7784 # Initializes 3 servers with srv1, srv2 and srv3 as names,
7785 # google.com as FQDN, and health-check enabled.
7786 server-template srv 1-3 google.com:80 check
7787
7788 # or
7789 server-template srv 3 google.com:80 check
7790
7791 # would be equivalent to:
7792 server srv1 google.com:80 check
7793 server srv2 google.com:80 check
7794 server srv3 google.com:80 check
7795
7796
7797
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007798source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02007799source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007800source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007801 Set the source address for outgoing connections
7802 May be used in sections : defaults | frontend | listen | backend
7803 yes | no | yes | yes
7804 Arguments :
7805 <addr> is the IPv4 address HAProxy will bind to before connecting to a
7806 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01007807
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007808 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01007809 the most appropriate address to reach its destination. Optionally
7810 an address family prefix may be used before the address to force
7811 the family regardless of the address format, which can be useful
7812 to specify a path to a unix socket with no slash ('/'). Currently
7813 supported prefixes are :
7814 - 'ipv4@' -> address is always IPv4
7815 - 'ipv6@' -> address is always IPv6
7816 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007817 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02007818 You may want to reference some environment variables in the
7819 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007820
7821 <port> is an optional port. It is normally not needed but may be useful
7822 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02007823 the system will select a free port. Note that port ranges are not
7824 supported in the backend. If you want to force port ranges, you
7825 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007826
7827 <addr2> is the IP address to present to the server when connections are
7828 forwarded in full transparent proxy mode. This is currently only
7829 supported on some patched Linux kernels. When this address is
7830 specified, clients connecting to the server will be presented
7831 with this address, while health checks will still use the address
7832 <addr>.
7833
7834 <port2> is the optional port to present to the server when connections
7835 are forwarded in full transparent proxy mode (see <addr2> above).
7836 The default value of zero means the system will select a free
7837 port.
7838
Willy Tarreaubce70882009-09-07 11:51:47 +02007839 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
7840 This is the name of a comma-separated header list which can
7841 contain multiple IP addresses. By default, the last occurrence is
7842 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01007843 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02007844 by previous proxy, typically Stunnel. In order to use another
7845 occurrence from the last one, please see the <occ> parameter
7846 below. When the header (or occurrence) is not found, no binding
7847 is performed so that the proxy's default IP address is used. Also
7848 keep in mind that the header name is case insensitive, as for any
7849 HTTP header.
7850
7851 <occ> is the occurrence number of a value to be used in a multi-value
7852 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007853 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02007854 address. Positive values indicate a position from the first
7855 occurrence, 1 being the first one. Negative values indicate
7856 positions relative to the last one, -1 being the last one. This
7857 is helpful for situations where an X-Forwarded-For header is set
7858 at the entry point of an infrastructure and must be used several
7859 proxy layers away. When this value is not specified, -1 is
7860 assumed. Passing a zero here disables the feature.
7861
Willy Tarreaud53f96b2009-02-04 18:46:54 +01007862 <name> is an optional interface name to which to bind to for outgoing
7863 traffic. On systems supporting this features (currently, only
7864 Linux), this allows one to bind all traffic to the server to
7865 this interface even if it is not the one the system would select
7866 based on routing tables. This should be used with extreme care.
7867 Note that using this option requires root privileges.
7868
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007869 The "source" keyword is useful in complex environments where a specific
7870 address only is allowed to connect to the servers. It may be needed when a
7871 private address must be used through a public gateway for instance, and it is
7872 known that the system cannot determine the adequate source address by itself.
7873
7874 An extension which is available on certain patched Linux kernels may be used
7875 through the "usesrc" optional keyword. It makes it possible to connect to the
7876 servers with an IP address which does not belong to the system itself. This
7877 is called "full transparent proxy mode". For this to work, the destination
7878 servers have to route their traffic back to this address through the machine
7879 running HAProxy, and IP forwarding must generally be enabled on this machine.
7880
7881 In this "full transparent proxy" mode, it is possible to force a specific IP
7882 address to be presented to the servers. This is not much used in fact. A more
7883 common use is to tell HAProxy to present the client's IP address. For this,
7884 there are two methods :
7885
7886 - present the client's IP and port addresses. This is the most transparent
7887 mode, but it can cause problems when IP connection tracking is enabled on
7888 the machine, because a same connection may be seen twice with different
7889 states. However, this solution presents the huge advantage of not
7890 limiting the system to the 64k outgoing address+port couples, because all
7891 of the client ranges may be used.
7892
7893 - present only the client's IP address and select a spare port. This
7894 solution is still quite elegant but slightly less transparent (downstream
7895 firewalls logs will not match upstream's). It also presents the downside
7896 of limiting the number of concurrent connections to the usual 64k ports.
7897 However, since the upstream and downstream ports are different, local IP
7898 connection tracking on the machine will not be upset by the reuse of the
7899 same session.
7900
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007901 This option sets the default source for all servers in the backend. It may
7902 also be specified in a "defaults" section. Finer source address specification
7903 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007904 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007905
Baptiste Assmann91bd3372015-07-17 21:59:42 +02007906 In order to work, "usesrc" requires root privileges.
7907
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007908 Examples :
7909 backend private
7910 # Connect to the servers using our 192.168.1.200 source address
7911 source 192.168.1.200
7912
7913 backend transparent_ssl1
7914 # Connect to the SSL farm from the client's source address
7915 source 192.168.1.200 usesrc clientip
7916
7917 backend transparent_ssl2
7918 # Connect to the SSL farm from the client's source address and port
7919 # not recommended if IP conntrack is present on the local machine.
7920 source 192.168.1.200 usesrc client
7921
7922 backend transparent_ssl3
7923 # Connect to the SSL farm from the client's source address. It
7924 # is more conntrack-friendly.
7925 source 192.168.1.200 usesrc clientip
7926
7927 backend transparent_smtp
7928 # Connect to the SMTP farm from the client's source address/port
7929 # with Tproxy version 4.
7930 source 0.0.0.0 usesrc clientip
7931
Willy Tarreaubce70882009-09-07 11:51:47 +02007932 backend transparent_http
7933 # Connect to the servers using the client's IP as seen by previous
7934 # proxy.
7935 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
7936
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007937 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007938 the Linux kernel on www.balabit.com, the "bind" keyword.
7939
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007940
Willy Tarreau844e3c52008-01-11 16:28:18 +01007941srvtimeout <timeout> (deprecated)
7942 Set the maximum inactivity time on the server side.
7943 May be used in sections : defaults | frontend | listen | backend
7944 yes | no | yes | yes
7945 Arguments :
7946 <timeout> is the timeout value specified in milliseconds by default, but
7947 can be in any other unit if the number is suffixed by the unit,
7948 as explained at the top of this document.
7949
7950 The inactivity timeout applies when the server is expected to acknowledge or
7951 send data. In HTTP mode, this timeout is particularly important to consider
7952 during the first phase of the server's response, when it has to send the
7953 headers, as it directly represents the server's processing time for the
7954 request. To find out what value to put there, it's often good to start with
7955 what would be considered as unacceptable response times, then check the logs
7956 to observe the response time distribution, and adjust the value accordingly.
7957
7958 The value is specified in milliseconds by default, but can be in any other
7959 unit if the number is suffixed by the unit, as specified at the top of this
7960 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
7961 recommended that the client timeout remains equal to the server timeout in
7962 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007963 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01007964 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01007965 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01007966
7967 This parameter is specific to backends, but can be specified once for all in
7968 "defaults" sections. This is in fact one of the easiest solutions not to
7969 forget about it. An unspecified timeout results in an infinite timeout, which
7970 is not recommended. Such a usage is accepted and works but reports a warning
7971 during startup because it may results in accumulation of expired sessions in
7972 the system if the system's timeouts are not configured either.
7973
7974 This parameter is provided for compatibility but is currently deprecated.
7975 Please use "timeout server" instead.
7976
Willy Tarreauce887fd2012-05-12 12:50:00 +02007977 See also : "timeout server", "timeout tunnel", "timeout client" and
7978 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01007979
7980
Cyril Bonté66c327d2010-10-12 00:14:37 +02007981stats admin { if | unless } <cond>
7982 Enable statistics admin level if/unless a condition is matched
7983 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02007984 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02007985
7986 This statement enables the statistics admin level if/unless a condition is
7987 matched.
7988
7989 The admin level allows to enable/disable servers from the web interface. By
7990 default, statistics page is read-only for security reasons.
7991
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01007992 Note : Consider not using this feature in multi-process mode (nbproc > 1)
7993 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007994 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01007995
Cyril Bonté23b39d92011-02-10 22:54:44 +01007996 Currently, the POST request is limited to the buffer size minus the reserved
7997 buffer space, which means that if the list of servers is too long, the
7998 request won't be processed. It is recommended to alter few servers at a
7999 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008000
8001 Example :
8002 # statistics admin level only for localhost
8003 backend stats_localhost
8004 stats enable
8005 stats admin if LOCALHOST
8006
8007 Example :
8008 # statistics admin level always enabled because of the authentication
8009 backend stats_auth
8010 stats enable
8011 stats auth admin:AdMiN123
8012 stats admin if TRUE
8013
8014 Example :
8015 # statistics admin level depends on the authenticated user
8016 userlist stats-auth
8017 group admin users admin
8018 user admin insecure-password AdMiN123
8019 group readonly users haproxy
8020 user haproxy insecure-password haproxy
8021
8022 backend stats_auth
8023 stats enable
8024 acl AUTH http_auth(stats-auth)
8025 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8026 stats http-request auth unless AUTH
8027 stats admin if AUTH_ADMIN
8028
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008029 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8030 "bind-process", section 3.4 about userlists and section 7 about
8031 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008032
8033
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008034stats auth <user>:<passwd>
8035 Enable statistics with authentication and grant access to an account
8036 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008037 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008038 Arguments :
8039 <user> is a user name to grant access to
8040
8041 <passwd> is the cleartext password associated to this user
8042
8043 This statement enables statistics with default settings, and restricts access
8044 to declared users only. It may be repeated as many times as necessary to
8045 allow as many users as desired. When a user tries to access the statistics
8046 without a valid account, a "401 Forbidden" response will be returned so that
8047 the browser asks the user to provide a valid user and password. The real
8048 which will be returned to the browser is configurable using "stats realm".
8049
8050 Since the authentication method is HTTP Basic Authentication, the passwords
8051 circulate in cleartext on the network. Thus, it was decided that the
8052 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008053 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008054
8055 It is also possible to reduce the scope of the proxies which appear in the
8056 report using "stats scope".
8057
8058 Though this statement alone is enough to enable statistics reporting, it is
8059 recommended to set all other settings in order to avoid relying on default
8060 unobvious parameters.
8061
8062 Example :
8063 # public access (limited to this backend only)
8064 backend public_www
8065 server srv1 192.168.0.1:80
8066 stats enable
8067 stats hide-version
8068 stats scope .
8069 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008070 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008071 stats auth admin1:AdMiN123
8072 stats auth admin2:AdMiN321
8073
8074 # internal monitoring access (unlimited)
8075 backend private_monitoring
8076 stats enable
8077 stats uri /admin?stats
8078 stats refresh 5s
8079
8080 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8081
8082
8083stats enable
8084 Enable statistics reporting with default settings
8085 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008086 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008087 Arguments : none
8088
8089 This statement enables statistics reporting with default settings defined
8090 at build time. Unless stated otherwise, these settings are used :
8091 - stats uri : /haproxy?stats
8092 - stats realm : "HAProxy Statistics"
8093 - stats auth : no authentication
8094 - stats scope : no restriction
8095
8096 Though this statement alone is enough to enable statistics reporting, it is
8097 recommended to set all other settings in order to avoid relying on default
8098 unobvious parameters.
8099
8100 Example :
8101 # public access (limited to this backend only)
8102 backend public_www
8103 server srv1 192.168.0.1:80
8104 stats enable
8105 stats hide-version
8106 stats scope .
8107 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008108 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008109 stats auth admin1:AdMiN123
8110 stats auth admin2:AdMiN321
8111
8112 # internal monitoring access (unlimited)
8113 backend private_monitoring
8114 stats enable
8115 stats uri /admin?stats
8116 stats refresh 5s
8117
8118 See also : "stats auth", "stats realm", "stats uri"
8119
8120
Willy Tarreaud63335a2010-02-26 12:56:52 +01008121stats hide-version
8122 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008123 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008124 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008125 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008126
Willy Tarreaud63335a2010-02-26 12:56:52 +01008127 By default, the stats page reports some useful status information along with
8128 the statistics. Among them is HAProxy's version. However, it is generally
8129 considered dangerous to report precise version to anyone, as it can help them
8130 target known weaknesses with specific attacks. The "stats hide-version"
8131 statement removes the version from the statistics report. This is recommended
8132 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008133
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008134 Though this statement alone is enough to enable statistics reporting, it is
8135 recommended to set all other settings in order to avoid relying on default
8136 unobvious parameters.
8137
Willy Tarreaud63335a2010-02-26 12:56:52 +01008138 Example :
8139 # public access (limited to this backend only)
8140 backend public_www
8141 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008142 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008143 stats hide-version
8144 stats scope .
8145 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008146 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008147 stats auth admin1:AdMiN123
8148 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008149
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008150 # internal monitoring access (unlimited)
8151 backend private_monitoring
8152 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008153 stats uri /admin?stats
8154 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008155
Willy Tarreaud63335a2010-02-26 12:56:52 +01008156 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008157
Willy Tarreau983e01e2010-01-11 18:42:06 +01008158
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008159stats http-request { allow | deny | auth [realm <realm>] }
8160 [ { if | unless } <condition> ]
8161 Access control for statistics
8162
8163 May be used in sections: defaults | frontend | listen | backend
8164 no | no | yes | yes
8165
8166 As "http-request", these set of options allow to fine control access to
8167 statistics. Each option may be followed by if/unless and acl.
8168 First option with matched condition (or option without condition) is final.
8169 For "deny" a 403 error will be returned, for "allow" normal processing is
8170 performed, for "auth" a 401/407 error code is returned so the client
8171 should be asked to enter a username and password.
8172
8173 There is no fixed limit to the number of http-request statements per
8174 instance.
8175
8176 See also : "http-request", section 3.4 about userlists and section 7
8177 about ACL usage.
8178
8179
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008180stats realm <realm>
8181 Enable statistics and set authentication realm
8182 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008183 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008184 Arguments :
8185 <realm> is the name of the HTTP Basic Authentication realm reported to
8186 the browser. The browser uses it to display it in the pop-up
8187 inviting the user to enter a valid username and password.
8188
8189 The realm is read as a single word, so any spaces in it should be escaped
8190 using a backslash ('\').
8191
8192 This statement is useful only in conjunction with "stats auth" since it is
8193 only related to authentication.
8194
8195 Though this statement alone is enough to enable statistics reporting, it is
8196 recommended to set all other settings in order to avoid relying on default
8197 unobvious parameters.
8198
8199 Example :
8200 # public access (limited to this backend only)
8201 backend public_www
8202 server srv1 192.168.0.1:80
8203 stats enable
8204 stats hide-version
8205 stats scope .
8206 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008207 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008208 stats auth admin1:AdMiN123
8209 stats auth admin2:AdMiN321
8210
8211 # internal monitoring access (unlimited)
8212 backend private_monitoring
8213 stats enable
8214 stats uri /admin?stats
8215 stats refresh 5s
8216
8217 See also : "stats auth", "stats enable", "stats uri"
8218
8219
8220stats refresh <delay>
8221 Enable statistics with automatic refresh
8222 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008223 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008224 Arguments :
8225 <delay> is the suggested refresh delay, specified in seconds, which will
8226 be returned to the browser consulting the report page. While the
8227 browser is free to apply any delay, it will generally respect it
8228 and refresh the page this every seconds. The refresh interval may
8229 be specified in any other non-default time unit, by suffixing the
8230 unit after the value, as explained at the top of this document.
8231
8232 This statement is useful on monitoring displays with a permanent page
8233 reporting the load balancer's activity. When set, the HTML report page will
8234 include a link "refresh"/"stop refresh" so that the user can select whether
8235 he wants automatic refresh of the page or not.
8236
8237 Though this statement alone is enough to enable statistics reporting, it is
8238 recommended to set all other settings in order to avoid relying on default
8239 unobvious parameters.
8240
8241 Example :
8242 # public access (limited to this backend only)
8243 backend public_www
8244 server srv1 192.168.0.1:80
8245 stats enable
8246 stats hide-version
8247 stats scope .
8248 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008249 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008250 stats auth admin1:AdMiN123
8251 stats auth admin2:AdMiN321
8252
8253 # internal monitoring access (unlimited)
8254 backend private_monitoring
8255 stats enable
8256 stats uri /admin?stats
8257 stats refresh 5s
8258
8259 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8260
8261
8262stats scope { <name> | "." }
8263 Enable statistics and limit access scope
8264 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008265 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008266 Arguments :
8267 <name> is the name of a listen, frontend or backend section to be
8268 reported. The special name "." (a single dot) designates the
8269 section in which the statement appears.
8270
8271 When this statement is specified, only the sections enumerated with this
8272 statement will appear in the report. All other ones will be hidden. This
8273 statement may appear as many times as needed if multiple sections need to be
8274 reported. Please note that the name checking is performed as simple string
8275 comparisons, and that it is never checked that a give section name really
8276 exists.
8277
8278 Though this statement alone is enough to enable statistics reporting, it is
8279 recommended to set all other settings in order to avoid relying on default
8280 unobvious parameters.
8281
8282 Example :
8283 # public access (limited to this backend only)
8284 backend public_www
8285 server srv1 192.168.0.1:80
8286 stats enable
8287 stats hide-version
8288 stats scope .
8289 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008290 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008291 stats auth admin1:AdMiN123
8292 stats auth admin2:AdMiN321
8293
8294 # internal monitoring access (unlimited)
8295 backend private_monitoring
8296 stats enable
8297 stats uri /admin?stats
8298 stats refresh 5s
8299
8300 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8301
Willy Tarreaud63335a2010-02-26 12:56:52 +01008302
Willy Tarreauc9705a12010-07-27 20:05:50 +02008303stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008304 Enable reporting of a description on the statistics page.
8305 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008306 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008307
Willy Tarreauc9705a12010-07-27 20:05:50 +02008308 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008309 description from global section is automatically used instead.
8310
8311 This statement is useful for users that offer shared services to their
8312 customers, where node or description should be different for each customer.
8313
8314 Though this statement alone is enough to enable statistics reporting, it is
8315 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008316 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008317
8318 Example :
8319 # internal monitoring access (unlimited)
8320 backend private_monitoring
8321 stats enable
8322 stats show-desc Master node for Europe, Asia, Africa
8323 stats uri /admin?stats
8324 stats refresh 5s
8325
8326 See also: "show-node", "stats enable", "stats uri" and "description" in
8327 global section.
8328
8329
8330stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008331 Enable reporting additional information on the statistics page
8332 May be used in sections : defaults | frontend | listen | backend
8333 yes | yes | yes | yes
8334 Arguments : none
8335
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008336 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008337 - cap: capabilities (proxy)
8338 - mode: one of tcp, http or health (proxy)
8339 - id: SNMP ID (proxy, socket, server)
8340 - IP (socket, server)
8341 - cookie (backend, server)
8342
8343 Though this statement alone is enough to enable statistics reporting, it is
8344 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008345 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008346
8347 See also: "stats enable", "stats uri".
8348
8349
8350stats show-node [ <name> ]
8351 Enable reporting of a host name on the statistics page.
8352 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008353 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008354 Arguments:
8355 <name> is an optional name to be reported. If unspecified, the
8356 node name from global section is automatically used instead.
8357
8358 This statement is useful for users that offer shared services to their
8359 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008360 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008361
8362 Though this statement alone is enough to enable statistics reporting, it is
8363 recommended to set all other settings in order to avoid relying on default
8364 unobvious parameters.
8365
8366 Example:
8367 # internal monitoring access (unlimited)
8368 backend private_monitoring
8369 stats enable
8370 stats show-node Europe-1
8371 stats uri /admin?stats
8372 stats refresh 5s
8373
8374 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8375 section.
8376
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008377
8378stats uri <prefix>
8379 Enable statistics and define the URI prefix to access them
8380 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008381 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008382 Arguments :
8383 <prefix> is the prefix of any URI which will be redirected to stats. This
8384 prefix may contain a question mark ('?') to indicate part of a
8385 query string.
8386
8387 The statistics URI is intercepted on the relayed traffic, so it appears as a
8388 page within the normal application. It is strongly advised to ensure that the
8389 selected URI will never appear in the application, otherwise it will never be
8390 possible to reach it in the application.
8391
8392 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008393 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008394 It is generally a good idea to include a question mark in the URI so that
8395 intermediate proxies refrain from caching the results. Also, since any string
8396 beginning with the prefix will be accepted as a stats request, the question
8397 mark helps ensuring that no valid URI will begin with the same words.
8398
8399 It is sometimes very convenient to use "/" as the URI prefix, and put that
8400 statement in a "listen" instance of its own. That makes it easy to dedicate
8401 an address or a port to statistics only.
8402
8403 Though this statement alone is enough to enable statistics reporting, it is
8404 recommended to set all other settings in order to avoid relying on default
8405 unobvious parameters.
8406
8407 Example :
8408 # public access (limited to this backend only)
8409 backend public_www
8410 server srv1 192.168.0.1:80
8411 stats enable
8412 stats hide-version
8413 stats scope .
8414 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008415 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008416 stats auth admin1:AdMiN123
8417 stats auth admin2:AdMiN321
8418
8419 # internal monitoring access (unlimited)
8420 backend private_monitoring
8421 stats enable
8422 stats uri /admin?stats
8423 stats refresh 5s
8424
8425 See also : "stats auth", "stats enable", "stats realm"
8426
8427
Willy Tarreaud63335a2010-02-26 12:56:52 +01008428stick match <pattern> [table <table>] [{if | unless} <cond>]
8429 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008430 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008431 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008432
8433 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008434 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008435 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008436 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008437 stickiness table. This rule is mandatory.
8438
8439 <table> is an optional stickiness table name. If unspecified, the same
8440 backend's table is used. A stickiness table is declared using
8441 the "stick-table" statement.
8442
8443 <cond> is an optional matching condition. It makes it possible to match
8444 on a certain criterion only when other conditions are met (or
8445 not met). For instance, it could be used to match on a source IP
8446 address except when a request passes through a known proxy, in
8447 which case we'd match on a header containing that IP address.
8448
8449 Some protocols or applications require complex stickiness rules and cannot
8450 always simply rely on cookies nor hashing. The "stick match" statement
8451 describes a rule to extract the stickiness criterion from an incoming request
8452 or connection. See section 7 for a complete list of possible patterns and
8453 transformation rules.
8454
8455 The table has to be declared using the "stick-table" statement. It must be of
8456 a type compatible with the pattern. By default it is the one which is present
8457 in the same backend. It is possible to share a table with other backends by
8458 referencing it using the "table" keyword. If another table is referenced,
8459 the server's ID inside the backends are used. By default, all server IDs
8460 start at 1 in each backend, so the server ordering is enough. But in case of
8461 doubt, it is highly recommended to force server IDs using their "id" setting.
8462
8463 It is possible to restrict the conditions where a "stick match" statement
8464 will apply, using "if" or "unless" followed by a condition. See section 7 for
8465 ACL based conditions.
8466
8467 There is no limit on the number of "stick match" statements. The first that
8468 applies and matches will cause the request to be directed to the same server
8469 as was used for the request which created the entry. That way, multiple
8470 matches can be used as fallbacks.
8471
8472 The stick rules are checked after the persistence cookies, so they will not
8473 affect stickiness if a cookie has already been used to select a server. That
8474 way, it becomes very easy to insert cookies and match on IP addresses in
8475 order to maintain stickiness between HTTP and HTTPS.
8476
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008477 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8478 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008479 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008480
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008481 Example :
8482 # forward SMTP users to the same server they just used for POP in the
8483 # last 30 minutes
8484 backend pop
8485 mode tcp
8486 balance roundrobin
8487 stick store-request src
8488 stick-table type ip size 200k expire 30m
8489 server s1 192.168.1.1:110
8490 server s2 192.168.1.1:110
8491
8492 backend smtp
8493 mode tcp
8494 balance roundrobin
8495 stick match src table pop
8496 server s1 192.168.1.1:25
8497 server s2 192.168.1.1:25
8498
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008499 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008500 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008501
8502
8503stick on <pattern> [table <table>] [{if | unless} <condition>]
8504 Define a request pattern to associate a user to a server
8505 May be used in sections : defaults | frontend | listen | backend
8506 no | no | yes | yes
8507
8508 Note : This form is exactly equivalent to "stick match" followed by
8509 "stick store-request", all with the same arguments. Please refer
8510 to both keywords for details. It is only provided as a convenience
8511 for writing more maintainable configurations.
8512
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008513 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8514 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008515 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008516
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008517 Examples :
8518 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008519 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008520
8521 # ...is strictly equivalent to this one :
8522 stick match src table pop if !localhost
8523 stick store-request src table pop if !localhost
8524
8525
8526 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8527 # well as HTTP without cookie. Share the same table between both accesses.
8528 backend http
8529 mode http
8530 balance roundrobin
8531 stick on src table https
8532 cookie SRV insert indirect nocache
8533 server s1 192.168.1.1:80 cookie s1
8534 server s2 192.168.1.1:80 cookie s2
8535
8536 backend https
8537 mode tcp
8538 balance roundrobin
8539 stick-table type ip size 200k expire 30m
8540 stick on src
8541 server s1 192.168.1.1:443
8542 server s2 192.168.1.1:443
8543
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008544 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008545
8546
8547stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8548 Define a request pattern used to create an entry in a stickiness table
8549 May be used in sections : defaults | frontend | listen | backend
8550 no | no | yes | yes
8551
8552 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008553 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008554 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008555 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008556 server is selected.
8557
8558 <table> is an optional stickiness table name. If unspecified, the same
8559 backend's table is used. A stickiness table is declared using
8560 the "stick-table" statement.
8561
8562 <cond> is an optional storage condition. It makes it possible to store
8563 certain criteria only when some conditions are met (or not met).
8564 For instance, it could be used to store the source IP address
8565 except when the request passes through a known proxy, in which
8566 case we'd store a converted form of a header containing that IP
8567 address.
8568
8569 Some protocols or applications require complex stickiness rules and cannot
8570 always simply rely on cookies nor hashing. The "stick store-request" statement
8571 describes a rule to decide what to extract from the request and when to do
8572 it, in order to store it into a stickiness table for further requests to
8573 match it using the "stick match" statement. Obviously the extracted part must
8574 make sense and have a chance to be matched in a further request. Storing a
8575 client's IP address for instance often makes sense. Storing an ID found in a
8576 URL parameter also makes sense. Storing a source port will almost never make
8577 any sense because it will be randomly matched. See section 7 for a complete
8578 list of possible patterns and transformation rules.
8579
8580 The table has to be declared using the "stick-table" statement. It must be of
8581 a type compatible with the pattern. By default it is the one which is present
8582 in the same backend. It is possible to share a table with other backends by
8583 referencing it using the "table" keyword. If another table is referenced,
8584 the server's ID inside the backends are used. By default, all server IDs
8585 start at 1 in each backend, so the server ordering is enough. But in case of
8586 doubt, it is highly recommended to force server IDs using their "id" setting.
8587
8588 It is possible to restrict the conditions where a "stick store-request"
8589 statement will apply, using "if" or "unless" followed by a condition. This
8590 condition will be evaluated while parsing the request, so any criteria can be
8591 used. See section 7 for ACL based conditions.
8592
8593 There is no limit on the number of "stick store-request" statements, but
8594 there is a limit of 8 simultaneous stores per request or response. This
8595 makes it possible to store up to 8 criteria, all extracted from either the
8596 request or the response, regardless of the number of rules. Only the 8 first
8597 ones which match will be kept. Using this, it is possible to feed multiple
8598 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008599 another protocol or access method. Using multiple store-request rules with
8600 the same table is possible and may be used to find the best criterion to rely
8601 on, by arranging the rules by decreasing preference order. Only the first
8602 extracted criterion for a given table will be stored. All subsequent store-
8603 request rules referencing the same table will be skipped and their ACLs will
8604 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008605
8606 The "store-request" rules are evaluated once the server connection has been
8607 established, so that the table will contain the real server that processed
8608 the request.
8609
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008610 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8611 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008612 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008613
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008614 Example :
8615 # forward SMTP users to the same server they just used for POP in the
8616 # last 30 minutes
8617 backend pop
8618 mode tcp
8619 balance roundrobin
8620 stick store-request src
8621 stick-table type ip size 200k expire 30m
8622 server s1 192.168.1.1:110
8623 server s2 192.168.1.1:110
8624
8625 backend smtp
8626 mode tcp
8627 balance roundrobin
8628 stick match src table pop
8629 server s1 192.168.1.1:25
8630 server s2 192.168.1.1:25
8631
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008632 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008633 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008634
8635
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008636stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008637 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8638 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008639 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008640 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008641 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008642
8643 Arguments :
8644 ip a table declared with "type ip" will only store IPv4 addresses.
8645 This form is very compact (about 50 bytes per entry) and allows
8646 very fast entry lookup and stores with almost no overhead. This
8647 is mainly used to store client source IP addresses.
8648
David du Colombier9a6d3c92011-03-17 10:40:24 +01008649 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8650 This form is very compact (about 60 bytes per entry) and allows
8651 very fast entry lookup and stores with almost no overhead. This
8652 is mainly used to store client source IP addresses.
8653
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008654 integer a table declared with "type integer" will store 32bit integers
8655 which can represent a client identifier found in a request for
8656 instance.
8657
8658 string a table declared with "type string" will store substrings of up
8659 to <len> characters. If the string provided by the pattern
8660 extractor is larger than <len>, it will be truncated before
8661 being stored. During matching, at most <len> characters will be
8662 compared between the string in the table and the extracted
8663 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008664 to 32 characters.
8665
8666 binary a table declared with "type binary" will store binary blocks
8667 of <len> bytes. If the block provided by the pattern
8668 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008669 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008670 is shorter than <len>, it will be padded by 0. When not
8671 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008672
8673 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008674 "string" type table (See type "string" above). Or the number
8675 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008676 changing this parameter as memory usage will proportionally
8677 increase.
8678
8679 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008680 value directly impacts memory usage. Count approximately
8681 50 bytes per entry, plus the size of a string if any. The size
8682 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008683
8684 [nopurge] indicates that we refuse to purge older entries when the table
8685 is full. When not specified and the table is full when haproxy
8686 wants to store an entry in it, it will flush a few of the oldest
8687 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008688 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008689 be desirable to refuse new entries instead of purging the older
8690 ones. That may be the case when the amount of data to store is
8691 far above the hardware limits and we prefer not to offer access
8692 to new clients than to reject the ones already connected. When
8693 using this parameter, be sure to properly set the "expire"
8694 parameter (see below).
8695
Emeric Brunf099e792010-09-27 12:05:28 +02008696 <peersect> is the name of the peers section to use for replication. Entries
8697 which associate keys to server IDs are kept synchronized with
8698 the remote peers declared in this section. All entries are also
8699 automatically learned from the local peer (old process) during a
8700 soft restart.
8701
Willy Tarreau1abc6732015-05-01 19:21:02 +02008702 NOTE : each peers section may be referenced only by tables
8703 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008704
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008705 <expire> defines the maximum duration of an entry in the table since it
8706 was last created, refreshed or matched. The expiration delay is
8707 defined using the standard time format, similarly as the various
8708 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008709 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008710 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008711 be removed once full. Be sure not to use the "nopurge" parameter
8712 if not expiration delay is specified.
8713
Willy Tarreau08d5f982010-06-06 13:34:54 +02008714 <data_type> is used to store additional information in the stick-table. This
8715 may be used by ACLs in order to control various criteria related
8716 to the activity of the client matching the stick-table. For each
8717 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008718 that the additional data can fit. Several data types may be
8719 stored with an entry. Multiple data types may be specified after
8720 the "store" keyword, as a comma-separated list. Alternatively,
8721 it is possible to repeat the "store" keyword followed by one or
8722 several data types. Except for the "server_id" type which is
8723 automatically detected and enabled, all data types must be
8724 explicitly declared to be stored. If an ACL references a data
8725 type which is not stored, the ACL will simply not match. Some
8726 data types require an argument which must be passed just after
8727 the type between parenthesis. See below for the supported data
8728 types and their arguments.
8729
8730 The data types that can be stored with an entry are the following :
8731 - server_id : this is an integer which holds the numeric ID of the server a
8732 request was assigned to. It is used by the "stick match", "stick store",
8733 and "stick on" rules. It is automatically enabled when referenced.
8734
8735 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
8736 integer which may be used for anything. Most of the time it will be used
8737 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008738 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02008739
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008740 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
8741 over a period. It is a positive 32-bit integer integer which may be used
8742 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01008743 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008744 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01008745 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02008746
Frédéric Lécaille6778b272018-01-29 15:22:53 +01008747 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
8748 integer which may be used for anything. Most of the time it will be used
8749 to put a special tag on some entries, for instance to note that a
8750 specific behavior was detected and must be known for future matches.
8751
8752 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
8753 over a period. It is a positive 32-bit integer integer which may be used
8754 for anything. Just like <gpc1>, it counts events, but instead of keeping
8755 a cumulative number, it maintains the rate at which the counter is
8756 incremented. Most of the time it will be used to measure the frequency of
8757 occurrence of certain events (e.g. requests to a specific URL).
8758
Willy Tarreauc9705a12010-07-27 20:05:50 +02008759 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
8760 the absolute number of connections received from clients which matched
8761 this entry. It does not mean the connections were accepted, just that
8762 they were received.
8763
8764 - conn_cur : Current Connections. It is a positive 32-bit integer which
8765 stores the concurrent connection counts for the entry. It is incremented
8766 once an incoming connection matches the entry, and decremented once the
8767 connection leaves. That way it is possible to know at any time the exact
8768 number of concurrent connections for an entry.
8769
8770 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8771 integer parameter <period> which indicates in milliseconds the length
8772 of the period over which the average is measured. It reports the average
8773 incoming connection rate over that period, in connections per period. The
8774 result is an integer which can be matched using ACLs.
8775
8776 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
8777 the absolute number of sessions received from clients which matched this
8778 entry. A session is a connection that was accepted by the layer 4 rules.
8779
8780 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8781 integer parameter <period> which indicates in milliseconds the length
8782 of the period over which the average is measured. It reports the average
8783 incoming session rate over that period, in sessions per period. The
8784 result is an integer which can be matched using ACLs.
8785
8786 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
8787 counts the absolute number of HTTP requests received from clients which
8788 matched this entry. It does not matter whether they are valid requests or
8789 not. Note that this is different from sessions when keep-alive is used on
8790 the client side.
8791
8792 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8793 integer parameter <period> which indicates in milliseconds the length
8794 of the period over which the average is measured. It reports the average
8795 HTTP request rate over that period, in requests per period. The result is
8796 an integer which can be matched using ACLs. It does not matter whether
8797 they are valid requests or not. Note that this is different from sessions
8798 when keep-alive is used on the client side.
8799
8800 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
8801 counts the absolute number of HTTP requests errors induced by clients
8802 which matched this entry. Errors are counted on invalid and truncated
8803 requests, as well as on denied or tarpitted requests, and on failed
8804 authentications. If the server responds with 4xx, then the request is
8805 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01008806 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02008807
8808 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8809 integer parameter <period> which indicates in milliseconds the length
8810 of the period over which the average is measured. It reports the average
8811 HTTP request error rate over that period, in requests per period (see
8812 http_err_cnt above for what is accounted as an error). The result is an
8813 integer which can be matched using ACLs.
8814
8815 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008816 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02008817 which matched this entry. Headers are included in the count. This may be
8818 used to limit abuse of upload features on photo or video servers.
8819
8820 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
8821 integer parameter <period> which indicates in milliseconds the length
8822 of the period over which the average is measured. It reports the average
8823 incoming bytes rate over that period, in bytes per period. It may be used
8824 to detect users which upload too much and too fast. Warning: with large
8825 uploads, it is possible that the amount of uploaded data will be counted
8826 once upon termination, thus causing spikes in the average transfer speed
8827 instead of having a smooth one. This may partially be smoothed with
8828 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
8829 recommended for better fairness.
8830
8831 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01008832 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02008833 matched this entry. Headers are included in the count. This may be used
8834 to limit abuse of bots sucking the whole site.
8835
8836 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
8837 an integer parameter <period> which indicates in milliseconds the length
8838 of the period over which the average is measured. It reports the average
8839 outgoing bytes rate over that period, in bytes per period. It may be used
8840 to detect users which download too much and too fast. Warning: with large
8841 transfers, it is possible that the amount of transferred data will be
8842 counted once upon termination, thus causing spikes in the average
8843 transfer speed instead of having a smooth one. This may partially be
8844 smoothed with "option contstats" though this is not perfect yet. Use of
8845 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02008846
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008847 There is only one stick-table per proxy. At the moment of writing this doc,
8848 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008849 to be required, simply create a dummy backend with a stick-table in it and
8850 reference it.
8851
8852 It is important to understand that stickiness based on learning information
8853 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01008854 lost upon restart unless peers are properly configured to transfer such
8855 information upon restart (recommended). In general it can be good as a
8856 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008857
Willy Tarreauc9705a12010-07-27 20:05:50 +02008858 Last, memory requirements may be important when storing many data types.
8859 Indeed, storing all indicators above at once in each entry requires 116 bytes
8860 per entry, or 116 MB for a 1-million entries table. This is definitely not
8861 something that can be ignored.
8862
8863 Example:
8864 # Keep track of counters of up to 1 million IP addresses over 5 minutes
8865 # and store a general purpose counter and the average connection rate
8866 # computed over a sliding window of 30 seconds.
8867 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
8868
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008869 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01008870 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008871
8872
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008873stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01008874 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008875 May be used in sections : defaults | frontend | listen | backend
8876 no | no | yes | yes
8877
8878 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008879 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008880 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008881 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008882 server is selected.
8883
8884 <table> is an optional stickiness table name. If unspecified, the same
8885 backend's table is used. A stickiness table is declared using
8886 the "stick-table" statement.
8887
8888 <cond> is an optional storage condition. It makes it possible to store
8889 certain criteria only when some conditions are met (or not met).
8890 For instance, it could be used to store the SSL session ID only
8891 when the response is a SSL server hello.
8892
8893 Some protocols or applications require complex stickiness rules and cannot
8894 always simply rely on cookies nor hashing. The "stick store-response"
8895 statement describes a rule to decide what to extract from the response and
8896 when to do it, in order to store it into a stickiness table for further
8897 requests to match it using the "stick match" statement. Obviously the
8898 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008899 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008900 See section 7 for a complete list of possible patterns and transformation
8901 rules.
8902
8903 The table has to be declared using the "stick-table" statement. It must be of
8904 a type compatible with the pattern. By default it is the one which is present
8905 in the same backend. It is possible to share a table with other backends by
8906 referencing it using the "table" keyword. If another table is referenced,
8907 the server's ID inside the backends are used. By default, all server IDs
8908 start at 1 in each backend, so the server ordering is enough. But in case of
8909 doubt, it is highly recommended to force server IDs using their "id" setting.
8910
8911 It is possible to restrict the conditions where a "stick store-response"
8912 statement will apply, using "if" or "unless" followed by a condition. This
8913 condition will be evaluated while parsing the response, so any criteria can
8914 be used. See section 7 for ACL based conditions.
8915
8916 There is no limit on the number of "stick store-response" statements, but
8917 there is a limit of 8 simultaneous stores per request or response. This
8918 makes it possible to store up to 8 criteria, all extracted from either the
8919 request or the response, regardless of the number of rules. Only the 8 first
8920 ones which match will be kept. Using this, it is possible to feed multiple
8921 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008922 another protocol or access method. Using multiple store-response rules with
8923 the same table is possible and may be used to find the best criterion to rely
8924 on, by arranging the rules by decreasing preference order. Only the first
8925 extracted criterion for a given table will be stored. All subsequent store-
8926 response rules referencing the same table will be skipped and their ACLs will
8927 not be evaluated. However, even if a store-request rule references a table, a
8928 store-response rule may also use the same table. This means that each table
8929 may learn exactly one element from the request and one element from the
8930 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008931
8932 The table will contain the real server that processed the request.
8933
8934 Example :
8935 # Learn SSL session ID from both request and response and create affinity.
8936 backend https
8937 mode tcp
8938 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02008939 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008940 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008941
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008942 acl clienthello req_ssl_hello_type 1
8943 acl serverhello rep_ssl_hello_type 2
8944
8945 # use tcp content accepts to detects ssl client and server hello.
8946 tcp-request inspect-delay 5s
8947 tcp-request content accept if clienthello
8948
8949 # no timeout on response inspect delay by default.
8950 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008951
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008952 # SSL session ID (SSLID) may be present on a client or server hello.
8953 # Its length is coded on 1 byte at offset 43 and its value starts
8954 # at offset 44.
8955
8956 # Match and learn on request if client hello.
8957 stick on payload_lv(43,1) if clienthello
8958
8959 # Learn on response if server hello.
8960 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02008961
Emeric Brun6a1cefa2010-09-24 18:15:17 +02008962 server s1 192.168.1.1:443
8963 server s2 192.168.1.1:443
8964
8965 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
8966 extraction.
8967
8968
Willy Tarreau938c7fe2014-04-25 14:21:39 +02008969tcp-check connect [params*]
8970 Opens a new connection
8971 May be used in sections: defaults | frontend | listen | backend
8972 no | no | yes | yes
8973
8974 When an application lies on more than a single TCP port or when HAProxy
8975 load-balance many services in a single backend, it makes sense to probe all
8976 the services individually before considering a server as operational.
8977
8978 When there are no TCP port configured on the server line neither server port
8979 directive, then the 'tcp-check connect port <port>' must be the first step
8980 of the sequence.
8981
8982 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
8983 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
8984 do.
8985
8986 Parameters :
8987 They are optional and can be used to describe how HAProxy should open and
8988 use the TCP connection.
8989
8990 port if not set, check port or server port is used.
8991 It tells HAProxy where to open the connection to.
8992 <port> must be a valid TCP port source integer, from 1 to 65535.
8993
8994 send-proxy send a PROXY protocol string
8995
8996 ssl opens a ciphered connection
8997
8998 Examples:
8999 # check HTTP and HTTPs services on a server.
9000 # first open port 80 thanks to server line port directive, then
9001 # tcp-check opens port 443, ciphered and run a request on it:
9002 option tcp-check
9003 tcp-check connect
9004 tcp-check send GET\ /\ HTTP/1.0\r\n
9005 tcp-check send Host:\ haproxy.1wt.eu\r\n
9006 tcp-check send \r\n
9007 tcp-check expect rstring (2..|3..)
9008 tcp-check connect port 443 ssl
9009 tcp-check send GET\ /\ HTTP/1.0\r\n
9010 tcp-check send Host:\ haproxy.1wt.eu\r\n
9011 tcp-check send \r\n
9012 tcp-check expect rstring (2..|3..)
9013 server www 10.0.0.1 check port 80
9014
9015 # check both POP and IMAP from a single server:
9016 option tcp-check
9017 tcp-check connect port 110
9018 tcp-check expect string +OK\ POP3\ ready
9019 tcp-check connect port 143
9020 tcp-check expect string *\ OK\ IMAP4\ ready
9021 server mail 10.0.0.1 check
9022
9023 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9024
9025
9026tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009027 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009028 May be used in sections: defaults | frontend | listen | backend
9029 no | no | yes | yes
9030
9031 Arguments :
9032 <match> is a keyword indicating how to look for a specific pattern in the
9033 response. The keyword may be one of "string", "rstring" or
9034 binary.
9035 The keyword may be preceded by an exclamation mark ("!") to negate
9036 the match. Spaces are allowed between the exclamation mark and the
9037 keyword. See below for more details on the supported keywords.
9038
9039 <pattern> is the pattern to look for. It may be a string or a regular
9040 expression. If the pattern contains spaces, they must be escaped
9041 with the usual backslash ('\').
9042 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009043 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009044 two digits will represent a byte. The hexadecimal digits may be
9045 used upper or lower case.
9046
9047
9048 The available matches are intentionally similar to their http-check cousins :
9049
9050 string <string> : test the exact string matches in the response buffer.
9051 A health check response will be considered valid if the
9052 response's buffer contains this exact string. If the
9053 "string" keyword is prefixed with "!", then the response
9054 will be considered invalid if the body contains this
9055 string. This can be used to look for a mandatory pattern
9056 in a protocol response, or to detect a failure when a
9057 specific error appears in a protocol banner.
9058
9059 rstring <regex> : test a regular expression on the response buffer.
9060 A health check response will be considered valid if the
9061 response's buffer matches this expression. If the
9062 "rstring" keyword is prefixed with "!", then the response
9063 will be considered invalid if the body matches the
9064 expression.
9065
9066 binary <hexstring> : test the exact string in its hexadecimal form matches
9067 in the response buffer. A health check response will
9068 be considered valid if the response's buffer contains
9069 this exact hexadecimal string.
9070 Purpose is to match data on binary protocols.
9071
9072 It is important to note that the responses will be limited to a certain size
9073 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9074 Thus, too large responses may not contain the mandatory pattern when using
9075 "string", "rstring" or binary. If a large response is absolutely required, it
9076 is possible to change the default max size by setting the global variable.
9077 However, it is worth keeping in mind that parsing very large responses can
9078 waste some CPU cycles, especially when regular expressions are used, and that
9079 it is always better to focus the checks on smaller resources. Also, in its
9080 current state, the check will not find any string nor regex past a null
9081 character in the response. Similarly it is not possible to request matching
9082 the null character.
9083
9084 Examples :
9085 # perform a POP check
9086 option tcp-check
9087 tcp-check expect string +OK\ POP3\ ready
9088
9089 # perform an IMAP check
9090 option tcp-check
9091 tcp-check expect string *\ OK\ IMAP4\ ready
9092
9093 # look for the redis master server
9094 option tcp-check
9095 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009096 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009097 tcp-check send info\ replication\r\n
9098 tcp-check expect string role:master
9099 tcp-check send QUIT\r\n
9100 tcp-check expect string +OK
9101
9102
9103 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9104 "tcp-check send-binary", "http-check expect", tune.chksize
9105
9106
9107tcp-check send <data>
9108 Specify a string to be sent as a question during a generic health check
9109 May be used in sections: defaults | frontend | listen | backend
9110 no | no | yes | yes
9111
9112 <data> : the data to be sent as a question during a generic health check
9113 session. For now, <data> must be a string.
9114
9115 Examples :
9116 # look for the redis master server
9117 option tcp-check
9118 tcp-check send info\ replication\r\n
9119 tcp-check expect string role:master
9120
9121 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9122 "tcp-check send-binary", tune.chksize
9123
9124
Davor Ocelice9ed2812017-12-25 17:49:28 +01009125tcp-check send-binary <hexstring>
9126 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009127 tcp health check
9128 May be used in sections: defaults | frontend | listen | backend
9129 no | no | yes | yes
9130
9131 <data> : the data to be sent as a question during a generic health check
9132 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009133 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009134 response buffer. A health check response will be considered
9135 valid if the response's buffer contains this exact
9136 hexadecimal string.
9137 Purpose is to send binary data to ask on binary protocols.
9138
9139 Examples :
9140 # redis check in binary
9141 option tcp-check
9142 tcp-check send-binary 50494e470d0a # PING\r\n
9143 tcp-check expect binary 2b504F4e47 # +PONG
9144
9145
9146 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9147 "tcp-check send", tune.chksize
9148
9149
Willy Tarreaue9656522010-08-17 15:40:09 +02009150tcp-request connection <action> [{if | unless} <condition>]
9151 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009152 May be used in sections : defaults | frontend | listen | backend
9153 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009154 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009155 <action> defines the action to perform if the condition applies. See
9156 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009157
Willy Tarreaue9656522010-08-17 15:40:09 +02009158 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009159
9160 Immediately after acceptance of a new incoming connection, it is possible to
9161 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009162 or dropped or have its counters tracked. Those conditions cannot make use of
9163 any data contents because the connection has not been read from yet, and the
9164 buffers are not yet allocated. This is used to selectively and very quickly
9165 accept or drop connections from various sources with a very low overhead. If
9166 some contents need to be inspected in order to take the decision, the
9167 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009168
Willy Tarreaue9656522010-08-17 15:40:09 +02009169 The "tcp-request connection" rules are evaluated in their exact declaration
9170 order. If no rule matches or if there is no rule, the default action is to
9171 accept the incoming connection. There is no specific limit to the number of
9172 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009173
Willy Tarreaua9083d02015-05-08 15:27:59 +02009174 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009175 - accept :
9176 accepts the connection if the condition is true (when used with "if")
9177 or false (when used with "unless"). The first such rule executed ends
9178 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009179
Willy Tarreaue9656522010-08-17 15:40:09 +02009180 - reject :
9181 rejects the connection if the condition is true (when used with "if")
9182 or false (when used with "unless"). The first such rule executed ends
9183 the rules evaluation. Rejected connections do not even become a
9184 session, which is why they are accounted separately for in the stats,
9185 as "denied connections". They are not considered for the session
9186 rate-limit and are not logged either. The reason is that these rules
9187 should only be used to filter extremely high connection rates such as
9188 the ones encountered during a massive DDoS attack. Under these extreme
9189 conditions, the simple action of logging each event would make the
9190 system collapse and would considerably lower the filtering capacity. If
9191 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009192 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009193
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009194 - expect-proxy layer4 :
9195 configures the client-facing connection to receive a PROXY protocol
9196 header before any byte is read from the socket. This is equivalent to
9197 having the "accept-proxy" keyword on the "bind" line, except that using
9198 the TCP rule allows the PROXY protocol to be accepted only for certain
9199 IP address ranges using an ACL. This is convenient when multiple layers
9200 of load balancers are passed through by traffic coming from public
9201 hosts.
9202
Bertrand Jacquin90759682016-06-06 15:35:39 +01009203 - expect-netscaler-cip layer4 :
9204 configures the client-facing connection to receive a NetScaler Client
9205 IP insertion protocol header before any byte is read from the socket.
9206 This is equivalent to having the "accept-netscaler-cip" keyword on the
9207 "bind" line, except that using the TCP rule allows the PROXY protocol
9208 to be accepted only for certain IP address ranges using an ACL. This
9209 is convenient when multiple layers of load balancers are passed
9210 through by traffic coming from public hosts.
9211
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009212 - capture <sample> len <length> :
9213 This only applies to "tcp-request content" rules. It captures sample
9214 expression <sample> from the request buffer, and converts it to a
9215 string of at most <len> characters. The resulting string is stored into
9216 the next request "capture" slot, so it will possibly appear next to
9217 some captured HTTP headers. It will then automatically appear in the
9218 logs, and it will be possible to extract it using sample fetch rules to
9219 feed it into headers or anything. The length should be limited given
9220 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009221 session life. Please check section 7.3 (Fetching samples) and "capture
9222 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009223
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009224 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009225 enables tracking of sticky counters from current connection. These
Willy Tarreau09448f72014-06-25 18:12:15 +02009226 rules do not stop evaluation and do not change default action. 3 sets
Willy Tarreaue9656522010-08-17 15:40:09 +02009227 of counters may be simultaneously tracked by the same connection. The
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009228 first "track-sc0" rule executed enables tracking of the counters of the
9229 specified table as the first set. The first "track-sc1" rule executed
Willy Tarreaue9656522010-08-17 15:40:09 +02009230 enables tracking of the counters of the specified table as the second
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009231 set. The first "track-sc2" rule executed enables tracking of the
9232 counters of the specified table as the third set. It is a recommended
9233 practice to use the first set of counters for the per-frontend counters
9234 and the second set for the per-backend ones. But this is just a
9235 guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009236
Willy Tarreaue9656522010-08-17 15:40:09 +02009237 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009238 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009239 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009240 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009241 and used to select which table entry to update the counters.
9242 Note that "tcp-request connection" cannot use content-based
9243 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009244
Willy Tarreaue9656522010-08-17 15:40:09 +02009245 <table> is an optional table to be used instead of the default one,
9246 which is the stick-table declared in the current proxy. All
9247 the counters for the matches and updates for the key will
9248 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009249
Willy Tarreaue9656522010-08-17 15:40:09 +02009250 Once a "track-sc*" rule is executed, the key is looked up in the table
9251 and if it is not found, an entry is allocated for it. Then a pointer to
9252 that entry is kept during all the session's life, and this entry's
9253 counters are updated as often as possible, every time the session's
9254 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009255 Counters are only updated for events that happen after the tracking has
9256 been started. For example, connection counters will not be updated when
9257 tracking layer 7 information, since the connection event happens before
9258 layer7 information is extracted.
9259
Willy Tarreaue9656522010-08-17 15:40:09 +02009260 If the entry tracks concurrent connection counters, one connection is
9261 counted for as long as the entry is tracked, and the entry will not
9262 expire during that time. Tracking counters also provides a performance
9263 advantage over just checking the keys, because only one table lookup is
9264 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009265
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009266 - sc-inc-gpc0(<sc-id>):
9267 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9268 counter designated by <sc-id>. If an error occurs, this action silently
9269 fails and the actions evaluation continues.
9270
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009271 - sc-inc-gpc1(<sc-id>):
9272 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9273 counter designated by <sc-id>. If an error occurs, this action silently
9274 fails and the actions evaluation continues.
9275
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009276 - sc-set-gpt0(<sc-id>) <int>:
9277 This action sets the GPT0 tag according to the sticky counter designated
9278 by <sc-id> and the value of <int>. The expected result is a boolean. If
9279 an error occurs, this action silently fails and the actions evaluation
9280 continues.
9281
William Lallemand2e785f22016-05-25 01:48:42 +02009282 - set-src <expr> :
9283 Is used to set the source IP address to the value of specified
9284 expression. Useful if you want to mask source IP for privacy.
9285 If you want to provide an IP from a HTTP header use "http-request
9286 set-src"
9287
9288 <expr> Is a standard HAProxy expression formed by a sample-fetch
9289 followed by some converters.
9290
9291 Example:
9292
9293 tcp-request connection set-src src,ipmask(24)
9294
Willy Tarreau0c630532016-10-21 17:52:58 +02009295 When possible, set-src preserves the original source port as long as the
9296 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009297
William Lallemand44be6402016-05-25 01:51:35 +02009298 - set-src-port <expr> :
9299 Is used to set the source port address to the value of specified
9300 expression.
9301
9302 <expr> Is a standard HAProxy expression formed by a sample-fetch
9303 followed by some converters.
9304
9305 Example:
9306
9307 tcp-request connection set-src-port int(4000)
9308
Willy Tarreau0c630532016-10-21 17:52:58 +02009309 When possible, set-src-port preserves the original source address as long
9310 as the address family supports a port, otherwise it forces the source
9311 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009312
William Lallemand13e9b0c2016-05-25 02:34:07 +02009313 - set-dst <expr> :
9314 Is used to set the destination IP address to the value of specified
9315 expression. Useful if you want to mask IP for privacy in log.
9316 If you want to provide an IP from a HTTP header use "http-request
9317 set-dst". If you want to connect to the new address/port, use
9318 '0.0.0.0:0' as a server address in the backend.
9319
9320 <expr> Is a standard HAProxy expression formed by a sample-fetch
9321 followed by some converters.
9322
9323 Example:
9324
9325 tcp-request connection set-dst dst,ipmask(24)
9326 tcp-request connection set-dst ipv4(10.0.0.1)
9327
Willy Tarreau0c630532016-10-21 17:52:58 +02009328 When possible, set-dst preserves the original destination port as long as
9329 the address family allows it, otherwise the destination port is set to 0.
9330
William Lallemand13e9b0c2016-05-25 02:34:07 +02009331 - set-dst-port <expr> :
9332 Is used to set the destination port address to the value of specified
9333 expression. If you want to connect to the new address/port, use
9334 '0.0.0.0:0' as a server address in the backend.
9335
9336
9337 <expr> Is a standard HAProxy expression formed by a sample-fetch
9338 followed by some converters.
9339
9340 Example:
9341
9342 tcp-request connection set-dst-port int(4000)
9343
Willy Tarreau0c630532016-10-21 17:52:58 +02009344 When possible, set-dst-port preserves the original destination address as
9345 long as the address family supports a port, otherwise it forces the
9346 destination address to IPv4 "0.0.0.0" before rewriting the port.
9347
Willy Tarreau2d392c22015-08-24 01:43:45 +02009348 - "silent-drop" :
9349 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009350 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009351 to prevent the client from being notified. The effect it then that the
9352 client still sees an established connection while there's none on
9353 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9354 except that it doesn't use any local resource at all on the machine
9355 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009356 slow down stronger attackers. It is important to understand the impact
9357 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009358 client and HAProxy (firewalls, proxies, load balancers) will also keep
9359 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009360 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009361 TCP_REPAIR socket option is used to block the emission of a TCP
9362 reset. On other systems, the socket's TTL is reduced to 1 so that the
9363 TCP reset doesn't pass the first router, though it's still delivered to
9364 local networks. Do not use it unless you fully understand how it works.
9365
Willy Tarreaue9656522010-08-17 15:40:09 +02009366 Note that the "if/unless" condition is optional. If no condition is set on
9367 the action, it is simply performed unconditionally. That can be useful for
9368 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009369
Willy Tarreaue9656522010-08-17 15:40:09 +02009370 Example: accept all connections from white-listed hosts, reject too fast
9371 connection without counting them, and track accepted connections.
9372 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009373
Willy Tarreaue9656522010-08-17 15:40:09 +02009374 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009375 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009376 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009377
Willy Tarreaue9656522010-08-17 15:40:09 +02009378 Example: accept all connections from white-listed hosts, count all other
9379 connections and reject too fast ones. This results in abusive ones
9380 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009381
Willy Tarreaue9656522010-08-17 15:40:09 +02009382 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009383 tcp-request connection track-sc0 src
9384 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009385
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009386 Example: enable the PROXY protocol for traffic coming from all known proxies.
9387
9388 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9389
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009390 See section 7 about ACL usage.
9391
Willy Tarreau4f614292016-10-21 17:49:36 +02009392 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009393
9394
Willy Tarreaue9656522010-08-17 15:40:09 +02009395tcp-request content <action> [{if | unless} <condition>]
9396 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009397 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009398 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009399 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009400 <action> defines the action to perform if the condition applies. See
9401 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009402
Willy Tarreaue9656522010-08-17 15:40:09 +02009403 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009404
Davor Ocelice9ed2812017-12-25 17:49:28 +01009405 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009406 called "TCP content inspection". During this stage, ACL-based rules are
9407 evaluated every time the request contents are updated, until either an
9408 "accept" or a "reject" rule matches, or the TCP request inspection delay
9409 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009410
Willy Tarreaue9656522010-08-17 15:40:09 +02009411 The first difference between these rules and "tcp-request connection" rules
9412 is that "tcp-request content" rules can make use of contents to take a
9413 decision. Most often, these decisions will consider a protocol recognition or
9414 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009415 both frontends and backends. In case of HTTP keep-alive with the client, all
9416 tcp-request content rules are evaluated again, so haproxy keeps a record of
9417 what sticky counters were assigned by a "tcp-request connection" versus a
9418 "tcp-request content" rule, and flushes all the content-related ones after
9419 processing an HTTP request, so that they may be evaluated again by the rules
9420 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009421 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009422 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009423
Willy Tarreaue9656522010-08-17 15:40:09 +02009424 Content-based rules are evaluated in their exact declaration order. If no
9425 rule matches or if there is no rule, the default action is to accept the
9426 contents. There is no specific limit to the number of rules which may be
9427 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009428
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009429 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009430 - accept : the request is accepted
9431 - reject : the request is rejected and the connection is closed
9432 - capture : the specified sample expression is captured
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009433 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009434 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009435 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009436 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009437 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009438 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009439 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009440 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009441
Willy Tarreaue9656522010-08-17 15:40:09 +02009442 They have the same meaning as their counter-parts in "tcp-request connection"
9443 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009444
Willy Tarreauf3338342014-01-28 21:40:28 +01009445 While there is nothing mandatory about it, it is recommended to use the
9446 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9447 content" rules in the frontend, and track-sc2 for "tcp-request content"
9448 rules in the backend, because that makes the configuration more readable
9449 and easier to troubleshoot, but this is just a guideline and all counters
9450 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009451
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009452 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009453 the action, it is simply performed unconditionally. That can be useful for
9454 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009455
Willy Tarreaue9656522010-08-17 15:40:09 +02009456 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009457 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9458 contents of a buffer before extracting the required data. If the buffered
9459 contents do not parse as a valid HTTP message, then the ACL does not match.
9460 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009461 processing, so there is no risk of parsing something differently. In an HTTP
9462 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9463 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009464
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009465 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009466 are present when the rule is processed. The rule processing engine is able to
9467 wait until the inspect delay expires when the data to be tracked is not yet
9468 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009469
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009470 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009471 declared inline. For "tcp-request session" rules, only session-level
9472 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009473
Daniel Schneller0b547052016-03-21 20:46:57 +01009474 <var-name> The name of the variable starts with an indication about
9475 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009476 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009477 "sess" : the variable is shared with the whole session
9478 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009479 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009480 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009481 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009482 "res" : the variable is shared only during response
9483 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009484 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009485 The name may only contain characters 'a-z', 'A-Z', '0-9',
9486 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009487
9488 <expr> Is a standard HAProxy expression formed by a sample-fetch
9489 followed by some converters.
9490
Christopher Faulet85d79c92016-11-09 16:54:56 +01009491 The "unset-var" is used to unset a variable. See above for details about
9492 <var-name>.
9493
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009494 The "send-spoe-group" is used to trigger sending of a group of SPOE
9495 messages. To do so, the SPOE engine used to send messages must be defined, as
9496 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9497 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9498 the SPOE agent name must be used.
9499
9500 <engine-name> The SPOE engine name.
9501
9502 <group-name> The SPOE group name as specified in the engine configuration.
9503
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009504 Example:
9505
9506 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009507 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009508
Willy Tarreau62644772008-07-16 18:36:06 +02009509 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009510 # Accept HTTP requests containing a Host header saying "example.com"
9511 # and reject everything else.
9512 acl is_host_com hdr(Host) -i example.com
9513 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009514 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009515 tcp-request content reject
9516
9517 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009518 # reject SMTP connection if client speaks first
9519 tcp-request inspect-delay 30s
9520 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009521 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009522
9523 # Forward HTTPS connection only if client speaks
9524 tcp-request inspect-delay 30s
9525 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009526 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009527 tcp-request content reject
9528
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009529 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009530 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009531 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009532 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009533 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9534 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009535
9536 Example:
9537 # track request counts per "base" (concatenation of Host+URL)
9538 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009539 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009540
Willy Tarreaue9656522010-08-17 15:40:09 +02009541 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009542 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009543
9544 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009545 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009546 # protecting all our sites
9547 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009548 tcp-request connection track-sc0 src
9549 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009550 ...
9551 use_backend http_dynamic if { path_end .php }
9552
9553 backend http_dynamic
9554 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009555 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009556 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009557 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009558 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009559 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009560 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009561
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009562 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009563
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009564 See also : "tcp-request connection", "tcp-request session",
9565 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009566
9567
9568tcp-request inspect-delay <timeout>
9569 Set the maximum allowed time to wait for data during content inspection
9570 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009571 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009572 Arguments :
9573 <timeout> is the timeout value specified in milliseconds by default, but
9574 can be in any other unit if the number is suffixed by the unit,
9575 as explained at the top of this document.
9576
9577 People using haproxy primarily as a TCP relay are often worried about the
9578 risk of passing any type of protocol to a server without any analysis. In
9579 order to be able to analyze the request contents, we must first withhold
9580 the data then analyze them. This statement simply enables withholding of
9581 data for at most the specified amount of time.
9582
Willy Tarreaufb356202010-08-03 14:02:05 +02009583 TCP content inspection applies very early when a connection reaches a
9584 frontend, then very early when the connection is forwarded to a backend. This
9585 means that a connection may experience a first delay in the frontend and a
9586 second delay in the backend if both have tcp-request rules.
9587
Willy Tarreau62644772008-07-16 18:36:06 +02009588 Note that when performing content inspection, haproxy will evaluate the whole
9589 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009590 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009591 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009592 contents are definitive. If no delay is set, haproxy will not wait at all
9593 and will immediately apply a verdict based on the available information.
9594 Obviously this is unlikely to be very useful and might even be racy, so such
9595 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009596
9597 As soon as a rule matches, the request is released and continues as usual. If
9598 the timeout is reached and no rule matches, the default policy will be to let
9599 it pass through unaffected.
9600
9601 For most protocols, it is enough to set it to a few seconds, as most clients
9602 send the full request immediately upon connection. Add 3 or more seconds to
9603 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009604 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009605 before the server (e.g. SMTP), or to wait for a client to talk before passing
9606 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009607 least the inspection delay, otherwise it will expire first. If the client
9608 closes the connection or if the buffer is full, the delay immediately expires
9609 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009610
Willy Tarreau55165fe2009-05-10 12:02:55 +02009611 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009612 "timeout client".
9613
9614
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009615tcp-response content <action> [{if | unless} <condition>]
9616 Perform an action on a session response depending on a layer 4-7 condition
9617 May be used in sections : defaults | frontend | listen | backend
9618 no | no | yes | yes
9619 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009620 <action> defines the action to perform if the condition applies. See
9621 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009622
9623 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9624
Davor Ocelice9ed2812017-12-25 17:49:28 +01009625 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009626 called "TCP content inspection". During this stage, ACL-based rules are
9627 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009628 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9629 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009630
9631 Most often, these decisions will consider a protocol recognition or validity.
9632
9633 Content-based rules are evaluated in their exact declaration order. If no
9634 rule matches or if there is no rule, the default action is to accept the
9635 contents. There is no specific limit to the number of rules which may be
9636 inserted.
9637
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009638 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009639 - accept :
9640 accepts the response if the condition is true (when used with "if")
9641 or false (when used with "unless"). The first such rule executed ends
9642 the rules evaluation.
9643
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009644 - close :
9645 immediately closes the connection with the server if the condition is
9646 true (when used with "if"), or false (when used with "unless"). The
9647 first such rule executed ends the rules evaluation. The main purpose of
9648 this action is to force a connection to be finished between a client
9649 and a server after an exchange when the application protocol expects
9650 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009651 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009652 protocols.
9653
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009654 - reject :
9655 rejects the response if the condition is true (when used with "if")
9656 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009657 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009658
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009659 - set-var(<var-name>) <expr>
9660 Sets a variable.
9661
Christopher Faulet85d79c92016-11-09 16:54:56 +01009662 - unset-var(<var-name>)
9663 Unsets a variable.
9664
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009665 - sc-inc-gpc0(<sc-id>):
9666 This action increments the GPC0 counter according to the sticky
9667 counter designated by <sc-id>. If an error occurs, this action fails
9668 silently and the actions evaluation continues.
9669
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009670 - sc-inc-gpc1(<sc-id>):
9671 This action increments the GPC1 counter according to the sticky
9672 counter designated by <sc-id>. If an error occurs, this action fails
9673 silently and the actions evaluation continues.
9674
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009675 - sc-set-gpt0(<sc-id>) <int> :
9676 This action sets the GPT0 tag according to the sticky counter designated
9677 by <sc-id> and the value of <int>. The expected result is a boolean. If
9678 an error occurs, this action silently fails and the actions evaluation
9679 continues.
9680
Willy Tarreau2d392c22015-08-24 01:43:45 +02009681 - "silent-drop" :
9682 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009683 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009684 to prevent the client from being notified. The effect it then that the
9685 client still sees an established connection while there's none on
9686 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9687 except that it doesn't use any local resource at all on the machine
9688 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009689 slow down stronger attackers. It is important to understand the impact
9690 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009691 client and HAProxy (firewalls, proxies, load balancers) will also keep
9692 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009693 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009694 TCP_REPAIR socket option is used to block the emission of a TCP
9695 reset. On other systems, the socket's TTL is reduced to 1 so that the
9696 TCP reset doesn't pass the first router, though it's still delivered to
9697 local networks. Do not use it unless you fully understand how it works.
9698
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009699 - send-spoe-group <engine-name> <group-name>
9700 Send a group of SPOE messages.
9701
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009702 Note that the "if/unless" condition is optional. If no condition is set on
9703 the action, it is simply performed unconditionally. That can be useful for
9704 for changing the default action to a reject.
9705
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009706 It is perfectly possible to match layer 7 contents with "tcp-response
9707 content" rules, but then it is important to ensure that a full response has
9708 been buffered, otherwise no contents will match. In order to achieve this,
9709 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009710 period.
9711
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009712 The "set-var" is used to set the content of a variable. The variable is
9713 declared inline.
9714
Daniel Schneller0b547052016-03-21 20:46:57 +01009715 <var-name> The name of the variable starts with an indication about
9716 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009717 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009718 "sess" : the variable is shared with the whole session
9719 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009720 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009721 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009722 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009723 "res" : the variable is shared only during response
9724 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009725 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009726 The name may only contain characters 'a-z', 'A-Z', '0-9',
9727 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009728
9729 <expr> Is a standard HAProxy expression formed by a sample-fetch
9730 followed by some converters.
9731
9732 Example:
9733
9734 tcp-request content set-var(sess.my_var) src
9735
Christopher Faulet85d79c92016-11-09 16:54:56 +01009736 The "unset-var" is used to unset a variable. See above for details about
9737 <var-name>.
9738
9739 Example:
9740
9741 tcp-request content unset-var(sess.my_var)
9742
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009743 The "send-spoe-group" is used to trigger sending of a group of SPOE
9744 messages. To do so, the SPOE engine used to send messages must be defined, as
9745 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9746 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9747 the SPOE agent name must be used.
9748
9749 <engine-name> The SPOE engine name.
9750
9751 <group-name> The SPOE group name as specified in the engine configuration.
9752
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009753 See section 7 about ACL usage.
9754
9755 See also : "tcp-request content", "tcp-response inspect-delay"
9756
9757
Willy Tarreau4f614292016-10-21 17:49:36 +02009758tcp-request session <action> [{if | unless} <condition>]
9759 Perform an action on a validated session depending on a layer 5 condition
9760 May be used in sections : defaults | frontend | listen | backend
9761 no | yes | yes | no
9762 Arguments :
9763 <action> defines the action to perform if the condition applies. See
9764 below.
9765
9766 <condition> is a standard layer5-only ACL-based condition (see section 7).
9767
Davor Ocelice9ed2812017-12-25 17:49:28 +01009768 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +02009769 it is possible to evaluate some conditions to decide whether this session
9770 must be accepted or dropped or have its counters tracked. Those conditions
9771 cannot make use of any data contents because no buffers are allocated yet and
9772 the processing cannot wait at this stage. The main use case it to copy some
9773 early information into variables (since variables are accessible in the
9774 session), or to keep track of some information collected after the handshake,
9775 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +01009776 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +02009777 extracted information can thus be copied to a variable or tracked using
9778 "track-sc" rules. Of course it is also possible to decide to accept/reject as
9779 with other rulesets. Most operations performed here could also be performed
9780 in "tcp-request content" rules, except that in HTTP these rules are evaluated
9781 for each new request, and that might not always be acceptable. For example a
9782 rule might increment a counter on each evaluation. It would also be possible
9783 that a country is resolved by geolocation from the source IP address,
9784 assigned to a session-wide variable, then the source address rewritten from
9785 an HTTP header for all requests. If some contents need to be inspected in
9786 order to take the decision, the "tcp-request content" statements must be used
9787 instead.
9788
9789 The "tcp-request session" rules are evaluated in their exact declaration
9790 order. If no rule matches or if there is no rule, the default action is to
9791 accept the incoming session. There is no specific limit to the number of
9792 rules which may be inserted.
9793
9794 Several types of actions are supported :
9795 - accept : the request is accepted
9796 - reject : the request is rejected and the connection is closed
9797 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
9798 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009799 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009800 - sc-set-gpt0(<sc-id>) <int>
9801 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009802 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +02009803 - silent-drop
9804
9805 These actions have the same meaning as their respective counter-parts in
9806 "tcp-request connection" and "tcp-request content", so please refer to these
9807 sections for a complete description.
9808
9809 Note that the "if/unless" condition is optional. If no condition is set on
9810 the action, it is simply performed unconditionally. That can be useful for
9811 "track-sc*" actions as well as for changing the default action to a reject.
9812
9813 Example: track the original source address by default, or the one advertised
9814 in the PROXY protocol header for connection coming from the local
9815 proxies. The first connection-level rule enables receipt of the
9816 PROXY protocol for these ones, the second rule tracks whatever
9817 address we decide to keep after optional decoding.
9818
9819 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9820 tcp-request session track-sc0 src
9821
9822 Example: accept all sessions from white-listed hosts, reject too fast
9823 sessions without counting them, and track accepted sessions.
9824 This results in session rate being capped from abusive sources.
9825
9826 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9827 tcp-request session reject if { src_sess_rate gt 10 }
9828 tcp-request session track-sc0 src
9829
9830 Example: accept all sessions from white-listed hosts, count all other
9831 sessions and reject too fast ones. This results in abusive ones
9832 being blocked as long as they don't slow down.
9833
9834 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
9835 tcp-request session track-sc0 src
9836 tcp-request session reject if { sc0_sess_rate gt 10 }
9837
9838 See section 7 about ACL usage.
9839
9840 See also : "tcp-request connection", "tcp-request content", "stick-table"
9841
9842
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009843tcp-response inspect-delay <timeout>
9844 Set the maximum allowed time to wait for a response during content inspection
9845 May be used in sections : defaults | frontend | listen | backend
9846 no | no | yes | yes
9847 Arguments :
9848 <timeout> is the timeout value specified in milliseconds by default, but
9849 can be in any other unit if the number is suffixed by the unit,
9850 as explained at the top of this document.
9851
9852 See also : "tcp-response content", "tcp-request inspect-delay".
9853
9854
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009855timeout check <timeout>
9856 Set additional check timeout, but only after a connection has been already
9857 established.
9858
9859 May be used in sections: defaults | frontend | listen | backend
9860 yes | no | yes | yes
9861 Arguments:
9862 <timeout> is the timeout value specified in milliseconds by default, but
9863 can be in any other unit if the number is suffixed by the unit,
9864 as explained at the top of this document.
9865
9866 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
9867 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009868 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009869 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +01009870 (Please also note that there is no valid reason to have such long connect
9871 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
9872 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009873
9874 If "timeout check" is not set haproxy uses "inter" for complete check
9875 timeout (connect + read) exactly like all <1.3.15 version.
9876
9877 In most cases check request is much simpler and faster to handle than normal
9878 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +01009879 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009880
9881 This parameter is specific to backends, but can be specified once for all in
9882 "defaults" sections. This is in fact one of the easiest solutions not to
9883 forget about it.
9884
Willy Tarreau41a340d2008-01-22 12:25:31 +01009885 See also: "timeout connect", "timeout queue", "timeout server",
9886 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009887
9888
Willy Tarreau0ba27502007-12-24 16:55:16 +01009889timeout client <timeout>
9890timeout clitimeout <timeout> (deprecated)
9891 Set the maximum inactivity time on the client side.
9892 May be used in sections : defaults | frontend | listen | backend
9893 yes | yes | yes | no
9894 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009895 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009896 can be in any other unit if the number is suffixed by the unit,
9897 as explained at the top of this document.
9898
9899 The inactivity timeout applies when the client is expected to acknowledge or
9900 send data. In HTTP mode, this timeout is particularly important to consider
9901 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +01009902 response while it is reading data sent by the server. That said, for the
9903 first phase, it is preferable to set the "timeout http-request" to better
9904 protect HAProxy from Slowloris like attacks. The value is specified in
9905 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +01009906 suffixed by the unit, as specified at the top of this document. In TCP mode
9907 (and to a lesser extent, in HTTP mode), it is highly recommended that the
9908 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009909 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01009910 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01009911 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
9912 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +02009913 which overrides "timeout client" and "timeout server" for tunnels, as well as
9914 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009915
9916 This parameter is specific to frontends, but can be specified once for all in
9917 "defaults" sections. This is in fact one of the easiest solutions not to
9918 forget about it. An unspecified timeout results in an infinite timeout, which
9919 is not recommended. Such a usage is accepted and works but reports a warning
9920 during startup because it may results in accumulation of expired sessions in
9921 the system if the system's timeouts are not configured either.
9922
Willy Tarreau95c4e142017-11-26 12:18:55 +01009923 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +01009924
Willy Tarreau0ba27502007-12-24 16:55:16 +01009925 This parameter replaces the old, deprecated "clitimeout". It is recommended
9926 to use it to write new configurations. The form "timeout clitimeout" is
9927 provided only by backwards compatibility but its use is strongly discouraged.
9928
Baptiste Assmann2e1941e2016-03-06 23:24:12 +01009929 See also : "clitimeout", "timeout server", "timeout tunnel",
9930 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009931
9932
Willy Tarreau05cdd962014-05-10 14:30:07 +02009933timeout client-fin <timeout>
9934 Set the inactivity timeout on the client side for half-closed connections.
9935 May be used in sections : defaults | frontend | listen | backend
9936 yes | yes | yes | no
9937 Arguments :
9938 <timeout> is the timeout value specified in milliseconds by default, but
9939 can be in any other unit if the number is suffixed by the unit,
9940 as explained at the top of this document.
9941
9942 The inactivity timeout applies when the client is expected to acknowledge or
9943 send data while one direction is already shut down. This timeout is different
9944 from "timeout client" in that it only applies to connections which are closed
9945 in one direction. This is particularly useful to avoid keeping connections in
9946 FIN_WAIT state for too long when clients do not disconnect cleanly. This
9947 problem is particularly common long connections such as RDP or WebSocket.
9948 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +01009949 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
9950 frame was sent, often indicating an expectation that the connection quickly
9951 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +02009952
9953 This parameter is specific to frontends, but can be specified once for all in
9954 "defaults" sections. By default it is not set, so half-closed connections
9955 will use the other timeouts (timeout.client or timeout.tunnel).
9956
9957 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
9958
9959
Willy Tarreau0ba27502007-12-24 16:55:16 +01009960timeout connect <timeout>
9961timeout contimeout <timeout> (deprecated)
9962 Set the maximum time to wait for a connection attempt to a server to succeed.
9963 May be used in sections : defaults | frontend | listen | backend
9964 yes | no | yes | yes
9965 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +01009966 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +01009967 can be in any other unit if the number is suffixed by the unit,
9968 as explained at the top of this document.
9969
9970 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01009971 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01009972 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01009973 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +01009974 connect timeout also presets both queue and tarpit timeouts to the same value
9975 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +01009976
9977 This parameter is specific to backends, but can be specified once for all in
9978 "defaults" sections. This is in fact one of the easiest solutions not to
9979 forget about it. An unspecified timeout results in an infinite timeout, which
9980 is not recommended. Such a usage is accepted and works but reports a warning
9981 during startup because it may results in accumulation of failed sessions in
9982 the system if the system's timeouts are not configured either.
9983
9984 This parameter replaces the old, deprecated "contimeout". It is recommended
9985 to use it to write new configurations. The form "timeout contimeout" is
9986 provided only by backwards compatibility but its use is strongly discouraged.
9987
Willy Tarreau41a340d2008-01-22 12:25:31 +01009988 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
9989 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +01009990
9991
Willy Tarreaub16a5742010-01-10 14:46:16 +01009992timeout http-keep-alive <timeout>
9993 Set the maximum allowed time to wait for a new HTTP request to appear
9994 May be used in sections : defaults | frontend | listen | backend
9995 yes | yes | yes | yes
9996 Arguments :
9997 <timeout> is the timeout value specified in milliseconds by default, but
9998 can be in any other unit if the number is suffixed by the unit,
9999 as explained at the top of this document.
10000
10001 By default, the time to wait for a new request in case of keep-alive is set
10002 by "timeout http-request". However this is not always convenient because some
10003 people want very short keep-alive timeouts in order to release connections
10004 faster, and others prefer to have larger ones but still have short timeouts
10005 once the request has started to present itself.
10006
10007 The "http-keep-alive" timeout covers these needs. It will define how long to
10008 wait for a new HTTP request to start coming after a response was sent. Once
10009 the first byte of request has been seen, the "http-request" timeout is used
10010 to wait for the complete request to come. Note that empty lines prior to a
10011 new request do not refresh the timeout and are not counted as a new request.
10012
10013 There is also another difference between the two timeouts : when a connection
10014 expires during timeout http-keep-alive, no error is returned, the connection
10015 just closes. If the connection expires in "http-request" while waiting for a
10016 connection to complete, a HTTP 408 error is returned.
10017
10018 In general it is optimal to set this value to a few tens to hundreds of
10019 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010020 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010021 1 millisecond) it will probably only accept pipelined requests but not the
10022 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010023 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010024
10025 If this parameter is not set, the "http-request" timeout applies, and if both
10026 are not set, "timeout client" still applies at the lower level. It should be
10027 set in the frontend to take effect, unless the frontend is in TCP mode, in
10028 which case the HTTP backend's timeout will be used.
10029
Willy Tarreau95c4e142017-11-26 12:18:55 +010010030 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10031 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010032 (where we only have one connection per client and a connection setup).
10033
Willy Tarreaub16a5742010-01-10 14:46:16 +010010034 See also : "timeout http-request", "timeout client".
10035
10036
Willy Tarreau036fae02008-01-06 13:24:40 +010010037timeout http-request <timeout>
10038 Set the maximum allowed time to wait for a complete HTTP request
10039 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010040 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010041 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010042 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010043 can be in any other unit if the number is suffixed by the unit,
10044 as explained at the top of this document.
10045
10046 In order to offer DoS protection, it may be required to lower the maximum
10047 accepted time to receive a complete HTTP request without affecting the client
10048 timeout. This helps protecting against established connections on which
10049 nothing is sent. The client timeout cannot offer a good protection against
10050 this abuse because it is an inactivity timeout, which means that if the
10051 attacker sends one character every now and then, the timeout will not
10052 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010053 types, the request will be aborted if it does not complete in time. When the
10054 timeout expires, an HTTP 408 response is sent to the client to inform it
10055 about the problem, and the connection is closed. The logs will report
10056 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010057 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010058 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10059 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010060
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010061 By default, this timeout only applies to the header part of the request,
10062 and not to any data. As soon as the empty line is received, this timeout is
10063 not used anymore. When combined with "option http-buffer-request", this
10064 timeout also applies to the body of the request..
10065 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010066 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010067
10068 Generally it is enough to set it to a few seconds, as most clients send the
10069 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010070 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010071 generally work on local networks as long as there are no packet losses. This
10072 will prevent people from sending bare HTTP requests using telnet.
10073
10074 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010075 chunk of the incoming request. It should be set in the frontend to take
10076 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10077 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010078
Willy Tarreau0f228a02015-05-01 15:37:53 +020010079 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010080 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010081
Willy Tarreau844e3c52008-01-11 16:28:18 +010010082
10083timeout queue <timeout>
10084 Set the maximum time to wait in the queue for a connection slot to be free
10085 May be used in sections : defaults | frontend | listen | backend
10086 yes | no | yes | yes
10087 Arguments :
10088 <timeout> is the timeout value specified in milliseconds by default, but
10089 can be in any other unit if the number is suffixed by the unit,
10090 as explained at the top of this document.
10091
10092 When a server's maxconn is reached, connections are left pending in a queue
10093 which may be server-specific or global to the backend. In order not to wait
10094 indefinitely, a timeout is applied to requests pending in the queue. If the
10095 timeout is reached, it is considered that the request will almost never be
10096 served, so it is dropped and a 503 error is returned to the client.
10097
10098 The "timeout queue" statement allows to fix the maximum time for a request to
10099 be left pending in a queue. If unspecified, the same value as the backend's
10100 connection timeout ("timeout connect") is used, for backwards compatibility
10101 with older versions with no "timeout queue" parameter.
10102
10103 See also : "timeout connect", "contimeout".
10104
10105
10106timeout server <timeout>
10107timeout srvtimeout <timeout> (deprecated)
10108 Set the maximum inactivity time on the server side.
10109 May be used in sections : defaults | frontend | listen | backend
10110 yes | no | yes | yes
10111 Arguments :
10112 <timeout> is the timeout value specified in milliseconds by default, but
10113 can be in any other unit if the number is suffixed by the unit,
10114 as explained at the top of this document.
10115
10116 The inactivity timeout applies when the server is expected to acknowledge or
10117 send data. In HTTP mode, this timeout is particularly important to consider
10118 during the first phase of the server's response, when it has to send the
10119 headers, as it directly represents the server's processing time for the
10120 request. To find out what value to put there, it's often good to start with
10121 what would be considered as unacceptable response times, then check the logs
10122 to observe the response time distribution, and adjust the value accordingly.
10123
10124 The value is specified in milliseconds by default, but can be in any other
10125 unit if the number is suffixed by the unit, as specified at the top of this
10126 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10127 recommended that the client timeout remains equal to the server timeout in
10128 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010129 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010130 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010131 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10132 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010133 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10134 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010135
10136 This parameter is specific to backends, but can be specified once for all in
10137 "defaults" sections. This is in fact one of the easiest solutions not to
10138 forget about it. An unspecified timeout results in an infinite timeout, which
10139 is not recommended. Such a usage is accepted and works but reports a warning
10140 during startup because it may results in accumulation of expired sessions in
10141 the system if the system's timeouts are not configured either.
10142
10143 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10144 to use it to write new configurations. The form "timeout srvtimeout" is
10145 provided only by backwards compatibility but its use is strongly discouraged.
10146
Willy Tarreauce887fd2012-05-12 12:50:00 +020010147 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010148
Willy Tarreau05cdd962014-05-10 14:30:07 +020010149
10150timeout server-fin <timeout>
10151 Set the inactivity timeout on the server side for half-closed connections.
10152 May be used in sections : defaults | frontend | listen | backend
10153 yes | no | yes | yes
10154 Arguments :
10155 <timeout> is the timeout value specified in milliseconds by default, but
10156 can be in any other unit if the number is suffixed by the unit,
10157 as explained at the top of this document.
10158
10159 The inactivity timeout applies when the server is expected to acknowledge or
10160 send data while one direction is already shut down. This timeout is different
10161 from "timeout server" in that it only applies to connections which are closed
10162 in one direction. This is particularly useful to avoid keeping connections in
10163 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10164 This problem is particularly common long connections such as RDP or WebSocket.
10165 Note that this timeout can override "timeout tunnel" when a connection shuts
10166 down in one direction. This setting was provided for completeness, but in most
10167 situations, it should not be needed.
10168
10169 This parameter is specific to backends, but can be specified once for all in
10170 "defaults" sections. By default it is not set, so half-closed connections
10171 will use the other timeouts (timeout.server or timeout.tunnel).
10172
10173 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10174
Willy Tarreau844e3c52008-01-11 16:28:18 +010010175
10176timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010177 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010178 May be used in sections : defaults | frontend | listen | backend
10179 yes | yes | yes | yes
10180 Arguments :
10181 <timeout> is the tarpit duration specified in milliseconds by default, but
10182 can be in any other unit if the number is suffixed by the unit,
10183 as explained at the top of this document.
10184
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010185 When a connection is tarpitted using "http-request tarpit" or
10186 "reqtarpit", it is maintained open with no activity for a certain
10187 amount of time, then closed. "timeout tarpit" defines how long it will
10188 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010189
10190 The value is specified in milliseconds by default, but can be in any other
10191 unit if the number is suffixed by the unit, as specified at the top of this
10192 document. If unspecified, the same value as the backend's connection timeout
10193 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010194 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010195
10196 See also : "timeout connect", "contimeout".
10197
10198
Willy Tarreauce887fd2012-05-12 12:50:00 +020010199timeout tunnel <timeout>
10200 Set the maximum inactivity time on the client and server side for tunnels.
10201 May be used in sections : defaults | frontend | listen | backend
10202 yes | no | yes | yes
10203 Arguments :
10204 <timeout> is the timeout value specified in milliseconds by default, but
10205 can be in any other unit if the number is suffixed by the unit,
10206 as explained at the top of this document.
10207
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010208 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010209 between a client and a server, and the connection remains inactive in both
10210 directions. This timeout supersedes both the client and server timeouts once
10211 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010212 analyzer remains attached to either connection (e.g. tcp content rules are
10213 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010214 when switching to the WebSocket protocol, or forwarding a CONNECT request
10215 to a proxy), or after the first response when no keepalive/close option is
10216 specified.
10217
Willy Tarreau05cdd962014-05-10 14:30:07 +020010218 Since this timeout is usually used in conjunction with long-lived connections,
10219 it usually is a good idea to also set "timeout client-fin" to handle the
10220 situation where a client suddenly disappears from the net and does not
10221 acknowledge a close, or sends a shutdown and does not acknowledge pending
10222 data anymore. This can happen in lossy networks where firewalls are present,
10223 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10224 state.
10225
Willy Tarreauce887fd2012-05-12 12:50:00 +020010226 The value is specified in milliseconds by default, but can be in any other
10227 unit if the number is suffixed by the unit, as specified at the top of this
10228 document. Whatever the expected normal idle time, it is a good practice to
10229 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010230 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010231
10232 This parameter is specific to backends, but can be specified once for all in
10233 "defaults" sections. This is in fact one of the easiest solutions not to
10234 forget about it.
10235
10236 Example :
10237 defaults http
10238 option http-server-close
10239 timeout connect 5s
10240 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010241 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010242 timeout server 30s
10243 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10244
Willy Tarreau05cdd962014-05-10 14:30:07 +020010245 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010246
10247
Willy Tarreau844e3c52008-01-11 16:28:18 +010010248transparent (deprecated)
10249 Enable client-side transparent proxying
10250 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010251 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010252 Arguments : none
10253
10254 This keyword was introduced in order to provide layer 7 persistence to layer
10255 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10256 connection for a remote address to a local process (here HAProxy), and let
10257 this process know what address was initially requested. When this option is
10258 used, sessions without cookies will be forwarded to the original destination
10259 IP address of the incoming request (which should match that of another
10260 equipment), while requests with cookies will still be forwarded to the
10261 appropriate server.
10262
10263 The "transparent" keyword is deprecated, use "option transparent" instead.
10264
10265 Note that contrary to a common belief, this option does NOT make HAProxy
10266 present the client's IP to the server when establishing the connection.
10267
Willy Tarreau844e3c52008-01-11 16:28:18 +010010268 See also: "option transparent"
10269
William Lallemanda73203e2012-03-12 12:48:57 +010010270unique-id-format <string>
10271 Generate a unique ID for each request.
10272 May be used in sections : defaults | frontend | listen | backend
10273 yes | yes | yes | no
10274 Arguments :
10275 <string> is a log-format string.
10276
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010277 This keyword creates a ID for each request using the custom log format. A
10278 unique ID is useful to trace a request passing through many components of
10279 a complex infrastructure. The newly created ID may also be logged using the
10280 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010281
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010282 The format should be composed from elements that are guaranteed to be
10283 unique when combined together. For instance, if multiple haproxy instances
10284 are involved, it might be important to include the node name. It is often
10285 needed to log the incoming connection's source and destination addresses
10286 and ports. Note that since multiple requests may be performed over the same
10287 connection, including a request counter may help differentiate them.
10288 Similarly, a timestamp may protect against a rollover of the counter.
10289 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010290
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010291 It is recommended to use hexadecimal notation for many fields since it
10292 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010293
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010294 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010295
Julien Vehentf21be322014-03-07 08:27:34 -050010296 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010297
10298 will generate:
10299
10300 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10301
10302 See also: "unique-id-header"
10303
10304unique-id-header <name>
10305 Add a unique ID header in the HTTP request.
10306 May be used in sections : defaults | frontend | listen | backend
10307 yes | yes | yes | no
10308 Arguments :
10309 <name> is the name of the header.
10310
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010311 Add a unique-id header in the HTTP request sent to the server, using the
10312 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010313
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010314 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010315
Julien Vehentf21be322014-03-07 08:27:34 -050010316 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010317 unique-id-header X-Unique-ID
10318
10319 will generate:
10320
10321 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10322
10323 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010324
Willy Tarreauf51658d2014-04-23 01:21:56 +020010325use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010326 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010327 May be used in sections : defaults | frontend | listen | backend
10328 no | yes | yes | no
10329 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010330 <backend> is the name of a valid backend or "listen" section, or a
10331 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010332
Willy Tarreauf51658d2014-04-23 01:21:56 +020010333 <condition> is a condition composed of ACLs, as described in section 7. If
10334 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010335
10336 When doing content-switching, connections arrive on a frontend and are then
10337 dispatched to various backends depending on a number of conditions. The
10338 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010339 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010340 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010341 source address validation) or combined with a "tcp-request" rule to wait for
10342 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010343
10344 There may be as many "use_backend" rules as desired. All of these rules are
10345 evaluated in their declaration order, and the first one which matches will
10346 assign the backend.
10347
10348 In the first form, the backend will be used if the condition is met. In the
10349 second form, the backend will be used if the condition is not met. If no
10350 condition is valid, the backend defined with "default_backend" will be used.
10351 If no default backend is defined, either the servers in the same section are
10352 used (in case of a "listen" section) or, in case of a frontend, no server is
10353 used and a 503 service unavailable response is returned.
10354
Willy Tarreau51aecc72009-07-12 09:47:04 +020010355 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010356 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010357 and backend processing will immediately follow, or the backend will wait for
10358 a complete HTTP request to get in. This feature is useful when a frontend
10359 must decode several protocols on a unique port, one of them being HTTP.
10360
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010361 When <backend> is a simple name, it is resolved at configuration time, and an
10362 error is reported if the specified backend does not exist. If <backend> is
10363 a log-format string instead, no check may be done at configuration time, so
10364 the backend name is resolved dynamically at run time. If the resulting
10365 backend name does not correspond to any valid backend, no other rule is
10366 evaluated, and the default_backend directive is applied instead. Note that
10367 when using dynamic backend names, it is highly recommended to use a prefix
10368 that no other backend uses in order to ensure that an unauthorized backend
10369 cannot be forced from the request.
10370
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010371 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010372 used to detect the association between frontends and backends to compute the
10373 backend's "fullconn" setting. This cannot be done for dynamic names.
10374
10375 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10376 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010377
Willy Tarreau036fae02008-01-06 13:24:40 +010010378
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010379use-server <server> if <condition>
10380use-server <server> unless <condition>
10381 Only use a specific server if/unless an ACL-based condition is matched.
10382 May be used in sections : defaults | frontend | listen | backend
10383 no | no | yes | yes
10384 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010385 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010386
10387 <condition> is a condition composed of ACLs, as described in section 7.
10388
10389 By default, connections which arrive to a backend are load-balanced across
10390 the available servers according to the configured algorithm, unless a
10391 persistence mechanism such as a cookie is used and found in the request.
10392
10393 Sometimes it is desirable to forward a particular request to a specific
10394 server without having to declare a dedicated backend for this server. This
10395 can be achieved using the "use-server" rules. These rules are evaluated after
10396 the "redirect" rules and before evaluating cookies, and they have precedence
10397 on them. There may be as many "use-server" rules as desired. All of these
10398 rules are evaluated in their declaration order, and the first one which
10399 matches will assign the server.
10400
10401 If a rule designates a server which is down, and "option persist" is not used
10402 and no force-persist rule was validated, it is ignored and evaluation goes on
10403 with the next rules until one matches.
10404
10405 In the first form, the server will be used if the condition is met. In the
10406 second form, the server will be used if the condition is not met. If no
10407 condition is valid, the processing continues and the server will be assigned
10408 according to other persistence mechanisms.
10409
10410 Note that even if a rule is matched, cookie processing is still performed but
10411 does not assign the server. This allows prefixed cookies to have their prefix
10412 stripped.
10413
10414 The "use-server" statement works both in HTTP and TCP mode. This makes it
10415 suitable for use with content-based inspection. For instance, a server could
10416 be selected in a farm according to the TLS SNI field. And if these servers
10417 have their weight set to zero, they will not be used for other traffic.
10418
10419 Example :
10420 # intercept incoming TLS requests based on the SNI field
10421 use-server www if { req_ssl_sni -i www.example.com }
10422 server www 192.168.0.1:443 weight 0
10423 use-server mail if { req_ssl_sni -i mail.example.com }
10424 server mail 192.168.0.1:587 weight 0
10425 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010426 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010427 # all the rest is forwarded to this server
10428 server default 192.168.0.2:443 check
10429
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010430 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010431
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010432
Davor Ocelice9ed2812017-12-25 17:49:28 +0100104335. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010434--------------------------
10435
10436The "bind", "server" and "default-server" keywords support a number of settings
10437depending on some build options and on the system HAProxy was built on. These
10438settings generally each consist in one word sometimes followed by a value,
10439written on the same line as the "bind" or "server" line. All these options are
10440described in this section.
10441
10442
104435.1. Bind options
10444-----------------
10445
10446The "bind" keyword supports a certain number of settings which are all passed
10447as arguments on the same line. The order in which those arguments appear makes
10448no importance, provided that they appear after the bind address. All of these
10449parameters are optional. Some of them consist in a single words (booleans),
10450while other ones expect a value after them. In this case, the value must be
10451provided immediately after the setting name.
10452
10453The currently supported settings are the following ones.
10454
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010455accept-netscaler-cip <magic number>
10456 Enforces the use of the NetScaler Client IP insertion protocol over any
10457 connection accepted by any of the TCP sockets declared on the same line. The
10458 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10459 the incoming connection to be used everywhere an address is used, with the
10460 only exception of "tcp-request connection" rules which will only see the
10461 real connection address. Logs will reflect the addresses indicated in the
10462 protocol, unless it is violated, in which case the real address will still
10463 be used. This keyword combined with support from external components can be
10464 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010465 mechanism which is not always reliable and not even always usable. See also
10466 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10467 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010468
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010469accept-proxy
10470 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010471 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10472 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010473 3/4 addresses of the incoming connection to be used everywhere an address is
10474 used, with the only exception of "tcp-request connection" rules which will
10475 only see the real connection address. Logs will reflect the addresses
10476 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010477 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010478 components can be used as an efficient and reliable alternative to the
10479 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010480 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10481 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010482
Olivier Houchardc2aae742017-09-22 18:26:28 +020010483allow-0rtt
10484 Allow receiving early data when using TLS 1.3. This is disabled by default,
10485 due to security considerations.
10486
Willy Tarreauab861d32013-04-02 02:30:41 +020010487alpn <protocols>
10488 This enables the TLS ALPN extension and advertises the specified protocol
10489 list as supported on top of ALPN. The protocol list consists in a comma-
10490 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10491 quotes). This requires that the SSL library is build with support for TLS
10492 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010493 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10494 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10495 now obsolete NPN extension. At the time of writing this, most browsers still
10496 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10497 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10498 are expected to be supported, both versions can be advertised, in order of
10499 preference, like below :
10500
10501 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010502
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010503backlog <backlog>
10504 Sets the socket's backlog to this value. If unspecified, the frontend's
10505 backlog is used instead, which generally defaults to the maxconn value.
10506
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010507curves <curves>
10508 This setting is only available when support for OpenSSL was built in. It sets
10509 the string describing the list of elliptic curves algorithms ("curve suite")
10510 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10511 string is a colon-delimited list of curve name.
10512 Example: "X25519:P-256" (without quote)
10513 When "curves" is set, "ecdhe" parameter is ignored.
10514
Emeric Brun7fb34422012-09-28 15:26:15 +020010515ecdhe <named curve>
10516 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010517 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10518 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010519
Emeric Brunfd33a262012-10-11 16:28:27 +020010520ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010521 This setting is only available when support for OpenSSL was built in. It
10522 designates a PEM file from which to load CA certificates used to verify
10523 client's certificate.
10524
Emeric Brunb6dc9342012-09-28 17:55:37 +020010525ca-ignore-err [all|<errorID>,...]
10526 This setting is only available when support for OpenSSL was built in.
10527 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10528 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10529 error is ignored.
10530
Christopher Faulet31af49d2015-06-09 17:29:50 +020010531ca-sign-file <cafile>
10532 This setting is only available when support for OpenSSL was built in. It
10533 designates a PEM file containing both the CA certificate and the CA private
10534 key used to create and sign server's certificates. This is a mandatory
10535 setting when the dynamic generation of certificates is enabled. See
10536 'generate-certificates' for details.
10537
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010538ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010539 This setting is only available when support for OpenSSL was built in. It is
10540 the CA private key passphrase. This setting is optional and used only when
10541 the dynamic generation of certificates is enabled. See
10542 'generate-certificates' for details.
10543
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010544ciphers <ciphers>
10545 This setting is only available when support for OpenSSL was built in. It sets
10546 the string describing the list of cipher algorithms ("cipher suite") that are
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010547 negotiated during the SSL/TLS handshake. The format of the string is defined
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010548 in "man 1 ciphers" from OpenSSL man pages, and can be for instance a string
10549 such as "AES:ALL:!aNULL:!eNULL:+RC4:@STRENGTH" (without quotes).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010550 Depending on the compatibility and security requirements, the list of suitable
Daniel Schneller87e43022017-09-01 19:29:57 +020010551 ciphers depends on a variety of variables. For background information and
10552 recommendations see e. g. (https://wiki.mozilla.org/Security/Server_Side_TLS)
10553 and (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010554
Emeric Brunfd33a262012-10-11 16:28:27 +020010555crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010556 This setting is only available when support for OpenSSL was built in. It
10557 designates a PEM file from which to load certificate revocation list used
10558 to verify client's certificate.
10559
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010560crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010561 This setting is only available when support for OpenSSL was built in. It
10562 designates a PEM file containing both the required certificates and any
10563 associated private keys. This file can be built by concatenating multiple
10564 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10565 requires an intermediate certificate, this can also be concatenated into this
10566 file.
10567
10568 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10569 are loaded.
10570
10571 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010572 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010573 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10574 specified multiple times in order to load certificates from multiple files or
10575 directories. The certificates will be presented to clients who provide a
10576 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010577 subjects. Wildcards are supported, where a wildcard character '*' is used
10578 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010579 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010580
10581 If no SNI is provided by the client or if the SSL library does not support
10582 TLS extensions, or if the client provides an SNI hostname which does not
10583 match any certificate, then the first loaded certificate will be presented.
10584 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010585 recommended to load the default one first as a file or to ensure that it will
10586 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010587
Emeric Brune032bfa2012-09-28 13:01:45 +020010588 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010589
Davor Ocelice9ed2812017-12-25 17:49:28 +010010590 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010591 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010592 choose a web server that the CA believes requires an intermediate CA (for
10593 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010594 others, e.g. nginx, result in a wrong bundle that will not work for some
10595 clients).
10596
Emeric Brun4147b2e2014-06-16 18:36:30 +020010597 For each PEM file, haproxy checks for the presence of file at the same path
10598 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10599 Status Request extension (also known as "OCSP stapling") is automatically
10600 enabled. The content of this file is optional. If not empty, it must contain
10601 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10602 must comply with the following rules: it has to indicate a good status,
10603 it has to be a single response for the certificate of the PEM file, and it
10604 has to be valid at the moment of addition. If these rules are not respected
10605 the OCSP Response is ignored and a warning is emitted. In order to identify
10606 which certificate an OCSP Response applies to, the issuer's certificate is
10607 necessary. If the issuer's certificate is not found in the PEM file, it will
10608 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10609 if it exists otherwise it will fail with an error.
10610
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010611 For each PEM file, haproxy also checks for the presence of file at the same
10612 path suffixed by ".sctl". If such file is found, support for Certificate
10613 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10614 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10615 to check basic syntax, but no signatures are verified.
10616
yanbzhu6c25e9e2016-01-05 12:52:02 -050010617 There are cases where it is desirable to support multiple key types, e.g. RSA
10618 and ECDSA in the cipher suites offered to the clients. This allows clients
10619 that support EC certificates to be able to use EC ciphers, while
10620 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010621
10622 In order to provide this functionality, multiple PEM files, each with a
10623 different key type, are required. To associate these PEM files into a
10624 "cert bundle" that is recognized by haproxy, they must be named in the
10625 following way: All PEM files that are to be bundled must have the same base
10626 name, with a suffix indicating the key type. Currently, three suffixes are
10627 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10628 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10629 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10630 suffix matters. To load this bundle into haproxy, specify the base name only:
10631
10632 Example : bind :8443 ssl crt example.pem
10633
yanbzhu6c25e9e2016-01-05 12:52:02 -050010634 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010635 a cert bundle.
10636
Davor Ocelice9ed2812017-12-25 17:49:28 +010010637 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010638 support multiple key types. PEM files are combined based on Common Name
10639 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10640 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10641 entries in the certificates in that bundle, haproxy will not be able to
10642 provide multi-cert support.
10643
10644 Assuming bundle in the example above contained the following:
10645
10646 Filename | CN | SAN
10647 -------------------+-----------------+-------------------
10648 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010649 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010650 example.pem.ecdsa | www.example.com | ecdsa.example.com
10651 -------------------+-----------------+-------------------
10652
10653 Users connecting with an SNI of "www.example.com" will be able
10654 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10655 "rsa.example.com" will only be able to use RSA cipher suites, and users
10656 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010657 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10658 no need to bundle certificates. ECDSA certificate will be preferred if client
10659 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010660
10661 If a directory name is given as the <cert> argument, haproxy will
10662 automatically search and load bundled files in that directory.
10663
10664 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10665 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10666 time, sctl is not supported in multi-certificate bundling.
10667
Emeric Brunb6dc9342012-09-28 17:55:37 +020010668crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010669 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010670 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010671 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010672 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010673
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010674crt-list <file>
10675 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010676 designates a list of PEM file with an optional ssl configuration and a SNI
10677 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010678
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010679 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10680
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010681 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10682 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010683 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010684 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010685
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010686 Wildcards are supported in the SNI filter. Negative filter are also supported,
10687 only useful in combination with a wildcard filter to exclude a particular SNI.
10688 The certificates will be presented to clients who provide a valid TLS Server
10689 Name Indication field matching one of the SNI filters. If no SNI filter is
10690 specified, the CN and alt subjects are used. This directive may be specified
10691 multiple times. See the "crt" option for more information. The default
10692 certificate is still needed to meet OpenSSL expectations. If it is not used,
10693 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010694
yanbzhu6c25e9e2016-01-05 12:52:02 -050010695 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010696 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010697 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10698 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10699 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010700
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010701 crt-list file example:
10702 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010703 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010704 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010010705 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010706
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010707defer-accept
10708 Is an optional keyword which is supported only on certain Linux kernels. It
10709 states that a connection will only be accepted once some data arrive on it,
10710 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010010711 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010712 performance by ensuring that most of the request is already available when
10713 the connection is accepted. On the other hand, it will not be able to detect
10714 connections which don't talk. It is important to note that this option is
10715 broken in all kernels up to 2.6.31, as the connection is never accepted until
10716 the client talks. This can cause issues with front firewalls which would see
10717 an established connection while the proxy will only see it in SYN_RECV. This
10718 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
10719
William Lallemandf6975e92017-05-26 17:42:10 +020010720expose-fd listeners
10721 This option is only usable with the stats socket. It gives your stats socket
10722 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020010723 During a reload with the master-worker mode, the process is automatically
10724 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010725 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020010726
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010727force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010728 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010729 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010730 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010731 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010732
10733force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010734 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010735 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010736 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010737
10738force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010739 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010740 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010741 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010742
10743force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010744 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010745 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010746 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010747
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010748force-tlsv13
10749 This option enforces use of TLSv1.3 only on SSL connections instantiated from
10750 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010751 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010752
Christopher Faulet31af49d2015-06-09 17:29:50 +020010753generate-certificates
10754 This setting is only available when support for OpenSSL was built in. It
10755 enables the dynamic SSL certificates generation. A CA certificate and its
10756 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
10757 a transparent forward proxy, SSL requests generate errors because of a common
10758 name mismatch on the certificate presented to the client. With this option
10759 enabled, HAProxy will try to forge a certificate using the SNI hostname
10760 indicated by the client. This is done only if no certificate matches the SNI
10761 hostname (see 'crt-list'). If an error occurs, the default certificate is
10762 used, else the 'strict-sni' option is set.
10763 It can also be used when HAProxy is configured as a reverse proxy to ease the
10764 deployment of an architecture with many backends.
10765
10766 Creating a SSL certificate is an expensive operation, so a LRU cache is used
10767 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010010768 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020010769 certificate is used many times.
10770
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010771gid <gid>
10772 Sets the group of the UNIX sockets to the designated system gid. It can also
10773 be set by default in the global section's "unix-bind" statement. Note that
10774 some platforms simply ignore this. This setting is equivalent to the "group"
10775 setting except that the group ID is used instead of its name. This setting is
10776 ignored by non UNIX sockets.
10777
10778group <group>
10779 Sets the group of the UNIX sockets to the designated system group. It can
10780 also be set by default in the global section's "unix-bind" statement. Note
10781 that some platforms simply ignore this. This setting is equivalent to the
10782 "gid" setting except that the group name is used instead of its gid. This
10783 setting is ignored by non UNIX sockets.
10784
10785id <id>
10786 Fixes the socket ID. By default, socket IDs are automatically assigned, but
10787 sometimes it is more convenient to fix them to ease monitoring. This value
10788 must be strictly positive and unique within the listener/frontend. This
10789 option can only be used when defining only a single socket.
10790
10791interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010010792 Restricts the socket to a specific interface. When specified, only packets
10793 received from that particular interface are processed by the socket. This is
10794 currently only supported on Linux. The interface must be a primary system
10795 interface, not an aliased interface. It is also possible to bind multiple
10796 frontends to the same address if they are bound to different interfaces. Note
10797 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010010798 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
10799 uses the same interface as inbound traffic, and its associated routing table,
10800 even if there are explicit routes through different interfaces configured.
10801 This can prove useful to address asymmetric routing issues when the same
10802 client IP addresses need to be able to reach frontends hosted on different
10803 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010804
Willy Tarreauabb175f2012-09-24 12:43:26 +020010805level <level>
10806 This setting is used with the stats sockets only to restrict the nature of
10807 the commands that can be issued on the socket. It is ignored by other
10808 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010010809 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020010810 read, and no change is allowed. It would make sense on systems where it
10811 is not easy to restrict access to the socket.
10812 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010813 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020010814 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010010815 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020010816 all counters).
10817
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020010818severity-output <format>
10819 This setting is used with the stats sockets only to configure severity
10820 level output prepended to informational feedback messages. Severity
10821 level of messages can range between 0 and 7, conforming to syslog
10822 rfc5424. Valid and successful socket commands requesting data
10823 (i.e. "show map", "get acl foo" etc.) will never have a severity level
10824 prepended. It is ignored by other sockets. <format> can be one of :
10825 - "none" (default) no severity level is prepended to feedback messages.
10826 - "number" severity level is prepended as a number.
10827 - "string" severity level is prepended as a string following the
10828 rfc5424 convention.
10829
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010830maxconn <maxconn>
10831 Limits the sockets to this number of concurrent connections. Extraneous
10832 connections will remain in the system's backlog until a connection is
10833 released. If unspecified, the limit will be the same as the frontend's
10834 maxconn. Note that in case of port ranges or multiple addresses, the same
10835 value will be applied to each socket. This setting enables different
10836 limitations on expensive sockets, for instance SSL entries which may easily
10837 eat all memory.
10838
10839mode <mode>
10840 Sets the octal mode used to define access permissions on the UNIX socket. It
10841 can also be set by default in the global section's "unix-bind" statement.
10842 Note that some platforms simply ignore this. This setting is ignored by non
10843 UNIX sockets.
10844
10845mss <maxseg>
10846 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
10847 connections. This can be used to force a lower MSS for certain specific
10848 ports, for instance for connections passing through a VPN. Note that this
10849 relies on a kernel feature which is theoretically supported under Linux but
10850 was buggy in all versions prior to 2.6.28. It may or may not work on other
10851 operating systems. It may also not change the advertised value but change the
10852 effective size of outgoing segments. The commonly advertised value for TCPv4
10853 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
10854 positive, it will be used as the advertised MSS. If it is negative, it will
10855 indicate by how much to reduce the incoming connection's advertised MSS for
10856 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
10857
10858name <name>
10859 Sets an optional name for these sockets, which will be reported on the stats
10860 page.
10861
Willy Tarreaud72f0f32015-10-13 14:50:22 +020010862namespace <name>
10863 On Linux, it is possible to specify which network namespace a socket will
10864 belong to. This directive makes it possible to explicitly bind a listener to
10865 a namespace different from the default one. Please refer to your operating
10866 system's documentation to find more details about network namespaces.
10867
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010868nice <nice>
10869 Sets the 'niceness' of connections initiated from the socket. Value must be
10870 in the range -1024..1024 inclusive, and defaults to zero. Positive values
10871 means that such connections are more friendly to others and easily offer
10872 their place in the scheduler. On the opposite, negative values mean that
10873 connections want to run with a higher priority than others. The difference
10874 only happens under high loads when the system is close to saturation.
10875 Negative values are appropriate for low-latency or administration services,
10876 and high values are generally recommended for CPU intensive tasks such as SSL
10877 processing or bulk transfers which are less sensible to latency. For example,
10878 it may make sense to use a positive value for an SMTP socket and a negative
10879 one for an RDP socket.
10880
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010881no-ca-names
10882 This setting is only available when support for OpenSSL was built in. It
10883 prevents from send CA names in server hello message when ca-file is used.
10884
Emeric Brun9b3009b2012-10-05 11:55:06 +020010885no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010886 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010887 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010888 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010889 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010890 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
10891 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010892
Emeric Brun90ad8722012-10-02 14:00:59 +020010893no-tls-tickets
10894 This setting is only available when support for OpenSSL was built in. It
10895 disables the stateless session resumption (RFC 5077 TLS Ticket
10896 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010897 session resumption is more expensive in CPU usage. This option is also
10898 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020010899
Emeric Brun9b3009b2012-10-05 11:55:06 +020010900no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010901 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010902 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010903 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010904 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010905 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10906 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010907
Emeric Brun9b3009b2012-10-05 11:55:06 +020010908no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020010909 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010910 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010911 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010912 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010913 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10914 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020010915
Emeric Brun9b3009b2012-10-05 11:55:06 +020010916no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020010917 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010918 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020010919 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010010920 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010921 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10922 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020010923
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010924no-tlsv13
10925 This setting is only available when support for OpenSSL was built in. It
10926 disables support for TLSv1.3 on any sockets instantiated from the listener
10927 when SSL is supported. Note that SSLv2 is forced disabled in the code and
10928 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010929 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
10930 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020010931
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020010932npn <protocols>
10933 This enables the NPN TLS extension and advertises the specified protocol list
10934 as supported on top of NPN. The protocol list consists in a comma-delimited
10935 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
10936 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020010937 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010010938 replaced with the ALPN extension (see the "alpn" keyword), though this one is
10939 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
10940 version of OpenSSL, NPN might still be used as most clients still support it
10941 at the time of writing this. It is possible to enable both NPN and ALPN
10942 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020010943
Lukas Tribus53ae85c2017-05-04 15:45:40 +000010944prefer-client-ciphers
10945 Use the client's preference when selecting the cipher suite, by default
10946 the server's preference is enforced. This option is also available on
10947 global statement "ssl-default-bind-options".
10948
Christopher Fauletc644fa92017-11-23 22:44:11 +010010949process <process-set>[/<thread-set>]
10950 This restricts the list of processes and/or threads on which this listener is
10951 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010010952 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010010953 between the two is applied. If in the end the listener is not allowed to run
10954 on any remaining process, a warning is emitted, and the listener will either
10955 run on the first process of the listener if a single process was specified,
10956 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010010957 set is specified, it limits the threads allowed to process incoming
Christopher Fauletc644fa92017-11-23 22:44:11 +010010958 connections for this listener, for the corresponding process set. For the
10959 unlikely case where several ranges are needed, this directive may be
10960 repeated. <process-set> and <thread-set> must use the format
10961
10962 all | odd | even | number[-[number]]
10963
10964 Ranges can be partially defined. The higher bound can be omitted. In such
10965 case, it is replaced by the corresponding maximum value. The main purpose of
10966 this directive is to be used with the stats sockets and have one different
10967 socket per process. The second purpose is to have multiple bind lines sharing
10968 the same IP:port but not the same process in a listener, so that the system
10969 can distribute the incoming connections into multiple queues and allow a
10970 smoother inter-process load balancing. Currently Linux 3.9 and above is known
10971 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020010972
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010973ssl
10974 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010975 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010976 certificate is necessary (see "crt" above). All contents in the buffers will
10977 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020010978 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
10979 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010980
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020010981ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
10982 This option enforces use of <version> or lower on SSL connections instantiated
10983 from this listener. This option is also available on global statement
10984 "ssl-default-bind-options". See also "ssl-min-ver".
10985
10986ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
10987 This option enforces use of <version> or upper on SSL connections instantiated
10988 from this listener. This option is also available on global statement
10989 "ssl-default-bind-options". See also "ssl-max-ver".
10990
Emmanuel Hocdet65623372013-01-24 17:17:15 +010010991strict-sni
10992 This setting is only available when support for OpenSSL was built in. The
10993 SSL/TLS negotiation is allow only if the client provided an SNI which match
10994 a certificate. The default certificate is not used.
10995 See the "crt" option for more information.
10996
Willy Tarreau2af207a2015-02-04 00:45:58 +010010997tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010010998 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010010999 listening socket. This option is available on Linux since version 2.6.37. It
11000 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011001 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011002 useful on long-lived connections experiencing long idle periods such as
11003 remote terminals or database connection pools, where the client and server
11004 timeouts must remain high to allow a long period of idle, but where it is
11005 important to detect that the client has disappeared in order to release all
11006 resources associated with its connection (and the server's session). The
11007 argument is a delay expressed in milliseconds by default. This only works
11008 for regular TCP connections, and is ignored for other protocols.
11009
Willy Tarreau1c862c52012-10-05 16:21:00 +020011010tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011011 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011012 enables TCP Fast Open on the listening socket, which means that clients which
11013 support this feature will be able to send a request and receive a response
11014 during the 3-way handshake starting from second connection, thus saving one
11015 round-trip after the first connection. This only makes sense with protocols
11016 that use high connection rates and where each round trip matters. This can
11017 possibly cause issues with many firewalls which do not accept data on SYN
11018 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011019 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11020 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11021 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011022
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011023tls-ticket-keys <keyfile>
11024 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
11025 bytes long, encoded with base64 (ex. openssl rand -base64 48). Number of keys
11026 is specified by the TLS_TICKETS_NO build option (default 3) and at least as
11027 many keys need to be present in the file. Last TLS_TICKETS_NO keys will be
11028 used for decryption and the penultimate one for encryption. This enables easy
11029 key rotation by just appending new key to the file and reloading the process.
11030 Keys must be periodically rotated (ex. every 12h) or Perfect Forward Secrecy
11031 is compromised. It is also a good idea to keep the keys off any permanent
11032 storage such as hard drives (hint: use tmpfs and don't swap those files).
11033 Lifetime hint can be changed using tune.ssl.timeout.
11034
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011035transparent
11036 Is an optional keyword which is supported only on certain Linux kernels. It
11037 indicates that the addresses will be bound even if they do not belong to the
11038 local machine, and that packets targeting any of these addresses will be
11039 intercepted just as if the addresses were locally configured. This normally
11040 requires that IP forwarding is enabled. Caution! do not use this with the
11041 default address '*', as it would redirect any traffic for the specified port.
11042 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11043 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11044 kernel version. Some distribution kernels include backports of the feature,
11045 so check for support with your vendor.
11046
Willy Tarreau77e3af92012-11-24 15:07:23 +010011047v4v6
11048 Is an optional keyword which is supported only on most recent systems
11049 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11050 and IPv6 when it uses the default address. Doing so is sometimes necessary
11051 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011052 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011053
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011054v6only
11055 Is an optional keyword which is supported only on most recent systems
11056 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11057 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011058 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11059 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011060
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011061uid <uid>
11062 Sets the owner of the UNIX sockets to the designated system uid. It can also
11063 be set by default in the global section's "unix-bind" statement. Note that
11064 some platforms simply ignore this. This setting is equivalent to the "user"
11065 setting except that the user numeric ID is used instead of its name. This
11066 setting is ignored by non UNIX sockets.
11067
11068user <user>
11069 Sets the owner of the UNIX sockets to the designated system user. It can also
11070 be set by default in the global section's "unix-bind" statement. Note that
11071 some platforms simply ignore this. This setting is equivalent to the "uid"
11072 setting except that the user name is used instead of its uid. This setting is
11073 ignored by non UNIX sockets.
11074
Emeric Brun1a073b42012-09-28 17:07:34 +020011075verify [none|optional|required]
11076 This setting is only available when support for OpenSSL was built in. If set
11077 to 'none', client certificate is not requested. This is the default. In other
11078 cases, a client certificate is requested. If the client does not provide a
11079 certificate after the request and if 'verify' is set to 'required', then the
11080 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011081 certificate provided by the client is always verified using CAs from
11082 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11083 is aborted, regardless of the 'verify' option, unless the error code exactly
11084 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011085
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200110865.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011087------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011088
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011089The "server" and "default-server" keywords support a certain number of settings
11090which are all passed as arguments on the server line. The order in which those
11091arguments appear does not count, and they are all optional. Some of those
11092settings are single words (booleans) while others expect one or several values
11093after them. In this case, the values must immediately follow the setting name.
11094Except default-server, all those settings must be specified after the server's
11095address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011096
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011097 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011098 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011099
Frédéric Lécailled2376272017-03-21 18:52:12 +010011100Note that all these settings are supported both by "server" and "default-server"
11101keywords, except "id" which is only supported by "server".
11102
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011103The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011104
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011105addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011106 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011107 to send health-checks or to probe the agent-check. On some servers, it may be
11108 desirable to dedicate an IP address to specific component able to perform
11109 complex tests which are more suitable to health-checks than the application.
11110 This parameter is ignored if the "check" parameter is not set. See also the
11111 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011112
Simon Hormand60d6912013-11-25 10:46:36 +090011113agent-check
11114 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011115 health check. An agent health check is performed by making a TCP connection
11116 to the port set by the "agent-port" parameter and reading an ASCII string.
11117 The string is made of a series of words delimited by spaces, tabs or commas
11118 in any order, optionally terminated by '\r' and/or '\n', each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011119
Willy Tarreau81f5d942013-12-09 20:51:51 +010011120 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011121 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011122 weight of a server as configured when haproxy starts. Note that a zero
11123 weight is reported on the stats page as "DRAIN" since it has the same
11124 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011125
Davor Ocelice9ed2812017-12-25 17:49:28 +010011126 - The string "maxconn:" followed by an integer (no space between). Values
11127 in this format will set the maxconn of a server. The maximum number of
11128 connections advertised needs to be multiplied by the number of load
11129 balancers and different backends that use this health check to get the
11130 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011131
Willy Tarreau81f5d942013-12-09 20:51:51 +010011132 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011133 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011134
Willy Tarreau81f5d942013-12-09 20:51:51 +010011135 - The word "drain". This will turn the server's administrative state to the
11136 DRAIN mode, thus it will not accept any new connections other than those
11137 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011138
Willy Tarreau81f5d942013-12-09 20:51:51 +010011139 - The word "maint". This will turn the server's administrative state to the
11140 MAINT mode, thus it will not accept any new connections at all, and health
11141 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011142
Willy Tarreau81f5d942013-12-09 20:51:51 +010011143 - The words "down", "failed", or "stopped", optionally followed by a
11144 description string after a sharp ('#'). All of these mark the server's
11145 operating state as DOWN, but since the word itself is reported on the stats
11146 page, the difference allows an administrator to know if the situation was
11147 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011148 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011149 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011150
Willy Tarreau81f5d942013-12-09 20:51:51 +010011151 - The word "up" sets back the server's operating state as UP if health checks
11152 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011153
Willy Tarreau81f5d942013-12-09 20:51:51 +010011154 Parameters which are not advertised by the agent are not changed. For
11155 example, an agent might be designed to monitor CPU usage and only report a
11156 relative weight and never interact with the operating status. Similarly, an
11157 agent could be designed as an end-user interface with 3 radio buttons
11158 allowing an administrator to change only the administrative state. However,
11159 it is important to consider that only the agent may revert its own actions,
11160 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11161 agent must implement the other equivalent actions to bring the service into
11162 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011163
Simon Horman2f1f9552013-11-25 10:46:37 +090011164 Failure to connect to the agent is not considered an error as connectivity
11165 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011166 parameter. Warning though, it is not a good idea to stop an agent after it
11167 reports "down", since only an agent reporting "up" will be able to turn the
11168 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011169 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011170
Willy Tarreau81f5d942013-12-09 20:51:51 +010011171 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011172 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011173
James Brown55f9ff12015-10-21 18:19:05 -070011174agent-send <string>
11175 If this option is specified, haproxy will send the given string (verbatim)
11176 to the agent server upon connection. You could, for example, encode
11177 the backend name into this string, which would enable your agent to send
11178 different responses based on the backend. Make sure to include a '\n' if
11179 you want to terminate your request with a newline.
11180
Simon Hormand60d6912013-11-25 10:46:36 +090011181agent-inter <delay>
11182 The "agent-inter" parameter sets the interval between two agent checks
11183 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11184
11185 Just as with every other time-based parameter, it may be entered in any
11186 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11187 parameter also serves as a timeout for agent checks "timeout check" is
11188 not set. In order to reduce "resonance" effects when multiple servers are
11189 hosted on the same hardware, the agent and health checks of all servers
11190 are started with a small time offset between them. It is also possible to
11191 add some random noise in the agent and health checks interval using the
11192 global "spread-checks" keyword. This makes sense for instance when a lot
11193 of backends use the same servers.
11194
11195 See also the "agent-check" and "agent-port" parameters.
11196
Misiek768d8602017-01-09 09:52:43 +010011197agent-addr <addr>
11198 The "agent-addr" parameter sets address for agent check.
11199
11200 You can offload agent-check to another target, so you can make single place
11201 managing status and weights of servers defined in haproxy in case you can't
11202 make self-aware and self-managing services. You can specify both IP or
11203 hostname, it will be resolved.
11204
Simon Hormand60d6912013-11-25 10:46:36 +090011205agent-port <port>
11206 The "agent-port" parameter sets the TCP port used for agent checks.
11207
11208 See also the "agent-check" and "agent-inter" parameters.
11209
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011210backup
11211 When "backup" is present on a server line, the server is only used in load
11212 balancing when all other non-backup servers are unavailable. Requests coming
11213 with a persistence cookie referencing the server will always be served
11214 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011215 the "allbackups" option is set in the backend. See also the "no-backup" and
11216 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011217
Emeric Brunef42d922012-10-11 16:11:36 +020011218ca-file <cafile>
11219 This setting is only available when support for OpenSSL was built in. It
11220 designates a PEM file from which to load CA certificates used to verify
11221 server's certificate.
11222
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011223check
11224 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011225 always considered available. If "check" is set, the server is available when
11226 accepting periodic TCP connections, to ensure that it is really able to serve
11227 requests. The default address and port to send the tests to are those of the
11228 server, and the default source is the same as the one defined in the
11229 backend. It is possible to change the address using the "addr" parameter, the
11230 port using the "port" parameter, the source address using the "source"
11231 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011232 parameters. The request method is define in the backend using the "httpchk",
11233 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011234 refer to those options and parameters for more information. See also
11235 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011236
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011237check-send-proxy
11238 This option forces emission of a PROXY protocol line with outgoing health
11239 checks, regardless of whether the server uses send-proxy or not for the
11240 normal traffic. By default, the PROXY protocol is enabled for health checks
11241 if it is already enabled for normal traffic and if no "port" nor "addr"
11242 directive is present. However, if such a directive is present, the
11243 "check-send-proxy" option needs to be used to force the use of the
11244 protocol. See also the "send-proxy" option for more information.
11245
Olivier Houchard9130a962017-10-17 17:33:43 +020011246check-sni
11247 This option allows you to specify the SNI to be used when doing health checks
11248 over SSL.
11249
Willy Tarreau763a95b2012-10-04 23:15:39 +020011250check-ssl
11251 This option forces encryption of all health checks over SSL, regardless of
11252 whether the server uses SSL or not for the normal traffic. This is generally
11253 used when an explicit "port" or "addr" directive is specified and SSL health
11254 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011255 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011256 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11257 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011258 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011259 See the "ssl" option for more information and "no-check-ssl" to disable
11260 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011261
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011262ciphers <ciphers>
11263 This option sets the string describing the list of cipher algorithms that is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011264 is negotiated during the SSL/TLS handshake with the server. The format of the
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011265 string is defined in "man 1 ciphers". When SSL is used to communicate with
11266 servers on the local network, it is common to see a weaker set of algorithms
11267 than what is used over the internet. Doing so reduces CPU usage on both the
11268 server and haproxy while still keeping it compatible with deployed software.
11269 Some algorithms such as RC4-SHA1 are reasonably cheap. If no security at all
11270 is needed and just connectivity, using DES can be appropriate.
11271
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011272cookie <value>
11273 The "cookie" parameter sets the cookie value assigned to the server to
11274 <value>. This value will be checked in incoming requests, and the first
11275 operational server possessing the same value will be selected. In return, in
11276 cookie insertion or rewrite modes, this value will be assigned to the cookie
11277 sent to the client. There is nothing wrong in having several servers sharing
11278 the same cookie value, and it is in fact somewhat common between normal and
11279 backup servers. See also the "cookie" keyword in backend section.
11280
Emeric Brunef42d922012-10-11 16:11:36 +020011281crl-file <crlfile>
11282 This setting is only available when support for OpenSSL was built in. It
11283 designates a PEM file from which to load certificate revocation list used
11284 to verify server's certificate.
11285
Emeric Bruna7aa3092012-10-26 12:58:00 +020011286crt <cert>
11287 This setting is only available when support for OpenSSL was built in.
11288 It designates a PEM file from which to load both a certificate and the
11289 associated private key. This file can be built by concatenating both PEM
11290 files into one. This certificate will be sent if the server send a client
11291 certificate request.
11292
Willy Tarreau96839092010-03-29 10:02:24 +020011293disabled
11294 The "disabled" keyword starts the server in the "disabled" state. That means
11295 that it is marked down in maintenance mode, and no connection other than the
11296 ones allowed by persist mode will reach it. It is very well suited to setup
11297 new servers, because normal traffic will never reach them, while it is still
11298 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011299 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011300
Frédéric Lécailled2376272017-03-21 18:52:12 +010011301enabled
11302 This option may be used as 'server' setting to reset any 'disabled'
11303 setting which would have been inherited from 'default-server' directive as
11304 default value.
11305 It may also be used as 'default-server' setting to reset any previous
11306 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011307
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011308error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011309 If health observing is enabled, the "error-limit" parameter specifies the
11310 number of consecutive errors that triggers event selected by the "on-error"
11311 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011312
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011313 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011314
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011315fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011316 The "fall" parameter states that a server will be considered as dead after
11317 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11318 unspecified. See also the "check", "inter" and "rise" parameters.
11319
Emeric Brun8694b9a2012-10-05 14:39:07 +020011320force-sslv3
11321 This option enforces use of SSLv3 only when SSL is used to communicate with
11322 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011323 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011324 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011325
11326force-tlsv10
11327 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011328 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011329 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011330
11331force-tlsv11
11332 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011333 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011334 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011335
11336force-tlsv12
11337 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011338 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011339 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011340
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011341force-tlsv13
11342 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11343 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011344 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011345
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011346id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011347 Set a persistent ID for the server. This ID must be positive and unique for
11348 the proxy. An unused ID will automatically be assigned if unset. The first
11349 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011350
Willy Tarreau6a031d12016-11-07 19:42:35 +010011351init-addr {last | libc | none | <ip>},[...]*
11352 Indicate in what order the server's address should be resolved upon startup
11353 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011354 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011355 method which succeeds is used. If the end of the list is reached without
11356 finding a working method, an error is thrown. Method "last" suggests to pick
11357 the address which appears in the state file (see "server-state-file"). Method
11358 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11359 depending on the operating system and build options). Method "none"
11360 specifically indicates that the server should start without any valid IP
11361 address in a down state. It can be useful to ignore some DNS issues upon
11362 startup, waiting for the situation to get fixed later. Finally, an IP address
11363 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011364 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011365 server used to catch old sessions and present them with a decent error
11366 message for example. When the "first" load balancing algorithm is used, this
11367 IP address could point to a fake server used to trigger the creation of new
11368 instances on the fly. This option defaults to "last,libc" indicating that the
11369 previous address found in the state file (if any) is used first, otherwise
11370 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011371 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011372
11373 Example:
11374 defaults
11375 # never fail on address resolution
11376 default-server init-addr last,libc,none
11377
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011378inter <delay>
11379fastinter <delay>
11380downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011381 The "inter" parameter sets the interval between two consecutive health checks
11382 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11383 It is also possible to use "fastinter" and "downinter" to optimize delays
11384 between checks depending on the server state :
11385
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011386 Server state | Interval used
11387 ----------------------------------------+----------------------------------
11388 UP 100% (non-transitional) | "inter"
11389 ----------------------------------------+----------------------------------
11390 Transitionally UP (going down "fall"), | "fastinter" if set,
11391 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11392 or yet unchecked. |
11393 ----------------------------------------+----------------------------------
11394 DOWN 100% (non-transitional) | "downinter" if set,
11395 | "inter" otherwise.
11396 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011397
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011398 Just as with every other time-based parameter, they can be entered in any
11399 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11400 serves as a timeout for health checks sent to servers if "timeout check" is
11401 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011402 hosted on the same hardware, the agent and health checks of all servers
11403 are started with a small time offset between them. It is also possible to
11404 add some random noise in the agent and health checks interval using the
11405 global "spread-checks" keyword. This makes sense for instance when a lot
11406 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011407
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011408maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011409 The "maxconn" parameter specifies the maximal number of concurrent
11410 connections that will be sent to this server. If the number of incoming
11411 concurrent requests goes higher than this value, they will be queued, waiting
11412 for a connection to be released. This parameter is very important as it can
11413 save fragile servers from going down under extreme loads. If a "minconn"
11414 parameter is specified, the limit becomes dynamic. The default value is "0"
11415 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11416 the backend's "fullconn" keyword.
11417
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011418maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011419 The "maxqueue" parameter specifies the maximal number of connections which
11420 will wait in the queue for this server. If this limit is reached, next
11421 requests will be redispatched to other servers instead of indefinitely
11422 waiting to be served. This will break persistence but may allow people to
11423 quickly re-log in when the server they try to connect to is dying. The
11424 default value is "0" which means the queue is unlimited. See also the
11425 "maxconn" and "minconn" parameters.
11426
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011427minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011428 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11429 limit following the backend's load. The server will always accept at least
11430 <minconn> connections, never more than <maxconn>, and the limit will be on
11431 the ramp between both values when the backend has less than <fullconn>
11432 concurrent connections. This makes it possible to limit the load on the
11433 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011434 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011435 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011436
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011437namespace <name>
11438 On Linux, it is possible to specify which network namespace a socket will
11439 belong to. This directive makes it possible to explicitly bind a server to
11440 a namespace different from the default one. Please refer to your operating
11441 system's documentation to find more details about network namespaces.
11442
Frédéric Lécailled2376272017-03-21 18:52:12 +010011443no-agent-check
11444 This option may be used as "server" setting to reset any "agent-check"
11445 setting which would have been inherited from "default-server" directive as
11446 default value.
11447 It may also be used as "default-server" setting to reset any previous
11448 "default-server" "agent-check" setting.
11449
11450no-backup
11451 This option may be used as "server" setting to reset any "backup"
11452 setting which would have been inherited from "default-server" directive as
11453 default value.
11454 It may also be used as "default-server" setting to reset any previous
11455 "default-server" "backup" setting.
11456
11457no-check
11458 This option may be used as "server" setting to reset any "check"
11459 setting which would have been inherited from "default-server" directive as
11460 default value.
11461 It may also be used as "default-server" setting to reset any previous
11462 "default-server" "check" setting.
11463
11464no-check-ssl
11465 This option may be used as "server" setting to reset any "check-ssl"
11466 setting which would have been inherited from "default-server" directive as
11467 default value.
11468 It may also be used as "default-server" setting to reset any previous
11469 "default-server" "check-ssl" setting.
11470
Frédéric Lécailled2376272017-03-21 18:52:12 +010011471no-send-proxy
11472 This option may be used as "server" setting to reset any "send-proxy"
11473 setting which would have been inherited from "default-server" directive as
11474 default value.
11475 It may also be used as "default-server" setting to reset any previous
11476 "default-server" "send-proxy" setting.
11477
11478no-send-proxy-v2
11479 This option may be used as "server" setting to reset any "send-proxy-v2"
11480 setting which would have been inherited from "default-server" directive as
11481 default value.
11482 It may also be used as "default-server" setting to reset any previous
11483 "default-server" "send-proxy-v2" setting.
11484
11485no-send-proxy-v2-ssl
11486 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11487 setting which would have been inherited from "default-server" directive as
11488 default value.
11489 It may also be used as "default-server" setting to reset any previous
11490 "default-server" "send-proxy-v2-ssl" setting.
11491
11492no-send-proxy-v2-ssl-cn
11493 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11494 setting which would have been inherited from "default-server" directive as
11495 default value.
11496 It may also be used as "default-server" setting to reset any previous
11497 "default-server" "send-proxy-v2-ssl-cn" setting.
11498
11499no-ssl
11500 This option may be used as "server" setting to reset any "ssl"
11501 setting which would have been inherited from "default-server" directive as
11502 default value.
11503 It may also be used as "default-server" setting to reset any previous
11504 "default-server" "ssl" setting.
11505
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011506no-ssl-reuse
11507 This option disables SSL session reuse when SSL is used to communicate with
11508 the server. It will force the server to perform a full handshake for every
11509 new connection. It's probably only useful for benchmarking, troubleshooting,
11510 and for paranoid users.
11511
Emeric Brun9b3009b2012-10-05 11:55:06 +020011512no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011513 This option disables support for SSLv3 when SSL is used to communicate with
11514 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011515 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011516
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011517 Supported in default-server: No
11518
Emeric Brunf9c5c472012-10-11 15:28:34 +020011519no-tls-tickets
11520 This setting is only available when support for OpenSSL was built in. It
11521 disables the stateless session resumption (RFC 5077 TLS Ticket
11522 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011523 session resumption is more expensive in CPU usage for servers. This option
11524 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011525 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011526
Emeric Brun9b3009b2012-10-05 11:55:06 +020011527no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011528 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011529 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11530 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011531 often makes sense to disable it when communicating with local servers. This
11532 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011533 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011534
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011535 Supported in default-server: No
11536
Emeric Brun9b3009b2012-10-05 11:55:06 +020011537no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011538 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011539 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11540 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011541 often makes sense to disable it when communicating with local servers. This
11542 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011543 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011544
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011545 Supported in default-server: No
11546
Emeric Brun9b3009b2012-10-05 11:55:06 +020011547no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011548 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011549 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11550 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011551 often makes sense to disable it when communicating with local servers. This
11552 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011553 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011554
11555 Supported in default-server: No
11556
11557no-tlsv13
11558 This option disables support for TLSv1.3 when SSL is used to communicate with
11559 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11560 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11561 often makes sense to disable it when communicating with local servers. This
11562 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011563 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011564
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011565 Supported in default-server: No
11566
Frédéric Lécailled2376272017-03-21 18:52:12 +010011567no-verifyhost
11568 This option may be used as "server" setting to reset any "verifyhost"
11569 setting which would have been inherited from "default-server" directive as
11570 default value.
11571 It may also be used as "default-server" setting to reset any previous
11572 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011573
Simon Hormanfa461682011-06-25 09:39:49 +090011574non-stick
11575 Never add connections allocated to this sever to a stick-table.
11576 This may be used in conjunction with backup to ensure that
11577 stick-table persistence is disabled for backup servers.
11578
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011579observe <mode>
11580 This option enables health adjusting based on observing communication with
11581 the server. By default this functionality is disabled and enabling it also
11582 requires to enable health checks. There are two supported modes: "layer4" and
11583 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11584 significant. In layer7, which is only allowed for http proxies, responses
11585 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011586 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011587
11588 See also the "check", "on-error" and "error-limit".
11589
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011590on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011591 Select what should happen when enough consecutive errors are detected.
11592 Currently, four modes are available:
11593 - fastinter: force fastinter
11594 - fail-check: simulate a failed check, also forces fastinter (default)
11595 - sudden-death: simulate a pre-fatal failed health check, one more failed
11596 check will mark a server down, forces fastinter
11597 - mark-down: mark the server immediately down and force fastinter
11598
11599 See also the "check", "observe" and "error-limit".
11600
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011601on-marked-down <action>
11602 Modify what occurs when a server is marked down.
11603 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011604 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11605 all connections to the server are immediately terminated when the server
11606 goes down. It might be used if the health check detects more complex cases
11607 than a simple connection status, and long timeouts would cause the service
11608 to remain unresponsive for too long a time. For instance, a health check
11609 might detect that a database is stuck and that there's no chance to reuse
11610 existing connections anymore. Connections killed this way are logged with
11611 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011612
11613 Actions are disabled by default
11614
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011615on-marked-up <action>
11616 Modify what occurs when a server is marked up.
11617 Currently one action is available:
11618 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11619 done only if the server is not in backup state and if it is not disabled
11620 (it must have an effective weight > 0). This can be used sometimes to force
11621 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011622 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11623 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011624 with extreme care. Sessions killed because a server comes up are logged
11625 with an 'U' termination code (for "Up").
11626
11627 Actions are disabled by default
11628
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011629port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011630 Using the "port" parameter, it becomes possible to use a different port to
11631 send health-checks. On some servers, it may be desirable to dedicate a port
11632 to a specific component able to perform complex tests which are more suitable
11633 to health-checks than the application. It is common to run a simple script in
11634 inetd for instance. This parameter is ignored if the "check" parameter is not
11635 set. See also the "addr" parameter.
11636
11637redir <prefix>
11638 The "redir" parameter enables the redirection mode for all GET and HEAD
11639 requests addressing this server. This means that instead of having HAProxy
11640 forward the request to the server, it will send an "HTTP 302" response with
11641 the "Location" header composed of this prefix immediately followed by the
11642 requested URI beginning at the leading '/' of the path component. That means
11643 that no trailing slash should be used after <prefix>. All invalid requests
11644 will be rejected, and all non-GET or HEAD requests will be normally served by
11645 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011646 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011647 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011648 users to a remote location in case of local disaster. Main use consists in
11649 increasing bandwidth for static servers by having the clients directly
11650 connect to them. Note: never use a relative location here, it would cause a
11651 loop between the client and HAProxy!
11652
11653 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
11654
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011655rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011656 The "rise" parameter states that a server will be considered as operational
11657 after <count> consecutive successful health checks. This value defaults to 2
11658 if unspecified. See also the "check", "inter" and "fall" parameters.
11659
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011660resolve-prefer <family>
11661 When DNS resolution is enabled for a server and multiple IP addresses from
11662 different families are returned, HAProxy will prefer using an IP address
11663 from the family mentioned in the "resolve-prefer" parameter.
11664 Available families: "ipv4" and "ipv6"
11665
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020011666 Default value: ipv6
11667
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011668 Example:
11669
11670 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011671
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011672resolve-net <network>[,<network[,...]]
11673 This options prioritize th choice of an ip address matching a network. This is
11674 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010011675 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010011676 different datacenters. The latency between datacenter is not negligible, so
11677 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011678 configured network, another address is selected.
11679
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011680 Example:
11681
11682 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010011683
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011684resolvers <id>
11685 Points to an existing "resolvers" section to resolve current server's
11686 hostname.
11687
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011688 Example:
11689
11690 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011691
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020011692 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011693
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011694send-proxy
11695 The "send-proxy" parameter enforces use of the PROXY protocol over any
11696 connection established to this server. The PROXY protocol informs the other
11697 end about the layer 3/4 addresses of the incoming connection, so that it can
11698 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011699 upper layer protocol. For connections accepted by an "accept-proxy" or
11700 "accept-netscaler-cip" listener, the advertised address will be used. Only
11701 TCPv4 and TCPv6 address families are supported. Other families such as
11702 Unix sockets, will report an UNKNOWN family. Servers using this option can
11703 fully be chained to another instance of haproxy listening with an
11704 "accept-proxy" setting. This setting must not be used if the server isn't
11705 aware of the protocol. When health checks are sent to the server, the PROXY
11706 protocol is automatically used when this option is set, unless there is an
11707 explicit "port" or "addr" directive, in which case an explicit
11708 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011709 See also the "no-send-proxy" option of this section and "accept-proxy" and
11710 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010011711
David Safb76832014-05-08 23:42:08 -040011712send-proxy-v2
11713 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
11714 over any connection established to this server. The PROXY protocol informs
11715 the other end about the layer 3/4 addresses of the incoming connection, so
11716 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020011717 whatever the upper layer protocol. It also send ALPN information if an alpn
11718 have been negotiated. This setting must not be used if the server isn't aware
11719 of this version of the protocol. See also the "no-send-proxy-v2" option of
11720 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011721
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011722proxy-v2-options <option>[,<option>]*
11723 The "proxy-v2-options" parameter add option to send in PROXY protocol version
11724 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010011725 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
11726 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010011727 certificate, "cert-key": key algorithm of the used certificate), "authority":
11728 host name value passed by the client (only sni from a tls connection is
11729 supported).
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010011730
David Safb76832014-05-08 23:42:08 -040011731send-proxy-v2-ssl
11732 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11733 2 over any connection established to this server. The PROXY protocol informs
11734 the other end about the layer 3/4 addresses of the incoming connection, so
11735 that it can know the client's address or the public address it accessed to,
11736 whatever the upper layer protocol. In addition, the SSL information extension
11737 of the PROXY protocol is added to the PROXY protocol header. This setting
11738 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011739 See also the "no-send-proxy-v2-ssl" option of this section and the
11740 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011741
11742send-proxy-v2-ssl-cn
11743 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
11744 2 over any connection established to this server. The PROXY protocol informs
11745 the other end about the layer 3/4 addresses of the incoming connection, so
11746 that it can know the client's address or the public address it accessed to,
11747 whatever the upper layer protocol. In addition, the SSL information extension
11748 of the PROXY protocol, along along with the Common Name from the subject of
11749 the client certificate (if any), is added to the PROXY protocol header. This
11750 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011751 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
11752 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040011753
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011754slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011755 The "slowstart" parameter for a server accepts a value in milliseconds which
11756 indicates after how long a server which has just come back up will run at
11757 full speed. Just as with every other time-based parameter, it can be entered
11758 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
11759 linearly from 0 to 100% during this time. The limitation applies to two
11760 parameters :
11761
11762 - maxconn: the number of connections accepted by the server will grow from 1
11763 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
11764
11765 - weight: when the backend uses a dynamic weighted algorithm, the weight
11766 grows linearly from 1 to 100%. In this case, the weight is updated at every
11767 health-check. For this reason, it is important that the "inter" parameter
11768 is smaller than the "slowstart", in order to maximize the number of steps.
11769
11770 The slowstart never applies when haproxy starts, otherwise it would cause
11771 trouble to running servers. It only applies when a server has been previously
11772 seen as failed.
11773
Willy Tarreau732eac42015-07-09 11:40:25 +020011774sni <expression>
11775 The "sni" parameter evaluates the sample fetch expression, converts it to a
11776 string and uses the result as the host name sent in the SNI TLS extension to
11777 the server. A typical use case is to send the SNI received from the client in
11778 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020011779 expression, though alternatives such as req.hdr(host) can also make sense. If
11780 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011781 name will also be matched against the server certificate's names. See the
11782 "verify" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020011783
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011784source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020011785source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011786source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011787 The "source" parameter sets the source address which will be used when
11788 connecting to the server. It follows the exact same parameters and principle
11789 as the backend "source" keyword, except that it only applies to the server
11790 referencing it. Please consult the "source" keyword for details.
11791
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020011792 Additionally, the "source" statement on a server line allows one to specify a
11793 source port range by indicating the lower and higher bounds delimited by a
11794 dash ('-'). Some operating systems might require a valid IP address when a
11795 source port range is specified. It is permitted to have the same IP/range for
11796 several servers. Doing so makes it possible to bypass the maximum of 64k
11797 total concurrent connections. The limit will then reach 64k connections per
11798 server.
11799
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000011800 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
11801 specifying the source address without port(s).
11802
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011803ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020011804 This option enables SSL ciphering on outgoing connections to the server. It
11805 is critical to verify server certificates using "verify" when using SSL to
11806 connect to servers, otherwise the communication is prone to trivial man in
11807 the-middle attacks rendering SSL useless. When this option is used, health
11808 checks are automatically sent in SSL too unless there is a "port" or an
11809 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011810 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
11811 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011812
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011813ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11814 This option enforces use of <version> or lower when SSL is used to communicate
11815 with the server. This option is also available on global statement
11816 "ssl-default-server-options". See also "ssl-min-ver".
11817
11818ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11819 This option enforces use of <version> or upper when SSL is used to communicate
11820 with the server. This option is also available on global statement
11821 "ssl-default-server-options". See also "ssl-max-ver".
11822
Frédéric Lécailled2376272017-03-21 18:52:12 +010011823ssl-reuse
11824 This option may be used as "server" setting to reset any "no-ssl-reuse"
11825 setting which would have been inherited from "default-server" directive as
11826 default value.
11827 It may also be used as "default-server" setting to reset any previous
11828 "default-server" "no-ssl-reuse" setting.
11829
11830stick
11831 This option may be used as "server" setting to reset any "non-stick"
11832 setting which would have been inherited from "default-server" directive as
11833 default value.
11834 It may also be used as "default-server" setting to reset any previous
11835 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011836
Willy Tarreau163d4622015-10-13 16:16:41 +020011837tcp-ut <delay>
11838 Sets the TCP User Timeout for all outgoing connections to this server. This
11839 option is available on Linux since version 2.6.37. It allows haproxy to
11840 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010011841 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020011842 long-lived connections experiencing long idle periods such as remote
11843 terminals or database connection pools, where the client and server timeouts
11844 must remain high to allow a long period of idle, but where it is important to
11845 detect that the server has disappeared in order to release all resources
11846 associated with its connection (and the client's session). One typical use
11847 case is also to force dead server connections to die when health checks are
11848 too slow or during a soft reload since health checks are then disabled. The
11849 argument is a delay expressed in milliseconds by default. This only works for
11850 regular TCP connections, and is ignored for other protocols.
11851
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011852track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020011853 This option enables ability to set the current state of the server by tracking
11854 another one. It is possible to track a server which itself tracks another
11855 server, provided that at the end of the chain, a server has health checks
11856 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011857 used, it has to be enabled on both proxies.
11858
Frédéric Lécailled2376272017-03-21 18:52:12 +010011859tls-tickets
11860 This option may be used as "server" setting to reset any "no-tls-tickets"
11861 setting which would have been inherited from "default-server" directive as
11862 default value.
11863 It may also be used as "default-server" setting to reset any previous
11864 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011865
Emeric Brunef42d922012-10-11 16:11:36 +020011866verify [none|required]
11867 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010011868 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011869 certificate provided by the server is verified using CAs from 'ca-file' and
11870 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011871 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011872 the name passed using the "sni" directive, or if not provided, the static
11873 host name passed using the "verifyhost" directive. When no name is found, the
11874 certificate's names are ignored. For this reason, without SNI it's important
11875 to use "verifyhost". On verification failure the handshake is aborted. It is
11876 critically important to verify server certificates when using SSL to connect
11877 to servers, otherwise the communication is prone to trivial man-in-the-middle
11878 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
11879 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020011880
Evan Broderbe554312013-06-27 00:05:25 -070011881verifyhost <hostname>
11882 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020011883 only takes effect if 'verify required' is also specified. This directive sets
11884 a default static hostname to check the server's certificate against when no
11885 SNI was used to connect to the server. If SNI is not used, this is the only
11886 way to enable hostname verification. This static hostname, when set, will
11887 also be used for health checks (which cannot provide an SNI value). If none
11888 of the hostnames in the certificate match the specified hostname, the
11889 handshake is aborted. The hostnames in the server-provided certificate may
11890 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070011891
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011892weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011893 The "weight" parameter is used to adjust the server's weight relative to
11894 other servers. All servers will receive a load proportional to their weight
11895 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020011896 load. The default weight is 1, and the maximal value is 256. A value of 0
11897 means the server will not participate in load-balancing but will still accept
11898 persistent connections. If this parameter is used to distribute the load
11899 according to server's capacity, it is recommended to start with values which
11900 can both grow and shrink, for instance between 10 and 100 to leave enough
11901 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011902
11903
Cyril Bonté46175dd2015-07-02 22:45:32 +0200119045.3. Server IP address resolution using DNS
11905-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011906
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011907HAProxy allows using a host name on the server line to retrieve its IP address
11908using name servers. By default, HAProxy resolves the name when parsing the
11909configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011910This is not sufficient in some cases, such as in Amazon where a server's IP
11911can change after a reboot or an ELB Virtual IP can change based on current
11912workload.
11913This chapter describes how HAProxy can be configured to process server's name
11914resolution at run time.
11915Whether run time server name resolution has been enable or not, HAProxy will
11916carry on doing the first resolution when parsing the configuration.
11917
11918
Cyril Bonté46175dd2015-07-02 22:45:32 +0200119195.3.1. Global overview
11920----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011921
11922As we've seen in introduction, name resolution in HAProxy occurs at two
11923different steps of the process life:
11924
11925 1. when starting up, HAProxy parses the server line definition and matches a
11926 host name. It uses libc functions to get the host name resolved. This
11927 resolution relies on /etc/resolv.conf file.
11928
Christopher Faulet67957bd2017-09-27 11:00:59 +020011929 2. at run time, HAProxy performs periodically name resolutions for servers
11930 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011931
11932A few other events can trigger a name resolution at run time:
11933 - when a server's health check ends up in a connection timeout: this may be
11934 because the server has a new IP address. So we need to trigger a name
11935 resolution to know this new IP.
11936
Christopher Faulet67957bd2017-09-27 11:00:59 +020011937When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011938HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020011939SRV label is specified, then the corresponding SRV records will be retrieved
11940from the DNS server, and the provided hostnames will be used. The SRV label
11941will be checked periodically, and if any server are added or removed, haproxy
11942will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020011943
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011944A few things important to notice:
11945 - all the name servers are queried in the mean time. HAProxy will process the
11946 first valid response.
11947
11948 - a resolution is considered as invalid (NX, timeout, refused), when all the
11949 servers return an error.
11950
11951
Cyril Bonté46175dd2015-07-02 22:45:32 +0200119525.3.2. The resolvers section
11953----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011954
11955This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020011956HAProxy. There can be as many as resolvers section as needed. Each section can
11957contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020011958
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011959When multiple name servers are configured in a resolvers section, then HAProxy
11960uses the first valid response. In case of invalid responses, only the last one
11961is treated. Purpose is to give the chance to a slow server to deliver a valid
11962answer after a fast faulty or outdated server.
11963
11964When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020011965used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011966
Christopher Faulet67957bd2017-09-27 11:00:59 +020011967 1. HAProxy retries the same DNS query with a new query type. The A queries are
11968 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
11969 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011970
Christopher Faulet67957bd2017-09-27 11:00:59 +020011971 2. When the fallback on the query type was done (or not applicable), HAProxy
11972 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011973
Christopher Faulet67957bd2017-09-27 11:00:59 +020011974 3. HAProxy retries previous steps <resolve_retires> times. If no valid
11975 response is received after that, it stops the DNS resolution and reports
11976 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011977
Christopher Faulet67957bd2017-09-27 11:00:59 +020011978For example, with 2 name servers configured in a resolvers section, the
11979following scenarios are possible:
11980
11981 - First response is valid and is applied directly, second response is
11982 ignored
11983
11984 - First response is invalid and second one is valid, then second response is
11985 applied
11986
11987 - First response is a NX domain and second one a truncated response, then
11988 HAProxy retries the query with a new type
11989
11990 - First response is a NX domain and second one is a timeout, then HAProxy
11991 retries the query with a new type
11992
11993 - Query timed out for both name servers, then HAProxy retries it with the
11994 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020011995
Olivier Houcharda8c6db82017-07-06 18:46:47 +020011996As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
11997a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020011998<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020011999
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012000
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012001resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012002 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012003
12004A resolvers section accept the following parameters:
12005
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012006accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012007 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012008 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012009 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12010 by RFC 6891)
12011
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012012 Note: the maximum allowed value is 8192.
12013
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012014nameserver <id> <ip>:<port>
12015 DNS server description:
12016 <id> : label of the server, should be unique
12017 <ip> : IP address of the server
12018 <port> : port where the DNS service actually runs
12019
12020hold <status> <period>
12021 Defines <period> during which the last name resolution should be kept based
12022 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012023 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012024 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012025 <period> : interval between two successive name resolution when the last
12026 answer was in <status>. It follows the HAProxy time format.
12027 <period> is in milliseconds by default.
12028
Baptiste Assmann686408b2017-08-18 10:15:42 +020012029 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012030
Christopher Faulet67957bd2017-09-27 11:00:59 +020012031resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012032 Defines the number of resolutions available in the pool for this resolvers.
12033 If not defines, it defaults to 64. If your configuration requires more than
12034 <nb>, then HAProxy will return an error when parsing the configuration.
12035
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012036resolve_retries <nb>
12037 Defines the number <nb> of queries to send to resolve a server name before
12038 giving up.
12039 Default value: 3
12040
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012041 A retry occurs on name server timeout or when the full sequence of DNS query
12042 type failover is over and we need to start up from the default ANY query
12043 type.
12044
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012045timeout <event> <time>
12046 Defines timeouts related to name resolution
12047 <event> : the event on which the <time> timeout period applies to.
12048 events available are:
Christopher Faulet67957bd2017-09-27 11:00:59 +020012049 - resolve : default time to trigger name resolutions when no
12050 other time applied.
12051 Default value: 1s
12052 - retry : time between two DNS queries, when no valid response
12053 have been received.
12054 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012055 <time> : time related to the event. It follows the HAProxy time format.
12056 <time> is expressed in milliseconds.
12057
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012058 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012059
12060 resolvers mydns
12061 nameserver dns1 10.0.0.1:53
12062 nameserver dns2 10.0.0.2:53
12063 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012064 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012065 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012066 hold other 30s
12067 hold refused 30s
12068 hold nx 30s
12069 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012070 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012071 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012072
12073
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200120746. HTTP header manipulation
12075---------------------------
12076
12077In HTTP mode, it is possible to rewrite, add or delete some of the request and
12078response headers based on regular expressions. It is also possible to block a
12079request or a response if a particular header matches a regular expression,
12080which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012081against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012082
Willy Tarreau70dffda2014-01-30 03:07:23 +010012083If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12084to process all rsp* rules which can allow, deny, rewrite or delete a header,
12085but it will refuse to add a header to any such messages as this is not
12086HTTP-compliant. The reason for still processing headers in such responses is to
12087stop and/or fix any possible information leak which may happen, for instance
12088because another downstream equipment would unconditionally add a header, or if
12089a server name appears there. When such messages are seen, normal processing
12090still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012091
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012092This section covers common usage of the following keywords, described in detail
12093in section 4.2 :
12094
12095 - reqadd <string>
12096 - reqallow <search>
12097 - reqiallow <search>
12098 - reqdel <search>
12099 - reqidel <search>
12100 - reqdeny <search>
12101 - reqideny <search>
12102 - reqpass <search>
12103 - reqipass <search>
12104 - reqrep <search> <replace>
12105 - reqirep <search> <replace>
12106 - reqtarpit <search>
12107 - reqitarpit <search>
12108 - rspadd <string>
12109 - rspdel <search>
12110 - rspidel <search>
12111 - rspdeny <search>
12112 - rspideny <search>
12113 - rsprep <search> <replace>
12114 - rspirep <search> <replace>
12115
12116With all these keywords, the same conventions are used. The <search> parameter
12117is a POSIX extended regular expression (regex) which supports grouping through
12118parenthesis (without the backslash). Spaces and other delimiters must be
12119prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12120Other characters may be prefixed with a backslash to change their meaning :
12121
12122 \t for a tab
12123 \r for a carriage return (CR)
12124 \n for a new line (LF)
12125 \ to mark a space and differentiate it from a delimiter
12126 \# to mark a sharp and differentiate it from a comment
12127 \\ to use a backslash in a regex
12128 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12129 \xXX to write the ASCII hex code XX as in the C language
12130
12131The <replace> parameter contains the string to be used to replace the largest
12132portion of text matching the regex. It can make use of the special characters
12133above, and can reference a substring which is delimited by parenthesis in the
12134regex, by writing a backslash ('\') immediately followed by one digit from 0 to
121359 indicating the group position (0 designating the entire line). This practice
12136is very common to users of the "sed" program.
12137
12138The <string> parameter represents the string which will systematically be added
12139after the last header line. It can also use special character sequences above.
12140
12141Notes related to these keywords :
12142---------------------------------
12143 - these keywords are not always convenient to allow/deny based on header
12144 contents. It is strongly recommended to use ACLs with the "block" keyword
12145 instead, resulting in far more flexible and manageable rules.
12146
12147 - lines are always considered as a whole. It is not possible to reference
12148 a header name only or a value only. This is important because of the way
12149 headers are written (notably the number of spaces after the colon).
12150
12151 - the first line is always considered as a header, which makes it possible to
12152 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12153 it harder to distinguish between headers and request line. The regex prefix
12154 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12155 ^[^ \t:]*: matches any header name followed by a colon.
12156
12157 - for performances reasons, the number of characters added to a request or to
12158 a response is limited at build time to values between 1 and 4 kB. This
12159 should normally be far more than enough for most usages. If it is too short
12160 on occasional usages, it is possible to gain some space by removing some
12161 useless headers before adding new ones.
12162
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012163 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012164 without the 'i' letter except that they ignore case when matching patterns.
12165
12166 - when a request passes through a frontend then a backend, all req* rules
12167 from the frontend will be evaluated, then all req* rules from the backend
12168 will be evaluated. The reverse path is applied to responses.
12169
12170 - req* statements are applied after "block" statements, so that "block" is
12171 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012172 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012173
12174
Willy Tarreau74ca5042013-06-11 23:12:07 +0200121757. Using ACLs and fetching samples
12176----------------------------------
12177
Davor Ocelice9ed2812017-12-25 17:49:28 +010012178HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012179client or server information, from tables, environmental information etc...
12180The action of extracting such data is called fetching a sample. Once retrieved,
12181these samples may be used for various purposes such as a key to a stick-table,
12182but most common usages consist in matching them against predefined constant
12183data called patterns.
12184
12185
121867.1. ACL basics
12187---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012188
12189The use of Access Control Lists (ACL) provides a flexible solution to perform
12190content switching and generally to take decisions based on content extracted
12191from the request, the response or any environmental status. The principle is
12192simple :
12193
Willy Tarreau74ca5042013-06-11 23:12:07 +020012194 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012195 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012196 - apply one or multiple pattern matching methods on this sample
12197 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012198
Willy Tarreau74ca5042013-06-11 23:12:07 +020012199The actions generally consist in blocking a request, selecting a backend, or
12200adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012201
12202In order to define a test, the "acl" keyword is used. The syntax is :
12203
Willy Tarreau74ca5042013-06-11 23:12:07 +020012204 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012205
12206This creates a new ACL <aclname> or completes an existing one with new tests.
12207Those tests apply to the portion of request/response specified in <criterion>
12208and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012209an operator which may be specified before the set of values. Optionally some
12210conversion operators may be applied to the sample, and they will be specified
12211as a comma-delimited list of keywords just after the first keyword. The values
12212are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012213
12214ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12215'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12216which means that "my_acl" and "My_Acl" are two different ACLs.
12217
12218There is no enforced limit to the number of ACLs. The unused ones do not affect
12219performance, they just consume a small amount of memory.
12220
Willy Tarreau74ca5042013-06-11 23:12:07 +020012221The criterion generally is the name of a sample fetch method, or one of its ACL
12222specific declinations. The default test method is implied by the output type of
12223this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012224methods of a same sample fetch method. The sample fetch methods are the only
12225ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012226
12227Sample fetch methods return data which can be of the following types :
12228 - boolean
12229 - integer (signed or unsigned)
12230 - IPv4 or IPv6 address
12231 - string
12232 - data block
12233
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012234Converters transform any of these data into any of these. For example, some
12235converters might convert a string to a lower-case string while other ones
12236would turn a string to an IPv4 address, or apply a netmask to an IP address.
12237The resulting sample is of the type of the last converter applied to the list,
12238which defaults to the type of the sample fetch method.
12239
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012240Each sample or converter returns data of a specific type, specified with its
12241keyword in this documentation. When an ACL is declared using a standard sample
12242fetch method, certain types automatically involved a default matching method
12243which are summarized in the table below :
12244
12245 +---------------------+-----------------+
12246 | Sample or converter | Default |
12247 | output type | matching method |
12248 +---------------------+-----------------+
12249 | boolean | bool |
12250 +---------------------+-----------------+
12251 | integer | int |
12252 +---------------------+-----------------+
12253 | ip | ip |
12254 +---------------------+-----------------+
12255 | string | str |
12256 +---------------------+-----------------+
12257 | binary | none, use "-m" |
12258 +---------------------+-----------------+
12259
12260Note that in order to match a binary samples, it is mandatory to specify a
12261matching method, see below.
12262
Willy Tarreau74ca5042013-06-11 23:12:07 +020012263The ACL engine can match these types against patterns of the following types :
12264 - boolean
12265 - integer or integer range
12266 - IP address / network
12267 - string (exact, substring, suffix, prefix, subdir, domain)
12268 - regular expression
12269 - hex block
12270
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012271The following ACL flags are currently supported :
12272
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012273 -i : ignore case during matching of all subsequent patterns.
12274 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012275 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012276 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012277 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012278 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012279 -- : force end of flags. Useful when a string looks like one of the flags.
12280
Willy Tarreau74ca5042013-06-11 23:12:07 +020012281The "-f" flag is followed by the name of a file from which all lines will be
12282read as individual values. It is even possible to pass multiple "-f" arguments
12283if the patterns are to be loaded from multiple files. Empty lines as well as
12284lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12285will be stripped. If it is absolutely necessary to insert a valid pattern
12286beginning with a sharp, just prefix it with a space so that it is not taken for
12287a comment. Depending on the data type and match method, haproxy may load the
12288lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12289exact string matching. In this case, duplicates will automatically be removed.
12290
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012291The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12292parsed as two column file. The first column contains the patterns used by the
12293ACL, and the second column contain the samples. The sample can be used later by
12294a map. This can be useful in some rare cases where an ACL would just be used to
12295check for the existence of a pattern in a map before a mapping is applied.
12296
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012297The "-u" flag forces the unique id of the ACL. This unique id is used with the
12298socket interface to identify ACL and dynamically change its values. Note that a
12299file is always identified by its name even if an id is set.
12300
Willy Tarreau74ca5042013-06-11 23:12:07 +020012301Also, note that the "-i" flag applies to subsequent entries and not to entries
12302loaded from files preceding it. For instance :
12303
12304 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12305
12306In this example, each line of "exact-ua.lst" will be exactly matched against
12307the "user-agent" header of the request. Then each line of "generic-ua" will be
12308case-insensitively matched. Then the word "test" will be insensitively matched
12309as well.
12310
12311The "-m" flag is used to select a specific pattern matching method on the input
12312sample. All ACL-specific criteria imply a pattern matching method and generally
12313do not need this flag. However, this flag is useful with generic sample fetch
12314methods to describe how they're going to be matched against the patterns. This
12315is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012316obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012317followed by a pattern matching method name, this method is used instead of the
12318default one for the criterion. This makes it possible to match contents in ways
12319that were not initially planned, or with sample fetch methods which return a
12320string. The matching method also affects the way the patterns are parsed.
12321
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012322The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12323By default, if the parser cannot parse ip address it considers that the parsed
12324string is maybe a domain name and try dns resolution. The flag "-n" disable this
12325resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12326server is not reachable, the haproxy configuration parsing may last many minutes
12327waiting fir the timeout. During this time no error messages are displayed. The
12328flag "-n" disable this behavior. Note also that during the runtime, this
12329function is disabled for the dynamic acl modifications.
12330
Willy Tarreau74ca5042013-06-11 23:12:07 +020012331There are some restrictions however. Not all methods can be used with all
12332sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12333be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012334
12335 - "found" : only check if the requested sample could be found in the stream,
12336 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012337 to pass any pattern to avoid confusion. This matching method is
12338 particularly useful to detect presence of certain contents such
12339 as headers, cookies, etc... even if they are empty and without
12340 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012341
12342 - "bool" : check the value as a boolean. It can only be applied to fetches
12343 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012344 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012345
12346 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012347 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012348
12349 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012350 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012351
Davor Ocelice9ed2812017-12-25 17:49:28 +010012352 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012353 binary sequence. This may be used with binary or string samples.
12354
12355 - "len" : match the sample's length as an integer. This may be used with
12356 binary or string samples.
12357
Willy Tarreau74ca5042013-06-11 23:12:07 +020012358 - "str" : exact match : match the contents against a string. This may be
12359 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012360
Willy Tarreau74ca5042013-06-11 23:12:07 +020012361 - "sub" : substring match : check that the contents contain at least one of
12362 the provided string patterns. This may be used with binary or
12363 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012364
Willy Tarreau74ca5042013-06-11 23:12:07 +020012365 - "reg" : regex match : match the contents against a list of regular
12366 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012367
Willy Tarreau74ca5042013-06-11 23:12:07 +020012368 - "beg" : prefix match : check that the contents begin like the provided
12369 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012370
Willy Tarreau74ca5042013-06-11 23:12:07 +020012371 - "end" : suffix match : check that the contents end like the provided
12372 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012373
Willy Tarreau74ca5042013-06-11 23:12:07 +020012374 - "dir" : subdir match : check that a slash-delimited portion of the
12375 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012376 This may be used with binary or string samples.
12377
Willy Tarreau74ca5042013-06-11 23:12:07 +020012378 - "dom" : domain match : check that a dot-delimited portion of the contents
12379 exactly match one of the provided string patterns. This may be
12380 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012381
12382For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12383request, it is possible to do :
12384
12385 acl jsess_present cook(JSESSIONID) -m found
12386
12387In order to apply a regular expression on the 500 first bytes of data in the
12388buffer, one would use the following acl :
12389
12390 acl script_tag payload(0,500) -m reg -i <script>
12391
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012392On systems where the regex library is much slower when using "-i", it is
12393possible to convert the sample to lowercase before matching, like this :
12394
12395 acl script_tag payload(0,500),lower -m reg <script>
12396
Willy Tarreau74ca5042013-06-11 23:12:07 +020012397All ACL-specific criteria imply a default matching method. Most often, these
12398criteria are composed by concatenating the name of the original sample fetch
12399method and the matching method. For example, "hdr_beg" applies the "beg" match
12400to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12401criteria rely on a sample fetch method, it is always possible instead to use
12402the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012403
Willy Tarreau74ca5042013-06-11 23:12:07 +020012404If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012405the matching method is simply applied to the underlying sample fetch method.
12406For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012407
Willy Tarreau74ca5042013-06-11 23:12:07 +020012408 acl short_form hdr_beg(host) www.
12409 acl alternate1 hdr_beg(host) -m beg www.
12410 acl alternate2 hdr_dom(host) -m beg www.
12411 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012412
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012413
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012414The table below summarizes the compatibility matrix between sample or converter
12415types and the pattern types to fetch against. It indicates for each compatible
12416combination the name of the matching method to be used, surrounded with angle
12417brackets ">" and "<" when the method is the default one and will work by
12418default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012419
Willy Tarreau74ca5042013-06-11 23:12:07 +020012420 +-------------------------------------------------+
12421 | Input sample type |
12422 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012423 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012424 +----------------------+---------+---------+---------+---------+---------+
12425 | none (presence only) | found | found | found | found | found |
12426 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012427 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012428 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012429 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012430 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012431 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012432 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012433 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012434 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012435 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012436 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012437 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012438 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012439 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012440 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012441 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012442 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012443 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012444 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012445 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012446 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012447 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012448 +----------------------+---------+---------+---------+---------+---------+
12449 | hex block | | | | bin | bin |
12450 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012451
12452
Willy Tarreau74ca5042013-06-11 23:12:07 +0200124537.1.1. Matching booleans
12454------------------------
12455
12456In order to match a boolean, no value is needed and all values are ignored.
12457Boolean matching is used by default for all fetch methods of type "boolean".
12458When boolean matching is used, the fetched value is returned as-is, which means
12459that a boolean "true" will always match and a boolean "false" will never match.
12460
12461Boolean matching may also be enforced using "-m bool" on fetch methods which
12462return an integer value. Then, integer value 0 is converted to the boolean
12463"false" and all other values are converted to "true".
12464
Willy Tarreau6a06a402007-07-15 20:15:28 +020012465
Willy Tarreau74ca5042013-06-11 23:12:07 +0200124667.1.2. Matching integers
12467------------------------
12468
12469Integer matching applies by default to integer fetch methods. It can also be
12470enforced on boolean fetches using "-m int". In this case, "false" is converted
12471to the integer 0, and "true" is converted to the integer 1.
12472
12473Integer matching also supports integer ranges and operators. Note that integer
12474matching only applies to positive values. A range is a value expressed with a
12475lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012476
12477For instance, "1024:65535" is a valid range to represent a range of
12478unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12479representation of privileged ports, and ":1023" would also work.
12480
Willy Tarreau62644772008-07-16 18:36:06 +020012481As a special case, some ACL functions support decimal numbers which are in fact
12482two integers separated by a dot. This is used with some version checks for
12483instance. All integer properties apply to those decimal numbers, including
12484ranges and operators.
12485
Willy Tarreau6a06a402007-07-15 20:15:28 +020012486For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012487operators with ranges does not make much sense and is strongly discouraged.
12488Similarly, it does not make much sense to perform order comparisons with a set
12489of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012490
Willy Tarreau0ba27502007-12-24 16:55:16 +010012491Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012492
12493 eq : true if the tested value equals at least one value
12494 ge : true if the tested value is greater than or equal to at least one value
12495 gt : true if the tested value is greater than at least one value
12496 le : true if the tested value is less than or equal to at least one value
12497 lt : true if the tested value is less than at least one value
12498
Willy Tarreau0ba27502007-12-24 16:55:16 +010012499For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012500
12501 acl negative-length hdr_val(content-length) lt 0
12502
Willy Tarreau62644772008-07-16 18:36:06 +020012503This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12504
12505 acl sslv3 req_ssl_ver 3:3.1
12506
Willy Tarreau6a06a402007-07-15 20:15:28 +020012507
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125087.1.3. Matching strings
12509-----------------------
12510
12511String matching applies to string or binary fetch methods, and exists in 6
12512different forms :
12513
12514 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012515 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012516
12517 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012518 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012519
12520 - prefix match (-m beg) : the patterns are compared with the beginning of
12521 the extracted string, and the ACL matches if any of them matches.
12522
12523 - suffix match (-m end) : the patterns are compared with the end of the
12524 extracted string, and the ACL matches if any of them matches.
12525
Baptiste Assmann33db6002016-03-06 23:32:10 +010012526 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012527 string, delimited with slashes ("/"), and the ACL matches if any of them
12528 matches.
12529
12530 - domain match (-m dom) : the patterns are looked up inside the extracted
12531 string, delimited with dots ("."), and the ACL matches if any of them
12532 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012533
12534String matching applies to verbatim strings as they are passed, with the
12535exception of the backslash ("\") which makes it possible to escape some
12536characters such as the space. If the "-i" flag is passed before the first
12537string, then the matching will be performed ignoring the case. In order
12538to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012539before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012540
12541
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125427.1.4. Matching regular expressions (regexes)
12543---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012544
12545Just like with string matching, regex matching applies to verbatim strings as
12546they are passed, with the exception of the backslash ("\") which makes it
12547possible to escape some characters such as the space. If the "-i" flag is
12548passed before the first regex, then the matching will be performed ignoring
12549the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012550the "--" flag before the first string. Same principle applies of course to
12551match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012552
12553
Willy Tarreau74ca5042013-06-11 23:12:07 +0200125547.1.5. Matching arbitrary data blocks
12555-------------------------------------
12556
12557It is possible to match some extracted samples against a binary block which may
12558not safely be represented as a string. For this, the patterns must be passed as
12559a series of hexadecimal digits in an even number, when the match method is set
12560to binary. Each sequence of two digits will represent a byte. The hexadecimal
12561digits may be used upper or lower case.
12562
12563Example :
12564 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12565 acl hello payload(0,6) -m bin 48656c6c6f0a
12566
12567
125687.1.6. Matching IPv4 and IPv6 addresses
12569---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012570
12571IPv4 addresses values can be specified either as plain addresses or with a
12572netmask appended, in which case the IPv4 address matches whenever it is
12573within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012574host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010012575difficult to read and debug configurations. If hostnames are used, you should
12576at least ensure that they are present in /etc/hosts so that the configuration
12577does not depend on any random DNS match at the moment the configuration is
12578parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012579
Daniel Schnellereba56342016-04-13 00:26:52 +020012580The dotted IPv4 address notation is supported in both regular as well as the
12581abbreviated form with all-0-octets omitted:
12582
12583 +------------------+------------------+------------------+
12584 | Example 1 | Example 2 | Example 3 |
12585 +------------------+------------------+------------------+
12586 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
12587 | 192.168.1 | 10.12 | 127.1 |
12588 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
12589 | 192.168.1/22 | 10.12/8 | 127.1/8 |
12590 +------------------+------------------+------------------+
12591
12592Notice that this is different from RFC 4632 CIDR address notation in which
12593192.168.42/24 would be equivalent to 192.168.42.0/24.
12594
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012595IPv6 may be entered in their usual form, with or without a netmask appended.
12596Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
12597trouble with randomly resolved IP addresses, host names are never allowed in
12598IPv6 patterns.
12599
12600HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
12601following situations :
12602 - tested address is IPv4, pattern address is IPv4, the match applies
12603 in IPv4 using the supplied mask if any.
12604 - tested address is IPv6, pattern address is IPv6, the match applies
12605 in IPv6 using the supplied mask if any.
12606 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
12607 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
12608 ::IPV4 or ::ffff:IPV4, otherwise it fails.
12609 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
12610 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
12611 applied in IPv6 using the supplied IPv6 mask.
12612
Willy Tarreau74ca5042013-06-11 23:12:07 +020012613
126147.2. Using ACLs to form conditions
12615----------------------------------
12616
12617Some actions are only performed upon a valid condition. A condition is a
12618combination of ACLs with operators. 3 operators are supported :
12619
12620 - AND (implicit)
12621 - OR (explicit with the "or" keyword or the "||" operator)
12622 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020012623
Willy Tarreau74ca5042013-06-11 23:12:07 +020012624A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012625
Willy Tarreau74ca5042013-06-11 23:12:07 +020012626 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020012627
Willy Tarreau74ca5042013-06-11 23:12:07 +020012628Such conditions are generally used after an "if" or "unless" statement,
12629indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020012630
Willy Tarreau74ca5042013-06-11 23:12:07 +020012631For instance, to block HTTP requests to the "*" URL with methods other than
12632"OPTIONS", as well as POST requests without content-length, and GET or HEAD
12633requests with a content-length greater than 0, and finally every request which
12634is not either GET/HEAD/POST/OPTIONS !
12635
12636 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012637 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
12638 http-request deny if METH_GET HTTP_CONTENT
12639 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020012640
12641To select a different backend for requests to static contents on the "www" site
12642and to every request on the "img", "video", "download" and "ftp" hosts :
12643
12644 acl url_static path_beg /static /images /img /css
12645 acl url_static path_end .gif .png .jpg .css .js
12646 acl host_www hdr_beg(host) -i www
12647 acl host_static hdr_beg(host) -i img. video. download. ftp.
12648
Davor Ocelice9ed2812017-12-25 17:49:28 +010012649 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020012650 # of host "www". Use backend "www" for the rest.
12651 use_backend static if host_static or host_www url_static
12652 use_backend www if host_www
12653
12654It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
12655expressions that are built on the fly without needing to be declared. They must
12656be enclosed between braces, with a space before and after each brace (because
12657the braces must be seen as independent words). Example :
12658
12659 The following rule :
12660
12661 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012662 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020012663
12664 Can also be written that way :
12665
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012666 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020012667
12668It is generally not recommended to use this construct because it's a lot easier
12669to leave errors in the configuration when written that way. However, for very
12670simple rules matching only one source IP address for instance, it can make more
12671sense to use them than to declare ACLs with random names. Another example of
12672good use is the following :
12673
12674 With named ACLs :
12675
12676 acl site_dead nbsrv(dynamic) lt 2
12677 acl site_dead nbsrv(static) lt 2
12678 monitor fail if site_dead
12679
12680 With anonymous ACLs :
12681
12682 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
12683
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030012684See section 4.2 for detailed help on the "http-request deny" and "use_backend"
12685keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012686
12687
126887.3. Fetching samples
12689---------------------
12690
12691Historically, sample fetch methods were only used to retrieve data to match
12692against patterns using ACLs. With the arrival of stick-tables, a new class of
12693sample fetch methods was created, most often sharing the same syntax as their
12694ACL counterpart. These sample fetch methods are also known as "fetches". As
12695of now, ACLs and fetches have converged. All ACL fetch methods have been made
12696available as fetch methods, and ACLs may use any sample fetch method as well.
12697
12698This section details all available sample fetch methods and their output type.
12699Some sample fetch methods have deprecated aliases that are used to maintain
12700compatibility with existing configurations. They are then explicitly marked as
12701deprecated and should not be used in new setups.
12702
12703The ACL derivatives are also indicated when available, with their respective
12704matching methods. These ones all have a well defined default pattern matching
12705method, so it is never necessary (though allowed) to pass the "-m" option to
12706indicate how the sample will be matched using ACLs.
12707
12708As indicated in the sample type versus matching compatibility matrix above,
12709when using a generic sample fetch method in an ACL, the "-m" option is
12710mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
12711the same keyword exists as an ACL keyword and as a standard fetch method, the
12712ACL engine will automatically pick the ACL-only one by default.
12713
12714Some of these keywords support one or multiple mandatory arguments, and one or
12715multiple optional arguments. These arguments are strongly typed and are checked
12716when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012717incorrect argument (e.g. an unresolved backend name). Fetch function arguments
12718are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020012719is optional, it will be indicated below between square brackets ('[ ]'). When
12720all arguments are optional, the parenthesis may be omitted.
12721
12722Thus, the syntax of a standard sample fetch method is one of the following :
12723 - name
12724 - name(arg1)
12725 - name(arg1,arg2)
12726
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012727
127287.3.1. Converters
12729-----------------
12730
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012731Sample fetch methods may be combined with transformations to be applied on top
12732of the fetched sample (also called "converters"). These combinations form what
12733is called "sample expressions" and the result is a "sample". Initially this
12734was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012735has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012736unique-id-format, add-header, ...).
12737
12738These transformations are enumerated as a series of specific keywords after the
12739sample fetch method. These keywords may equally be appended immediately after
12740the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010012741support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012742
Willy Tarreau97707872015-01-27 15:12:13 +010012743A certain category of converters are bitwise and arithmetic operators which
12744support performing basic operations on integers. Some bitwise operations are
12745supported (and, or, xor, cpl) and some arithmetic operations are supported
12746(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
12747bool) which make it possible to report a match without having to write an ACL.
12748
Willy Tarreau74ca5042013-06-11 23:12:07 +020012749The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010012750
Ben Shillitof25e8e52016-12-02 14:25:37 +00001275151d.single(<prop>[,<prop>*])
12752 Returns values for the properties requested as a string, where values are
12753 separated by the delimiter specified with "51degrees-property-separator".
12754 The device is identified using the User-Agent header passed to the
12755 converter. The function can be passed up to five property names, and if a
12756 property name can't be found, the value "NoData" is returned.
12757
12758 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010012759 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
12760 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000012761 # User-Agent passed to the converter.
12762 frontend http-in
12763 bind *:8081
12764 default_backend servers
12765 http-request set-header X-51D-DeviceTypeMobileTablet \
12766 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
12767
Willy Tarreau97707872015-01-27 15:12:13 +010012768add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012769 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012770 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010012771 name. The name of the variable starts with an indication about its scope. The
12772 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012773 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012774 "sess" : the variable is shared with the whole session
12775 "txn" : the variable is shared with the transaction (request and response)
12776 "req" : the variable is shared only during request processing
12777 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012778 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012779 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012780
12781and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012782 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012783 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010012784 numeric value or a variable name. The name of the variable starts with an
12785 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012786 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012787 "sess" : the variable is shared with the whole session
12788 "txn" : the variable is shared with the transaction (request and response)
12789 "req" : the variable is shared only during request processing
12790 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012791 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012792 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012793
Holger Just1bfc24b2017-05-06 00:56:53 +020012794b64dec
12795 Converts (decodes) a base64 encoded input string to its binary
12796 representation. It performs the inverse operation of base64().
12797
Emeric Brun53d1a982014-04-30 18:21:37 +020012798base64
12799 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010012800 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020012801 an SSL ID can be copied in a header).
12802
Willy Tarreau97707872015-01-27 15:12:13 +010012803bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012804 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010012805 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010012806 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010012807 presence of a flag).
12808
Emeric Brun54c4ac82014-11-03 15:32:43 +010012809bytes(<offset>[,<length>])
12810 Extracts some bytes from an input binary sample. The result is a binary
12811 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010012812 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010012813
Willy Tarreau280f42b2018-02-19 15:34:12 +010012814concat([<start>],[<var>],[<end>])
12815 Concatenates up to 3 fields after the current sample which is then turned to
12816 a string. The first one, <start>, is a constant string, that will be appended
12817 immediately after the existing sample. It may be omitted if not used. The
12818 second one, <var>, is a variable name. The variable will be looked up, its
12819 contents converted to a string, and it will be appended immediately after the
12820 <first> part. If the variable is not found, nothing is appended. It may be
12821 omitted as well. The third field, <end> is a constant string that will be
12822 appended after the variable. It may also be omitted. Together, these elements
12823 allow to concatenate variables with delimiters to an existing set of
12824 variables. This can be used to build new variables made of a succession of
12825 other variables, such as colon-delimited varlues. Note that due to the config
12826 parser, it is not possible to use a comma nor a closing parenthesis as
12827 delimitors.
12828
12829 Example:
12830 tcp-request session set-var(sess.src) src
12831 tcp-request session set-var(sess.dn) ssl_c_s_dn
12832 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
12833 http-request set-header x-hap-sig %[var(txn.sig)]
12834
Willy Tarreau97707872015-01-27 15:12:13 +010012835cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012836 Takes the input value of type signed integer, applies a ones-complement
12837 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010012838
Willy Tarreau80599772015-01-20 19:35:24 +010012839crc32([<avalanche>])
12840 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
12841 hash function. Optionally, it is possible to apply a full avalanche hash
12842 function to the output if the optional <avalanche> argument equals 1. This
12843 converter uses the same functions as used by the various hash-based load
12844 balancing algorithms, so it will provide exactly the same results. It is
12845 provided for compatibility with other software which want a CRC32 to be
12846 computed on some input keys, so it follows the most common implementation as
12847 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
12848 but may provide a better or at least less predictable distribution. It must
12849 not be used for security purposes as a 32-bit hash is trivial to break. See
12850 also "djb2", "sdbm", "wt6" and the "hash-type" directive.
12851
David Carlier29b3ca32015-09-25 14:09:21 +010012852da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020012853 Asks the DeviceAtlas converter to identify the User Agent string passed on
12854 input, and to emit a string made of the concatenation of the properties
12855 enumerated in argument, delimited by the separator defined by the global
12856 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000012857 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020012858 configuration language.
12859
12860 Example:
12861 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020012862 bind *:8881
12863 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000012864 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 +020012865
Thierry FOURNIER9687c772015-05-07 15:46:29 +020012866debug
12867 This converter is used as debug tool. It dumps on screen the content and the
12868 type of the input sample. The sample is returned as is on its output. This
12869 converter only exists when haproxy was built with debugging enabled.
12870
Willy Tarreau97707872015-01-27 15:12:13 +010012871div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012872 Divides the input value of type signed integer by <value>, and returns the
12873 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020012874 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010012875 variable name. The name of the variable starts with an indication about its
12876 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012877 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012878 "sess" : the variable is shared with the whole session
12879 "txn" : the variable is shared with the transaction (request and response)
12880 "req" : the variable is shared only during request processing
12881 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012882 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010012883 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010012884
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012885djb2([<avalanche>])
12886 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
12887 hash function. Optionally, it is possible to apply a full avalanche hash
12888 function to the output if the optional <avalanche> argument equals 1. This
12889 converter uses the same functions as used by the various hash-based load
12890 balancing algorithms, so it will provide exactly the same results. It is
12891 mostly intended for debugging, but can be used as a stick-table entry to
12892 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010012893 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6" and the
12894 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020012895
Willy Tarreau97707872015-01-27 15:12:13 +010012896even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020012897 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010012898 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
12899
Emeric Brunf399b0d2014-11-03 17:07:03 +010012900field(<index>,<delimiters>)
12901 Extracts the substring at the given index considering given delimiters from
12902 an input string. Indexes start at 1 and delimiters are a string formatted
12903 list of chars.
12904
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012905hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010012906 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012907 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010012908 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 +020012909 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010012910
Dragan Dosen3f957b22017-10-24 09:27:34 +020012911hex2i
12912 Converts a hex string containing two hex digits per input byte to an
12913 integer. If the input value can not be converted, then zero is returned.
12914
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012915http_date([<offset>])
12916 Converts an integer supposed to contain a date since epoch to a string
12917 representing this date in a format suitable for use in HTTP header fields. If
12918 an offset value is specified, then it is a number of seconds that is added to
12919 the date before the conversion is operated. This is particularly useful to
12920 emit Date header fields, Expires values in responses when combined with a
12921 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012922
Willy Tarreaud9f316a2014-07-10 14:03:38 +020012923in_table(<table>)
12924 Uses the string representation of the input sample to perform a look up in
12925 the specified table. If the key is not found in the table, a boolean false
12926 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010012927 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020012928 or not a source IP address or an Authorization header was already seen).
12929
Tim Duesterhus1478aa72018-01-25 16:24:51 +010012930ipmask(<mask4>, [<mask6>])
12931 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020012932 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010012933 table entries and as such use the same server. The mask4 can be passed in
12934 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
12935 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
12936 If no mask6 is given IPv6 addresses will fail to convert for backwards
12937 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020012938
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012939json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010012940 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012941 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020012942 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012943 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
12944 of errors:
12945 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
12946 bytes, ...)
12947 - invalid range (the decoded value is within a UTF-8 prohibited range),
12948 - code overlong (the value is encoded with more bytes than necessary).
12949
12950 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
12951 character is greater than 0xffff because the JSON string escape specification
12952 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
12953 in 4 variants designated by a combination of two suffix letters : "p" for
12954 "permissive" and "s" for "silently ignore". The behaviors of the decoders
12955 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010012956 - "ascii" : never fails;
12957 - "utf8" : fails on any detected errors;
12958 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012959 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010012960 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012961 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
12962 characters corresponding to the other errors.
12963
12964 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010012965 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012966
12967 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012968 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020012969 capture request header user-agent len 150
12970 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020012971
12972 Input request from client 127.0.0.1:
12973 GET / HTTP/1.0
12974 User-Agent: Very "Ugly" UA 1/2
12975
12976 Output log:
12977 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
12978
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012979language(<value>[,<default>])
12980 Returns the value with the highest q-factor from a list as extracted from the
12981 "accept-language" header using "req.fhdr". Values with no q-factor have a
12982 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
12983 belong to the list of semi-colon delimited <values> will be considered. The
12984 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
12985 given list and a default value is provided, it is returned. Note that language
12986 names may have a variant after a dash ('-'). If this variant is present in the
12987 list, it will be matched, but if it is not, only the base language is checked.
12988 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010012989 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012990 ordering of the values in the request counts, as the first value among
12991 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020012992
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012993 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020012994
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012995 # this configuration switches to the backend matching a
12996 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020012997
Thierry FOURNIER060762e2014-04-23 13:29:15 +020012998 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
12999 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13000 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13001 use_backend spanish if es
13002 use_backend french if fr
13003 use_backend english if en
13004 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013005
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013006length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013007 Get the length of the string. This can only be placed after a string
13008 sample fetch function or after a transformation keyword returning a string
13009 type. The result is of type integer.
13010
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013011lower
13012 Convert a string sample to lower case. This can only be placed after a string
13013 sample fetch function or after a transformation keyword returning a string
13014 type. The result is of type string.
13015
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013016ltime(<format>[,<offset>])
13017 Converts an integer supposed to contain a date since epoch to a string
13018 representing this date in local time using a format defined by the <format>
13019 string using strftime(3). The purpose is to allow any date format to be used
13020 in logs. An optional <offset> in seconds may be applied to the input date
13021 (positive or negative). See the strftime() man page for the format supported
13022 by your operating system. See also the utime converter.
13023
13024 Example :
13025
13026 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013027 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013028 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13029
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013030map(<map_file>[,<default_value>])
13031map_<match_type>(<map_file>[,<default_value>])
13032map_<match_type>_<output_type>(<map_file>[,<default_value>])
13033 Search the input value from <map_file> using the <match_type> matching method,
13034 and return the associated value converted to the type <output_type>. If the
13035 input value cannot be found in the <map_file>, the converter returns the
13036 <default_value>. If the <default_value> is not set, the converter fails and
13037 acts as if no input value could be fetched. If the <match_type> is not set, it
13038 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13039 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13040 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013041
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013042 It is important to avoid overlapping between the keys : IP addresses and
13043 strings are stored in trees, so the first of the finest match will be used.
13044 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013045
Tim Düsterhus4896c442016-11-29 02:15:19 +010013046 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013047 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013048
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013049 input type | match method | output type str | output type int | output type ip
13050 -----------+--------------+-----------------+-----------------+---------------
13051 str | str | map_str | map_str_int | map_str_ip
13052 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013053 str | beg | map_beg | map_beg_int | map_end_ip
13054 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013055 str | sub | map_sub | map_sub_int | map_sub_ip
13056 -----------+--------------+-----------------+-----------------+---------------
13057 str | dir | map_dir | map_dir_int | map_dir_ip
13058 -----------+--------------+-----------------+-----------------+---------------
13059 str | dom | map_dom | map_dom_int | map_dom_ip
13060 -----------+--------------+-----------------+-----------------+---------------
13061 str | end | map_end | map_end_int | map_end_ip
13062 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013063 str | reg | map_reg | map_reg_int | map_reg_ip
13064 -----------+--------------+-----------------+-----------------+---------------
13065 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013066 -----------+--------------+-----------------+-----------------+---------------
13067 int | int | map_int | map_int_int | map_int_ip
13068 -----------+--------------+-----------------+-----------------+---------------
13069 ip | ip | map_ip | map_ip_int | map_ip_ip
13070 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013071
Thierry Fournier8feaa662016-02-10 22:55:20 +010013072 The special map called "map_regm" expect matching zone in the regular
13073 expression and modify the output replacing back reference (like "\1") by
13074 the corresponding match text.
13075
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013076 The file contains one key + value per line. Lines which start with '#' are
13077 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13078 is then the first "word" (series of non-space/tabs characters), and the value
13079 is what follows this series of space/tab till the end of the line excluding
13080 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013081
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013082 Example :
13083
13084 # this is a comment and is ignored
13085 2.22.246.0/23 United Kingdom \n
13086 <-><-----------><--><------------><---->
13087 | | | | `- trailing spaces ignored
13088 | | | `---------- value
13089 | | `-------------------- middle spaces ignored
13090 | `---------------------------- key
13091 `------------------------------------ leading spaces ignored
13092
Willy Tarreau97707872015-01-27 15:12:13 +010013093mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013094 Divides the input value of type signed integer by <value>, and returns the
13095 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013096 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013097 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013098 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013099 "sess" : the variable is shared with the whole session
13100 "txn" : the variable is shared with the transaction (request and response)
13101 "req" : the variable is shared only during request processing
13102 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013103 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013104 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013105
13106mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013107 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013108 the product as an signed integer. In case of overflow, the largest possible
13109 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013110 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013111 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013112 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013113 "sess" : the variable is shared with the whole session
13114 "txn" : the variable is shared with the transaction (request and response)
13115 "req" : the variable is shared only during request processing
13116 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013117 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013118 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013119
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013120nbsrv
13121 Takes an input value of type string, interprets it as a backend name and
13122 returns the number of usable servers in that backend. Can be used in places
13123 where we want to look up a backend from a dynamic name, like a result of a
13124 map lookup.
13125
Willy Tarreau97707872015-01-27 15:12:13 +010013126neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013127 Takes the input value of type signed integer, computes the opposite value,
13128 and returns the remainder as an signed integer. 0 is identity. This operator
13129 is provided for reversed subtracts : in order to subtract the input from a
13130 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013131
13132not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013133 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013134 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013135 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013136 absence of a flag).
13137
13138odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013139 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013140 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13141
13142or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013143 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013144 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013145 numeric value or a variable name. The name of the variable starts with an
13146 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013147 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013148 "sess" : the variable is shared with the whole session
13149 "txn" : the variable is shared with the transaction (request and response)
13150 "req" : the variable is shared only during request processing
13151 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013152 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013153 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013154
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013155regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013156 Applies a regex-based substitution to the input string. It does the same
13157 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13158 default it will replace in the input string the first occurrence of the
13159 largest part matching the regular expression <regex> with the substitution
13160 string <subst>. It is possible to replace all occurrences instead by adding
13161 the flag "g" in the third argument <flags>. It is also possible to make the
13162 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13163 string, it is made up from the concatenation of all desired flags. Thus if
13164 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13165 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013166 configuration parser, some characters such as closing parenthesis, closing
13167 square brackets or comma are not possible to use in the arguments. The first
13168 use of this converter is to replace certain characters or sequence of
13169 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013170
13171 Example :
13172
13173 # de-duplicate "/" in header "x-path".
13174 # input: x-path: /////a///b/c/xzxyz/
13175 # output: x-path: /a/b/c/xzxyz/
13176 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13177
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013178capture-req(<id>)
13179 Capture the string entry in the request slot <id> and returns the entry as
13180 is. If the slot doesn't exist, the capture fails silently.
13181
13182 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013183 "http-response capture", "capture.req.hdr" and
13184 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013185
13186capture-res(<id>)
13187 Capture the string entry in the response slot <id> and returns the entry as
13188 is. If the slot doesn't exist, the capture fails silently.
13189
13190 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013191 "http-response capture", "capture.req.hdr" and
13192 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013193
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013194sdbm([<avalanche>])
13195 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13196 hash function. Optionally, it is possible to apply a full avalanche hash
13197 function to the output if the optional <avalanche> argument equals 1. This
13198 converter uses the same functions as used by the various hash-based load
13199 balancing algorithms, so it will provide exactly the same results. It is
13200 mostly intended for debugging, but can be used as a stick-table entry to
13201 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010013202 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6" and the
13203 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013204
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013205set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013206 Sets a variable with the input content and returns the content on the output
13207 as-is. The variable keeps the value and the associated input type. The name of
13208 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013209 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013210 "sess" : the variable is shared with the whole session
13211 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013212 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013213 "req" : the variable is shared only during request processing,
13214 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013215 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013216 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013217
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013218sha1
13219 Converts a binary input sample to a SHA1 digest. The result is a binary
13220 sample with length of 20 bytes.
13221
Willy Tarreau97707872015-01-27 15:12:13 +010013222sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013223 Subtracts <value> from the input value of type signed integer, and returns
13224 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013225 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013226 or a variable name. The name of the variable starts with an indication about
13227 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013228 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013229 "sess" : the variable is shared with the whole session
13230 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013231 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013232 "req" : the variable is shared only during request processing,
13233 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013234 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013235 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013236
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013237table_bytes_in_rate(<table>)
13238 Uses the string representation of the input sample to perform a look up in
13239 the specified table. If the key is not found in the table, integer value zero
13240 is returned. Otherwise the converter returns the average client-to-server
13241 bytes rate associated with the input sample in the designated table, measured
13242 in amount of bytes over the period configured in the table. See also the
13243 sc_bytes_in_rate sample fetch keyword.
13244
13245
13246table_bytes_out_rate(<table>)
13247 Uses the string representation of the input sample to perform a look up in
13248 the specified table. If the key is not found in the table, integer value zero
13249 is returned. Otherwise the converter returns the average server-to-client
13250 bytes rate associated with the input sample in the designated table, measured
13251 in amount of bytes over the period configured in the table. See also the
13252 sc_bytes_out_rate sample fetch keyword.
13253
13254table_conn_cnt(<table>)
13255 Uses the string representation of the input sample to perform a look up in
13256 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013257 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013258 connections associated with the input sample in the designated table. See
13259 also the sc_conn_cnt sample fetch keyword.
13260
13261table_conn_cur(<table>)
13262 Uses the string representation of the input sample to perform a look up in
13263 the specified table. If the key is not found in the table, integer value zero
13264 is returned. Otherwise the converter returns the current amount of concurrent
13265 tracked connections associated with the input sample in the designated table.
13266 See also the sc_conn_cur sample fetch keyword.
13267
13268table_conn_rate(<table>)
13269 Uses the string representation of the input sample to perform a look up in
13270 the specified table. If the key is not found in the table, integer value zero
13271 is returned. Otherwise the converter returns the average incoming connection
13272 rate associated with the input sample in the designated table. See also the
13273 sc_conn_rate sample fetch keyword.
13274
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013275table_gpt0(<table>)
13276 Uses the string representation of the input sample to perform a look up in
13277 the specified table. If the key is not found in the table, boolean value zero
13278 is returned. Otherwise the converter returns the current value of the first
13279 general purpose tag associated with the input sample in the designated table.
13280 See also the sc_get_gpt0 sample fetch keyword.
13281
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013282table_gpc0(<table>)
13283 Uses the string representation of the input sample to perform a look up in
13284 the specified table. If the key is not found in the table, integer value zero
13285 is returned. Otherwise the converter returns the current value of the first
13286 general purpose counter associated with the input sample in the designated
13287 table. See also the sc_get_gpc0 sample fetch keyword.
13288
13289table_gpc0_rate(<table>)
13290 Uses the string representation of the input sample to perform a look up in
13291 the specified table. If the key is not found in the table, integer value zero
13292 is returned. Otherwise the converter returns the frequency which the gpc0
13293 counter was incremented over the configured period in the table, associated
13294 with the input sample in the designated table. See also the sc_get_gpc0_rate
13295 sample fetch keyword.
13296
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013297table_gpc1(<table>)
13298 Uses the string representation of the input sample to perform a look up in
13299 the specified table. If the key is not found in the table, integer value zero
13300 is returned. Otherwise the converter returns the current value of the second
13301 general purpose counter associated with the input sample in the designated
13302 table. See also the sc_get_gpc1 sample fetch keyword.
13303
13304table_gpc1_rate(<table>)
13305 Uses the string representation of the input sample to perform a look up in
13306 the specified table. If the key is not found in the table, integer value zero
13307 is returned. Otherwise the converter returns the frequency which the gpc1
13308 counter was incremented over the configured period in the table, associated
13309 with the input sample in the designated table. See also the sc_get_gpc1_rate
13310 sample fetch keyword.
13311
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013312table_http_err_cnt(<table>)
13313 Uses the string representation of the input sample to perform a look up in
13314 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013315 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013316 errors associated with the input sample in the designated table. See also the
13317 sc_http_err_cnt sample fetch keyword.
13318
13319table_http_err_rate(<table>)
13320 Uses the string representation of the input sample to perform a look up in
13321 the specified table. If the key is not found in the table, integer value zero
13322 is returned. Otherwise the average rate of HTTP errors associated with the
13323 input sample in the designated table, measured in amount of errors over the
13324 period configured in the table. See also the sc_http_err_rate sample fetch
13325 keyword.
13326
13327table_http_req_cnt(<table>)
13328 Uses the string representation of the input sample to perform a look up in
13329 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013330 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013331 requests associated with the input sample in the designated table. See also
13332 the sc_http_req_cnt sample fetch keyword.
13333
13334table_http_req_rate(<table>)
13335 Uses the string representation of the input sample to perform a look up in
13336 the specified table. If the key is not found in the table, integer value zero
13337 is returned. Otherwise the average rate of HTTP requests associated with the
13338 input sample in the designated table, measured in amount of requests over the
13339 period configured in the table. See also the sc_http_req_rate sample fetch
13340 keyword.
13341
13342table_kbytes_in(<table>)
13343 Uses the string representation of the input sample to perform a look up in
13344 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013345 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013346 to-server data associated with the input sample in the designated table,
13347 measured in kilobytes. The test is currently performed on 32-bit integers,
13348 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13349 keyword.
13350
13351table_kbytes_out(<table>)
13352 Uses the string representation of the input sample to perform a look up in
13353 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013354 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013355 to-client data associated with the input sample in the designated table,
13356 measured in kilobytes. The test is currently performed on 32-bit integers,
13357 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13358 keyword.
13359
13360table_server_id(<table>)
13361 Uses the string representation of the input sample to perform a look up in
13362 the specified table. If the key is not found in the table, integer value zero
13363 is returned. Otherwise the converter returns the server ID associated with
13364 the input sample in the designated table. A server ID is associated to a
13365 sample by a "stick" rule when a connection to a server succeeds. A server ID
13366 zero means that no server is associated with this key.
13367
13368table_sess_cnt(<table>)
13369 Uses the string representation of the input sample to perform a look up in
13370 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013371 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013372 sessions associated with the input sample in the designated table. Note that
13373 a session here refers to an incoming connection being accepted by the
13374 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13375 keyword.
13376
13377table_sess_rate(<table>)
13378 Uses the string representation of the input sample to perform a look up in
13379 the specified table. If the key is not found in the table, integer value zero
13380 is returned. Otherwise the converter returns the average incoming session
13381 rate associated with the input sample in the designated table. Note that a
13382 session here refers to an incoming connection being accepted by the
13383 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13384 keyword.
13385
13386table_trackers(<table>)
13387 Uses the string representation of the input sample to perform a look up in
13388 the specified table. If the key is not found in the table, integer value zero
13389 is returned. Otherwise the converter returns the current amount of concurrent
13390 connections tracking the same key as the input sample in the designated
13391 table. It differs from table_conn_cur in that it does not rely on any stored
13392 information but on the table's reference count (the "use" value which is
13393 returned by "show table" on the CLI). This may sometimes be more suited for
13394 layer7 tracking. It can be used to tell a server how many concurrent
13395 connections there are from a given address for example. See also the
13396 sc_trackers sample fetch keyword.
13397
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013398upper
13399 Convert a string sample to upper case. This can only be placed after a string
13400 sample fetch function or after a transformation keyword returning a string
13401 type. The result is of type string.
13402
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013403url_dec
13404 Takes an url-encoded string provided as input and returns the decoded
13405 version as output. The input and the output are of type string.
13406
Christopher Faulet85d79c92016-11-09 16:54:56 +010013407unset-var(<var name>)
13408 Unsets a variable if the input content is defined. The name of the variable
13409 starts with an indication about its scope. The scopes allowed are:
13410 "proc" : the variable is shared with the whole process
13411 "sess" : the variable is shared with the whole session
13412 "txn" : the variable is shared with the transaction (request and
13413 response),
13414 "req" : the variable is shared only during request processing,
13415 "res" : the variable is shared only during response processing.
13416 This prefix is followed by a name. The separator is a '.'. The name may only
13417 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13418
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013419utime(<format>[,<offset>])
13420 Converts an integer supposed to contain a date since epoch to a string
13421 representing this date in UTC time using a format defined by the <format>
13422 string using strftime(3). The purpose is to allow any date format to be used
13423 in logs. An optional <offset> in seconds may be applied to the input date
13424 (positive or negative). See the strftime() man page for the format supported
13425 by your operating system. See also the ltime converter.
13426
13427 Example :
13428
13429 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013430 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013431 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13432
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013433word(<index>,<delimiters>)
13434 Extracts the nth word considering given delimiters from an input string.
13435 Indexes start at 1 and delimiters are a string formatted list of chars.
13436
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013437wt6([<avalanche>])
13438 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13439 hash function. Optionally, it is possible to apply a full avalanche hash
13440 function to the output if the optional <avalanche> argument equals 1. This
13441 converter uses the same functions as used by the various hash-based load
13442 balancing algorithms, so it will provide exactly the same results. It is
13443 mostly intended for debugging, but can be used as a stick-table entry to
13444 collect rough statistics. It must not be used for security purposes as a
Willy Tarreau80599772015-01-20 19:35:24 +010013445 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", and the
13446 "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013447
Willy Tarreau97707872015-01-27 15:12:13 +010013448xor(<value>)
13449 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013450 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013451 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013452 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013453 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013454 "sess" : the variable is shared with the whole session
13455 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013456 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013457 "req" : the variable is shared only during request processing,
13458 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013459 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013460 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013461
Thierry FOURNIER01e09742016-12-26 11:46:11 +010013462xxh32([<seed>])
13463 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
13464 variant of the XXHash hash function. This hash supports a seed which defaults
13465 to zero but a different value maybe passed as the <seed> argument. This hash
13466 is known to be very good and very fast so it can be used to hash URLs and/or
13467 URL parameters for use as stick-table keys to collect statistics with a low
13468 collision rate, though care must be taken as the algorithm is not considered
13469 as cryptographically secure.
13470
13471xxh64([<seed>])
13472 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
13473 variant of the XXHash hash function. This hash supports a seed which defaults
13474 to zero but a different value maybe passed as the <seed> argument. This hash
13475 is known to be very good and very fast so it can be used to hash URLs and/or
13476 URL parameters for use as stick-table keys to collect statistics with a low
13477 collision rate, though care must be taken as the algorithm is not considered
13478 as cryptographically secure.
13479
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013480
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200134817.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020013482--------------------------------------------
13483
13484A first set of sample fetch methods applies to internal information which does
13485not even relate to any client information. These ones are sometimes used with
13486"monitor-fail" directives to report an internal status to external watchers.
13487The sample fetch methods described in this section are usable anywhere.
13488
13489always_false : boolean
13490 Always returns the boolean "false" value. It may be used with ACLs as a
13491 temporary replacement for another one when adjusting configurations.
13492
13493always_true : boolean
13494 Always returns the boolean "true" value. It may be used with ACLs as a
13495 temporary replacement for another one when adjusting configurations.
13496
13497avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013498 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020013499 divided by the number of active servers. The current backend is used if no
13500 backend is specified. This is very similar to "queue" except that the size of
13501 the farm is considered, in order to give a more accurate measurement of the
13502 time it may take for a new connection to be processed. The main usage is with
13503 ACL to return a sorry page to new users when it becomes certain they will get
13504 a degraded service, or to pass to the backend servers in a header so that
13505 they decide to work in degraded mode or to disable some functions to speed up
13506 the processing a bit. Note that in the event there would not be any active
13507 server anymore, twice the number of queued connections would be considered as
13508 the measured value. This is a fair estimate, as we expect one server to get
13509 back soon anyway, but we still prefer to send new traffic to another backend
13510 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
13511 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010013512
Willy Tarreau74ca5042013-06-11 23:12:07 +020013513be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020013514 Applies to the number of currently established connections on the backend,
13515 possibly including the connection being evaluated. If no backend name is
13516 specified, the current one is used. But it is also possible to check another
13517 backend. It can be used to use a specific farm when the nominal one is full.
13518 See also the "fe_conn", "queue" and "be_sess_rate" criteria.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013519
Willy Tarreau74ca5042013-06-11 23:12:07 +020013520be_sess_rate([<backend>]) : integer
13521 Returns an integer value corresponding to the sessions creation rate on the
13522 backend, in number of new sessions per second. This is used with ACLs to
13523 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010013524 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020013525 online dictionary). It can also be useful to add this element to logs using a
13526 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013527
13528 Example :
13529 # Redirect to an error page if the dictionary is requested too often
13530 backend dynamic
13531 mode http
13532 acl being_scanned be_sess_rate gt 100
13533 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010013534
Davor Ocelice9ed2812017-12-25 17:49:28 +010013535bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013536 Returns a binary chain. The input is the hexadecimal representation
13537 of the string.
13538
13539bool(<bool>) : bool
13540 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
13541 'false' and '0' are the same. 'true' and '1' are the same.
13542
Willy Tarreau74ca5042013-06-11 23:12:07 +020013543connslots([<backend>]) : integer
13544 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013545 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020013546 connections on all servers and the maximum queue size. This is probably only
13547 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050013548
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013549 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020013550 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013551 usage; see "use_backend" keyword) can be redirected to a different backend.
13552
Willy Tarreau55165fe2009-05-10 12:02:55 +020013553 'connslots' = number of available server connection slots, + number of
13554 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013555
Willy Tarreaua36af912009-10-10 12:02:45 +020013556 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020013557 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020013558 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020013559 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010013560 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020013561 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020013562 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013563
Willy Tarreau55165fe2009-05-10 12:02:55 +020013564 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
13565 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020013566 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020013567 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080013568
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013569date([<offset>]) : integer
13570 Returns the current date as the epoch (number of seconds since 01/01/1970).
13571 If an offset value is specified, then it is a number of seconds that is added
13572 to the current date before returning the value. This is particularly useful
13573 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020013574 It is useful combined with the http_date converter.
13575
13576 Example :
13577
13578 # set an expires header to now+1 hour in every response
13579 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020013580
Etienne Carrierea792a0a2018-01-17 13:43:24 +010013581date_us : integer
13582 Return the microseconds part of the date (the "second" part is returned by
13583 date sample). This sample is coherent with the date sample as it is comes
13584 from the same timeval structure.
13585
Willy Tarreaud716f9b2017-10-13 11:03:15 +020013586distcc_body(<token>[,<occ>]) : binary
13587 Parses a distcc message and returns the body associated to occurrence #<occ>
13588 of the token <token>. Occurrences start at 1, and when unspecified, any may
13589 match though in practice only the first one is checked for now. This can be
13590 used to extract file names or arguments in files built using distcc through
13591 haproxy. Please refer to distcc's protocol documentation for the complete
13592 list of supported tokens.
13593
13594distcc_param(<token>[,<occ>]) : integer
13595 Parses a distcc message and returns the parameter associated to occurrence
13596 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
13597 any may match though in practice only the first one is checked for now. This
13598 can be used to extract certain information such as the protocol version, the
13599 file size or the argument in files built using distcc through haproxy.
13600 Another use case consists in waiting for the start of the preprocessed file
13601 contents before connecting to the server to avoid keeping idle connections.
13602 Please refer to distcc's protocol documentation for the complete list of
13603 supported tokens.
13604
13605 Example :
13606 # wait up to 20s for the pre-processed file to be uploaded
13607 tcp-request inspect-delay 20s
13608 tcp-request content accept if { distcc_param(DOTI) -m found }
13609 # send large files to the big farm
13610 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
13611
Willy Tarreau595ec542013-06-12 21:34:28 +020013612env(<name>) : string
13613 Returns a string containing the value of environment variable <name>. As a
13614 reminder, environment variables are per-process and are sampled when the
13615 process starts. This can be useful to pass some information to a next hop
13616 server, or with ACLs to take specific action when the process is started a
13617 certain way.
13618
13619 Examples :
13620 # Pass the Via header to next hop with the local hostname in it
13621 http-request add-header Via 1.1\ %[env(HOSTNAME)]
13622
13623 # reject cookie-less requests when the STOP environment variable is set
13624 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
13625
Willy Tarreau74ca5042013-06-11 23:12:07 +020013626fe_conn([<frontend>]) : integer
13627 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010013628 possibly including the connection being evaluated. If no frontend name is
13629 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013630 frontend. It can be used to return a sorry page before hard-blocking, or to
13631 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010013632 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020013633 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
13634 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020013635
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020013636fe_req_rate([<frontend>]) : integer
13637 Returns an integer value corresponding to the number of HTTP requests per
13638 second sent to a frontend. This number can differ from "fe_sess_rate" in
13639 situations where client-side keep-alive is enabled.
13640
Willy Tarreau74ca5042013-06-11 23:12:07 +020013641fe_sess_rate([<frontend>]) : integer
13642 Returns an integer value corresponding to the sessions creation rate on the
13643 frontend, in number of new sessions per second. This is used with ACLs to
13644 limit the incoming session rate to an acceptable range in order to prevent
13645 abuse of service at the earliest moment, for example when combined with other
13646 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
13647 down below the limit. It can also be useful to add this element to logs using
13648 a log-format directive. See also the "rate-limit sessions" directive for use
13649 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013650
13651 Example :
13652 # This frontend limits incoming mails to 10/s with a max of 100
13653 # concurrent connections. We accept any connection below 10/s, and
13654 # force excess clients to wait for 100 ms. Since clients are limited to
13655 # 100 max, there cannot be more than 10 incoming mails per second.
13656 frontend mail
13657 bind :25
13658 mode tcp
13659 maxconn 100
13660 acl too_fast fe_sess_rate ge 10
13661 tcp-request inspect-delay 100ms
13662 tcp-request content accept if ! too_fast
13663 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010013664
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010013665hostname : string
13666 Returns the system hostname.
13667
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013668int(<integer>) : signed integer
13669 Returns a signed integer.
13670
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013671ipv4(<ipv4>) : ipv4
13672 Returns an ipv4.
13673
13674ipv6(<ipv6>) : ipv6
13675 Returns an ipv6.
13676
13677meth(<method>) : method
13678 Returns a method.
13679
Willy Tarreau0f30d262014-11-24 16:02:05 +010013680nbproc : integer
13681 Returns an integer value corresponding to the number of processes that were
13682 started (it equals the global "nbproc" setting). This is useful for logging
13683 and debugging purposes.
13684
Willy Tarreau74ca5042013-06-11 23:12:07 +020013685nbsrv([<backend>]) : integer
13686 Returns an integer value corresponding to the number of usable servers of
13687 either the current backend or the named backend. This is mostly used with
13688 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010013689 switch to an alternate backend when the number of servers is too low to
13690 to handle some load. It is useful to report a failure when combined with
13691 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010013692
Willy Tarreau0f30d262014-11-24 16:02:05 +010013693proc : integer
13694 Returns an integer value corresponding to the position of the process calling
13695 the function, between 1 and global.nbproc. This is useful for logging and
13696 debugging purposes.
13697
Willy Tarreau74ca5042013-06-11 23:12:07 +020013698queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010013699 Returns the total number of queued connections of the designated backend,
13700 including all the connections in server queues. If no backend name is
13701 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020013702 one. This is useful with ACLs or to pass statistics to backend servers. This
13703 can be used to take actions when queuing goes above a known level, generally
13704 indicating a surge of traffic or a massive slowdown on the servers. One
13705 possible action could be to reject new users but still accept old ones. See
13706 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
13707
Willy Tarreau84310e22014-02-14 11:59:04 +010013708rand([<range>]) : integer
13709 Returns a random integer value within a range of <range> possible values,
13710 starting at zero. If the range is not specified, it defaults to 2^32, which
13711 gives numbers between 0 and 4294967295. It can be useful to pass some values
13712 needed to take some routing decisions for example, or just for debugging
13713 purposes. This random must not be used for security purposes.
13714
Willy Tarreau74ca5042013-06-11 23:12:07 +020013715srv_conn([<backend>/]<server>) : integer
13716 Returns an integer value corresponding to the number of currently established
13717 connections on the designated server, possibly including the connection being
13718 evaluated. If <backend> is omitted, then the server is looked up in the
13719 current backend. It can be used to use a specific farm when one server is
13720 full, or to inform the server about our view of the number of active
13721 connections with it. See also the "fe_conn", "be_conn" and "queue" fetch
13722 methods.
13723
13724srv_is_up([<backend>/]<server>) : boolean
13725 Returns true when the designated server is UP, and false when it is either
13726 DOWN or in maintenance mode. If <backend> is omitted, then the server is
13727 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010013728 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020013729 availability). Another possible use which is more of a hack consists in
13730 using dummy servers as boolean variables that can be enabled or disabled from
13731 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
13732
Willy Tarreauff2b7af2017-10-13 11:46:26 +020013733srv_queue([<backend>/]<server>) : integer
13734 Returns an integer value corresponding to the number of connections currently
13735 pending in the designated server's queue. If <backend> is omitted, then the
13736 server is looked up in the current backend. It can sometimes be used together
13737 with the "use-server" directive to force to use a known faster server when it
13738 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
13739 fetch methods.
13740
Willy Tarreau74ca5042013-06-11 23:12:07 +020013741srv_sess_rate([<backend>/]<server>) : integer
13742 Returns an integer corresponding to the sessions creation rate on the
13743 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013744 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020013745 used with ACLs but can make sense with logs too. This is used to switch to an
13746 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010013747 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020013748 overloading servers).
13749
13750 Example :
13751 # Redirect to a separate back
13752 acl srv1_full srv_sess_rate(be1/srv1) gt 50
13753 acl srv2_full srv_sess_rate(be1/srv2) gt 50
13754 use_backend be2 if srv1_full or srv2_full
13755
Willy Tarreau0f30d262014-11-24 16:02:05 +010013756stopping : boolean
13757 Returns TRUE if the process calling the function is currently stopping. This
13758 can be useful for logging, or for relaxing certain checks or helping close
13759 certain connections upon graceful shutdown.
13760
Thierry FOURNIERcc103292015-06-06 19:30:17 +020013761str(<string>) : string
13762 Returns a string.
13763
Willy Tarreau74ca5042013-06-11 23:12:07 +020013764table_avl([<table>]) : integer
13765 Returns the total number of available entries in the current proxy's
13766 stick-table or in the designated stick-table. See also table_cnt.
13767
13768table_cnt([<table>]) : integer
13769 Returns the total number of entries currently in use in the current proxy's
13770 stick-table or in the designated stick-table. See also src_conn_cnt and
13771 table_avl for other entry counting methods.
13772
Christopher Faulet34adb2a2017-11-21 21:45:38 +010013773thread : integer
13774 Returns an integer value corresponding to the position of the thread calling
13775 the function, between 0 and (global.nbthread-1). This is useful for logging
13776 and debugging purposes.
13777
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013778var(<var-name>) : undefined
13779 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010013780 sample fetch fails. The name of the variable starts with an indication
13781 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013782 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013783 "sess" : the variable is shared with the whole session
13784 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013785 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013786 "req" : the variable is shared only during request processing,
13787 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013788 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013789 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013790
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200137917.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020013792----------------------------------
13793
13794The layer 4 usually describes just the transport layer which in haproxy is
13795closest to the connection, where no content is yet made available. The fetch
13796methods described here are usable as low as the "tcp-request connection" rule
13797sets unless they require some future information. Those generally include
13798TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013799the incoming connection. For retrieving a value from a sticky counters, the
13800counter number can be explicitly set as 0, 1, or 2 using the pre-defined
13801"sc0_", "sc1_", or "sc2_" prefix, or it can be specified as the first integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013802argument when using the "sc_" prefix. An optional table may be specified with
13803the "sc*" form, in which case the currently tracked key will be looked up into
13804this alternate table instead of the table currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013805
13806be_id : integer
13807 Returns an integer containing the current backend's id. It can be used in
13808 frontends with responses to check which backend processed the request.
13809
Marcin Deranekd2471c22016-12-12 14:08:05 +010013810be_name : string
13811 Returns a string containing the current backend's name. It can be used in
13812 frontends with responses to check which backend processed the request.
13813
Willy Tarreau74ca5042013-06-11 23:12:07 +020013814dst : ip
13815 This is the destination IPv4 address of the connection on the client side,
13816 which is the address the client connected to. It can be useful when running
13817 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
13818 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
13819 RFC 4291.
13820
13821dst_conn : integer
13822 Returns an integer value corresponding to the number of currently established
13823 connections on the same socket including the one being evaluated. It is
13824 normally used with ACLs but can as well be used to pass the information to
13825 servers in an HTTP header or in logs. It can be used to either return a sorry
13826 page before hard-blocking, or to use a specific backend to drain new requests
13827 when the socket is considered saturated. This offers the ability to assign
13828 different limits to different listening ports or addresses. See also the
13829 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010013830
Willy Tarreau16e01562016-08-09 16:46:18 +020013831dst_is_local : boolean
13832 Returns true if the destination address of the incoming connection is local
13833 to the system, or false if the address doesn't exist on the system, meaning
13834 that it was intercepted in transparent mode. It can be useful to apply
13835 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010013836 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020013837 be delivered only on this address, or SSH access could be locally redirected.
13838 Please note that the check involves a few system calls, so it's better to do
13839 it only once per connection.
13840
Willy Tarreau74ca5042013-06-11 23:12:07 +020013841dst_port : integer
13842 Returns an integer value corresponding to the destination TCP port of the
13843 connection on the client side, which is the port the client connected to.
13844 This might be used when running in transparent mode, when assigning dynamic
13845 ports to some clients for a whole application session, to stick all users to
13846 a same server, or to pass the destination port information to a server using
13847 an HTTP header.
13848
Willy Tarreau60ca10a2017-08-18 15:26:54 +020013849fc_http_major : integer
13850 Reports the front connection's HTTP major version encoding, which may be 1
13851 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
13852 encoding and not on the version present in the request header.
13853
Emeric Brun4f603012017-01-05 15:11:44 +010013854fc_rcvd_proxy : boolean
13855 Returns true if the client initiated the connection with a PROXY protocol
13856 header.
13857
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020013858fc_rtt(<unit>) : integer
13859 Returns the Round Trip Time (RTT) measured by the kernel for the client
13860 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
13861 can be set to "ms" for milliseconds or "us" for microseconds. If the server
13862 connection is not established, if the connection is not TCP or if the
13863 operating system does not support TCP_INFO, for example Linux kernels before
13864 2.4, the sample fetch fails.
13865
13866fc_rttvar(<unit>) : integer
13867 Returns the Round Trip Time (RTT) variance measured by the kernel for the
13868 client connection. <unit> is facultative, by default the unit is milliseconds.
13869 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
13870 server connection is not established, if the connection is not TCP or if the
13871 operating system does not support TCP_INFO, for example Linux kernels before
13872 2.4, the sample fetch fails.
13873
Joe Williams30fcd392016-08-10 07:06:44 -070013874fc_unacked(<unit>) : integer
13875 Returns the unacked counter measured by the kernel for the client connection.
13876 If the server connection is not established, if the connection is not TCP or
13877 if the operating system does not support TCP_INFO, for example Linux kernels
13878 before 2.4, the sample fetch fails.
13879
13880fc_sacked(<unit>) : integer
13881 Returns the sacked counter measured by the kernel for the client connection.
13882 If the server connection is not established, if the connection is not TCP or
13883 if the operating system does not support TCP_INFO, for example Linux kernels
13884 before 2.4, the sample fetch fails.
13885
13886fc_retrans(<unit>) : integer
13887 Returns the retransmits counter measured by the kernel for the client
13888 connection. If the server connection is not established, if the connection is
13889 not TCP or if the operating system does not support TCP_INFO, for example
13890 Linux kernels before 2.4, the sample fetch fails.
13891
13892fc_fackets(<unit>) : integer
13893 Returns the fack counter measured by the kernel for the client
13894 connection. If the server connection is not established, if the connection is
13895 not TCP or if the operating system does not support TCP_INFO, for example
13896 Linux kernels before 2.4, the sample fetch fails.
13897
13898fc_lost(<unit>) : integer
13899 Returns the lost counter measured by the kernel for the client
13900 connection. If the server connection is not established, if the connection is
13901 not TCP or if the operating system does not support TCP_INFO, for example
13902 Linux kernels before 2.4, the sample fetch fails.
13903
13904fc_reordering(<unit>) : integer
13905 Returns the reordering counter measured by the kernel for the client
13906 connection. If the server connection is not established, if the connection is
13907 not TCP or if the operating system does not support TCP_INFO, for example
13908 Linux kernels before 2.4, the sample fetch fails.
13909
Willy Tarreau74ca5042013-06-11 23:12:07 +020013910fe_id : integer
13911 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010013912 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020013913 coming via a same frontend to the same server.
13914
Marcin Deranekd2471c22016-12-12 14:08:05 +010013915fe_name : string
13916 Returns a string containing the current frontend's name. It can be used in
13917 backends to check from which frontend it was called, or to stick all users
13918 coming via a same frontend to the same server.
13919
Cyril Bonté62ba8702014-04-22 23:52:25 +020013920sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013921sc0_bytes_in_rate([<table>]) : integer
13922sc1_bytes_in_rate([<table>]) : integer
13923sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013924 Returns the average client-to-server bytes rate from the currently tracked
13925 counters, measured in amount of bytes over the period configured in the
13926 table. See also src_bytes_in_rate.
13927
Cyril Bonté62ba8702014-04-22 23:52:25 +020013928sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013929sc0_bytes_out_rate([<table>]) : integer
13930sc1_bytes_out_rate([<table>]) : integer
13931sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013932 Returns the average server-to-client bytes rate from the currently tracked
13933 counters, measured in amount of bytes over the period configured in the
13934 table. See also src_bytes_out_rate.
13935
Cyril Bonté62ba8702014-04-22 23:52:25 +020013936sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013937sc0_clr_gpc0([<table>]) : integer
13938sc1_clr_gpc0([<table>]) : integer
13939sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020013940 Clears the first General Purpose Counter associated to the currently tracked
13941 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010013942 stored value is zero, so first invocation will always return zero. This is
13943 typically used as a second ACL in an expression in order to mark a connection
13944 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020013945
Jarno Huuskonen676f6222017-03-30 09:19:45 +030013946 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020013947 # block if 5 consecutive requests continue to come faster than 10 sess
13948 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020013949 acl abuse sc0_http_req_rate gt 10
13950 acl kill sc0_inc_gpc0 gt 5
13951 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020013952 tcp-request connection accept if !abuse save
13953 tcp-request connection reject if abuse kill
13954
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013955sc_clr_gpc1(<ctr>[,<table>]) : integer
13956sc0_clr_gpc1([<table>]) : integer
13957sc1_clr_gpc1([<table>]) : integer
13958sc2_clr_gpc1([<table>]) : integer
13959 Clears the second General Purpose Counter associated to the currently tracked
13960 counters, and returns its previous value. Before the first invocation, the
13961 stored value is zero, so first invocation will always return zero. This is
13962 typically used as a second ACL in an expression in order to mark a connection
13963 when a first ACL was verified.
13964
Cyril Bonté62ba8702014-04-22 23:52:25 +020013965sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013966sc0_conn_cnt([<table>]) : integer
13967sc1_conn_cnt([<table>]) : integer
13968sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010013969 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020013970 counters. See also src_conn_cnt.
13971
Cyril Bonté62ba8702014-04-22 23:52:25 +020013972sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013973sc0_conn_cur([<table>]) : integer
13974sc1_conn_cur([<table>]) : integer
13975sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013976 Returns the current amount of concurrent connections tracking the same
13977 tracked counters. This number is automatically incremented when tracking
13978 begins and decremented when tracking stops. See also src_conn_cur.
13979
Cyril Bonté62ba8702014-04-22 23:52:25 +020013980sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013981sc0_conn_rate([<table>]) : integer
13982sc1_conn_rate([<table>]) : integer
13983sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013984 Returns the average connection rate from the currently tracked counters,
13985 measured in amount of connections over the period configured in the table.
13986 See also src_conn_rate.
13987
Cyril Bonté62ba8702014-04-22 23:52:25 +020013988sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020013989sc0_get_gpc0([<table>]) : integer
13990sc1_get_gpc0([<table>]) : integer
13991sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020013992 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020013993 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020013994
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013995sc_get_gpc1(<ctr>[,<table>]) : integer
13996sc0_get_gpc1([<table>]) : integer
13997sc1_get_gpc1([<table>]) : integer
13998sc2_get_gpc1([<table>]) : integer
13999 Returns the value of the second General Purpose Counter associated to the
14000 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14001
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014002sc_get_gpt0(<ctr>[,<table>]) : integer
14003sc0_get_gpt0([<table>]) : integer
14004sc1_get_gpt0([<table>]) : integer
14005sc2_get_gpt0([<table>]) : integer
14006 Returns the value of the first General Purpose Tag associated to the
14007 currently tracked counters. See also src_get_gpt0.
14008
Cyril Bonté62ba8702014-04-22 23:52:25 +020014009sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014010sc0_gpc0_rate([<table>]) : integer
14011sc1_gpc0_rate([<table>]) : integer
14012sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014013 Returns the average increment rate of the first General Purpose Counter
14014 associated to the currently tracked counters. It reports the frequency
14015 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014016 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14017 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14018 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014019
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014020sc_gpc1_rate(<ctr>[,<table>]) : integer
14021sc0_gpc1_rate([<table>]) : integer
14022sc1_gpc1_rate([<table>]) : integer
14023sc2_gpc1_rate([<table>]) : integer
14024 Returns the average increment rate of the second General Purpose Counter
14025 associated to the currently tracked counters. It reports the frequency
14026 which the gpc1 counter was incremented over the configured period. See also
14027 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14028 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14029 be returned, as "gpc1" only holds the event count.
14030
Cyril Bonté62ba8702014-04-22 23:52:25 +020014031sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014032sc0_http_err_cnt([<table>]) : integer
14033sc1_http_err_cnt([<table>]) : integer
14034sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014035 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014036 counters. This includes the both request errors and 4xx error responses.
14037 See also src_http_err_cnt.
14038
Cyril Bonté62ba8702014-04-22 23:52:25 +020014039sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014040sc0_http_err_rate([<table>]) : integer
14041sc1_http_err_rate([<table>]) : integer
14042sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014043 Returns the average rate of HTTP errors from the currently tracked counters,
14044 measured in amount of errors over the period configured in the table. This
14045 includes the both request errors and 4xx error responses. See also
14046 src_http_err_rate.
14047
Cyril Bonté62ba8702014-04-22 23:52:25 +020014048sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014049sc0_http_req_cnt([<table>]) : integer
14050sc1_http_req_cnt([<table>]) : integer
14051sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014052 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014053 counters. This includes every started request, valid or not. See also
14054 src_http_req_cnt.
14055
Cyril Bonté62ba8702014-04-22 23:52:25 +020014056sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014057sc0_http_req_rate([<table>]) : integer
14058sc1_http_req_rate([<table>]) : integer
14059sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014060 Returns the average rate of HTTP requests from the currently tracked
14061 counters, measured in amount of requests over the period configured in
14062 the table. This includes every started request, valid or not. See also
14063 src_http_req_rate.
14064
Cyril Bonté62ba8702014-04-22 23:52:25 +020014065sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014066sc0_inc_gpc0([<table>]) : integer
14067sc1_inc_gpc0([<table>]) : integer
14068sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014069 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014070 tracked counters, and returns its new value. Before the first invocation,
14071 the stored value is zero, so first invocation will increase it to 1 and will
14072 return 1. This is typically used as a second ACL in an expression in order
14073 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014074
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014075 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014076 acl abuse sc0_http_req_rate gt 10
14077 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014078 tcp-request connection reject if abuse kill
14079
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014080sc_inc_gpc1(<ctr>[,<table>]) : integer
14081sc0_inc_gpc1([<table>]) : integer
14082sc1_inc_gpc1([<table>]) : integer
14083sc2_inc_gpc1([<table>]) : integer
14084 Increments the second General Purpose Counter associated to the currently
14085 tracked counters, and returns its new value. Before the first invocation,
14086 the stored value is zero, so first invocation will increase it to 1 and will
14087 return 1. This is typically used as a second ACL in an expression in order
14088 to mark a connection when a first ACL was verified.
14089
Cyril Bonté62ba8702014-04-22 23:52:25 +020014090sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014091sc0_kbytes_in([<table>]) : integer
14092sc1_kbytes_in([<table>]) : integer
14093sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014094 Returns the total amount of client-to-server data from the currently tracked
14095 counters, measured in kilobytes. The test is currently performed on 32-bit
14096 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014097
Cyril Bonté62ba8702014-04-22 23:52:25 +020014098sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014099sc0_kbytes_out([<table>]) : integer
14100sc1_kbytes_out([<table>]) : integer
14101sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014102 Returns the total amount of server-to-client data from the currently tracked
14103 counters, measured in kilobytes. The test is currently performed on 32-bit
14104 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014105
Cyril Bonté62ba8702014-04-22 23:52:25 +020014106sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014107sc0_sess_cnt([<table>]) : integer
14108sc1_sess_cnt([<table>]) : integer
14109sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014110 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014111 into sessions, which means that they were accepted by a "tcp-request
14112 connection" rule, from the currently tracked counters. A backend may count
14113 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014114 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014115 with the client. See also src_sess_cnt.
14116
Cyril Bonté62ba8702014-04-22 23:52:25 +020014117sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014118sc0_sess_rate([<table>]) : integer
14119sc1_sess_rate([<table>]) : integer
14120sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014121 Returns the average session rate from the currently tracked counters,
14122 measured in amount of sessions over the period configured in the table. A
14123 session is a connection that got past the early "tcp-request connection"
14124 rules. A backend may count more sessions than connections because each
14125 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014126 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014127
Cyril Bonté62ba8702014-04-22 23:52:25 +020014128sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014129sc0_tracked([<table>]) : boolean
14130sc1_tracked([<table>]) : boolean
14131sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014132 Returns true if the designated session counter is currently being tracked by
14133 the current session. This can be useful when deciding whether or not we want
14134 to set some values in a header passed to the server.
14135
Cyril Bonté62ba8702014-04-22 23:52:25 +020014136sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014137sc0_trackers([<table>]) : integer
14138sc1_trackers([<table>]) : integer
14139sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014140 Returns the current amount of concurrent connections tracking the same
14141 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014142 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014143 that it does not rely on any stored information but on the table's reference
14144 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014145 may sometimes be more suited for layer7 tracking. It can be used to tell a
14146 server how many concurrent connections there are from a given address for
14147 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014148
Willy Tarreau74ca5042013-06-11 23:12:07 +020014149so_id : integer
14150 Returns an integer containing the current listening socket's id. It is useful
14151 in frontends involving many "bind" lines, or to stick all users coming via a
14152 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014153
Willy Tarreau74ca5042013-06-11 23:12:07 +020014154src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014155 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014156 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14157 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14158 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014159 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14160 directive is used, it can be the address of a client behind another
14161 PROXY-protocol compatible component for all rule sets except
14162 "tcp-request connection" which sees the real address.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014163
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014164 Example:
14165 # add an HTTP header in requests with the originating address' country
14166 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14167
Willy Tarreau74ca5042013-06-11 23:12:07 +020014168src_bytes_in_rate([<table>]) : integer
14169 Returns the average bytes rate from the incoming connection's source address
14170 in the current proxy's stick-table or in the designated stick-table, measured
14171 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014172 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014173
Willy Tarreau74ca5042013-06-11 23:12:07 +020014174src_bytes_out_rate([<table>]) : integer
14175 Returns the average bytes rate to the incoming connection's source address in
14176 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014177 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014178 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014179
Willy Tarreau74ca5042013-06-11 23:12:07 +020014180src_clr_gpc0([<table>]) : integer
14181 Clears the first General Purpose Counter associated to the incoming
14182 connection's source address in the current proxy's stick-table or in the
14183 designated stick-table, and returns its previous value. If the address is not
14184 found, an entry is created and 0 is returned. This is typically used as a
14185 second ACL in an expression in order to mark a connection when a first ACL
14186 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014187
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014188 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014189 # block if 5 consecutive requests continue to come faster than 10 sess
14190 # per second, and reset the counter as soon as the traffic slows down.
14191 acl abuse src_http_req_rate gt 10
14192 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014193 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014194 tcp-request connection accept if !abuse save
14195 tcp-request connection reject if abuse kill
14196
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014197src_clr_gpc1([<table>]) : integer
14198 Clears the second General Purpose Counter associated to the incoming
14199 connection's source address in the current proxy's stick-table or in the
14200 designated stick-table, and returns its previous value. If the address is not
14201 found, an entry is created and 0 is returned. This is typically used as a
14202 second ACL in an expression in order to mark a connection when a first ACL
14203 was verified.
14204
Willy Tarreau74ca5042013-06-11 23:12:07 +020014205src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014206 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014207 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014208 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014209 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014210
Willy Tarreau74ca5042013-06-11 23:12:07 +020014211src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014212 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014213 current incoming connection's source address in the current proxy's
14214 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014215 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014216
Willy Tarreau74ca5042013-06-11 23:12:07 +020014217src_conn_rate([<table>]) : integer
14218 Returns the average connection rate from the incoming connection's source
14219 address in the current proxy's stick-table or in the designated stick-table,
14220 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014221 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014222
Willy Tarreau74ca5042013-06-11 23:12:07 +020014223src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014224 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014225 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014226 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014227 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014228
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014229src_get_gpc1([<table>]) : integer
14230 Returns the value of the second General Purpose Counter associated to the
14231 incoming connection's source address in the current proxy's stick-table or in
14232 the designated stick-table. If the address is not found, zero is returned.
14233 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14234
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014235src_get_gpt0([<table>]) : integer
14236 Returns the value of the first General Purpose Tag associated to the
14237 incoming connection's source address in the current proxy's stick-table or in
14238 the designated stick-table. If the address is not found, zero is returned.
14239 See also sc/sc0/sc1/sc2_get_gpt0.
14240
Willy Tarreau74ca5042013-06-11 23:12:07 +020014241src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014242 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014243 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014244 stick-table or in the designated stick-table. It reports the frequency
14245 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014246 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14247 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14248 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014249
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014250src_gpc1_rate([<table>]) : integer
14251 Returns the average increment rate of the second General Purpose Counter
14252 associated to the incoming connection's source address in the current proxy's
14253 stick-table or in the designated stick-table. It reports the frequency
14254 which the gpc1 counter was incremented over the configured period. See also
14255 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14256 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14257 be returned, as "gpc1" only holds the event count.
14258
Willy Tarreau74ca5042013-06-11 23:12:07 +020014259src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014260 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014261 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014262 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014263 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014264 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014265
Willy Tarreau74ca5042013-06-11 23:12:07 +020014266src_http_err_rate([<table>]) : integer
14267 Returns the average rate of HTTP errors from the incoming connection's source
14268 address in the current proxy's stick-table or in the designated stick-table,
14269 measured in amount of errors over the period configured in the table. This
14270 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014271 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014272
Willy Tarreau74ca5042013-06-11 23:12:07 +020014273src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014274 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014275 source address in the current proxy's stick-table or in the designated stick-
14276 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014277 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014278
Willy Tarreau74ca5042013-06-11 23:12:07 +020014279src_http_req_rate([<table>]) : integer
14280 Returns the average rate of HTTP requests from the incoming connection's
14281 source address in the current proxy's stick-table or in the designated stick-
14282 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014283 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014284 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014285
Willy Tarreau74ca5042013-06-11 23:12:07 +020014286src_inc_gpc0([<table>]) : integer
14287 Increments the first General Purpose Counter associated to the incoming
14288 connection's source address in the current proxy's stick-table or in the
14289 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014290 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014291 This is typically used as a second ACL in an expression in order to mark a
14292 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014293
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014294 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014295 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014296 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014297 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014298
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014299src_inc_gpc1([<table>]) : integer
14300 Increments the second General Purpose Counter associated to the incoming
14301 connection's source address in the current proxy's stick-table or in the
14302 designated stick-table, and returns its new value. If the address is not
14303 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14304 This is typically used as a second ACL in an expression in order to mark a
14305 connection when a first ACL was verified.
14306
Willy Tarreau16e01562016-08-09 16:46:18 +020014307src_is_local : boolean
14308 Returns true if the source address of the incoming connection is local to the
14309 system, or false if the address doesn't exist on the system, meaning that it
14310 comes from a remote machine. Note that UNIX addresses are considered local.
14311 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014312 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014313 note that the check involves a few system calls, so it's better to do it only
14314 once per connection.
14315
Willy Tarreau74ca5042013-06-11 23:12:07 +020014316src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014317 Returns the total amount of data received from the incoming connection's
14318 source address in the current proxy's stick-table or in the designated
14319 stick-table, measured in kilobytes. If the address is not found, zero is
14320 returned. The test is currently performed on 32-bit integers, which limits
14321 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014322
Willy Tarreau74ca5042013-06-11 23:12:07 +020014323src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014324 Returns the total amount of data sent to the incoming connection's source
14325 address in the current proxy's stick-table or in the designated stick-table,
14326 measured in kilobytes. If the address is not found, zero is returned. The
14327 test is currently performed on 32-bit integers, which limits values to 4
14328 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014329
Willy Tarreau74ca5042013-06-11 23:12:07 +020014330src_port : integer
14331 Returns an integer value corresponding to the TCP source port of the
14332 connection on the client side, which is the port the client connected from.
14333 Usage of this function is very limited as modern protocols do not care much
14334 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014335
Willy Tarreau74ca5042013-06-11 23:12:07 +020014336src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014337 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020014338 connection's source IPv4 address in the current proxy's stick-table or in the
14339 designated stick-table, that were transformed into sessions, which means that
14340 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014341 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014342
Willy Tarreau74ca5042013-06-11 23:12:07 +020014343src_sess_rate([<table>]) : integer
14344 Returns the average session rate from the incoming connection's source
14345 address in the current proxy's stick-table or in the designated stick-table,
14346 measured in amount of sessions over the period configured in the table. A
14347 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014348 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014349
Willy Tarreau74ca5042013-06-11 23:12:07 +020014350src_updt_conn_cnt([<table>]) : integer
14351 Creates or updates the entry associated to the incoming connection's source
14352 address in the current proxy's stick-table or in the designated stick-table.
14353 This table must be configured to store the "conn_cnt" data type, otherwise
14354 the match will be ignored. The current count is incremented by one, and the
14355 expiration timer refreshed. The updated count is returned, so this match
14356 can't return zero. This was used to reject service abusers based on their
14357 source address. Note: it is recommended to use the more complete "track-sc*"
14358 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014359
14360 Example :
14361 # This frontend limits incoming SSH connections to 3 per 10 second for
14362 # each source address, and rejects excess connections until a 10 second
14363 # silence is observed. At most 20 addresses are tracked.
14364 listen ssh
14365 bind :22
14366 mode tcp
14367 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020014368 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020014369 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020014370 server local 127.0.0.1:22
14371
Willy Tarreau74ca5042013-06-11 23:12:07 +020014372srv_id : integer
14373 Returns an integer containing the server's id when processing the response.
14374 While it's almost only used with ACLs, it may be used for logging or
14375 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020014376
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200143777.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020014378----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020014379
Willy Tarreau74ca5042013-06-11 23:12:07 +020014380The layer 5 usually describes just the session layer which in haproxy is
14381closest to the session once all the connection handshakes are finished, but
14382when no content is yet made available. The fetch methods described here are
14383usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014384future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020014385
Ben Shillitof25e8e52016-12-02 14:25:37 +00001438651d.all(<prop>[,<prop>*]) : string
14387 Returns values for the properties requested as a string, where values are
14388 separated by the delimiter specified with "51degrees-property-separator".
14389 The device is identified using all the important HTTP headers from the
14390 request. The function can be passed up to five property names, and if a
14391 property name can't be found, the value "NoData" is returned.
14392
14393 Example :
14394 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
14395 # containing the three properties requested using all relevant headers from
14396 # the request.
14397 frontend http-in
14398 bind *:8081
14399 default_backend servers
14400 http-request set-header X-51D-DeviceTypeMobileTablet \
14401 %[51d.all(DeviceType,IsMobile,IsTablet)]
14402
Emeric Brun645ae792014-04-30 14:21:06 +020014403ssl_bc : boolean
14404 Returns true when the back connection was made via an SSL/TLS transport
14405 layer and is locally deciphered. This means the outgoing connection was made
14406 other a server with the "ssl" option.
14407
14408ssl_bc_alg_keysize : integer
14409 Returns the symmetric cipher key size supported in bits when the outgoing
14410 connection was made over an SSL/TLS transport layer.
14411
14412ssl_bc_cipher : string
14413 Returns the name of the used cipher when the outgoing connection was made
14414 over an SSL/TLS transport layer.
14415
Emeric Brun74f7ffa2018-02-19 16:14:12 +010014416ssl_bc_is_resumed : boolean
14417 Returns true when the back connection was made over an SSL/TLS transport
14418 layer and the newly created SSL session was resumed using a cached
14419 session or a TLS ticket.
14420
Emeric Brun645ae792014-04-30 14:21:06 +020014421ssl_bc_protocol : string
14422 Returns the name of the used protocol when the outgoing connection was made
14423 over an SSL/TLS transport layer.
14424
Emeric Brunb73a9b02014-04-30 18:49:19 +020014425ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020014426 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014427 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14428 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020014429
14430ssl_bc_session_id : binary
14431 Returns the SSL ID of the back connection when the outgoing connection was
14432 made over an SSL/TLS transport layer. It is useful to log if we want to know
14433 if session was reused or not.
14434
14435ssl_bc_use_keysize : integer
14436 Returns the symmetric cipher key size used in bits when the outgoing
14437 connection was made over an SSL/TLS transport layer.
14438
Willy Tarreau74ca5042013-06-11 23:12:07 +020014439ssl_c_ca_err : integer
14440 When the incoming connection was made over an SSL/TLS transport layer,
14441 returns the ID of the first error detected during verification of the client
14442 certificate at depth > 0, or 0 if no error was encountered during this
14443 verification process. Please refer to your SSL library's documentation to
14444 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020014445
Willy Tarreau74ca5042013-06-11 23:12:07 +020014446ssl_c_ca_err_depth : integer
14447 When the incoming connection was made over an SSL/TLS transport layer,
14448 returns the depth in the CA chain of the first error detected during the
14449 verification of the client certificate. If no error is encountered, 0 is
14450 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014451
Emeric Brun43e79582014-10-29 19:03:26 +010014452ssl_c_der : binary
14453 Returns the DER formatted certificate presented by the client when the
14454 incoming connection was made over an SSL/TLS transport layer. When used for
14455 an ACL, the value(s) to match against can be passed in hexadecimal form.
14456
Willy Tarreau74ca5042013-06-11 23:12:07 +020014457ssl_c_err : integer
14458 When the incoming connection was made over an SSL/TLS transport layer,
14459 returns the ID of the first error detected during verification at depth 0, or
14460 0 if no error was encountered during this verification process. Please refer
14461 to your SSL library's documentation to find the exhaustive list of error
14462 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020014463
Willy Tarreau74ca5042013-06-11 23:12:07 +020014464ssl_c_i_dn([<entry>[,<occ>]]) : string
14465 When the incoming connection was made over an SSL/TLS transport layer,
14466 returns the full distinguished name of the issuer of the certificate
14467 presented by the client when no <entry> is specified, or the value of the
14468 first given entry found from the beginning of the DN. If a positive/negative
14469 occurrence number is specified as the optional second argument, it returns
14470 the value of the nth given entry value from the beginning/end of the DN.
14471 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
14472 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020014473
Willy Tarreau74ca5042013-06-11 23:12:07 +020014474ssl_c_key_alg : string
14475 Returns the name of the algorithm used to generate the key of the certificate
14476 presented by the client when the incoming connection was made over an SSL/TLS
14477 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020014478
Willy Tarreau74ca5042013-06-11 23:12:07 +020014479ssl_c_notafter : string
14480 Returns the end date presented by the client as a formatted string
14481 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14482 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020014483
Willy Tarreau74ca5042013-06-11 23:12:07 +020014484ssl_c_notbefore : string
14485 Returns the start date presented by the client as a formatted string
14486 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14487 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014488
Willy Tarreau74ca5042013-06-11 23:12:07 +020014489ssl_c_s_dn([<entry>[,<occ>]]) : string
14490 When the incoming connection was made over an SSL/TLS transport layer,
14491 returns the full distinguished name of the subject of the certificate
14492 presented by the client when no <entry> is specified, or the value of the
14493 first given entry found from the beginning of the DN. If a positive/negative
14494 occurrence number is specified as the optional second argument, it returns
14495 the value of the nth given entry value from the beginning/end of the DN.
14496 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
14497 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010014498
Willy Tarreau74ca5042013-06-11 23:12:07 +020014499ssl_c_serial : binary
14500 Returns the serial of the certificate presented by the client when the
14501 incoming connection was made over an SSL/TLS transport layer. When used for
14502 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014503
Willy Tarreau74ca5042013-06-11 23:12:07 +020014504ssl_c_sha1 : binary
14505 Returns the SHA-1 fingerprint of the certificate presented by the client when
14506 the incoming connection was made over an SSL/TLS transport layer. This can be
14507 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014508 Note that the output is binary, so if you want to pass that signature to the
14509 server, you need to encode it in hex or base64, such as in the example below:
14510
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014511 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020014512 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020014513
Willy Tarreau74ca5042013-06-11 23:12:07 +020014514ssl_c_sig_alg : string
14515 Returns the name of the algorithm used to sign the certificate presented by
14516 the client when the incoming connection was made over an SSL/TLS transport
14517 layer.
Emeric Brun87855892012-10-17 17:39:35 +020014518
Willy Tarreau74ca5042013-06-11 23:12:07 +020014519ssl_c_used : boolean
14520 Returns true if current SSL session uses a client certificate even if current
14521 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020014522
Willy Tarreau74ca5042013-06-11 23:12:07 +020014523ssl_c_verify : integer
14524 Returns the verify result error ID when the incoming connection was made over
14525 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
14526 refer to your SSL library's documentation for an exhaustive list of error
14527 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020014528
Willy Tarreau74ca5042013-06-11 23:12:07 +020014529ssl_c_version : integer
14530 Returns the version of the certificate presented by the client when the
14531 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020014532
Emeric Brun43e79582014-10-29 19:03:26 +010014533ssl_f_der : binary
14534 Returns the DER formatted certificate presented by the frontend when the
14535 incoming connection was made over an SSL/TLS transport layer. When used for
14536 an ACL, the value(s) to match against can be passed in hexadecimal form.
14537
Willy Tarreau74ca5042013-06-11 23:12:07 +020014538ssl_f_i_dn([<entry>[,<occ>]]) : string
14539 When the incoming connection was made over an SSL/TLS transport layer,
14540 returns the full distinguished name of the issuer of the certificate
14541 presented by the frontend when no <entry> is specified, or the value of the
14542 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020014543 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020014544 the value of the nth given entry value from the beginning/end of the DN.
14545 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
14546 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020014547
Willy Tarreau74ca5042013-06-11 23:12:07 +020014548ssl_f_key_alg : string
14549 Returns the name of the algorithm used to generate the key of the certificate
14550 presented by the frontend when the incoming connection was made over an
14551 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014552
Willy Tarreau74ca5042013-06-11 23:12:07 +020014553ssl_f_notafter : string
14554 Returns the end date presented by the frontend as a formatted string
14555 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14556 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014557
Willy Tarreau74ca5042013-06-11 23:12:07 +020014558ssl_f_notbefore : string
14559 Returns the start date presented by the frontend as a formatted string
14560 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
14561 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020014562
Willy Tarreau74ca5042013-06-11 23:12:07 +020014563ssl_f_s_dn([<entry>[,<occ>]]) : string
14564 When the incoming connection was made over an SSL/TLS transport layer,
14565 returns the full distinguished name of the subject of the certificate
14566 presented by the frontend when no <entry> is specified, or the value of the
14567 first given entry found from the beginning of the DN. If a positive/negative
14568 occurrence number is specified as the optional second argument, it returns
14569 the value of the nth given entry value from the beginning/end of the DN.
14570 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
14571 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020014572
Willy Tarreau74ca5042013-06-11 23:12:07 +020014573ssl_f_serial : binary
14574 Returns the serial of the certificate presented by the frontend when the
14575 incoming connection was made over an SSL/TLS transport layer. When used for
14576 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020014577
Emeric Brun55f4fa82014-04-30 17:11:25 +020014578ssl_f_sha1 : binary
14579 Returns the SHA-1 fingerprint of the certificate presented by the frontend
14580 when the incoming connection was made over an SSL/TLS transport layer. This
14581 can be used to know which certificate was chosen using SNI.
14582
Willy Tarreau74ca5042013-06-11 23:12:07 +020014583ssl_f_sig_alg : string
14584 Returns the name of the algorithm used to sign the certificate presented by
14585 the frontend when the incoming connection was made over an SSL/TLS transport
14586 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020014587
Willy Tarreau74ca5042013-06-11 23:12:07 +020014588ssl_f_version : integer
14589 Returns the version of the certificate presented by the frontend when the
14590 incoming connection was made over an SSL/TLS transport layer.
14591
14592ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014593 Returns true when the front connection was made via an SSL/TLS transport
14594 layer and is locally deciphered. This means it has matched a socket declared
14595 with a "bind" line having the "ssl" option.
14596
Willy Tarreau74ca5042013-06-11 23:12:07 +020014597 Example :
14598 # This passes "X-Proto: https" to servers when client connects over SSL
14599 listen http-https
14600 bind :80
14601 bind :443 ssl crt /etc/haproxy.pem
14602 http-request add-header X-Proto https if { ssl_fc }
14603
14604ssl_fc_alg_keysize : integer
14605 Returns the symmetric cipher key size supported in bits when the incoming
14606 connection was made over an SSL/TLS transport layer.
14607
14608ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014609 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014610 incoming connection made via a TLS transport layer and locally deciphered by
14611 haproxy. The result is a string containing the protocol name advertised by
14612 the client. The SSL library must have been built with support for TLS
14613 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
14614 not advertised unless the "alpn" keyword on the "bind" line specifies a
14615 protocol list. Also, nothing forces the client to pick a protocol from this
14616 list, any other one may be requested. The TLS ALPN extension is meant to
14617 replace the TLS NPN extension. See also "ssl_fc_npn".
14618
Willy Tarreau74ca5042013-06-11 23:12:07 +020014619ssl_fc_cipher : string
14620 Returns the name of the used cipher when the incoming connection was made
14621 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020014622
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014623ssl_fc_cipherlist_bin : binary
14624 Returns the binary form of the client hello cipher list. The maximum returned
14625 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014626 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014627
14628ssl_fc_cipherlist_hex : string
14629 Returns the binary form of the client hello cipher list encoded as
14630 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014631 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014632
14633ssl_fc_cipherlist_str : string
14634 Returns the decoded text form of the client hello cipher list. The maximum
14635 number of ciphers returned is according with the value of
14636 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010014637 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020014638 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014639
14640ssl_fc_cipherlist_xxh : integer
14641 Returns a xxh64 of the cipher list. This hash can be return only is the value
14642 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010014643 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010014644
Willy Tarreau74ca5042013-06-11 23:12:07 +020014645ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020014646 Returns true if a client certificate is present in an incoming connection over
14647 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010014648 Note: on SSL session resumption with Session ID or TLS ticket, client
14649 certificate is not present in the current connection but may be retrieved
14650 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
14651 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020014652
Olivier Houchardccaa7de2017-10-02 11:51:03 +020014653ssl_fc_has_early : boolean
14654 Returns true if early data were sent, and the handshake didn't happen yet. As
14655 it has security implications, it is useful to be able to refuse those, or
14656 wait until the handshake happened.
14657
Willy Tarreau74ca5042013-06-11 23:12:07 +020014658ssl_fc_has_sni : boolean
14659 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020014660 in an incoming connection was made over an SSL/TLS transport layer. Returns
14661 true when the incoming connection presents a TLS SNI field. This requires
14662 that the SSL library is build with support for TLS extensions enabled (check
14663 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020014664
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020014665ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020014666 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010014667 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
14668 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020014669
Willy Tarreau74ca5042013-06-11 23:12:07 +020014670ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014671 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020014672 made via a TLS transport layer and locally deciphered by haproxy. The result
14673 is a string containing the protocol name advertised by the client. The SSL
14674 library must have been built with support for TLS extensions enabled (check
14675 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
14676 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
14677 forces the client to pick a protocol from this list, any other one may be
14678 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020014679
Willy Tarreau74ca5042013-06-11 23:12:07 +020014680ssl_fc_protocol : string
14681 Returns the name of the used protocol when the incoming connection was made
14682 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020014683
Emeric Brunb73a9b02014-04-30 18:49:19 +020014684ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040014685 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020014686 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
14687 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040014688
Willy Tarreau74ca5042013-06-11 23:12:07 +020014689ssl_fc_session_id : binary
14690 Returns the SSL ID of the front connection when the incoming connection was
14691 made over an SSL/TLS transport layer. It is useful to stick a given client to
14692 a server. It is important to note that some browsers refresh their session ID
14693 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020014694
Willy Tarreau74ca5042013-06-11 23:12:07 +020014695ssl_fc_sni : string
14696 This extracts the Server Name Indication TLS extension (SNI) field from an
14697 incoming connection made via an SSL/TLS transport layer and locally
14698 deciphered by haproxy. The result (when present) typically is a string
14699 matching the HTTPS host name (253 chars or less). The SSL library must have
14700 been built with support for TLS extensions enabled (check haproxy -vv).
14701
14702 This fetch is different from "req_ssl_sni" above in that it applies to the
14703 connection being deciphered by haproxy and not to SSL contents being blindly
14704 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020014705 requires that the SSL library is build with support for TLS extensions
14706 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020014707
Willy Tarreau74ca5042013-06-11 23:12:07 +020014708 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020014709 ssl_fc_sni_end : suffix match
14710 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020014711
Willy Tarreau74ca5042013-06-11 23:12:07 +020014712ssl_fc_use_keysize : integer
14713 Returns the symmetric cipher key size used in bits when the incoming
14714 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014715
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014716
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200147177.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020014718------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020014719
Willy Tarreau74ca5042013-06-11 23:12:07 +020014720Fetching samples from buffer contents is a bit different from the previous
14721sample fetches above because the sampled data are ephemeral. These data can
14722only be used when they're available and will be lost when they're forwarded.
14723For this reason, samples fetched from buffer contents during a request cannot
14724be used in a response for example. Even while the data are being fetched, they
14725can change. Sometimes it is necessary to set some delays or combine multiple
14726sample fetch methods to ensure that the expected data are complete and usable,
14727for example through TCP request content inspection. Please see the "tcp-request
14728content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020014729
Willy Tarreau74ca5042013-06-11 23:12:07 +020014730payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010014731 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014732 "stick on", "stick match"), and for "res.payload" when used in the context of
14733 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014734
Willy Tarreau74ca5042013-06-11 23:12:07 +020014735payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
14736 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010014737 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014738 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014739
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020014740req.hdrs : string
14741 Returns the current request headers as string including the last empty line
14742 separating headers from the request body. The last empty line can be used to
14743 detect a truncated header block. This sample fetch is useful for some SPOE
14744 headers analyzers and for advanced logging.
14745
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020014746req.hdrs_bin : binary
14747 Returns the current request headers contained in preparsed binary form. This
14748 is useful for offloading some processing with SPOE. Each string is described
14749 by a length followed by the number of bytes indicated in the length. The
14750 length is represented using the variable integer encoding detailed in the
14751 SPOE documentation. The end of the list is marked by a couple of empty header
14752 names and values (length of 0 for both).
14753
14754 *(<str:header-name><str:header-value>)<empty string><empty string>
14755
14756 int: refer to the SPOE documentation for the encoding
14757 str: <int:length><bytes>
14758
Willy Tarreau74ca5042013-06-11 23:12:07 +020014759req.len : integer
14760req_len : integer (deprecated)
14761 Returns an integer value corresponding to the number of bytes present in the
14762 request buffer. This is mostly used in ACL. It is important to understand
14763 that this test does not return false as long as the buffer is changing. This
14764 means that a check with equality to zero will almost always immediately match
14765 at the beginning of the session, while a test for more data will wait for
14766 that data to come in and return false only when haproxy is certain that no
14767 more data will come in. This test was designed to be used with TCP request
14768 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014769
Willy Tarreau74ca5042013-06-11 23:12:07 +020014770req.payload(<offset>,<length>) : binary
14771 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020014772 in the request buffer. As a special case, if the <length> argument is zero,
14773 the the whole buffer from <offset> to the end is extracted. This can be used
14774 with ACLs in order to check for the presence of some content in a buffer at
14775 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014776
Willy Tarreau74ca5042013-06-11 23:12:07 +020014777 ACL alternatives :
14778 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014779
Willy Tarreau74ca5042013-06-11 23:12:07 +020014780req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
14781 This extracts a binary block whose size is specified at <offset1> for <length>
14782 bytes, and which starts at <offset2> if specified or just after the length in
14783 the request buffer. The <offset2> parameter also supports relative offsets if
14784 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014785
Willy Tarreau74ca5042013-06-11 23:12:07 +020014786 ACL alternatives :
14787 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014788
Willy Tarreau74ca5042013-06-11 23:12:07 +020014789 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014790
Willy Tarreau74ca5042013-06-11 23:12:07 +020014791req.proto_http : boolean
14792req_proto_http : boolean (deprecated)
14793 Returns true when data in the request buffer look like HTTP and correctly
14794 parses as such. It is the same parser as the common HTTP request parser which
14795 is used so there should be no surprises. The test does not match until the
14796 request is complete, failed or timed out. This test may be used to report the
14797 protocol in TCP logs, but the biggest use is to block TCP request analysis
14798 until a complete HTTP request is present in the buffer, for example to track
14799 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014800
Willy Tarreau74ca5042013-06-11 23:12:07 +020014801 Example:
14802 # track request counts per "base" (concatenation of Host+URL)
14803 tcp-request inspect-delay 10s
14804 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014805 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020014806
Willy Tarreau74ca5042013-06-11 23:12:07 +020014807req.rdp_cookie([<name>]) : string
14808rdp_cookie([<name>]) : string (deprecated)
14809 When the request buffer looks like the RDP protocol, extracts the RDP cookie
14810 <name>, or any cookie if unspecified. The parser only checks for the first
14811 cookie, as illustrated in the RDP protocol specification. The cookie name is
14812 case insensitive. Generally the "MSTS" cookie name will be used, as it can
14813 contain the user name of the client connecting to the server if properly
14814 configured on the client. The "MSTSHASH" cookie is often used as well for
14815 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014816
Willy Tarreau74ca5042013-06-11 23:12:07 +020014817 This differs from "balance rdp-cookie" in that any balancing algorithm may be
14818 used and thus the distribution of clients to backend servers is not linked to
14819 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
14820 such as "balance roundrobin" or "balance leastconn" will lead to a more even
14821 distribution of clients to backend servers than the hash used by "balance
14822 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014823
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824 ACL derivatives :
14825 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014826
Willy Tarreau74ca5042013-06-11 23:12:07 +020014827 Example :
14828 listen tse-farm
14829 bind 0.0.0.0:3389
14830 # wait up to 5s for an RDP cookie in the request
14831 tcp-request inspect-delay 5s
14832 tcp-request content accept if RDP_COOKIE
14833 # apply RDP cookie persistence
14834 persist rdp-cookie
14835 # Persist based on the mstshash cookie
14836 # This is only useful makes sense if
14837 # balance rdp-cookie is not used
14838 stick-table type string size 204800
14839 stick on req.rdp_cookie(mstshash)
14840 server srv1 1.1.1.1:3389
14841 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014842
Willy Tarreau74ca5042013-06-11 23:12:07 +020014843 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
14844 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014845
Willy Tarreau74ca5042013-06-11 23:12:07 +020014846req.rdp_cookie_cnt([name]) : integer
14847rdp_cookie_cnt([name]) : integer (deprecated)
14848 Tries to parse the request buffer as RDP protocol, then returns an integer
14849 corresponding to the number of RDP cookies found. If an optional cookie name
14850 is passed, only cookies matching this name are considered. This is mostly
14851 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014852
Willy Tarreau74ca5042013-06-11 23:12:07 +020014853 ACL derivatives :
14854 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014855
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020014856req.ssl_ec_ext : boolean
14857 Returns a boolean identifying if client sent the Supported Elliptic Curves
14858 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020014859 message. This can be used to present ECC compatible clients with EC
14860 certificate and to use RSA for all others, on the same IP address. Note that
14861 this only applies to raw contents found in the request buffer and not to
14862 contents deciphered via an SSL data layer, so this will not work with "bind"
14863 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020014864
Willy Tarreau74ca5042013-06-11 23:12:07 +020014865req.ssl_hello_type : integer
14866req_ssl_hello_type : integer (deprecated)
14867 Returns an integer value containing the type of the SSL hello message found
14868 in the request buffer if the buffer contains data that parse as a complete
14869 SSL (v3 or superior) client hello message. Note that this only applies to raw
14870 contents found in the request buffer and not to contents deciphered via an
14871 SSL data layer, so this will not work with "bind" lines having the "ssl"
14872 option. This is mostly used in ACL to detect presence of an SSL hello message
14873 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014874
Willy Tarreau74ca5042013-06-11 23:12:07 +020014875req.ssl_sni : string
14876req_ssl_sni : string (deprecated)
14877 Returns a string containing the value of the Server Name TLS extension sent
14878 by a client in a TLS stream passing through the request buffer if the buffer
14879 contains data that parse as a complete SSL (v3 or superior) client hello
14880 message. Note that this only applies to raw contents found in the request
14881 buffer and not to contents deciphered via an SSL data layer, so this will not
14882 work with "bind" lines having the "ssl" option. SNI normally contains the
14883 name of the host the client tries to connect to (for recent browsers). SNI is
14884 useful for allowing or denying access to certain hosts when SSL/TLS is used
14885 by the client. This test was designed to be used with TCP request content
14886 inspection. If content switching is needed, it is recommended to first wait
14887 for a complete client hello (type 1), like in the example below. See also
14888 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014889
Willy Tarreau74ca5042013-06-11 23:12:07 +020014890 ACL derivatives :
14891 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014892
Willy Tarreau74ca5042013-06-11 23:12:07 +020014893 Examples :
14894 # Wait for a client hello for at most 5 seconds
14895 tcp-request inspect-delay 5s
14896 tcp-request content accept if { req_ssl_hello_type 1 }
14897 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
14898 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020014899
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053014900req.ssl_st_ext : integer
14901 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
14902 Returns 1 if the client sent SessionTicket TLS Extension
14903 Returns 2 if the client also sent non-zero length TLS SessionTicket
14904 Note that this only applies to raw contents found in the request buffer and
14905 not to contents deciphered via an SSL data layer, so this will not work with
14906 "bind" lines having the "ssl" option. This can for example be used to detect
14907 whether the client sent a SessionTicket or not and stick it accordingly, if
14908 no SessionTicket then stick on SessionID or don't stick as there's no server
14909 side state is there when SessionTickets are in use.
14910
Willy Tarreau74ca5042013-06-11 23:12:07 +020014911req.ssl_ver : integer
14912req_ssl_ver : integer (deprecated)
14913 Returns an integer value containing the version of the SSL/TLS protocol of a
14914 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
14915 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
14916 composed of the major version multiplied by 65536, added to the minor
14917 version. Note that this only applies to raw contents found in the request
14918 buffer and not to contents deciphered via an SSL data layer, so this will not
14919 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010014920 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014921 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014922
Willy Tarreau74ca5042013-06-11 23:12:07 +020014923 ACL derivatives :
14924 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010014925
Willy Tarreau47e8eba2013-09-11 23:28:46 +020014926res.len : integer
14927 Returns an integer value corresponding to the number of bytes present in the
14928 response buffer. This is mostly used in ACL. It is important to understand
14929 that this test does not return false as long as the buffer is changing. This
14930 means that a check with equality to zero will almost always immediately match
14931 at the beginning of the session, while a test for more data will wait for
14932 that data to come in and return false only when haproxy is certain that no
14933 more data will come in. This test was designed to be used with TCP response
14934 content inspection.
14935
Willy Tarreau74ca5042013-06-11 23:12:07 +020014936res.payload(<offset>,<length>) : binary
14937 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020014938 in the response buffer. As a special case, if the <length> argument is zero,
14939 the the whole buffer from <offset> to the end is extracted. This can be used
14940 with ACLs in order to check for the presence of some content in a buffer at
14941 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014942
Willy Tarreau74ca5042013-06-11 23:12:07 +020014943res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
14944 This extracts a binary block whose size is specified at <offset1> for <length>
14945 bytes, and which starts at <offset2> if specified or just after the length in
14946 the response buffer. The <offset2> parameter also supports relative offsets
14947 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014948
Willy Tarreau74ca5042013-06-11 23:12:07 +020014949 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014950
Willy Tarreau971f7b62015-09-29 14:06:59 +020014951res.ssl_hello_type : integer
14952rep_ssl_hello_type : integer (deprecated)
14953 Returns an integer value containing the type of the SSL hello message found
14954 in the response buffer if the buffer contains data that parses as a complete
14955 SSL (v3 or superior) hello message. Note that this only applies to raw
14956 contents found in the response buffer and not to contents deciphered via an
14957 SSL data layer, so this will not work with "server" lines having the "ssl"
14958 option. This is mostly used in ACL to detect presence of an SSL hello message
14959 that is supposed to contain an SSL session ID usable for stickiness.
14960
Willy Tarreau74ca5042013-06-11 23:12:07 +020014961wait_end : boolean
14962 This fetch either returns true when the inspection period is over, or does
14963 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010014964 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014965 actions, such as a delayed reject for some special addresses. Since it either
14966 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010014967 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020014968 "WAIT_END" is always usable without prior declaration. This test was designed
14969 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014970
Willy Tarreau74ca5042013-06-11 23:12:07 +020014971 Examples :
14972 # delay every incoming request by 2 seconds
14973 tcp-request inspect-delay 2s
14974 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010014975
Willy Tarreau74ca5042013-06-11 23:12:07 +020014976 # don't immediately tell bad guys they are rejected
14977 tcp-request inspect-delay 10s
14978 acl goodguys src 10.0.0.0/24
14979 acl badguys src 10.0.1.0/24
14980 tcp-request content accept if goodguys
14981 tcp-request content reject if badguys WAIT_END
14982 tcp-request content reject
14983
14984
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200149857.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020014986--------------------------------------
14987
14988It is possible to fetch samples from HTTP contents, requests and responses.
14989This application layer is also called layer 7. It is only possible to fetch the
14990data in this section when a full HTTP request or response has been parsed from
14991its respective request or response buffer. This is always the case with all
14992HTTP specific rules and for sections running with "mode http". When using TCP
14993content inspection, it may be necessary to support an inspection delay in order
14994to let the request or response come in first. These fetches may require a bit
14995more CPU resources than the layer 4 ones, but not much since the request and
14996response are indexed.
14997
14998base : string
14999 This returns the concatenation of the first Host header and the path part of
15000 the request, which starts at the first slash and ends before the question
15001 mark. It can be useful in virtual hosted environments to detect URL abuses as
15002 well as to improve shared caches efficiency. Using this with a limited size
15003 stick table also allows one to collect statistics about most commonly
15004 requested objects by host/path. With ACLs it can allow simple content
15005 switching rules involving the host and the path at the same time, such as
15006 "www.example.com/favicon.ico". See also "path" and "uri".
15007
15008 ACL derivatives :
15009 base : exact string match
15010 base_beg : prefix match
15011 base_dir : subdir match
15012 base_dom : domain match
15013 base_end : suffix match
15014 base_len : length match
15015 base_reg : regex match
15016 base_sub : substring match
15017
15018base32 : integer
15019 This returns a 32-bit hash of the value returned by the "base" fetch method
15020 above. This is useful to track per-URL activity on high traffic sites without
15021 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015022 memory. The output type is an unsigned integer. The hash function used is
15023 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15024 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015025
15026base32+src : binary
15027 This returns the concatenation of the base32 fetch above and the src fetch
15028 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15029 depending on the source address family. This can be used to track per-IP,
15030 per-URL counters.
15031
William Lallemand65ad6e12014-01-31 15:08:02 +010015032capture.req.hdr(<idx>) : string
15033 This extracts the content of the header captured by the "capture request
15034 header", idx is the position of the capture keyword in the configuration.
15035 The first entry is an index of 0. See also: "capture request header".
15036
15037capture.req.method : string
15038 This extracts the METHOD of an HTTP request. It can be used in both request
15039 and response. Unlike "method", it can be used in both request and response
15040 because it's allocated.
15041
15042capture.req.uri : string
15043 This extracts the request's URI, which starts at the first slash and ends
15044 before the first space in the request (without the host part). Unlike "path"
15045 and "url", it can be used in both request and response because it's
15046 allocated.
15047
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015048capture.req.ver : string
15049 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15050 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15051 logs because it relies on a persistent flag.
15052
William Lallemand65ad6e12014-01-31 15:08:02 +010015053capture.res.hdr(<idx>) : string
15054 This extracts the content of the header captured by the "capture response
15055 header", idx is the position of the capture keyword in the configuration.
15056 The first entry is an index of 0.
15057 See also: "capture response header"
15058
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015059capture.res.ver : string
15060 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15061 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15062 persistent flag.
15063
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015064req.body : binary
15065 This returns the HTTP request's available body as a block of data. It
15066 requires that the request body has been buffered made available using
15067 "option http-buffer-request". In case of chunked-encoded body, currently only
15068 the first chunk is analyzed.
15069
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015070req.body_param([<name>) : string
15071 This fetch assumes that the body of the POST request is url-encoded. The user
15072 can check if the "content-type" contains the value
15073 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15074 parameter <name> in the body, which ends before '&'. The parameter name is
15075 case-sensitive. If no name is given, any parameter will match, and the first
15076 one will be returned. The result is a string corresponding to the value of the
15077 parameter <name> as presented in the request body (no URL decoding is
15078 performed). Note that the ACL version of this fetch iterates over multiple
15079 parameters and will iteratively report all parameters values if no name is
15080 given.
15081
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015082req.body_len : integer
15083 This returns the length of the HTTP request's available body in bytes. It may
15084 be lower than the advertised length if the body is larger than the buffer. It
15085 requires that the request body has been buffered made available using
15086 "option http-buffer-request".
15087
15088req.body_size : integer
15089 This returns the advertised length of the HTTP request's body in bytes. It
15090 will represent the advertised Content-Length header, or the size of the first
15091 chunk in case of chunked encoding. In order to parse the chunks, it requires
15092 that the request body has been buffered made available using
15093 "option http-buffer-request".
15094
Willy Tarreau74ca5042013-06-11 23:12:07 +020015095req.cook([<name>]) : string
15096cook([<name>]) : string (deprecated)
15097 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15098 header line from the request, and returns its value as string. If no name is
15099 specified, the first cookie value is returned. When used with ACLs, all
15100 matching cookies are evaluated. Spaces around the name and the value are
15101 ignored as requested by the Cookie header specification (RFC6265). The cookie
15102 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15103 well return an empty value if it is present. Use the "found" match to detect
15104 presence. Use the res.cook() variant for response cookies sent by the server.
15105
15106 ACL derivatives :
15107 cook([<name>]) : exact string match
15108 cook_beg([<name>]) : prefix match
15109 cook_dir([<name>]) : subdir match
15110 cook_dom([<name>]) : domain match
15111 cook_end([<name>]) : suffix match
15112 cook_len([<name>]) : length match
15113 cook_reg([<name>]) : regex match
15114 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015115
Willy Tarreau74ca5042013-06-11 23:12:07 +020015116req.cook_cnt([<name>]) : integer
15117cook_cnt([<name>]) : integer (deprecated)
15118 Returns an integer value representing the number of occurrences of the cookie
15119 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015120
Willy Tarreau74ca5042013-06-11 23:12:07 +020015121req.cook_val([<name>]) : integer
15122cook_val([<name>]) : integer (deprecated)
15123 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15124 header line from the request, and converts its value to an integer which is
15125 returned. If no name is specified, the first cookie value is returned. When
15126 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015127
Willy Tarreau74ca5042013-06-11 23:12:07 +020015128cookie([<name>]) : string (deprecated)
15129 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15130 header line from the request, or a "Set-Cookie" header from the response, and
15131 returns its value as a string. A typical use is to get multiple clients
15132 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015133 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015134 multi-peer synchronization and state keeping across restarts. If no name is
15135 specified, the first cookie value is returned. This fetch should not be used
15136 anymore and should be replaced by req.cook() or res.cook() instead as it
15137 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015138
Willy Tarreau74ca5042013-06-11 23:12:07 +020015139hdr([<name>[,<occ>]]) : string
15140 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15141 used on responses. Please refer to these respective fetches for more details.
15142 In case of doubt about the fetch direction, please use the explicit ones.
15143 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015144 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015145
Willy Tarreau74ca5042013-06-11 23:12:07 +020015146req.fhdr(<name>[,<occ>]) : string
15147 This extracts the last occurrence of header <name> in an HTTP request. When
15148 used from an ACL, all occurrences are iterated over until a match is found.
15149 Optionally, a specific occurrence might be specified as a position number.
15150 Positive values indicate a position from the first occurrence, with 1 being
15151 the first one. Negative values indicate positions relative to the last one,
15152 with -1 being the last one. It differs from req.hdr() in that any commas
15153 present in the value are returned and are not used as delimiters. This is
15154 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015155
Willy Tarreau74ca5042013-06-11 23:12:07 +020015156req.fhdr_cnt([<name>]) : integer
15157 Returns an integer value representing the number of occurrences of request
15158 header field name <name>, or the total number of header fields if <name> is
15159 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15160 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015161
Willy Tarreau74ca5042013-06-11 23:12:07 +020015162req.hdr([<name>[,<occ>]]) : string
15163 This extracts the last occurrence of header <name> in an HTTP request. When
15164 used from an ACL, all occurrences are iterated over until a match is found.
15165 Optionally, a specific occurrence might be specified as a position number.
15166 Positive values indicate a position from the first occurrence, with 1 being
15167 the first one. Negative values indicate positions relative to the last one,
15168 with -1 being the last one. A typical use is with the X-Forwarded-For header
15169 once converted to IP, associated with an IP stick-table. The function
15170 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015171 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015172 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015173 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015174
Willy Tarreau74ca5042013-06-11 23:12:07 +020015175 ACL derivatives :
15176 hdr([<name>[,<occ>]]) : exact string match
15177 hdr_beg([<name>[,<occ>]]) : prefix match
15178 hdr_dir([<name>[,<occ>]]) : subdir match
15179 hdr_dom([<name>[,<occ>]]) : domain match
15180 hdr_end([<name>[,<occ>]]) : suffix match
15181 hdr_len([<name>[,<occ>]]) : length match
15182 hdr_reg([<name>[,<occ>]]) : regex match
15183 hdr_sub([<name>[,<occ>]]) : substring match
15184
15185req.hdr_cnt([<name>]) : integer
15186hdr_cnt([<header>]) : integer (deprecated)
15187 Returns an integer value representing the number of occurrences of request
15188 header field name <name>, or the total number of header field values if
15189 <name> is not specified. It is important to remember that one header line may
15190 count as several headers if it has several values. The function considers any
15191 comma as a delimiter for distinct values. If full-line headers are desired
15192 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15193 detect presence, absence or abuse of a specific header, as well as to block
15194 request smuggling attacks by rejecting requests which contain more than one
15195 of certain headers. See "req.hdr" for more information on header matching.
15196
15197req.hdr_ip([<name>[,<occ>]]) : ip
15198hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15199 This extracts the last occurrence of header <name> in an HTTP request,
15200 converts it to an IPv4 or IPv6 address and returns this address. When used
15201 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15202 of every header is checked. Optionally, a specific occurrence might be
15203 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015204 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015205 positions relative to the last one, with -1 being the last one. A typical use
15206 is with the X-Forwarded-For and X-Client-IP headers.
15207
15208req.hdr_val([<name>[,<occ>]]) : integer
15209hdr_val([<name>[,<occ>]]) : integer (deprecated)
15210 This extracts the last occurrence of header <name> in an HTTP request, and
15211 converts it to an integer value. When used with ACLs, all occurrences are
15212 checked, and if <name> is omitted, every value of every header is checked.
15213 Optionally, a specific occurrence might be specified as a position number.
15214 Positive values indicate a position from the first occurrence, with 1 being
15215 the first one. Negative values indicate positions relative to the last one,
15216 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15217
15218http_auth(<userlist>) : boolean
15219 Returns a boolean indicating whether the authentication data received from
15220 the client match a username & password stored in the specified userlist. This
15221 fetch function is not really useful outside of ACLs. Currently only http
15222 basic auth is supported.
15223
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015224http_auth_group(<userlist>) : string
15225 Returns a string corresponding to the user name found in the authentication
15226 data received from the client if both the user name and password are valid
15227 according to the specified userlist. The main purpose is to use it in ACLs
15228 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015229 This fetch function is not really useful outside of ACLs. Currently only http
15230 basic auth is supported.
15231
15232 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015233 http_auth_group(<userlist>) : group ...
15234 Returns true when the user extracted from the request and whose password is
15235 valid according to the specified userlist belongs to at least one of the
15236 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015237
15238http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015239 Returns true when the request being processed is the first one of the
15240 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015241 from some requests when a request is not the first one, or to help grouping
15242 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015243
Willy Tarreau74ca5042013-06-11 23:12:07 +020015244method : integer + string
15245 Returns an integer value corresponding to the method in the HTTP request. For
15246 example, "GET" equals 1 (check sources to establish the matching). Value 9
15247 means "other method" and may be converted to a string extracted from the
15248 stream. This should not be used directly as a sample, this is only meant to
15249 be used from ACLs, which transparently convert methods from patterns to these
15250 integer + string values. Some predefined ACL already check for most common
15251 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015252
Willy Tarreau74ca5042013-06-11 23:12:07 +020015253 ACL derivatives :
15254 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015255
Willy Tarreau74ca5042013-06-11 23:12:07 +020015256 Example :
15257 # only accept GET and HEAD requests
15258 acl valid_method method GET HEAD
15259 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015260
Willy Tarreau74ca5042013-06-11 23:12:07 +020015261path : string
15262 This extracts the request's URL path, which starts at the first slash and
15263 ends before the question mark (without the host part). A typical use is with
15264 prefetch-capable caches, and with portals which need to aggregate multiple
15265 information from databases and keep them in caches. Note that with outgoing
15266 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015267 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015268 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015269
Willy Tarreau74ca5042013-06-11 23:12:07 +020015270 ACL derivatives :
15271 path : exact string match
15272 path_beg : prefix match
15273 path_dir : subdir match
15274 path_dom : domain match
15275 path_end : suffix match
15276 path_len : length match
15277 path_reg : regex match
15278 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015279
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015280query : string
15281 This extracts the request's query string, which starts after the first
15282 question mark. If no question mark is present, this fetch returns nothing. If
15283 a question mark is present but nothing follows, it returns an empty string.
15284 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010015285 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010015286 which stops before the question mark.
15287
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015288req.hdr_names([<delim>]) : string
15289 This builds a string made from the concatenation of all header names as they
15290 appear in the request when the rule is evaluated. The default delimiter is
15291 the comma (',') but it may be overridden as an optional argument <delim>. In
15292 this case, only the first character of <delim> is considered.
15293
Willy Tarreau74ca5042013-06-11 23:12:07 +020015294req.ver : string
15295req_ver : string (deprecated)
15296 Returns the version string from the HTTP request, for example "1.1". This can
15297 be useful for logs, but is mostly there for ACL. Some predefined ACL already
15298 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015299
Willy Tarreau74ca5042013-06-11 23:12:07 +020015300 ACL derivatives :
15301 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020015302
Willy Tarreau74ca5042013-06-11 23:12:07 +020015303res.comp : boolean
15304 Returns the boolean "true" value if the response has been compressed by
15305 HAProxy, otherwise returns boolean "false". This may be used to add
15306 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015307
Willy Tarreau74ca5042013-06-11 23:12:07 +020015308res.comp_algo : string
15309 Returns a string containing the name of the algorithm used if the response
15310 was compressed by HAProxy, for example : "deflate". This may be used to add
15311 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015312
Willy Tarreau74ca5042013-06-11 23:12:07 +020015313res.cook([<name>]) : string
15314scook([<name>]) : string (deprecated)
15315 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15316 header line from the response, and returns its value as string. If no name is
15317 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015318
Willy Tarreau74ca5042013-06-11 23:12:07 +020015319 ACL derivatives :
15320 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020015321
Willy Tarreau74ca5042013-06-11 23:12:07 +020015322res.cook_cnt([<name>]) : integer
15323scook_cnt([<name>]) : integer (deprecated)
15324 Returns an integer value representing the number of occurrences of the cookie
15325 <name> in the response, or all cookies if <name> is not specified. This is
15326 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015327
Willy Tarreau74ca5042013-06-11 23:12:07 +020015328res.cook_val([<name>]) : integer
15329scook_val([<name>]) : integer (deprecated)
15330 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15331 header line from the response, and converts its value to an integer which is
15332 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015333
Willy Tarreau74ca5042013-06-11 23:12:07 +020015334res.fhdr([<name>[,<occ>]]) : string
15335 This extracts the last occurrence of header <name> in an HTTP response, or of
15336 the last header if no <name> is specified. Optionally, a specific occurrence
15337 might be specified as a position number. Positive values indicate a position
15338 from the first occurrence, with 1 being the first one. Negative values
15339 indicate positions relative to the last one, with -1 being the last one. It
15340 differs from res.hdr() in that any commas present in the value are returned
15341 and are not used as delimiters. If this is not desired, the res.hdr() fetch
15342 should be used instead. This is sometimes useful with headers such as Date or
15343 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015344
Willy Tarreau74ca5042013-06-11 23:12:07 +020015345res.fhdr_cnt([<name>]) : integer
15346 Returns an integer value representing the number of occurrences of response
15347 header field name <name>, or the total number of header fields if <name> is
15348 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
15349 the number of full line headers and does not stop on commas. If this is not
15350 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015351
Willy Tarreau74ca5042013-06-11 23:12:07 +020015352res.hdr([<name>[,<occ>]]) : string
15353shdr([<name>[,<occ>]]) : string (deprecated)
15354 This extracts the last occurrence of header <name> in an HTTP response, or of
15355 the last header if no <name> is specified. Optionally, a specific occurrence
15356 might be specified as a position number. Positive values indicate a position
15357 from the first occurrence, with 1 being the first one. Negative values
15358 indicate positions relative to the last one, with -1 being the last one. This
15359 can be useful to learn some data into a stick-table. The function considers
15360 any comma as a delimiter for distinct values. If this is not desired, the
15361 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015362
Willy Tarreau74ca5042013-06-11 23:12:07 +020015363 ACL derivatives :
15364 shdr([<name>[,<occ>]]) : exact string match
15365 shdr_beg([<name>[,<occ>]]) : prefix match
15366 shdr_dir([<name>[,<occ>]]) : subdir match
15367 shdr_dom([<name>[,<occ>]]) : domain match
15368 shdr_end([<name>[,<occ>]]) : suffix match
15369 shdr_len([<name>[,<occ>]]) : length match
15370 shdr_reg([<name>[,<occ>]]) : regex match
15371 shdr_sub([<name>[,<occ>]]) : substring match
15372
15373res.hdr_cnt([<name>]) : integer
15374shdr_cnt([<name>]) : integer (deprecated)
15375 Returns an integer value representing the number of occurrences of response
15376 header field name <name>, or the total number of header fields if <name> is
15377 not specified. The function considers any comma as a delimiter for distinct
15378 values. If this is not desired, the res.fhdr_cnt() fetch should be used
15379 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015380
Willy Tarreau74ca5042013-06-11 23:12:07 +020015381res.hdr_ip([<name>[,<occ>]]) : ip
15382shdr_ip([<name>[,<occ>]]) : ip (deprecated)
15383 This extracts the last occurrence of header <name> in an HTTP response,
15384 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
15385 specific occurrence might be specified as a position number. Positive values
15386 indicate a position from the first occurrence, with 1 being the first one.
15387 Negative values indicate positions relative to the last one, with -1 being
15388 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015389
Willy Tarreaueb27ec72015-02-20 13:55:29 +010015390res.hdr_names([<delim>]) : string
15391 This builds a string made from the concatenation of all header names as they
15392 appear in the response when the rule is evaluated. The default delimiter is
15393 the comma (',') but it may be overridden as an optional argument <delim>. In
15394 this case, only the first character of <delim> is considered.
15395
Willy Tarreau74ca5042013-06-11 23:12:07 +020015396res.hdr_val([<name>[,<occ>]]) : integer
15397shdr_val([<name>[,<occ>]]) : integer (deprecated)
15398 This extracts the last occurrence of header <name> in an HTTP response, and
15399 converts it to an integer value. Optionally, a specific occurrence might be
15400 specified as a position number. Positive values indicate a position from the
15401 first occurrence, with 1 being the first one. Negative values indicate
15402 positions relative to the last one, with -1 being the last one. This can be
15403 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015404
Willy Tarreau74ca5042013-06-11 23:12:07 +020015405res.ver : string
15406resp_ver : string (deprecated)
15407 Returns the version string from the HTTP response, for example "1.1". This
15408 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020015409
Willy Tarreau74ca5042013-06-11 23:12:07 +020015410 ACL derivatives :
15411 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010015412
Willy Tarreau74ca5042013-06-11 23:12:07 +020015413set-cookie([<name>]) : string (deprecated)
15414 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
15415 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020015416 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020015417 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015418
Willy Tarreau74ca5042013-06-11 23:12:07 +020015419 This fetch function is deprecated and has been superseded by the "res.cook"
15420 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010015421
Willy Tarreau74ca5042013-06-11 23:12:07 +020015422status : integer
15423 Returns an integer containing the HTTP status code in the HTTP response, for
15424 example, 302. It is mostly used within ACLs and integer ranges, for example,
15425 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015426
Thierry Fournier0e00dca2016-04-07 15:47:40 +020015427unique-id : string
15428 Returns the unique-id attached to the request. The directive
15429 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
15430 fails. Note that the unique-id is usually used with HTTP requests, however this
15431 sample fetch can be used with other protocols. Obviously, if it is used with
15432 other protocols than HTTP, the unique-id-format directive must not contain
15433 HTTP parts. See: unique-id-format and unique-id-header
15434
Willy Tarreau74ca5042013-06-11 23:12:07 +020015435url : string
15436 This extracts the request's URL as presented in the request. A typical use is
15437 with prefetch-capable caches, and with portals which need to aggregate
15438 multiple information from databases and keep them in caches. With ACLs, using
15439 "path" is preferred over using "url", because clients may send a full URL as
15440 is normally done with proxies. The only real use is to match "*" which does
15441 not match in "path", and for which there is already a predefined ACL. See
15442 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015443
Willy Tarreau74ca5042013-06-11 23:12:07 +020015444 ACL derivatives :
15445 url : exact string match
15446 url_beg : prefix match
15447 url_dir : subdir match
15448 url_dom : domain match
15449 url_end : suffix match
15450 url_len : length match
15451 url_reg : regex match
15452 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015453
Willy Tarreau74ca5042013-06-11 23:12:07 +020015454url_ip : ip
15455 This extracts the IP address from the request's URL when the host part is
15456 presented as an IP address. Its use is very limited. For instance, a
15457 monitoring system might use this field as an alternative for the source IP in
15458 order to test what path a given source address would follow, or to force an
15459 entry in a table for a given source address. With ACLs it can be used to
15460 restrict access to certain systems through a proxy, for example when combined
15461 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015462
Willy Tarreau74ca5042013-06-11 23:12:07 +020015463url_port : integer
15464 This extracts the port part from the request's URL. Note that if the port is
15465 not specified in the request, port 80 is assumed. With ACLs it can be used to
15466 restrict access to certain systems through a proxy, for example when combined
15467 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015468
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015469urlp([<name>[,<delim>]]) : string
15470url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020015471 This extracts the first occurrence of the parameter <name> in the query
15472 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015473 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
15474 any parameter will match, and the first one will be returned. The result is
15475 a string corresponding to the value of the parameter <name> as presented in
15476 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020015477 stickiness based on a client ID, to extract an application cookie passed as a
15478 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020015479 this fetch iterates over multiple parameters and will iteratively report all
15480 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015481
Willy Tarreau74ca5042013-06-11 23:12:07 +020015482 ACL derivatives :
15483 urlp(<name>[,<delim>]) : exact string match
15484 urlp_beg(<name>[,<delim>]) : prefix match
15485 urlp_dir(<name>[,<delim>]) : subdir match
15486 urlp_dom(<name>[,<delim>]) : domain match
15487 urlp_end(<name>[,<delim>]) : suffix match
15488 urlp_len(<name>[,<delim>]) : length match
15489 urlp_reg(<name>[,<delim>]) : regex match
15490 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015491
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015492
Willy Tarreau74ca5042013-06-11 23:12:07 +020015493 Example :
15494 # match http://example.com/foo?PHPSESSIONID=some_id
15495 stick on urlp(PHPSESSIONID)
15496 # match http://example.com/foo;JSESSIONID=some_id
15497 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020015498
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015499urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020015500 See "urlp" above. This one extracts the URL parameter <name> in the request
15501 and converts it to an integer value. This can be used for session stickiness
15502 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020015503
Dragan Dosen0070cd52016-06-16 12:19:49 +020015504url32 : integer
15505 This returns a 32-bit hash of the value obtained by concatenating the first
15506 Host header and the whole URL including parameters (not only the path part of
15507 the request, as in the "base32" fetch above). This is useful to track per-URL
15508 activity. A shorter hash is stored, saving a lot of memory. The output type
15509 is an unsigned integer.
15510
15511url32+src : binary
15512 This returns the concatenation of the "url32" fetch and the "src" fetch. The
15513 resulting type is of type binary, with a size of 8 or 20 bytes depending on
15514 the source address family. This can be used to track per-IP, per-URL counters.
15515
Willy Tarreau198a7442008-01-17 12:05:32 +010015516
Willy Tarreau74ca5042013-06-11 23:12:07 +0200155177.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015518---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015519
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015520Some predefined ACLs are hard-coded so that they do not have to be declared in
15521every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020015522order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010015523
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015524ACL name Equivalent to Usage
15525---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015526FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020015527HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015528HTTP_1.0 req_ver 1.0 match HTTP version 1.0
15529HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010015530HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
15531HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
15532HTTP_URL_SLASH url_beg / match URL beginning with "/"
15533HTTP_URL_STAR url * match URL equal to "*"
15534LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015535METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015536METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015537METH_GET method GET HEAD match HTTP GET or HEAD method
15538METH_HEAD method HEAD match HTTP HEAD method
15539METH_OPTIONS method OPTIONS match HTTP OPTIONS method
15540METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020015541METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015542METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020015543RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015544REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015545TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015546WAIT_END wait_end wait for end of content analysis
15547---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010015548
Willy Tarreaub937b7e2010-01-12 15:27:54 +010015549
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200155508. Logging
15551----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010015552
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015553One of HAProxy's strong points certainly lies is its precise logs. It probably
15554provides the finest level of information available for such a product, which is
15555very important for troubleshooting complex environments. Standard information
15556provided in logs include client ports, TCP/HTTP state timers, precise session
15557state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010015558to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015559headers.
15560
15561In order to improve administrators reactivity, it offers a great transparency
15562about encountered problems, both internal and external, and it is possible to
15563send logs to different sources at the same time with different level filters :
15564
15565 - global process-level logs (system errors, start/stop, etc..)
15566 - per-instance system and internal errors (lack of resource, bugs, ...)
15567 - per-instance external troubles (servers up/down, max connections)
15568 - per-instance activity (client connections), either at the establishment or
15569 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010015570 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060015571 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015572
15573The ability to distribute different levels of logs to different log servers
15574allow several production teams to interact and to fix their problems as soon
15575as possible. For example, the system team might monitor system-wide errors,
15576while the application team might be monitoring the up/down for their servers in
15577real time, and the security team might analyze the activity logs with one hour
15578delay.
15579
15580
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200155818.1. Log levels
15582---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015583
Simon Hormandf791f52011-05-29 15:01:10 +090015584TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015585source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090015586HTTP request, HTTP return code, number of bytes transmitted, conditions
15587in which the session ended, and even exchanged cookies values. For example
15588track a particular user's problems. All messages may be sent to up to two
15589syslog servers. Check the "log" keyword in section 4.2 for more information
15590about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015591
15592
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200155938.2. Log formats
15594----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015595
William Lallemand48940402012-01-30 16:47:22 +010015596HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090015597and will be detailed in the following sections. A few of them may vary
15598slightly with the configuration, due to indicators specific to certain
15599options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015600
15601 - the default format, which is very basic and very rarely used. It only
15602 provides very basic information about the incoming connection at the moment
15603 it is accepted : source IP:port, destination IP:port, and frontend-name.
15604 This mode will eventually disappear so it will not be described to great
15605 extents.
15606
15607 - the TCP format, which is more advanced. This format is enabled when "option
15608 tcplog" is set on the frontend. HAProxy will then usually wait for the
15609 connection to terminate before logging. This format provides much richer
15610 information, such as timers, connection counts, queue size, etc... This
15611 format is recommended for pure TCP proxies.
15612
15613 - the HTTP format, which is the most advanced for HTTP proxying. This format
15614 is enabled when "option httplog" is set on the frontend. It provides the
15615 same information as the TCP format with some HTTP-specific fields such as
15616 the request, the status code, and captures of headers and cookies. This
15617 format is recommended for HTTP proxies.
15618
Emeric Brun3a058f32009-06-30 18:26:00 +020015619 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
15620 fields arranged in the same order as the CLF format. In this mode, all
15621 timers, captures, flags, etc... appear one per field after the end of the
15622 common fields, in the same order they appear in the standard HTTP format.
15623
William Lallemand48940402012-01-30 16:47:22 +010015624 - the custom log format, allows you to make your own log line.
15625
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015626Next sections will go deeper into details for each of these formats. Format
15627specification will be performed on a "field" basis. Unless stated otherwise, a
15628field is a portion of text delimited by any number of spaces. Since syslog
15629servers are susceptible of inserting fields at the beginning of a line, it is
15630always assumed that the first field is the one containing the process name and
15631identifier.
15632
15633Note : Since log lines may be quite long, the log examples in sections below
15634 might be broken into multiple lines. The example log lines will be
15635 prefixed with 3 closing angle brackets ('>>>') and each time a log is
15636 broken into multiple lines, each non-final line will end with a
15637 backslash ('\') and the next line will start indented by two characters.
15638
15639
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200156408.2.1. Default log format
15641-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015642
15643This format is used when no specific option is set. The log is emitted as soon
15644as the connection is accepted. One should note that this currently is the only
15645format which logs the request's destination IP and ports.
15646
15647 Example :
15648 listen www
15649 mode http
15650 log global
15651 server srv1 127.0.0.1:8000
15652
15653 >>> Feb 6 12:12:09 localhost \
15654 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
15655 (www/HTTP)
15656
15657 Field Format Extract from the example above
15658 1 process_name '[' pid ']:' haproxy[14385]:
15659 2 'Connect from' Connect from
15660 3 source_ip ':' source_port 10.0.1.2:33312
15661 4 'to' to
15662 5 destination_ip ':' destination_port 10.0.3.31:8012
15663 6 '(' frontend_name '/' mode ')' (www/HTTP)
15664
15665Detailed fields description :
15666 - "source_ip" is the IP address of the client which initiated the connection.
15667 - "source_port" is the TCP port of the client which initiated the connection.
15668 - "destination_ip" is the IP address the client connected to.
15669 - "destination_port" is the TCP port the client connected to.
15670 - "frontend_name" is the name of the frontend (or listener) which received
15671 and processed the connection.
15672 - "mode is the mode the frontend is operating (TCP or HTTP).
15673
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015674In case of a UNIX socket, the source and destination addresses are marked as
15675"unix:" and the ports reflect the internal ID of the socket which accepted the
15676connection (the same ID as reported in the stats).
15677
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015678It is advised not to use this deprecated format for newer installations as it
15679will eventually disappear.
15680
15681
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200156828.2.2. TCP log format
15683---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015684
15685The TCP format is used when "option tcplog" is specified in the frontend, and
15686is the recommended format for pure TCP proxies. It provides a lot of precious
15687information for troubleshooting. Since this format includes timers and byte
15688counts, the log is normally emitted at the end of the session. It can be
15689emitted earlier if "option logasap" is specified, which makes sense in most
15690environments with long sessions such as remote terminals. Sessions which match
15691the "monitor" rules are never logged. It is also possible not to emit logs for
15692sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015693specifying "option dontlognull" in the frontend. Successful connections will
15694not be logged if "option dontlog-normal" is specified in the frontend. A few
15695fields may slightly vary depending on some configuration options, those are
15696marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015697
15698 Example :
15699 frontend fnt
15700 mode tcp
15701 option tcplog
15702 log global
15703 default_backend bck
15704
15705 backend bck
15706 server srv1 127.0.0.1:8000
15707
15708 >>> Feb 6 12:12:56 localhost \
15709 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
15710 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
15711
15712 Field Format Extract from the example above
15713 1 process_name '[' pid ']:' haproxy[14387]:
15714 2 client_ip ':' client_port 10.0.1.2:33313
15715 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
15716 4 frontend_name fnt
15717 5 backend_name '/' server_name bck/srv1
15718 6 Tw '/' Tc '/' Tt* 0/0/5007
15719 7 bytes_read* 212
15720 8 termination_state --
15721 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
15722 10 srv_queue '/' backend_queue 0/0
15723
15724Detailed fields description :
15725 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015726 connection to haproxy. If the connection was accepted on a UNIX socket
15727 instead, the IP address would be replaced with the word "unix". Note that
15728 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015729 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010015730 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015731 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015732
15733 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015734 If the connection was accepted on a UNIX socket instead, the port would be
15735 replaced with the ID of the accepting socket, which is also reported in the
15736 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015737
15738 - "accept_date" is the exact date when the connection was received by haproxy
15739 (which might be very slightly different from the date observed on the
15740 network if there was some queuing in the system's backlog). This is usually
15741 the same date which may appear in any upstream firewall's log.
15742
15743 - "frontend_name" is the name of the frontend (or listener) which received
15744 and processed the connection.
15745
15746 - "backend_name" is the name of the backend (or listener) which was selected
15747 to manage the connection to the server. This will be the same as the
15748 frontend if no switching rule has been applied, which is common for TCP
15749 applications.
15750
15751 - "server_name" is the name of the last server to which the connection was
15752 sent, which might differ from the first one if there were connection errors
15753 and a redispatch occurred. Note that this server belongs to the backend
15754 which processed the request. If the connection was aborted before reaching
15755 a server, "<NOSRV>" is indicated instead of a server name.
15756
15757 - "Tw" is the total time in milliseconds spent waiting in the various queues.
15758 It can be "-1" if the connection was aborted before reaching the queue.
15759 See "Timers" below for more details.
15760
15761 - "Tc" is the total time in milliseconds spent waiting for the connection to
15762 establish to the final server, including retries. It can be "-1" if the
15763 connection was aborted before a connection could be established. See
15764 "Timers" below for more details.
15765
15766 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015767 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015768 "option logasap" was specified, then the time counting stops at the moment
15769 the log is emitted. In this case, a '+' sign is prepended before the value,
15770 indicating that the final one will be larger. See "Timers" below for more
15771 details.
15772
15773 - "bytes_read" is the total number of bytes transmitted from the server to
15774 the client when the log is emitted. If "option logasap" is specified, the
15775 this value will be prefixed with a '+' sign indicating that the final one
15776 may be larger. Please note that this value is a 64-bit counter, so log
15777 analysis tools must be able to handle it without overflowing.
15778
15779 - "termination_state" is the condition the session was in when the session
15780 ended. This indicates the session state, which side caused the end of
15781 session to happen, and for what reason (timeout, error, ...). The normal
15782 flags should be "--", indicating the session was closed by either end with
15783 no data remaining in buffers. See below "Session state at disconnection"
15784 for more details.
15785
15786 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015787 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015788 limits have been reached. For instance, if actconn is close to 512 when
15789 multiple connection errors occur, chances are high that the system limits
15790 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015791 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015792
15793 - "feconn" is the total number of concurrent connections on the frontend when
15794 the session was logged. It is useful to estimate the amount of resource
15795 required to sustain high loads, and to detect when the frontend's "maxconn"
15796 has been reached. Most often when this value increases by huge jumps, it is
15797 because there is congestion on the backend servers, but sometimes it can be
15798 caused by a denial of service attack.
15799
15800 - "beconn" is the total number of concurrent connections handled by the
15801 backend when the session was logged. It includes the total number of
15802 concurrent connections active on servers as well as the number of
15803 connections pending in queues. It is useful to estimate the amount of
15804 additional servers needed to support high loads for a given application.
15805 Most often when this value increases by huge jumps, it is because there is
15806 congestion on the backend servers, but sometimes it can be caused by a
15807 denial of service attack.
15808
15809 - "srv_conn" is the total number of concurrent connections still active on
15810 the server when the session was logged. It can never exceed the server's
15811 configured "maxconn" parameter. If this value is very often close or equal
15812 to the server's "maxconn", it means that traffic regulation is involved a
15813 lot, meaning that either the server's maxconn value is too low, or that
15814 there aren't enough servers to process the load with an optimal response
15815 time. When only one of the server's "srv_conn" is high, it usually means
15816 that this server has some trouble causing the connections to take longer to
15817 be processed than on other servers.
15818
15819 - "retries" is the number of connection retries experienced by this session
15820 when trying to connect to the server. It must normally be zero, unless a
15821 server is being stopped at the same moment the connection was attempted.
15822 Frequent retries generally indicate either a network problem between
15823 haproxy and the server, or a misconfigured system backlog on the server
15824 preventing new connections from being queued. This field may optionally be
15825 prefixed with a '+' sign, indicating that the session has experienced a
15826 redispatch after the maximal retry count has been reached on the initial
15827 server. In this case, the server name appearing in the log is the one the
15828 connection was redispatched to, and not the first one, though both may
15829 sometimes be the same in case of hashing for instance. So as a general rule
15830 of thumb, when a '+' is present in front of the retry count, this count
15831 should not be attributed to the logged server.
15832
15833 - "srv_queue" is the total number of requests which were processed before
15834 this one in the server queue. It is zero when the request has not gone
15835 through the server queue. It makes it possible to estimate the approximate
15836 server's response time by dividing the time spent in queue by the number of
15837 requests in the queue. It is worth noting that if a session experiences a
15838 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015839 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015840 backend queue unless a redispatch occurs.
15841
15842 - "backend_queue" is the total number of requests which were processed before
15843 this one in the backend's global queue. It is zero when the request has not
15844 gone through the global queue. It makes it possible to estimate the average
15845 queue length, which easily translates into a number of missing servers when
15846 divided by a server's "maxconn" parameter. It is worth noting that if a
15847 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010015848 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015849 through both the server queue and the backend queue unless a redispatch
15850 occurs.
15851
15852
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158538.2.3. HTTP log format
15854----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015855
15856The HTTP format is the most complete and the best suited for HTTP proxies. It
15857is enabled by when "option httplog" is specified in the frontend. It provides
15858the same level of information as the TCP format with additional features which
15859are specific to the HTTP protocol. Just like the TCP format, the log is usually
15860emitted at the end of the session, unless "option logasap" is specified, which
15861generally only makes sense for download sites. A session which matches the
15862"monitor" rules will never logged. It is also possible not to log sessions for
15863which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020015864frontend. Successful connections will not be logged if "option dontlog-normal"
15865is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015866
15867Most fields are shared with the TCP log, some being different. A few fields may
15868slightly vary depending on some configuration options. Those ones are marked
15869with a star ('*') after the field name below.
15870
15871 Example :
15872 frontend http-in
15873 mode http
15874 option httplog
15875 log global
15876 default_backend bck
15877
15878 backend static
15879 server srv1 127.0.0.1:8000
15880
15881 >>> Feb 6 12:14:14 localhost \
15882 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
15883 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 +010015884 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015885
15886 Field Format Extract from the example above
15887 1 process_name '[' pid ']:' haproxy[14389]:
15888 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015889 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015890 4 frontend_name http-in
15891 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015892 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015893 7 status_code 200
15894 8 bytes_read* 2750
15895 9 captured_request_cookie -
15896 10 captured_response_cookie -
15897 11 termination_state ----
15898 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
15899 13 srv_queue '/' backend_queue 0/0
15900 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
15901 15 '{' captured_response_headers* '}' {}
15902 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010015903
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015904Detailed fields description :
15905 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015906 connection to haproxy. If the connection was accepted on a UNIX socket
15907 instead, the IP address would be replaced with the word "unix". Note that
15908 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015909 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010015910 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015911 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015912
15913 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010015914 If the connection was accepted on a UNIX socket instead, the port would be
15915 replaced with the ID of the accepting socket, which is also reported in the
15916 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015917
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015918 - "request_date" is the exact date when the first byte of the HTTP request
15919 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015920
15921 - "frontend_name" is the name of the frontend (or listener) which received
15922 and processed the connection.
15923
15924 - "backend_name" is the name of the backend (or listener) which was selected
15925 to manage the connection to the server. This will be the same as the
15926 frontend if no switching rule has been applied.
15927
15928 - "server_name" is the name of the last server to which the connection was
15929 sent, which might differ from the first one if there were connection errors
15930 and a redispatch occurred. Note that this server belongs to the backend
15931 which processed the request. If the request was aborted before reaching a
15932 server, "<NOSRV>" is indicated instead of a server name. If the request was
15933 intercepted by the stats subsystem, "<STATS>" is indicated instead.
15934
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015935 - "TR" is the total time in milliseconds spent waiting for a full HTTP
15936 request from the client (not counting body) after the first byte was
15937 received. It can be "-1" if the connection was aborted before a complete
15938 request could be received or the a bad request was received. It should
15939 always be very small because a request generally fits in one single packet.
15940 Large times here generally indicate network issues between the client and
15941 haproxy or requests being typed by hand. See "Timers" below for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015942
15943 - "Tw" is the total time in milliseconds spent waiting in the various queues.
15944 It can be "-1" if the connection was aborted before reaching the queue.
15945 See "Timers" below for more details.
15946
15947 - "Tc" is the total time in milliseconds spent waiting for the connection to
15948 establish to the final server, including retries. It can be "-1" if the
15949 request was aborted before a connection could be established. See "Timers"
15950 below for more details.
15951
15952 - "Tr" is the total time in milliseconds spent waiting for the server to send
15953 a full HTTP response, not counting data. It can be "-1" if the request was
15954 aborted before a complete response could be received. It generally matches
15955 the server's processing time for the request, though it may be altered by
15956 the amount of data sent by the client to the server. Large times here on
15957 "GET" requests generally indicate an overloaded server. See "Timers" below
15958 for more details.
15959
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020015960 - "Ta" is the time the request remained active in haproxy, which is the total
15961 time in milliseconds elapsed between the first byte of the request was
15962 received and the last byte of response was sent. It covers all possible
15963 processing except the handshake (see Th) and idle time (see Ti). There is
15964 one exception, if "option logasap" was specified, then the time counting
15965 stops at the moment the log is emitted. In this case, a '+' sign is
15966 prepended before the value, indicating that the final one will be larger.
15967 See "Timers" below for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010015968
15969 - "status_code" is the HTTP status code returned to the client. This status
15970 is generally set by the server, but it might also be set by haproxy when
15971 the server cannot be reached or when its response is blocked by haproxy.
15972
15973 - "bytes_read" is the total number of bytes transmitted to the client when
15974 the log is emitted. This does include HTTP headers. If "option logasap" is
15975 specified, the this value will be prefixed with a '+' sign indicating that
15976 the final one may be larger. Please note that this value is a 64-bit
15977 counter, so log analysis tools must be able to handle it without
15978 overflowing.
15979
15980 - "captured_request_cookie" is an optional "name=value" entry indicating that
15981 the client had this cookie in the request. The cookie name and its maximum
15982 length are defined by the "capture cookie" statement in the frontend
15983 configuration. The field is a single dash ('-') when the option is not
15984 set. Only one cookie may be captured, it is generally used to track session
15985 ID exchanges between a client and a server to detect session crossing
15986 between clients due to application bugs. For more details, please consult
15987 the section "Capturing HTTP headers and cookies" below.
15988
15989 - "captured_response_cookie" is an optional "name=value" entry indicating
15990 that the server has returned a cookie with its response. The cookie name
15991 and its maximum length are defined by the "capture cookie" statement in the
15992 frontend configuration. The field is a single dash ('-') when the option is
15993 not set. Only one cookie may be captured, it is generally used to track
15994 session ID exchanges between a client and a server to detect session
15995 crossing between clients due to application bugs. For more details, please
15996 consult the section "Capturing HTTP headers and cookies" below.
15997
15998 - "termination_state" is the condition the session was in when the session
15999 ended. This indicates the session state, which side caused the end of
16000 session to happen, for what reason (timeout, error, ...), just like in TCP
16001 logs, and information about persistence operations on cookies in the last
16002 two characters. The normal flags should begin with "--", indicating the
16003 session was closed by either end with no data remaining in buffers. See
16004 below "Session state at disconnection" for more details.
16005
16006 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016007 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016008 limits have been reached. For instance, if actconn is close to 512 or 1024
16009 when multiple connection errors occur, chances are high that the system
16010 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016011 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016012 system.
16013
16014 - "feconn" is the total number of concurrent connections on the frontend when
16015 the session was logged. It is useful to estimate the amount of resource
16016 required to sustain high loads, and to detect when the frontend's "maxconn"
16017 has been reached. Most often when this value increases by huge jumps, it is
16018 because there is congestion on the backend servers, but sometimes it can be
16019 caused by a denial of service attack.
16020
16021 - "beconn" is the total number of concurrent connections handled by the
16022 backend when the session was logged. It includes the total number of
16023 concurrent connections active on servers as well as the number of
16024 connections pending in queues. It is useful to estimate the amount of
16025 additional servers needed to support high loads for a given application.
16026 Most often when this value increases by huge jumps, it is because there is
16027 congestion on the backend servers, but sometimes it can be caused by a
16028 denial of service attack.
16029
16030 - "srv_conn" is the total number of concurrent connections still active on
16031 the server when the session was logged. It can never exceed the server's
16032 configured "maxconn" parameter. If this value is very often close or equal
16033 to the server's "maxconn", it means that traffic regulation is involved a
16034 lot, meaning that either the server's maxconn value is too low, or that
16035 there aren't enough servers to process the load with an optimal response
16036 time. When only one of the server's "srv_conn" is high, it usually means
16037 that this server has some trouble causing the requests to take longer to be
16038 processed than on other servers.
16039
16040 - "retries" is the number of connection retries experienced by this session
16041 when trying to connect to the server. It must normally be zero, unless a
16042 server is being stopped at the same moment the connection was attempted.
16043 Frequent retries generally indicate either a network problem between
16044 haproxy and the server, or a misconfigured system backlog on the server
16045 preventing new connections from being queued. This field may optionally be
16046 prefixed with a '+' sign, indicating that the session has experienced a
16047 redispatch after the maximal retry count has been reached on the initial
16048 server. In this case, the server name appearing in the log is the one the
16049 connection was redispatched to, and not the first one, though both may
16050 sometimes be the same in case of hashing for instance. So as a general rule
16051 of thumb, when a '+' is present in front of the retry count, this count
16052 should not be attributed to the logged server.
16053
16054 - "srv_queue" is the total number of requests which were processed before
16055 this one in the server queue. It is zero when the request has not gone
16056 through the server queue. It makes it possible to estimate the approximate
16057 server's response time by dividing the time spent in queue by the number of
16058 requests in the queue. It is worth noting that if a session experiences a
16059 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016060 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016061 backend queue unless a redispatch occurs.
16062
16063 - "backend_queue" is the total number of requests which were processed before
16064 this one in the backend's global queue. It is zero when the request has not
16065 gone through the global queue. It makes it possible to estimate the average
16066 queue length, which easily translates into a number of missing servers when
16067 divided by a server's "maxconn" parameter. It is worth noting that if a
16068 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016069 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016070 through both the server queue and the backend queue unless a redispatch
16071 occurs.
16072
16073 - "captured_request_headers" is a list of headers captured in the request due
16074 to the presence of the "capture request header" statement in the frontend.
16075 Multiple headers can be captured, they will be delimited by a vertical bar
16076 ('|'). When no capture is enabled, the braces do not appear, causing a
16077 shift of remaining fields. It is important to note that this field may
16078 contain spaces, and that using it requires a smarter log parser than when
16079 it's not used. Please consult the section "Capturing HTTP headers and
16080 cookies" below for more details.
16081
16082 - "captured_response_headers" is a list of headers captured in the response
16083 due to the presence of the "capture response header" statement in the
16084 frontend. Multiple headers can be captured, they will be delimited by a
16085 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16086 causing a shift of remaining fields. It is important to note that this
16087 field may contain spaces, and that using it requires a smarter log parser
16088 than when it's not used. Please consult the section "Capturing HTTP headers
16089 and cookies" below for more details.
16090
16091 - "http_request" is the complete HTTP request line, including the method,
16092 request and HTTP version string. Non-printable characters are encoded (see
16093 below the section "Non-printable characters"). This is always the last
16094 field, and it is always delimited by quotes and is the only one which can
16095 contain quotes. If new fields are added to the log format, they will be
16096 added before this field. This field might be truncated if the request is
16097 huge and does not fit in the standard syslog buffer (1024 characters). This
16098 is the reason why this field must always remain the last one.
16099
16100
Cyril Bontédc4d9032012-04-08 21:57:39 +0200161018.2.4. Custom log format
16102------------------------
William Lallemand48940402012-01-30 16:47:22 +010016103
Willy Tarreau2beef582012-12-20 17:22:52 +010016104The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016105mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016106
Davor Ocelice9ed2812017-12-25 17:49:28 +010016107HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016108Variables can take arguments using braces ('{}'), and multiple arguments are
16109separated by commas within the braces. Flags may be added or removed by
16110prefixing them with a '+' or '-' sign.
16111
16112Special variable "%o" may be used to propagate its flags to all other
16113variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016114("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016115
Willy Tarreauc8368452012-12-21 00:09:23 +010016116If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016117as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016118less common information such as the client's SSL certificate's DN, or to log
16119the key that would be used to store an entry into a stick table.
16120
William Lallemand48940402012-01-30 16:47:22 +010016121Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016122In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016123in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016124
Dragan Dosen835b9212016-02-12 13:23:03 +010016125Note: when using the RFC5424 syslog message format, the characters '"',
16126'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16127https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16128such cases, the use of the flag "E" should be considered.
16129
William Lallemand48940402012-01-30 16:47:22 +010016130Flags are :
16131 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016132 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016133 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16134 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016135
16136 Example:
16137
16138 log-format %T\ %t\ Some\ Text
16139 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16140
Dragan Dosen835b9212016-02-12 13:23:03 +010016141 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16142
William Lallemand48940402012-01-30 16:47:22 +010016143At the moment, the default HTTP format is defined this way :
16144
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016145 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16146 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016147
William Lallemandbddd4fd2012-02-27 11:23:10 +010016148the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016149
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016150 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16151 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16152 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016153
William Lallemandbddd4fd2012-02-27 11:23:10 +010016154and the default TCP format is defined this way :
16155
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016156 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16157 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016158
William Lallemand48940402012-01-30 16:47:22 +010016159Please refer to the table below for currently defined variables :
16160
William Lallemandbddd4fd2012-02-27 11:23:10 +010016161 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016162 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016163 +---+------+-----------------------------------------------+-------------+
16164 | | %o | special variable, apply flags on all next var | |
16165 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016166 | | %B | bytes_read (from server to client) | numeric |
16167 | H | %CC | captured_request_cookie | string |
16168 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016169 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016170 | H | %HM | HTTP method (ex: POST) | string |
16171 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016172 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016173 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16174 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016175 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016176 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016177 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016178 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016179 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016180 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016181 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016182 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16183 | H | %Ti | idle time before the HTTP request | numeric |
16184 | H | %Tq | Th + Ti + TR | numeric |
16185 | H | %TR | time to receive the full request from 1st byte| numeric |
16186 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016187 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016188 | | %Tt | Tt | numeric |
16189 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016190 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016191 | | %ac | actconn | numeric |
16192 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016193 | | %bc | beconn (backend concurrent connections) | numeric |
16194 | | %bi | backend_source_ip (connecting address) | IP |
16195 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016196 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016197 | | %ci | client_ip (accepted address) | IP |
16198 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016199 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016200 | | %fc | feconn (frontend concurrent connections) | numeric |
16201 | | %fi | frontend_ip (accepting address) | IP |
16202 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016203 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016204 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016205 | | %hr | captured_request_headers default style | string |
16206 | | %hrl | captured_request_headers CLF style | string list |
16207 | | %hs | captured_response_headers default style | string |
16208 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016209 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016210 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016211 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016212 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016213 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016214 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016215 | | %sc | srv_conn (server concurrent connections) | numeric |
16216 | | %si | server_IP (target address) | IP |
16217 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016218 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016219 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16220 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016221 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016222 | H | %tr | date_time of HTTP request | date |
16223 | H | %trg | gmt_date_time of start of HTTP request | date |
16224 | H | %trl | locla_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016225 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016226 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016227 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016228
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016229 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016230
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016231
162328.2.5. Error log format
16233-----------------------
16234
16235When an incoming connection fails due to an SSL handshake or an invalid PROXY
16236protocol header, haproxy will log the event using a shorter, fixed line format.
16237By default, logs are emitted at the LOG_INFO level, unless the option
16238"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016239will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016240logged if the "dontlognull" option is set.
16241
16242The format looks like this :
16243
16244 >>> Dec 3 18:27:14 localhost \
16245 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16246 Connection error during SSL handshake
16247
16248 Field Format Extract from the example above
16249 1 process_name '[' pid ']:' haproxy[6103]:
16250 2 client_ip ':' client_port 127.0.0.1:56059
16251 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16252 4 frontend_name "/" bind_name ":" frt/f1:
16253 5 message Connection error during SSL handshake
16254
16255These fields just provide minimal information to help debugging connection
16256failures.
16257
16258
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162598.3. Advanced logging options
16260-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016261
16262Some advanced logging options are often looked for but are not easy to find out
16263just by looking at the various options. Here is an entry point for the few
16264options which can enable better logging. Please refer to the keywords reference
16265for more information about their usage.
16266
16267
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162688.3.1. Disabling logging of external tests
16269------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016270
16271It is quite common to have some monitoring tools perform health checks on
16272haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
16273commercial load-balancer, and sometimes it will simply be a more complete
16274monitoring system such as Nagios. When the tests are very frequent, users often
16275ask how to disable logging for those checks. There are three possibilities :
16276
16277 - if connections come from everywhere and are just TCP probes, it is often
16278 desired to simply disable logging of connections without data exchange, by
16279 setting "option dontlognull" in the frontend. It also disables logging of
16280 port scans, which may or may not be desired.
16281
16282 - if the connection come from a known source network, use "monitor-net" to
16283 declare this network as monitoring only. Any host in this network will then
16284 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016285 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016286 such as other load-balancers.
16287
16288 - if the tests are performed on a known URI, use "monitor-uri" to declare
16289 this URI as dedicated to monitoring. Any host sending this request will
16290 only get the result of a health-check, and the request will not be logged.
16291
16292
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162938.3.2. Logging before waiting for the session to terminate
16294----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016295
16296The problem with logging at end of connection is that you have no clue about
16297what is happening during very long sessions, such as remote terminal sessions
16298or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010016299"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016300just before data transfer begins. This means that in case of TCP, it will still
16301log the connection status to the server, and in case of HTTP, it will log just
16302after processing the server headers. In this case, the number of bytes reported
16303is the number of header bytes sent to the client. In order to avoid confusion
16304with normal logs, the total time field and the number of bytes are prefixed
16305with a '+' sign which means that real numbers are certainly larger.
16306
16307
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163088.3.3. Raising log level upon errors
16309------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016310
16311Sometimes it is more convenient to separate normal traffic from errors logs,
16312for instance in order to ease error monitoring from log files. When the option
16313"log-separate-errors" is used, connections which experience errors, timeouts,
16314retries, redispatches or HTTP status codes 5xx will see their syslog level
16315raised from "info" to "err". This will help a syslog daemon store the log in
16316a separate file. It is very important to keep the errors in the normal traffic
16317file too, so that log ordering is not altered. You should also be careful if
16318you already have configured your syslog daemon to store all logs higher than
16319"notice" in an "admin" file, because the "err" level is higher than "notice".
16320
16321
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163228.3.4. Disabling logging of successful connections
16323--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016324
16325Although this may sound strange at first, some large sites have to deal with
16326multiple thousands of logs per second and are experiencing difficulties keeping
16327them intact for a long time or detecting errors within them. If the option
16328"dontlog-normal" is set on the frontend, all normal connections will not be
16329logged. In this regard, a normal connection is defined as one without any
16330error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
16331and a response with a status 5xx is not considered normal and will be logged
16332too. Of course, doing is is really discouraged as it will remove most of the
16333useful information from the logs. Do this only if you have no other
16334alternative.
16335
16336
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163378.4. Timing events
16338------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016339
16340Timers provide a great help in troubleshooting network problems. All values are
16341reported in milliseconds (ms). These timers should be used in conjunction with
16342the session termination flags. In TCP mode with "option tcplog" set on the
16343frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016344mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
16345addition, three other measures are provided, "Th", "Ti", and "Tq".
16346
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010016347Timings events in HTTP mode:
16348
16349 first request 2nd request
16350 |<-------------------------------->|<-------------- ...
16351 t tr t tr ...
16352 ---|----|----|----|----|----|----|----|----|--
16353 : Th Ti TR Tw Tc Tr Td : Ti ...
16354 :<---- Tq ---->: :
16355 :<-------------- Tt -------------->:
16356 :<--------- Ta --------->:
16357
16358Timings events in TCP mode:
16359
16360 TCP session
16361 |<----------------->|
16362 t t
16363 ---|----|----|----|----|---
16364 | Th Tw Tc Td |
16365 |<------ Tt ------->|
16366
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016367 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016368 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016369 only happen once during the whole connection's lifetime. A large time here
16370 may indicate that the client only pre-established the connection without
16371 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010016372 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016373 SSL handshake was very expensive to compute.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016374
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016375 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
16376 counts between the end of the handshakes and the first byte of the HTTP
16377 request. When dealing with a second request in keep-alive mode, it starts
16378 to count after the end of the transmission the previous response. Some
16379 browsers pre-establish connections to a server in order to reduce the
16380 latency of a future request, and keep them pending until they need it. This
16381 delay will be reported as the idle time. A value of -1 indicates that
16382 nothing was received on the connection.
16383
16384 - TR: total time to get the client request (HTTP mode only). It's the time
16385 elapsed between the first bytes received and the moment the proxy received
16386 the empty line marking the end of the HTTP headers. The value "-1"
16387 indicates that the end of headers has never been seen. This happens when
16388 the client closes prematurely or times out. This time is usually very short
16389 since most requests fit in a single packet. A large time may indicate a
16390 request typed by hand during a test.
16391
16392 - Tq: total time to get the client request from the accept date or since the
16393 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010016394 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 +020016395 returns -1 as well. This timer used to be very useful before the arrival of
16396 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
16397 it in favor of TR nowadays, as the idle time adds a lot of noise to the
16398 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016399
16400 - Tw: total time spent in the queues waiting for a connection slot. It
16401 accounts for backend queue as well as the server queues, and depends on the
16402 queue size, and the time needed for the server to complete previous
16403 requests. The value "-1" means that the request was killed before reaching
16404 the queue, which is generally what happens with invalid or denied requests.
16405
16406 - Tc: total time to establish the TCP connection to the server. It's the time
16407 elapsed between the moment the proxy sent the connection request, and the
16408 moment it was acknowledged by the server, or between the TCP SYN packet and
16409 the matching SYN/ACK packet in return. The value "-1" means that the
16410 connection never established.
16411
16412 - Tr: server response time (HTTP mode only). It's the time elapsed between
16413 the moment the TCP connection was established to the server and the moment
16414 the server sent its complete response headers. It purely shows its request
16415 processing time, without the network overhead due to the data transmission.
16416 It is worth noting that when the client has data to send to the server, for
16417 instance during a POST request, the time already runs, and this can distort
16418 apparent response time. For this reason, it's generally wise not to trust
16419 too much this field for POST requests initiated from clients behind an
16420 untrusted network. A value of "-1" here means that the last the response
16421 header (empty line) was never seen, most likely because the server timeout
16422 stroke before the server managed to process the request.
16423
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016424 - Ta: total active time for the HTTP request, between the moment the proxy
16425 received the first byte of the request header and the emission of the last
16426 byte of the response body. The exception is when the "logasap" option is
16427 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
16428 a '+' sign. From this field, we can deduce "Td", the data transmission time,
16429 by subtracting other timers when valid :
16430
16431 Td = Ta - (TR + Tw + Tc + Tr)
16432
16433 Timers with "-1" values have to be excluded from this equation. Note that
16434 "Ta" can never be negative.
16435
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016436 - Tt: total session duration time, between the moment the proxy accepted it
16437 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016438 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
16439 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016440 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016441
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016442 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016443
16444 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016445 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
16446 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016447
16448These timers provide precious indications on trouble causes. Since the TCP
16449protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
16450that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016451due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
16452"Tt" is close to a timeout value specified in the configuration, it often means
16453that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016454
16455Most common cases :
16456
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016457 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
16458 the client and the proxy. This is very rare on local networks but might
16459 happen when clients are on far remote networks and send large requests. It
16460 may happen that values larger than usual appear here without any network
16461 cause. Sometimes, during an attack or just after a resource starvation has
16462 ended, haproxy may accept thousands of connections in a few milliseconds.
16463 The time spent accepting these connections will inevitably slightly delay
16464 processing of other connections, and it can happen that request times in the
16465 order of a few tens of milliseconds are measured after a few thousands of
16466 new connections have been accepted at once. Using one of the keep-alive
16467 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020016468 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016469
16470 - If "Tc" is close to 3000, a packet has probably been lost between the
16471 server and the proxy during the server connection phase. This value should
16472 always be very low, such as 1 ms on local networks and less than a few tens
16473 of ms on remote networks.
16474
Willy Tarreau55165fe2009-05-10 12:02:55 +020016475 - If "Tr" is nearly always lower than 3000 except some rare values which seem
16476 to be the average majored by 3000, there are probably some packets lost
16477 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016478
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016479 - If "Ta" is large even for small byte counts, it generally is because
16480 neither the client nor the server decides to close the connection while
16481 haproxy is running in tunnel mode and both have agreed on a keep-alive
16482 connection mode. In order to solve this issue, it will be needed to specify
16483 one of the HTTP options to manipulate keep-alive or close options on either
16484 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
16485 important when connection regulation is used with the "maxconn" option on
16486 the servers, since no new connection will be sent to the server until
16487 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016488
16489Other noticeable HTTP log cases ('xx' means any value to be ignored) :
16490
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016491 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016492 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016493 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016494
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016495 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016496 or it aborted too early. Check the session termination flags
16497 then "timeout http-request" and "timeout client" settings.
16498
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016499 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016500 servers were out of order, because the request was invalid
16501 or forbidden by ACL rules. Check the session termination
16502 flags.
16503
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016504 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
16505 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016506 Check the session termination flags, then check the
16507 "timeout connect" setting. Note that the tarpit action might
16508 return similar-looking patterns, with "Tw" equal to the time
16509 the client connection was maintained open.
16510
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016511 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016512 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016513 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016514 termination flags, then check the "timeout server" setting.
16515
16516
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165178.5. Session state at disconnection
16518-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016519
16520TCP and HTTP logs provide a session termination indicator in the
16521"termination_state" field, just before the number of active connections. It is
165222-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
16523each of which has a special meaning :
16524
16525 - On the first character, a code reporting the first event which caused the
16526 session to terminate :
16527
16528 C : the TCP session was unexpectedly aborted by the client.
16529
16530 S : the TCP session was unexpectedly aborted by the server, or the
16531 server explicitly refused it.
16532
16533 P : the session was prematurely aborted by the proxy, because of a
16534 connection limit enforcement, because a DENY filter was matched,
16535 because of a security check which detected and blocked a dangerous
16536 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010016537 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020016538
16539 L : the session was locally processed by haproxy and was not passed to
16540 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016541
16542 R : a resource on the proxy has been exhausted (memory, sockets, source
16543 ports, ...). Usually, this appears during the connection phase, and
16544 system logs should contain a copy of the precise error. If this
16545 happens, it must be considered as a very serious anomaly which
16546 should be fixed as soon as possible by any means.
16547
16548 I : an internal error was identified by the proxy during a self-check.
16549 This should NEVER happen, and you are encouraged to report any log
16550 containing this, because this would almost certainly be a bug. It
16551 would be wise to preventively restart the process after such an
16552 event too, in case it would be caused by memory corruption.
16553
Simon Horman752dc4a2011-06-21 14:34:59 +090016554 D : the session was killed by haproxy because the server was detected
16555 as down and was configured to kill all connections when going down.
16556
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070016557 U : the session was killed by haproxy on this backup server because an
16558 active server was detected as up and was configured to kill all
16559 backup connections when going up.
16560
Willy Tarreaua2a64e92011-09-07 23:01:56 +020016561 K : the session was actively killed by an admin operating on haproxy.
16562
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016563 c : the client-side timeout expired while waiting for the client to
16564 send or receive data.
16565
16566 s : the server-side timeout expired while waiting for the server to
16567 send or receive data.
16568
16569 - : normal session completion, both the client and the server closed
16570 with nothing left in the buffers.
16571
16572 - on the second character, the TCP or HTTP session state when it was closed :
16573
Willy Tarreauf7b30a92010-12-06 22:59:17 +010016574 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016575 (HTTP mode only). Nothing was sent to any server.
16576
16577 Q : the proxy was waiting in the QUEUE for a connection slot. This can
16578 only happen when servers have a 'maxconn' parameter set. It can
16579 also happen in the global queue after a redispatch consecutive to
16580 a failed attempt to connect to a dying server. If no redispatch is
16581 reported, then no connection attempt was made to any server.
16582
16583 C : the proxy was waiting for the CONNECTION to establish on the
16584 server. The server might at most have noticed a connection attempt.
16585
16586 H : the proxy was waiting for complete, valid response HEADERS from the
16587 server (HTTP only).
16588
16589 D : the session was in the DATA phase.
16590
16591 L : the proxy was still transmitting LAST data to the client while the
16592 server had already finished. This one is very rare as it can only
16593 happen when the client dies while receiving the last packets.
16594
16595 T : the request was tarpitted. It has been held open with the client
16596 during the whole "timeout tarpit" duration or until the client
16597 closed, both of which will be reported in the "Tw" timer.
16598
16599 - : normal session completion after end of data transfer.
16600
16601 - the third character tells whether the persistence cookie was provided by
16602 the client (only in HTTP mode) :
16603
16604 N : the client provided NO cookie. This is usually the case for new
16605 visitors, so counting the number of occurrences of this flag in the
16606 logs generally indicate a valid trend for the site frequentation.
16607
16608 I : the client provided an INVALID cookie matching no known server.
16609 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020016610 cookies between HTTP/HTTPS sites, persistence conditionally
16611 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016612
16613 D : the client provided a cookie designating a server which was DOWN,
16614 so either "option persist" was used and the client was sent to
16615 this server, or it was not set and the client was redispatched to
16616 another server.
16617
Willy Tarreau996a92c2010-10-13 19:30:47 +020016618 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016619 server.
16620
Willy Tarreau996a92c2010-10-13 19:30:47 +020016621 E : the client provided a valid cookie, but with a last date which was
16622 older than what is allowed by the "maxidle" cookie parameter, so
16623 the cookie is consider EXPIRED and is ignored. The request will be
16624 redispatched just as if there was no cookie.
16625
16626 O : the client provided a valid cookie, but with a first date which was
16627 older than what is allowed by the "maxlife" cookie parameter, so
16628 the cookie is consider too OLD and is ignored. The request will be
16629 redispatched just as if there was no cookie.
16630
Willy Tarreauc89ccb62012-04-05 21:18:22 +020016631 U : a cookie was present but was not used to select the server because
16632 some other server selection mechanism was used instead (typically a
16633 "use-server" rule).
16634
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016635 - : does not apply (no cookie set in configuration).
16636
16637 - the last character reports what operations were performed on the persistence
16638 cookie returned by the server (only in HTTP mode) :
16639
16640 N : NO cookie was provided by the server, and none was inserted either.
16641
16642 I : no cookie was provided by the server, and the proxy INSERTED one.
16643 Note that in "cookie insert" mode, if the server provides a cookie,
16644 it will still be overwritten and reported as "I" here.
16645
Willy Tarreau996a92c2010-10-13 19:30:47 +020016646 U : the proxy UPDATED the last date in the cookie that was presented by
16647 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016648 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020016649 date indicated in the cookie. If any other change happens, such as
16650 a redispatch, then the cookie will be marked as inserted instead.
16651
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016652 P : a cookie was PROVIDED by the server and transmitted as-is.
16653
16654 R : the cookie provided by the server was REWRITTEN by the proxy, which
16655 happens in "cookie rewrite" or "cookie prefix" modes.
16656
16657 D : the cookie provided by the server was DELETED by the proxy.
16658
16659 - : does not apply (no cookie set in configuration).
16660
Willy Tarreau996a92c2010-10-13 19:30:47 +020016661The combination of the two first flags gives a lot of information about what
16662was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016663helpful to detect server saturation, network troubles, local system resource
16664starvation, attacks, etc...
16665
16666The most common termination flags combinations are indicated below. They are
16667alphabetically sorted, with the lowercase set just after the upper case for
16668easier finding and understanding.
16669
16670 Flags Reason
16671
16672 -- Normal termination.
16673
16674 CC The client aborted before the connection could be established to the
16675 server. This can happen when haproxy tries to connect to a recently
16676 dead (or unchecked) server, and the client aborts while haproxy is
16677 waiting for the server to respond or for "timeout connect" to expire.
16678
16679 CD The client unexpectedly aborted during data transfer. This can be
16680 caused by a browser crash, by an intermediate equipment between the
16681 client and haproxy which decided to actively break the connection,
16682 by network routing issues between the client and haproxy, or by a
16683 keep-alive session between the server and the client terminated first
16684 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010016685
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016686 cD The client did not send nor acknowledge any data for as long as the
16687 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020016688 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016689
16690 CH The client aborted while waiting for the server to start responding.
16691 It might be the server taking too long to respond or the client
16692 clicking the 'Stop' button too fast.
16693
16694 cH The "timeout client" stroke while waiting for client data during a
16695 POST request. This is sometimes caused by too large TCP MSS values
16696 for PPPoE networks which cannot transport full-sized packets. It can
16697 also happen when client timeout is smaller than server timeout and
16698 the server takes too long to respond.
16699
16700 CQ The client aborted while its session was queued, waiting for a server
16701 with enough empty slots to accept it. It might be that either all the
16702 servers were saturated or that the assigned server was taking too
16703 long a time to respond.
16704
16705 CR The client aborted before sending a full HTTP request. Most likely
16706 the request was typed by hand using a telnet client, and aborted
16707 too early. The HTTP status code is likely a 400 here. Sometimes this
16708 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020016709 and the client. "option http-ignore-probes" can be used to ignore
16710 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016711
16712 cR The "timeout http-request" stroke before the client sent a full HTTP
16713 request. This is sometimes caused by too large TCP MSS values on the
16714 client side for PPPoE networks which cannot transport full-sized
16715 packets, or by clients sending requests by hand and not typing fast
16716 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020016717 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020016718 some browsers started to implement a "pre-connect" feature consisting
16719 in speculatively connecting to some recently visited web sites just
16720 in case the user would like to visit them. This results in many
16721 connections being established to web sites, which end up in 408
16722 Request Timeout if the timeout strikes first, or 400 Bad Request when
16723 the browser decides to close them first. These ones pollute the log
16724 and feed the error counters. Some versions of some browsers have even
16725 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010016726 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020016727 http-ignore-probes" in the frontend, resulting in connections with
16728 zero data transfer to be totally ignored. This will definitely hide
16729 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016730
16731 CT The client aborted while its session was tarpitted. It is important to
16732 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020016733 wrong tarpit rules have been written. If a lot of them happen, it
16734 might make sense to lower the "timeout tarpit" value to something
16735 closer to the average reported "Tw" timer, in order not to consume
16736 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016737
Willy Tarreau570f2212013-06-10 16:42:09 +020016738 LR The request was intercepted and locally handled by haproxy. Generally
16739 it means that this was a redirect or a stats request.
16740
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016741 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016742 the TCP connection (the proxy received a TCP RST or an ICMP message
16743 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010016744 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016745 or no ARP response on local network). When this happens in HTTP mode,
16746 the status code is likely a 502 or 503 here.
16747
16748 sC The "timeout connect" stroke before a connection to the server could
16749 complete. When this happens in HTTP mode, the status code is likely a
16750 503 or 504 here.
16751
16752 SD The connection to the server died with an error during the data
16753 transfer. This usually means that haproxy has received an RST from
16754 the server or an ICMP message from an intermediate equipment while
16755 exchanging data with the server. This can be caused by a server crash
16756 or by a network issue on an intermediate equipment.
16757
16758 sD The server did not send nor acknowledge any data for as long as the
16759 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010016760 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016761 load-balancers, ...), as well as keep-alive sessions maintained
16762 between the client and the server expiring first on haproxy.
16763
16764 SH The server aborted before sending its full HTTP response headers, or
16765 it crashed while processing the request. Since a server aborting at
16766 this moment is very rare, it would be wise to inspect its logs to
16767 control whether it crashed and why. The logged request may indicate a
16768 small set of faulty requests, demonstrating bugs in the application.
16769 Sometimes this might also be caused by an IDS killing the connection
16770 between haproxy and the server.
16771
16772 sH The "timeout server" stroke before the server could return its
16773 response headers. This is the most common anomaly, indicating too
16774 long transactions, probably caused by server or database saturation.
16775 The immediate workaround consists in increasing the "timeout server"
16776 setting, but it is important to keep in mind that the user experience
16777 will suffer from these long response times. The only long term
16778 solution is to fix the application.
16779
16780 sQ The session spent too much time in queue and has been expired. See
16781 the "timeout queue" and "timeout connect" settings to find out how to
16782 fix this if it happens too often. If it often happens massively in
16783 short periods, it may indicate general problems on the affected
16784 servers due to I/O or database congestion, or saturation caused by
16785 external attacks.
16786
16787 PC The proxy refused to establish a connection to the server because the
16788 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020016789 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016790 so that it does not happen anymore. This status is very rare and
16791 might happen when the global "ulimit-n" parameter is forced by hand.
16792
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016793 PD The proxy blocked an incorrectly formatted chunked encoded message in
16794 a request or a response, after the server has emitted its headers. In
16795 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016796 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020016797 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016798
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016799 PH The proxy blocked the server's response, because it was invalid,
16800 incomplete, dangerous (cache control), or matched a security filter.
16801 In any case, an HTTP 502 error is sent to the client. One possible
16802 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010016803 containing unauthorized characters. It is also possible but quite
16804 rare, that the proxy blocked a chunked-encoding request from the
16805 client due to an invalid syntax, before the server responded. In this
16806 case, an HTTP 400 error is sent to the client and reported in the
16807 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016808
16809 PR The proxy blocked the client's HTTP request, either because of an
16810 invalid HTTP syntax, in which case it returned an HTTP 400 error to
16811 the client, or because a deny filter matched, in which case it
16812 returned an HTTP 403 error.
16813
16814 PT The proxy blocked the client's request and has tarpitted its
16815 connection before returning it a 500 server error. Nothing was sent
16816 to the server. The connection was maintained open for as long as
16817 reported by the "Tw" timer field.
16818
16819 RC A local resource has been exhausted (memory, sockets, source ports)
16820 preventing the connection to the server from establishing. The error
16821 logs will tell precisely what was missing. This is very rare and can
16822 only be solved by proper system tuning.
16823
Willy Tarreau996a92c2010-10-13 19:30:47 +020016824The combination of the two last flags gives a lot of information about how
16825persistence was handled by the client, the server and by haproxy. This is very
16826important to troubleshoot disconnections, when users complain they have to
16827re-authenticate. The commonly encountered flags are :
16828
16829 -- Persistence cookie is not enabled.
16830
16831 NN No cookie was provided by the client, none was inserted in the
16832 response. For instance, this can be in insert mode with "postonly"
16833 set on a GET request.
16834
16835 II A cookie designating an invalid server was provided by the client,
16836 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016837 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020016838 value can be presented by a client when no other server knows it.
16839
16840 NI No cookie was provided by the client, one was inserted in the
16841 response. This typically happens for first requests from every user
16842 in "insert" mode, which makes it an easy way to count real users.
16843
16844 VN A cookie was provided by the client, none was inserted in the
16845 response. This happens for most responses for which the client has
16846 already got a cookie.
16847
16848 VU A cookie was provided by the client, with a last visit date which is
16849 not completely up-to-date, so an updated cookie was provided in
16850 response. This can also happen if there was no date at all, or if
16851 there was a date but the "maxidle" parameter was not set, so that the
16852 cookie can be switched to unlimited time.
16853
16854 EI A cookie was provided by the client, with a last visit date which is
16855 too old for the "maxidle" parameter, so the cookie was ignored and a
16856 new cookie was inserted in the response.
16857
16858 OI A cookie was provided by the client, with a first visit date which is
16859 too old for the "maxlife" parameter, so the cookie was ignored and a
16860 new cookie was inserted in the response.
16861
16862 DI The server designated by the cookie was down, a new server was
16863 selected and a new cookie was emitted in the response.
16864
16865 VI The server designated by the cookie was not marked dead but could not
16866 be reached. A redispatch happened and selected another one, which was
16867 then advertised in the response.
16868
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016869
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168708.6. Non-printable characters
16871-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016872
16873In order not to cause trouble to log analysis tools or terminals during log
16874consulting, non-printable characters are not sent as-is into log files, but are
16875converted to the two-digits hexadecimal representation of their ASCII code,
16876prefixed by the character '#'. The only characters that can be logged without
16877being escaped are comprised between 32 and 126 (inclusive). Obviously, the
16878escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
16879is the same for the character '"' which becomes "#22", as well as '{', '|' and
16880'}' when logging headers.
16881
16882Note that the space character (' ') is not encoded in headers, which can cause
16883issues for tools relying on space count to locate fields. A typical header
16884containing spaces is "User-Agent".
16885
16886Last, it has been observed that some syslog daemons such as syslog-ng escape
16887the quote ('"') with a backslash ('\'). The reverse operation can safely be
16888performed since no quote may appear anywhere else in the logs.
16889
16890
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168918.7. Capturing HTTP cookies
16892---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016893
16894Cookie capture simplifies the tracking a complete user session. This can be
16895achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016896section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016897cookie will simultaneously be checked in the request ("Cookie:" header) and in
16898the response ("Set-Cookie:" header). The respective values will be reported in
16899the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016900locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016901not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
16902user switches to a new session for example, because the server will reassign it
16903a new cookie. It is also possible to detect if a server unexpectedly sets a
16904wrong cookie to a client, leading to session crossing.
16905
16906 Examples :
16907 # capture the first cookie whose name starts with "ASPSESSION"
16908 capture cookie ASPSESSION len 32
16909
16910 # capture the first cookie whose name is exactly "vgnvisitor"
16911 capture cookie vgnvisitor= len 32
16912
16913
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169148.8. Capturing HTTP headers
16915---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016916
16917Header captures are useful to track unique request identifiers set by an upper
16918proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
16919the response, one can search for information about the response length, how the
16920server asked the cache to behave, or an object location during a redirection.
16921
16922Header captures are performed using the "capture request header" and "capture
16923response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016924section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016925
16926It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016927time. Non-existent headers are logged as empty strings, and if one header
16928appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016929are grouped within braces '{' and '}' in the same order as they were declared,
16930and delimited with a vertical bar '|' without any space. Response headers
16931follow the same representation, but are displayed after a space following the
16932request headers block. These blocks are displayed just before the HTTP request
16933in the logs.
16934
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016935As a special case, it is possible to specify an HTTP header capture in a TCP
16936frontend. The purpose is to enable logging of headers which will be parsed in
16937an HTTP backend if the request is then switched to this HTTP backend.
16938
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016939 Example :
16940 # This instance chains to the outgoing proxy
16941 listen proxy-out
16942 mode http
16943 option httplog
16944 option logasap
16945 log global
16946 server cache1 192.168.1.1:3128
16947
16948 # log the name of the virtual server
16949 capture request header Host len 20
16950
16951 # log the amount of data uploaded during a POST
16952 capture request header Content-Length len 10
16953
16954 # log the beginning of the referrer
16955 capture request header Referer len 20
16956
16957 # server name (useful for outgoing proxies only)
16958 capture response header Server len 20
16959
16960 # logging the content-length is useful with "option logasap"
16961 capture response header Content-Length len 10
16962
Davor Ocelice9ed2812017-12-25 17:49:28 +010016963 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016964 capture response header Cache-Control len 8
16965
16966 # the Via header will report the next proxy's name
16967 capture response header Via len 20
16968
16969 # log the URL location during a redirection
16970 capture response header Location len 20
16971
16972 >>> Aug 9 20:26:09 localhost \
16973 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
16974 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
16975 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
16976 "GET http://fr.adserver.yahoo.com/"
16977
16978 >>> Aug 9 20:30:46 localhost \
16979 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
16980 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
16981 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016982 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016983
16984 >>> Aug 9 20:30:46 localhost \
16985 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
16986 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
16987 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
16988 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010016989 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016990
16991
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169928.9. Examples of logs
16993---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016994
16995These are real-world examples of logs accompanied with an explanation. Some of
16996them have been made up by hand. The syslog part has been removed for better
16997reading. Their sole purpose is to explain how to decipher them.
16998
16999 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17000 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17001 "HEAD / HTTP/1.0"
17002
17003 => long request (6.5s) entered by hand through 'telnet'. The server replied
17004 in 147 ms, and the session ended normally ('----')
17005
17006 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17007 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17008 0/9 "HEAD / HTTP/1.0"
17009
17010 => Idem, but the request was queued in the global queue behind 9 other
17011 requests, and waited there for 1230 ms.
17012
17013 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17014 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17015 "GET /image.iso HTTP/1.0"
17016
17017 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017018 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017019 14 ms, 243 bytes of headers were sent to the client, and total time from
17020 accept to first data byte is 30 ms.
17021
17022 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17023 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17024 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17025
17026 => the proxy blocked a server response either because of an "rspdeny" or
17027 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017028 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017029 risked being cached. In this case, the response is replaced with a "502
17030 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17031 to return the 502 and not the server.
17032
17033 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017034 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 +010017035
17036 => the client never completed its request and aborted itself ("C---") after
17037 8.5s, while the proxy was waiting for the request headers ("-R--").
17038 Nothing was sent to any server.
17039
17040 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17041 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17042
17043 => The client never completed its request, which was aborted by the
17044 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017045 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017046 send a 408 return code to the client.
17047
17048 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17049 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17050
17051 => This log was produced with "option tcplog". The client timed out after
17052 5 seconds ("c----").
17053
17054 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17055 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017056 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017057
17058 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017059 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017060 (config says 'retries 3'), and no redispatch (otherwise we would have
17061 seen "/+3"). Status code 503 was returned to the client. There were 115
17062 connections on this server, 202 connections on this proxy, and 205 on
17063 the global process. It is possible that the server refused the
17064 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017065
Willy Tarreau52b2d222011-09-07 23:48:48 +020017066
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200170679. Supported filters
17068--------------------
17069
17070Here are listed officially supported filters with the list of parameters they
17071accept. Depending on compile options, some of these filters might be
17072unavailable. The list of available filters is reported in haproxy -vv.
17073
17074See also : "filter"
17075
170769.1. Trace
17077----------
17078
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017079filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017080
17081 Arguments:
17082 <name> is an arbitrary name that will be reported in
17083 messages. If no name is provided, "TRACE" is used.
17084
17085 <random-parsing> enables the random parsing of data exchanged between
17086 the client and the server. By default, this filter
17087 parses all available data. With this parameter, it
17088 only parses a random amount of the available data.
17089
Davor Ocelice9ed2812017-12-25 17:49:28 +010017090 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017091 default, this filter forwards all previously parsed
17092 data. With this parameter, it only forwards a random
17093 amount of the parsed data.
17094
Davor Ocelice9ed2812017-12-25 17:49:28 +010017095 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017096
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017097This filter can be used as a base to develop new filters. It defines all
17098callbacks and print a message on the standard error stream (stderr) with useful
17099information for all of them. It may be useful to debug the activity of other
17100filters or, quite simply, HAProxy's activity.
17101
17102Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17103tests the behavior of a filter that parses data exchanged between a client and
17104a server by adding some latencies in the processing.
17105
17106
171079.2. HTTP compression
17108---------------------
17109
17110filter compression
17111
17112The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17113keyword must still be used to enable and configure the HTTP compression. And
17114when no other filter is used, it is enough. But it is mandatory to explicitly
17115use a filter line to enable the HTTP compression when two or more filters are
17116used for the same listener/frontend/backend. This is important to know the
17117filters evaluation order.
17118
17119See also : "compression"
17120
17121
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200171229.3. Stream Processing Offload Engine (SPOE)
17123--------------------------------------------
17124
17125filter spoe [engine <name>] config <file>
17126
17127 Arguments :
17128
17129 <name> is the engine name that will be used to find the right scope in
17130 the configuration file. If not provided, all the file will be
17131 parsed.
17132
17133 <file> is the path of the engine configuration file. This file can
17134 contain configuration of several engines. In this case, each
17135 part must be placed in its own scope.
17136
17137The Stream Processing Offload Engine (SPOE) is a filter communicating with
17138external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017139streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017140exchanged with them are configured in dedicated files, for the main part. It
17141also requires dedicated backends, defined in HAProxy configuration.
17142
17143SPOE communicates with external components using an in-house binary protocol,
17144the Stream Processing Offload Protocol (SPOP).
17145
Tim Düsterhus4896c442016-11-29 02:15:19 +010017146For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017147"doc/SPOE.txt".
17148
17149Important note:
17150 The SPOE filter is highly experimental for now and was not heavily
17151 tested. It is really not production ready. So use it carefully.
17152
William Lallemand86d0df02017-11-24 21:36:45 +01001715310. Cache
17154---------
17155
17156HAProxy provides a cache, which was designed to perform cache on small objects
17157(favicon, css...). This is a minimalist low-maintenance cache which runs in
17158RAM.
17159
17160The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017161this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017162
17163If an object is not used anymore, it can be deleted to store a new object
17164independently of its expiration date. The oldest objects are deleted first
17165when we try to allocate a new one.
17166
Cyril Bonté7b888f12017-11-26 22:24:31 +010017167The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017168
17169It's possible to view the status of a cache using the Unix socket command
17170"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17171for more details.
17172
17173When an object is delivered from the cache, the server name in the log is
17174replaced by "<CACHE>".
17175
Cyril Bonté7b888f12017-11-26 22:24:31 +01001717610.1. Limitation
17177----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017178
17179The cache won't store and won't deliver objects in these cases:
17180
17181- If the response is not a 200
17182- If the response contains a Vary header
17183- If the response does not contain a Content-Length header or if the
17184 Content-Length + the headers size is greater than a buffer size - the
17185 reserve.
17186- If the response is not cacheable
17187
17188- If the request is not a GET
17189- If the HTTP version of the request is smaller than 1.1
17190
17191Caution!: Due to the current limitation of the filters, it is not recommended
17192to use the cache with other filters. Using them can cause undefined behavior
Cyril Bonté7b888f12017-11-26 22:24:31 +010017193if they modify the response (compression for example).
William Lallemand86d0df02017-11-24 21:36:45 +010017194
Cyril Bonté7b888f12017-11-26 22:24:31 +01001719510.2. Setup
17196-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017197
17198To setup a cache, you must define a cache section and use it in a proxy with
17199the corresponding http-request and response actions.
17200
Cyril Bonté7b888f12017-11-26 22:24:31 +01001720110.2.1. Cache section
17202---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017203
17204cache <name>
17205 Declare a cache section, allocate a shared cache memory named <name>, the
17206 size of cache is mandatory.
17207
17208total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017209 Define the size in RAM of the cache in megabytes. This size is split in
17210 blocks of 1kB which are used by the cache entries.
William Lallemand86d0df02017-11-24 21:36:45 +010017211
17212max-age <seconds>
17213 Define the maximum expiration duration. The expiration is set has the lowest
17214 value between the s-maxage or max-age (in this order) directive in the
17215 Cache-Control response header and this value. The default value is 60
17216 seconds, which means that you can't cache an object more than 60 seconds by
17217 default.
17218
Cyril Bonté7b888f12017-11-26 22:24:31 +01001721910.2.2. Proxy section
17220---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017221
17222http-request cache-use <name>
17223 Try to deliver a cached object from the cache <name>. This directive is also
17224 mandatory to store the cache as it calculates the cache hash. If you want to
17225 use a condition for both storage and delivering that's a good idea to put it
17226 after this one.
17227
17228http-response cache-store <name>
17229 Store an http-response within the cache. The storage of the response headers
17230 is done at this step, which means you can use others http-response actions
17231 to modify headers before or after the storage of the response. This action
17232 is responsible for the setup of the cache storage filter.
17233
17234
17235Example:
17236
17237 backend bck1
17238 mode http
17239
17240 http-request cache-use foobar
17241 http-response cache-store foobar
17242 server srv1 127.0.0.1:80
17243
17244 cache foobar
17245 total-max-size 4
17246 max-age 240
17247
Willy Tarreau0ba27502007-12-24 16:55:16 +010017248/*
17249 * Local variables:
17250 * fill-column: 79
17251 * End:
17252 */